Home

Further development of a pulse magnetizer - IEA

1. test Pause and resume sequence On the front panel there are buttons to pause and resume the execution of a sequence Too implement that the Servo Change position checks 1f this should be done If so it enqueues a task called System Clear new tasks to schedule This task stores variables that determines the program flow of the sequence and sets them to do nothing The same task can be enqueued again to restore these variables Safety in software There are a number of safety precautions in software Both the embedded computer and the FPGA are responsible for the safety The utmost responsibility lies on the FPGA since it is considered to be most stable and besides has the fastest response time The shake to wake signal is generated in the FPGA so it is also logical to put it in the FPGA for that reason The program in the embedded computer however can unload the FPGA program from the FPGA Therefore the program in the embedded computer prevents itself from do that until the DC link voltage has dropped below 20 V This is done in the System Idle task There is also a task called System Activate which sends a signal from the embedded computer to the FPGA that it is okay to turn on the shake to wake signal The rest of the safety precautions for the embedded computer are in the System 49 Diagnostics task It checks that the supply voltages are within a user defined tolerance and sends a boolean to the FPGA if the result
2. wden resetname eu m eee A e Le 97 divide index divide index nbr of index 2 end rounds n current current pulses number of pulses result zeros 0 14 invalid result NaN 0 14 invalid Current cei1l1 20 001 current0 delta current delta current invalid pulse ceil 300 number of pulses delta number of pulses delta number of pulses tor n Liroungs resetname genvarname workname reset num2str n eval reset data resetname l dataname genvarname workname data num2str n eval pulse data dataname slk datanumbername genvarname workname datanumber num2str n eval iterations datanumbername result part invalid result part Current pulses PeakNDeltaFinderOneRound reset data pulse data usedposrLrions iterations SsCaningres pulse width current delta current pulses delta number of pulses number of pulses max number of pulses invalid Current invalid pulse result result resulti part invalid result invalid result invalid result part end max current Current Chis L9 Used in Peak urreni 0 nbr of pulse types ceil max number of pulses number of pulsestl delta number of pulses if nbr or pulse types Int abr Of pulse types Lr end if slope 1 x result l nbr of pulse types 2 10 else m pesult linbE of pulse types A end y fTesultili or or pulse types end 1 7
3. The used operational amplifier is a LF347 10 The used operational amplifier is a LM339 11 One of the channels of the NI 9215 is used to sample this signal As mentioned this signal 1s sampled for monitoring purposes only It is used when the current controller is to be tuned As an extra monitor and a remained there is a voltmeter next to the power electronics in the feet of the cabinet Level converter to the IGBT s The IGBT driver is configurable to handle both 5 V and 15 V signal levels on the input however the manufacturer recommends the use of 15 V level when the signal cable is as long as for the construction 12 The output on the NI 9401 is a 5 V TTL output hence a level converter is needed The level converter card does not only convert the signal levels It inverts the signal to the lower IGBT in the bridge and has the ability to put zero volts on all outputs which 1s used to reset errors reported by the driver The complete circuit is in llustration 4 below There are four inputs and one output to the cRIO The upper most input 1s the enable input If it 1s enabled the control signals will reach the drivers If it disables all outputs will be zero which as mentioned resets the driver The first inverter buffers and inverts the signal The transistor coupling makes the conversion from 5 V to 15 V and inverts one more time This signal 1s used to the output AND gates The next two inputs are for the bridge controlling signa
4. ripple on the current becomes more and more significantly due to the decreasing magnitude of the references Where the current slope is steep one can see that the current actually is crossing the reference This 1s due to the delay in the power electronics In graph 6 one can see that the ripple performance is improved This is due to the variable inductance which has increased the inductance In graph 9 is a triangular current shot of 300 A and in graph 10 is the corresponding switch pattern On the switch pattern one can see that both output combinations resulting in zero output voltage 1s used to distribute the switching load On graph 11 the corresponding time between each switch with each IGBT is shown Around 1 2 5 ms at the time axis the times between the switches are close to 10 us which corresponds to 10 KHz which in turn 1s the threshold for the high frequency protection Graph 12 14 shows to the same thing as graph 9 11 but with an incorrectly tuned controller the tolerances are too tight This means that the switching frequency will be too high which trips the high frequency protection The green and the blue line show where the high frequency protection is engaged A cursor is showing one of the interventions and cursors are found at the corresponding time in the other graphs At the time where the high frequency is active the switch time is down to 10 us which as mentioned is the threshold of it It is also clear that to tig
5. 250 Current i ry Estimated distance 280 SES 293 297 20 f S j vf Width 72 J20 J20 2 Resolution mm Pulse time 2 vE 10 005 Sensor position correction mm USE 279 283 5 291 73 296 Number of rounds a dl 4 Pulse position 150 4 OBS Values are only used in the final plot when the USE button is pressed Number of pulses position res ap DEE mm ndex Pulse direction 1 positive J 1 v 10 25 a v Pulsmag 2 0 lvproj cRIO Pulsmag 4 z Front panel 6 Magnetization Position calibration The next sub tab is seen above and it 1s used for the next step in the creation of a working magnetization system and that is to calibrate the positions of the hall sensors When the servo is at 0 the position 0 mm is defined as the middle of the magnetization head When this position of the magnet band is scanned the servo motor will not be at 0 for example it could be at 180 in the case that the hall sensors is at the opposite of the magnetization head This offset needs to be calibrated using this sub tab The method for this calibration is to have a band with a constant magnetization level and then magnetize one position in the band This position can then be found by scanning the band and find the position of the peak value of the magnetization level This sub tab gives the opportunity to do a
6. Peakvalue RelPeakVaduce Deltax scaningres invalid resulctin i2 14 end end pd pulse data index halt widbhsaindexchalt width 5 7 Ed reset data index halt widthsrznudexthalf width 9 dE current gt 12 PeakValue RelPeakValue FirstIndex Secondlndex DeltaX PeakNDeltaFinder pd rd else mid floor pulse data length pd 2 PeakValue pd mid RelPeakValue pd mid rd mid DeltaX 0 end result n result n 1 11 PeakValue RelPeakValue DeltaX scaningres ir Current lt invalid current iL pulses gt invalid pulse invalid result n invalid result ny 2211 PeakValue RelPeakValue DeltaX scaningres end end pulses pulses delta number of pulses if pulses max number of pulses pulses number of pulses Current u ime F gelte cine end end SULFeNL ou Current pulses out pulses 102 PeakNDeltaFinder m 6 5 5555 550500000400400040040040400400040040000040040020400400040040004004000020400000400000100400 5 5 5 Pulse magnetizer 2001 05 17 5 5 5 oo oo oo oo oo function PeakValue RelPeakValue FirstIndex SecondIndex DeltaX PeakNDeltaFinder PeakData ResetData NbrOfElements length PeakData NormData PeakData ResetData ResetMean mean ResetData if ResetMean lt 0 PeakValue I max PeakData else PeakValue I min PeakData end if ResetMean lt 0 RelPeakValue I max NormData else RelPeakValue FA min No
7. also used here which leads to similar logic This state also shoots pulses but in 43 this case it is done on over and over again on different positions until it has iterated over the whole magnet band Scan state machine Once again there is a good example how to implement a state machine with the Magnetization Calibration scan I state The corresponding flowchart is seen below Start x I First position LES Increase reset pulses Pasitian gt YES stop position di Calculate new position Engueue Magnetization and update poition Position calibratiar Enqueue Sera Change position Flowchart 4 Magnetization Calibration scan 1 Once again the winning concept to utilize the chain reaction is used and the logic of the state has similarities with the other two 44 Magnet diagnostics The same thinking when implementing the state machine for the position calibration has been used when implementing the state machine for the magnet diagnostics procedure This makes the structures of the state machines very similar A state machine for the magnet diagnostics can be seen below Start Magnetization Magnet diagnostics Magnetization Diagnostics array init Magnetization Diag reset band positive Magnetization Diag scan positive Magnetization Diag reset band negative Magnetization Diag scan negative Magnetization Calibration plot State
8. an amplitude of 85 A and a current slope of 96 5 kA s Magnetization G 300 5 500 400 300 200 100 100 200 70 Position mm Pulse 4 The accomplished magnetization of triangular pulse with an amplitude of 150 A and a current slope of 1 5 kA s 84 Magnetization G 500 400 300 200 100 100 200 205 210 215 220 Position mm 225 230 Pulse 5 The accomplished magnetization of triangular pulse with an amplitude of 150 A and a current slope of 96 5 kA s Magnetization G 500 400 300 200 100 100 200 100 Position mm 105 110 115 Pulse 6 The accomplished magnetization of triangular pulse with an amplitude of 250 A and a current slope of 100 kA s 85 5 3 Time usage The time to perform a magnetization pulse is less than s the exception is when the pulse time is close to or more than 1 s That includes calculating the current reference shooting the pulse and transferring the data to the embedded computer When the current controller is tuned the embedded computer makes an analysis of the resulting current and plots it The analysis is also done within this second for pulses with a pulse time less than a tens of ms In cases with a longer pulse time the calculation time for the analysis makes it take longer time The time to position the servo depends on the distance to travel It basically consists of two parts the time it takes to communica
9. and a more suitable output voltage could be applied This controller has its strength in its simplicity and insensitivity for outer variations The choice of control system It is desirable to use currents the order of 300 A and it is known that at least parts of the magnetic circuit is saturating at the flux densities that these currents will result in This means the inductance will vary to and hence the sampled current controller is not the wisest choice Instead the choice is a direct current controller or a tolerance band controller as it is going to be called in this text Today there are mainly two different ways when designing a tolerance band current controller The old way is to use analog electronics for the comparisons and digital circuits to choose action The new way is to sample the current extremely fast and practically implement the same thing in software The new way requires an extremely fast hundreds of KHz AD converter and calculating performance that can handle the amount of data But in return it 1s more flexible Our choice is to make a hybrid of these to variants The bands are going to be generated digitally 3 and hence the controller will be as flexible as the digital version But the comparisons are going to be made with analog electronics and hence there will be no need for an extreme AD converter 3 Hardware 3 1 Existing hardware As you may have noticed this thesis is not about building a magnetizing ma
10. care of which will be explained later Since this comparator circuit only does the comparison around zero volts the measured current is subtracted from the tolerance band level and feed to the comparator This is done by the operational amplifiers in the middle When the current is approaching the the tolerance band level the output of this operational amplifier will approach zero which if it continues will result in a comparator output change The comparator output signal is buffered and converted in TTL level signals with the transistor couplings to the right of them Variable inductance The power electronic converter and its control system are designed to handle in the order of hundreds of amperes with small amounts of current ripple but still acceptable rise times This means that if the current reference is in the order of ten amperes instead the ripple will probably be unacceptable To come around this problem an extra inductor is connected in series with the flux providing coils when the current peak value is 20 A or less The coupling is very simple and is shown below 13 Digital out from NI 9205 D extra turns around the current trancducer 16 mH Low current mode contactor PEEPS RPP ZIL ag Li Li LI Li Cooling fan L Lili High current mode contactor EEEEEEEEEEEKEEEEEEEL 0 REEREEEEREEEEEET Flux providing cails Illustration 2 Variable inductance electronics In high curren
11. current and the current reference are feed to a regulator which creates the voltage reference to the modulator It 1s possible to calculate the ideal regulator parameters analytically from the inductance and the resistance of the load and the time between the samples These parameters are called dead beat parameters and provide a new voltage reference that will eliminate all the error in the current This controller has many advantages for instance that the switching frequency can be chosen to a known value The direct current controller or the tolerance band current controller is much more simple In its most simple form it compares the actual current with two references which forms the tolerance band If the current is below the lower band the voltage is turned on If the current is above the upper band the voltage is turned off This creates a self oscillating controller which is very simple but also stable When using a four quadrant converter there are more choices than only on and off the voltage can be reversed to This means that there are three different current slopes to choose between hence a little refinement is needed The solution is to add two more bands outside the earlier two bands to find out if the chosen action was sufficient For example the current crosses the upper inner band and the controller changes the output voltage to zero but this is not enough to make the current fall Hence the current will cross the outer upper band
12. earlier mention emergency stop called shake to wake has the same effect as the first emergency stop But instead of a button it is a relay connected to a circuit board connected to the cRIO The main function is to detect if the FPGA freezes and if it is happening it trips the emergency stop This is done by a VI that continuously generates square wave via one of the digital outputs This square wave is filtered by a low pass filter to obtain a signal around 2 5 volts which is compared with 2 and 3 volts If the comparisons show that the signal is between 2 and 3 volts it will close the relay Of course this leaves the opportunity to trip the emergency stop from the software as well This is used by diagnostics the activate button and the emergency stop on the front panel The activate button turns on the shake to wake if the control system is in ready mode This occurs if the diagnostics part of the program does not find any faults The coupling is shown below and the used circuits were HEF4030 18 and LF347 10 Ca LAP GND gt l gt n IP oye i x R2 R4 x 10k oys GND D3 D4 X2 1 R1 fi E a 347N 10K agn C2 CI d es BC547B 10k GND RB GND i LF347N Illustration 7 Shake to wake electronics Diagnostics The diagnostics consists of three parts One that 1s continuously sampling all supply voltages for the peripheral electronics and checking that they are within allowed limits The next part
13. function of the field strength in the measuring direction and the supply voltage Hence the supply voltage is sampled to and taken into account in the software The coupling is shown below Illustration 5 Hall sensor voltage regulator electronics 18 3 2 6 3 2 7 Cooling of the flux providing coils When using current greater than 100 A and shooting many and longer pulses the flux providing coils get hot And in order to not take any risks with the coil insulation these has to able to cool down after more intense periods for example a reset To make the cool down time shorter a fan 1s placed close to the yoke and the air flow is guided through it to cool the coils as effectively as possible The fan is activated when the inductance is set to high current mode This leaves the opportunity to leave the fan on during a cool of pause as well as turning it off during a hall measurement Safety The DC link capacitors of the power electronics are able to store a lot of energy If something goes wrong in the control system it can result in quite dangerous situations where hardware as expensive IGBTS or as well humans can take damage Hence a little extra effort was paid when the control panel of the magnetizer was made Control panel To begin with there are four groups of power supplies each group has its own circuit breaker The first switch turns on the power supply to the cRIO and the power supply to the servo controller This make
14. in depth in 21 It is a protocol where the computer is the master and the servo controller is the slave The master is always the initiator of communication This means that to get information from the servo controller to the embedded computer the embedded computer needs to ask for it first When a message is sent to the servo controller it will always answer If a value was requested that value will be sent back If a value or command was sent to the servo controller it will respond with a message that tells if 1t was okay or not The request used continuously in this project 1s requests for errors and a request for the current position of the servo motor The commands sent to the servo controller are enable disable clear errors change to position control change to speed control and go to new position Also two commands that sends a value to the servo controller 1s used these values are new position and speed set points The nit task also uses some commands not mentioned above for the initialization of the communication with the servo controller A FIFO queue called send queue contains commands to send This is a good solution since it 1s tasks in the operating system that know when to send a command and it 1s the while loop parallel to the operating system that sends the message containing the command It will also be easy to 38 send commands from a task since it 1s just to enqueue the command to the queue The rest of the job is handled by th
15. it could be seen in the state machine below 46 Start Magnetization Surface init Magnetization Surface pulse reset Magnetization Surface slope reset Magnetization Magnetization Surface slope reset scan Surface pulse reset scan Magnetization Surface pulse reset scan analysis Magnetization Surface slope reset scan analysis Magnetization Surface pulse reset scan store Magnetization Surface slope reset scan store Magnetization Surface pulse first pulse Magnetization Surface slope first pulse Magnetization Surface pulse pulse Magnetization Surface slope pulse Magnetization Surface pulse pulse scan Magnetization Surface slope pulse scan Magnetization Surface pulse pulse scan store Magnetization Surface slope pulse scan store State machine 5 Performance measurements The Magnetization Surface init state initiates variables and creates a file name for the file to store data in It also creates an array with positions to use from the controls on the front panel and saves it to a file with the created file name The Magnetization Surface pulse reset and Magnetization Surface pulse reset scan states are implemented in the same way as the reset and scan states described above except that the reset starts a timer so the flux providing coils can cool down An analysis of the achi
16. loop in LabVIEW How the operating system Is used The operating system structure described above was considered to be a good starting point for the program for the embedded computer and has been that throughout the project The heart of the program developed is a state machine looking much like the state machine with two states above but it 1s expanded to four states There are four tasks that implement these four states and these tasks are dle Check front panel Diagnostics and Background activities The Idle task enqueues the other tasks and itself as well This will work since they are enqueued in a FIFO queue It can also stop the program by dequeueing all elements if a stop condition is met The last purpose of the dle task is to keep the cycle time of the state machine constant When the dle task is finished next up is the Check front panel task The name describes it very well since its job 1s to read the front panel controls and start the corresponding state machine if a control implies that It does this for all controls on the front panel that could imply that a state machine should be started Next up is the Diagnostics which gathers information from the FPGA program and presents this in the Diagnostics tab in the main tab interface It also checks so that the monitored power supply 36 voltages are okay and reports this to the FPGA program Finally Background activities have various tasks to do hence the little more generic name It
17. machine 3 Magnet diagnostics It has the supervising state Magnetization Magnet diagnostics that determines the execution order of the other states It also initializes variables in the variable cluster before the other states are entered 45 Scan magnet band and data save This is a rather simple state machine as seen below Start Start Magnetization Magnetization Init scan band Scan band save Magnetization Scan band Magnetization Scan band plot State machine 4 Left Scan band Right Scan band save It has no supervising state but that is because it does not need to be more advanced than it is The state Magnetization Init scan band initializes an array to save the measurement data into The next state is Magnetization Scan band It utilizes the fact that Servo Change position will start a chain reaction that will enqueue the task that implements this state machine again Finally the result is presented in a plot in the Magnetization Scan band plot state As seen in the state machine the there is a separate state for saving the data to a file Performance measurements As described in 4 3 1 there is a test procedure that tests the performance of the machine The test conditions are different combinations of current and pulses slopes and the program in the embedded computer performs the tests and saves the resulting magnetization The state machine that implements this is rather large but has a simple structure
18. mains 1 e it will light up if the power to the supply is on and the fuse is intact As mentioned the machine should be started in a certain order The same applies when shutting it down To make sure that the controller software is not shut down while the DC link is charged the stop execution button on front panel 1s disabled while the DC link voltage 1s above 20 volts The instrumentation also includes an emergency stop The emergency stop does not turn of the 19 power completely since this is not the wisest action in case of an emergency As mentioned before the emergency stops only disables the Operational mode which in turn disables the servo power and the DC link if they are turned on of course This means that the IGBT drivers will remain on so that the DC link can be safely discharged A picture of the control panel and the wiring diagram is shown below Image 6 Control panel 20 DC link inrush current limiter Servo power power supply Servo power pilot lamp Dc link capacitors VTT DT TT Fast stop discharging larnps Operational lamp power supply VERO pilot lamp Servo logic power supply Servo logic pilot lamp Power supply Servo power Shake to wake electronics cRIO pilot lam Ei 230 We di cRIO Interface Emergency link unununnunnuunnuune Dc lin servo logic electronics stop Shake to wake relay lustration 6 Control panel 21 Shake to wake The second
19. mechanical unevennesses Discussion about the surfaces The Surfaces 1 12 especially 2 and 8 tells us that the number of pulses does not affect the magnetizing result in any significant way Another obvious conclusion is that the resulting flux density 1s correlated with the current magnitude until the magnet is saturated By trying to magnetize the magnet more the resulting flux density is decreasing which invites to discussion As one can see the magnetizing width is consequently which could be explained by that the leakage flux will increase with the total flux Another interesting trend is that triangular shaped current pulses results in both higher resulting flux and narrower magnetizing pulses only benefits in other words Likely it depends on that the flux collecting lenses is doing a better work when flux derivative and hence the eddy currents in them are kept at a more constant level The foregoing reasoning 1s supported by for example surface 16 where one can see that steeper current slope results in narrower pulses In the same way the maximum achievable flux in surface 14 increases with steeper current slopes A set of magnetizing pulses are shown in attachment Magnetizing results as Pulses 1 6 Pulse 1 shows the result of a relatively small current pulse while pulse 2 shows the result of a bigger pulse but still in the linear area of magnetization Pulse 3 shows the beginning of the saturation of the magnet The pulses 5 and 6
20. mid DeltaX 0 end resultin t result n 122 PeakValue RelPeakValue DeltaX scaningres result n 6 14 AL Current lt invalid current if pulses gt invalid pulse invalid resultin invalid result n 1 2 PeakValue RelPeakValue Deltax scaningres invalid result n 6 14 end end pd pulse data index half width indexthalf width 3 rd reset data index half widcCh Indexth lf width 3 7 SE curren 1 PeakValue RelPeakValue FirstIndex Second ndex DeltaX PeakNDeltaFinder pd rd else mid floor pulse data length pd 2 PeakValue pd mid RelPeakValue pd mid rd mid DeltaX 0 end result n i result in ls 5 PeakValue RelPeakValue DeltaX scaningres result n 9 14 13 AL Current lt invalid currence IL pulses gt invalid pulse invalid resultin invalid result n 1 5 PeakValue RelPeakValue Deltax scaningres invalid resulti n 9 14 end end 101 pd pulse data index half width indexthalf width 4 Ed reset data index half width indexthalf width 4 ir current PeakValue RelPeakValue FirstIndex Secondlndex DeltaX PeakNDeltaFinder pd rd else mid floor pulse data length pd 2 PeakValue pd mid RelPeakValue pd mid rd mid DeltaX 0 end result n result n 128 PeakValue RelPeakValue DeltaX scaningres resultin l2sl14 5 iL current lt invalid current Lt pulses gt anvabrd pulse invalid pesultin s invalid resultin 1 9
21. minimizes the risk for damage To provide power to the cRIO a power supply delivered by National instruments is used To provide power to all peripheral electronics related to the current control and power electronics two different power supplies are used The first one delivered by VERO power is providing 15 and 15 volts to the analog electronics 16 It 1s also providing 5 volts to the digital circuits used in the level conversion stage the calibration equipment controller and the shake to wake Another 15 voltage is supplied by a power supply delivered by XP power 17 This is used for the IGBT 17 3 2 5 drivers the level conversion stage the relays to the calibrator and the shake to wake This means that the analog 15 volt supply is separated from the digital one A separate power supply was used for the current control calibration device There are different reasons for this first of all when the device is in use it requires more current that any of the other supplies can deliver Since it is used in relation to the current sensor the main part of it is placed in the high voltage volume of the cabinet By using a different power supply EMC problem is avoided The used power supply is an ordinary computer power supply which is modified to start when it is connected to mains To make the installation more clean all unused output cables are cut The used voltages are 5 V for creating the reference current and 5 and 12 V for operate t
22. name disp mget cRIO fileslist file name end end close cRIO clear cRIO Clear fileslist clear n clear file 95 MATLAB file loader m fileslist2 ls ps size fileslist2 n 1 while n lt ps 1 if striind fileslist2 n i xla8 disp Strcat Llnumezstrin rxloeslristz l n ntl else fileslist2 n Il ps 1 ps 1 1 end end file 1 while file gt O file input Select file enter 0 to close if file gt 0 filesformat strread fileslist2 file e current0 str2double filesformat 2 delta Current ir2leonble t lestormat 17 number of pulses str2double filesformat 4 delta number or pulses Str2doubleirilesrormat io max number of pulses str2double filesformat 6 pulse width str2double rilesformat 7 time 1000 str2double filesformat 9 workname cell2mat filesformat 10 workname workname 1 length workname 4 workname genvarname workname plotname strcat cell2mat filesformat 8 workname eval workname load fileslist2 file end end clear fileslist2 clear ps clear n clear filesformat clear file clear time 96 num2str time ms Data_divider m oe These variables should be changed by the user to fit the measurment that is analyzed o slope 1 5 0 curent and pulse sensor 1 Which Sensor to look at l zy 3 or 4 figure start 1 Figure number to Start at
23. project The current controller is built as planned but a lot of safety devices are added to minimize the risk for damage on hardware as well as humans and hence minimize the number of delays in the work No advanced measuring head is developed Instead hall sensors placed next to the magnet band are used Instead of studying the possibilities of automatic iterative magnetization the performance and properties of the magnetizer is studied Only small spots were magnetized and the peak magnetization and pulse width is studied as functions of magnitude time and shape of the current Due to this delimitation the possibilities to industrialize the process 1s not studied 2 1 PN Theory Current control with inductive loads There are two main types of current controllers the sampled current controller and the direct current controller 2 This book 1s the basis for our view of how these two controllers work but to make this text more readable here is a short summary It is assumed that the reader is familiar with switching electronics and the voltage current characteristics of inductive loads The sampled current controller uses a triangular wave modulator to create a switch pattern which result in a mean output voltage that corresponds to the desired voltage reference The current through the load is sampled in the middle of the current rise or the current fall to obtain a mean value of the current without using any averaging The sampled
24. px sc sS 2 cut 1 else cut 0 end sensor sensor 3 zl vec2mat result sensor length x zl vecdmat l nval d result sensor length x Z2 vec2mat result sensor 1 length x 221 Vecemat invalid rfesulLti sensorvl Lemgthix 3 tmp result sensort2 tmpt invalid resultiiySensort2 lf cut gt 1 y3 y 2 1ength y 98 Z3 vec2mat tmp length x l length result 231 VeCc2mat tmp lengehix tl lengch invalld resulr lengthi x else y3 yi Z3 vec2mat tmp length x z31 vec2mat tmpi length x end if slope xlabel Slope kA s else xlabel Number of pulses end ylabel Current A zlabel Maximum magnetization G Figure T g re Start SULL 2 Y 2L hold on Hio Burt x 7 211 4 set h FaceColor 0 3804 0 36 04 0 3604 hold rt if slope xlabel Slope kA s else xlabel Number of pulses end ylabpel turrenLi zlabel Maximum magnetization G Irgure rigure start l Sure 2 y 22 hold on N gt Surf x y 221 5 set bh FaceColor 0 3804 0 3804 0 3804 hold orii if slope xlabel Slope kA s else xlabel Number of pulses end ylaoel Currenrt A zlabel Maximum magnetization G I3gure rigure Start 2 Surfi x y3 235 hold on Ij SURE X Vo aol 7 set h FaceColor 0 3804 0 3804 0 3804 hold eff if slope xlabel Slope kA s else xlabel Number of pulses e
25. reset of the band which will be explained later a pulse at a specific position in the band and of course scan the band and find the peaks For each sensor the estimated distance and a search width is specified this is to make sure that only a healthy part of the magnet band is used in the analysis of the peak value The value for the Estimated distance is 32 the distance between the magnetization head and hall sensors along the magnet band When the calibration is done the correction values are updated and the uncalibrated result for the four sensors is plotted in the four graphs in the Hall sensor position calibration tab in the graph tabs When the Use button is pressed the calibration values will be used in the rest of the program and the four plots will be updated using these correction values So if the pulse was done at the position 100 mm all plots should show a peak at 100 mm The earlier mentioned reset method has been found experimentally It uses a large current that saturates the magnet band on a wider distance than the width of the magnetization slit due to the leakage flow This makes it possible to use a greater distance between the magnetization pulses than the width of the magnetization slit It is good to control that the magnet band is in good shape and if it is not it is good to know where and how it is differing from the rest of the band This is done in the third sub tab Magnet diagnostics The diagnostics is done with a re
26. stores which variable that where changed to which value it was changed and on which clock cycle it changed into the FIFO queue called Digital samples This allows the embedded computer to analyze the data and reconstruct the switching pattern both of the IGBT and the desired values from the current controller logic This means that is also can detect when the HF protection was engaged 4 5 MATLAB programs The purpose of the programs in MATLAB is to compile the results into information that could be analyzed by the user The result is presented as a surface with the test conditions on the x and y axis and the analyzed parameter on the z axis The analyzed parameters are the maximum resulting magnetization and width of the magnetized segment The width is from 50 on the positive edge to 50 of the negative edge The first m script downloads files from the cRIO to the working directory of the computer it 1s running on It does this by presenting a list of the files in the folder Surface results in the cRIO When the script runs the user can then choose a file to download and when done the script can be stopped The script is called cR O_result_files_getterm and can be found in the attachment MATLAB code 58 The next m script loads a file from the working directory into a matrix in the MATLAB workspace It also interprets the file name that has the format presented in table 1 in the attachments and saves the values from segment 2 to 7 and 10 int
27. the machine halts on some kind of fault to The philosophy of the construction EMC During the design phase of the machine EMC problems was avoided as far as possible To begin with the control system and its peripheral electronics are placed in a separate volume electrically shielded from the power electronics Furthermore the peripheral electronics is placed in a screening box of aluminum All cables related to the control system are shielded as well and the shields are only connected to the screening box containing the peripheral electronics to avoid ground loops For the same reason is the screening box the only connection between the earth of the control system and protective earth More specific solutions will be described under the heading for each part Choice of components The attentive reader or maybe the next person who are going to develop the machine further will probable find the choice of components a little strange in some parts of the construction Therefore we want to point out that we used parts that we found in stock at IEA to minimize the number of orders and delays The parts are selected to not compromise the function or the result but as said some of the choices may seem a little illogical 24 4 4 1 Software The basis of our understanding how to use the program language LabVIEW and the cRIO is the course manuals 19 and 20 Distribution of tasks The tasks of the software in this project ranges from
28. 04 0 0045 0 005 0 0055 K Time s Pulsmag 2 0 lvproj cRIO Pulsmag Front panel 4 Current control The next tab in the main tab interface 1s the seen above namely the Current control tab It 1s important that the performance of the current controller is as good as it can be and this tab 1s used to test and trim the performance To do a test of the controller a set of controller parameters and settings for a test pulse 1s entered Then the RUN button is pressed and the current controller parameters are loaded and the system changes the current reference to create the test pulse The current controller follows the reference according to the controller parameters and the resulting current can be seen in a graph in the main tab interface For further evaluation of the performance of the current controller there are four other graphs available in the Current control analysis tab in the graph tabs for The first graph shows the switch time the inverse of the switching frequency 30 for each switching event The second graph shows the resulting switching pattern of switch 1 and 2 it also tells 1f and when the high switching frequency protection was engaged The third graph shows the DC link voltage during the pulse The fourth graph shows a numerical calculation on the inductance based on the sampled current and voltage The calculation of the inductance is only valid during the time while there is an active vector applied and no switching occur
29. CODEN LUTEDX TEIE 5286 1 111 201 1 Further development of a pulse magnetizer JUS 4v O 4 gt lt O C D e d e e LLI be Y JU U E LLI Jon Axelsson Per S derberg Division of Industrial Electrical Engineering and Automation Industrial Faculty of Engineering Lund University Abstract A machine for magnetizing permanent magnets is under development at LTH The idea of the machine is to affect only small pieces of a magnet at the time to obtain magnets with an arbitrary flux density of course with respect to the magnetic material The final machine is supposed to work in a closed loop fashion Every part of the magnet is magnetized measured and magnetized again with a flux density based on the desired flux density and the measured magnetizing result until the final result is satisfactory This thesis covers the development of a new control system for the machine including a new current controller for the coils providing the flux It also includes performance tests of the magnetizing head and a parameter study of the magnetization result as a function of current magnitude shape and pulse time Preface This thesis has been made at LTH for the division of Industrial Electrical Engineering and Automation IEA in cooperation with Industrial Production IPROD and is the final part of our education We want to thank both IEA and IPROD for their help and special thanks to Mats Al
30. Deg 565 42 mm Diagnostics Calibration of comparators Current control Magnetization Hardware Servo Software Tolerance P Wa 3 Comparator values Nominal value Calibration scal factor A within coarse limits A Comparator error Driver error analog 415y 15 0109 7 5 42 0003 comp A Analog 15 v 15 0245 o J 15 1 9989 EDEA 51 at error M z Y TCR xi 0 Digtal 5v 490863 s 21 9987 32 at error Digtal 15 v 150776 is 1 9996 Ack Hall supply 5 91612 6 hall samples A 10 Hall sensor sensitivity 1 V G Offset trim 1 G Average of hall samples 1 Hall supply sample 10 003125 M 197 67 E j 5 018076419830322260000 S i Hall sample 1 d sensor sensitivity 2 v G EE 2 G Average of hal samples 2 1 88091278076 1718750000 4 0 003125 E 210 878 Hall sample 2 Hall sensor sensitivity 3 V G Offset trim 3 G Average of hal samples 3 3 154695510864257810000 0 0013 SE 209 604 Hal sample 3 12 231373310089111330000 Hall sensor sensitivity 4 V G Offset rim 4 G Average of hall samples 4 i Hall sample 4 0 0013 E 218 651 y 12 790790081024 169920000 Pulsmag 2 0 lvproj cRIO Pulsmag 4 Front panel 2 Diagnostics The Diagnostics tab seen above presents information and controls that are useful when the user wants to see if the system reports anything that 1s wrong There are three sub tabs in the diagnostics tab to limit the information presented to the specific interes
31. FIFO player Analog Outer band distance A Reference Reference leve reference calculator levels Analog out Embedded computer Comparator cal start Comparator calibration Tolerance band size A plexers Measurments Compensation values Compo Compl Comp2 Comp3 Enable drivers output Digital output Block diagram 5 FPGA code In the block diagram above the Error watcher block has been added It ensures that the IGBT drivers is only enabled when it is needed it 1s also responsible for shutting down the IGBT driver if an errors occurs It the driver itself that detects an error the Error watcher gets to know this by the signal Driver error If driver causes the error the S7 and 52 signals are saved to help the error search afterwards 55 4 4 7 Data sampling Min ticks S1TB pi protection S1 HF protection S2 Comp Sl mm estan E omparator ector type se bo Digital input interpreter selector zero vectors Driver error A 4 Start playback Sequence Clear play FIFO player Outer band distance A Reference Tolerance band size A calculator Analog reference levels plexers Measurments Compensation values Embedded computer Compo Compl Comp2 Comps Comparator cal start Comparator calibration Analog out Take hall sample Enable drivers output Measurments Digital output Voltages OK Block diagram 6 FP
32. GA code In the block diagram above the Analog input block has been added It samples the analog power supplies and sends the measured data to the embedded computer The embedded computer analyzes the samples and returns the Voltages OK signal to the FPGA program When the Joke hall sample signal is activated the Analog input block changes the sampling tactics and only samples the hall sensors for a while along with the supply voltage to the hall sensors The samples are sent to the embedded computer which uses them to take a mean value of the magnetization level When the hall sensors are sampled the driver is disabled with the Disable drivers hall signal This is only a safety precaution because the driver should not be enabled when the hall sensors 1s sampled The Sequence player also takes analog samples That is samples of the current in the flux providing coils and the DC link voltage when a current pulse is generated The comparator calibration also samples current when it is doing the calibration procedure 56 4 4 8 Safety in software Comparator error Min ticks S1TB H HF protection S1 Vector type Use both selector Dec zerovectors soTp HF protection S2 Comparators at erro Comp Compte Digital input Comp2 p Ki Comp3 Start playback Sequence Clear play FIFO player Analog Outer band distance A Reference reference calculator levels Analog out Comparator inter
33. GND 3 ii 2 D c EH GND Illustration 1 Current and voltage measuring electronics 12 Current measuring The current transducer is delivered by LEM It is a LA 306 S SPI and it is capable of measuring 640Aproperly configured 8 The conversion ratio is 1 5000 and the measuring resistance is selected to 25 ohms Since the output from the power electronics only passes through the transducer once this gives a total conversion ratio of 0 005V A This signal is inverted and amplified four times by the JC D which implies that the conversion ratio will be 9 024 V This signal is used in the comparison with the tolerance bands and for monitoring after being inverted another time by ICIA One of the channels of the NI 9215 are used to sample this signal Notice that this signal is sampled for monitoring purposes only It is used when the current controller 1s to be tuned Comparison electronics As mentioned the decision if the current is within tolerance or not is made with analog electronics There are four identical couplings based on IC2 IC3 and IC4 The left most operational amplifier buffers the output signal from the analog output module The right most IC is a comparator that is provided with a hysteresis band to minimize the risk flickering The hysteresis band corresponds of of the total output voltage swing of the comparator which is equal to0 03V The consequences of this hysteresis is taken
34. Min ticks S1TB Ir protection S1 HF protection S2 Sl Compo m m Digital input ne erior ype se bo interpreter selector zero vectors S2 TB Analog Outer band distance A Reference Reference leve Saale calculator evels Embedded computer Tolerance band size A Multi plexers Analog out Measurments Compensation values Compo Compl Comp2 Comps Reference levels Comparator cal start Digital output Block diagram 3 FPGA code When the Comparator calibration is added the resulting block diagram is seen above The Comparator calibration performs a calibration by sweeping the Reference levels and check when the comparators change To do this the Reference levels it generates needs to be sent to the Analog out block here the Multiplexers block comes in handy With the input Mode it can choose to use the Reference levels from the Reference calculator or the Comparator calibration block When the calibration is started the Comparator calibration block will as said sweep the Reference levels and check when the comparators change It does this for two different test currents one positive and one negative The measurements are then processed by the embedded computer which calculates the Compensation values 53 4 4 5 Current pulse generation Min ticks l gt VD p TB hrs protection S1 Compl e Digital input Re Vector type Use both H H gt gt interpreter Z
35. PRE ee ee 13 KEE 14 a TE re 16 25 EE tothe IGB See are 16 pd TOW CE SUP S YENE NE EE 17 3 23 e 1 ONS OLS EN 18 3 2 6 Cooling ot the flux providing coils ee en 19 e cn 19 ore AINE NT m 19 Shake 10 Wake Gelee 22 DAE SOS EE 23 Fast Sophie 23 3 2 0 Ihe philosophy ofthe ONS EML LLM aaa 24 ENG HH 24 Choice Of EENEG 24 e EE 25 41 Dist li be e EE 25 42 Communication between the different parts 26 4 21 Ee EE 26 a MEN VO EE 26 42 Eeer 26 4 3 The structure of the program on the embedded computer 27 A ON MEE BC WS NEN Um Top 27 2 9 2 SO DETAUNS Sy EE 35 SIPUC O er re Eee 35 Implementing a State Wachtel 36 How the Operating systems USCC EE 36 Current puke EE 37 Hall sensor samping erse a de Ba ni nnd lal ue en ede label eli eee lay 38 Serial communication withtheservVo 38 Positonie Me Servo eege 39 Positions Call EAM OM am a OO aia ri Gn ca eh m Seren 4 Pulse state machine e e 42 Reset state machte een an 43 Cab Stale Mache uses aueh 44 Mache e e ee re ee 45 Scan magnet band and data save ame Elite 46 Performance EE TE ne aa 46 Pause and Ee UE 49 GENEE 49 4 Dhe Pn DEDE EE 50 41 OV CIE Ure III ME TT TM 50 44 2 Current control e 50 Ac SGeneradon or tolerance bands ended tud de itn eet eee ama keza ale aki torent 22 4 44 Comparator Calibra on zu iret yine ad ari aH EEN 53 445 G ren pulse SENCTANION EE 54 TEG Eror MES OS NU RR 55 dak EE 56 A
36. Seen from the right 80 ee p ye s 25 20 Wu Width mm o Current A Slope kA s Surface 16 Pulse width with triangular shaped current pulses with varied pulse time to obtain current slopes according to the slope axis Width mm 40 60 80 100 120 Slope kA s Surface 17 Pulse width with triangular shaped current pulses with varied pulse time to obtain current slopes according to the slope axis Seen from the front 81 Width mm Current A Surface 18 Pulse width with triangular shaped current pulses with varied pulse time to obtain current slopes according to the slope axis Seen from the right To facilitate an attempt to explain the shapes of these surfaces here are six pictures on carefully selected magnetizing pulses 82 Magnetization G 500 400 300 200 100 100 200 145 150 155 160 Position mm 165 170 175 Pulse 1 The accomplished magnetization of triangular pulse with an amplitude of 20 A and a current slope of 16 5 kA s Magnetization G 500 400 300 200 100 100 200 505 510 515 520 Position mm 525 530 Pulse 2 The accomplished magnetization of triangular pulse with an amplitude of 55 A and a current slope of 101 5 kA s 83 Magnetization G 500 400 300 200 100 100 200 480 485 490 Position mm 495 500 Pulse 3 The accomplished magnetization of triangular pulse with
37. a user friendly measurement system with settings graphs and the possibility to save measurements to a file down to the current controller with response times that are less than one us To accomplish this the cRIO system has been used since it can provide both ends on this scale and of course everything 1n between To understand how to distribute tasks between the FPGA and the embedded computer in the cRIO system a little more information about the cooperation between the embedded computer FPGA and the I O modules than given in 3 2 1 1s needed The I O modules communicate with the FPGA which in turn communicates with the embedded computer If the embedded computer wants to communicate with the I O modules this is done through a program on the FPGA The embedded computer runs a real time operating system RTOS made by National Instruments This makes the cRIO to a stable controller with features found in a normal operating system for a PC This includes a file structure ethernet connections ftp server capabilities and hardware resource monitoring The sequential nature of executing a program for a CPU leads to latency between input and output which can be a limitation when implementing controllers with the embedded computer Furthermore an increased workload can decrease the performance In that essence the FPGA has an advantage since the program is literally implemented in hardware which makes the execution parallel of its nature Thereby an increa
38. ak la our supervisor for his dedication inspiration and imaginativeness which has led the project to interesting conclusions Getachew Darge at LTH for his knowledge and experience all his hints and his almost inexhaustible supply of good to have stuff which altogether has saved us a lot of time and made this project possible Kenneth Frogner at IPROD for his knowledge and hints about the machine as well as his knowledge about everything from magnets to electronics Andreas J nsson and Christoffer Orndal former DLAB for introduction and support in labVIEW programming Svante Bouvin at IPROD for his knowledge and experience in mechanical workshops Contents NR Kai ue e E I LL e NET ars eects I PME OUI C T P UH 2 J E e M EYE 2 PANIQUE T M 3 2 1 Currentcontrol with inductive loads a aea eaei 3 202 Line CHOICe OF COBDEOLSV SEN NEE 3 SEMEN TL 3 2b ee e 5 SN MEN e T UU m ee en rn ee 5 S NE M A TE 7 S Ne E Ed e 7 o4 TOW a e 7 3 153 Int sh curreht e E 8 ILO Ee ERE ee 8 Pe b VOC ge e b iy A A ANN EE 9 321 be are Ru Te d NE NY NE NK EY 9 10 ZABOHLTHE C Ompase e I nee A e e 10 Input and Opt Modul a ais al ei alan o lam lala a lal 10 rte 11 32 2 vr CON E 11 uu aaa 11 C EE e 13 E 10 Ee Te 13 ARIAS DAU Lane PN ME ee ande d um
39. ations NI 9215 pp 1 21 National Instruments Corp Operating instructions and specifications NI 9263 pp 1 15 National Instruments Corp Operating instructions and specifications NI 9401 pp 1 14 LEM Datasheet for Current Transducer LA 306 S SP1 pp 1 3 July 2007 LEM Datasheet for Voltage transducer LV 25 P PS2 pp 1 2 Mars 2006 National Semiconductor Datasheet for LF347 pp 1 2 August 2000 National Semiconductor Datasheet for LM339 pp 1 4 March 2004 Semikron Datasheet for SKHI 23 17 R pp 1 10 June 2006 Philips Semiconductor Datasheet for 74HCUO4 pp 3 5 September 1993 Toshiba Datasheet for TC4049 pp 1 3 February 1998 Philips Semiconductor Datasheet for HEF4081 pp 2 3 January 1995 VERO POWER Datasheet for EPLAX PK serie PK60 pp 10 11 XP POWER Datasheet for VCS Series VCSS0USIS pp 1 2 May 2011 Philips Semiconductor Datasheet for HEF4030 pp 2 3 January 1995 National Instruments Corporation LabVIEW Core 1 Course Manual pp 1 1 9 24 August 2010 National Instruments Corporation CompactRIO and LabVIEW Development Fundamentals Course Manual pp 1 1 9 49 September 2008 89 21 Cooper PowerTools User Manual DIS 2 48 10 IC Metronix Messgerate und Elektronik GmbH pp 13 138 2006 90 Attachments Hardware Connection lis
40. checks the calibration data after a calibration of the analog electronics this makes it impossible to turn on the DC link without valid calibration parameters since a fault in the diagnostics will stop the shake to wake which in turn trips the emergency stop Of course a supply failure voltage will have the same effect The last part is listening to the error signal from the drivers If a driver announces an error it freezes the bridge position to make it possible for the user to see what caused the error Of course even this error will result in an emergency stop The voltage dividers are shown below X 1 1 Ac Giel 1 edes Ci x2 SND Illustration 8 Diagnostics electronics Fast stop If one of the emergency stops trips the power to the DC link and servo will be cut but normally the DC link capacitors will still be charged The only thing discharging in the normal case 1s the bleeder resistors connected across them As mentioned before it is not possible to stop the controller program before the DC link is discharged properly this means a lot of waiting during debugging for example Hence a fast stop device is implemented It 1s simply three light bulbs of 60 watts at 230 volts connected in series that is connected in parallel with the DC link using a contactor The contactor is activated closed when the machine leaves operational mode 1 e 23 3 2 8 when one of the emergency buttons trips This means that a fast stop will occur if
41. chine from scratch Most of the hardware for the prototype was already built but in need for some improvements But to give the reader of this report a fair picture of the conditions for our work here is a description of the hardware that we recycled for our work This chapter will only cover a description of the hardware not necessarily an explanation about why it 1s built as it 1s 3 1 1 Enclosure Image 1 Pulse magnetizer machine In the image above the pulse magnetizer machine is seen It is built in and on top of a metal box The inside is split into two different volumes where the big one is inside the feet of the cabinet On top of this part the magnet holder and the magnetizing head 1s placed In front of the top of this part is a pult shaped enclosure This part 1s electrically shielded from the rest of the machine and represents the smaller volume On top of it the control panel 1s placed Image 2 The big volume inside the feet of the pulse magnetizer machine In the big volume all high voltage devices such as power electronics etc is placed This is seen in the image above and gives the opportunity to connect the power electronics to the magnetizing head with short cables In the smaller volume the control system is placed as far as possible protected against electromagnetic interference Nevertheless since the mains instrumentation and the power supplies for the cRIO and the peripheral electronics are placed here to it occurs
42. d in the lenses is forcing the flux through the slit The flux is provided by two copper coils in which it is driven a pulse shaped current The equipment before this thesis was not working satisfactory and was in need for improvements The achieved magnetization was not as precise as wanted and the reason for this is not directly specified A possible reason is that the ripple of the current driven through the flux providing coils was to large hence a new current controller is to be implemented To gain better control of the machine a new modern control system has been purchased The new control system is meant to replace the old one and allow development of a more advanced control system with better performance but still be easy to use 1 2 Objectives e Replace the current controller to a control system based on NI CompactRIO and introduce a method of modulation that minimizes the ripple of the current e Improve the usage of sensors for measuring the accomplished magnetization The measuring width should be the same as the magnetizing width e Further improvement for automatic iterative magnetization of discrete magnets e Initial study of the possibility to industrialize the process in terms of the cycle time and the quality of the final magnets 1 3 Scope The target of this thesis is to meet all the objectives of the project plan Since the conditions for some of the parts of the project was unknown the objectives was adjusted during the
43. e card to the right of the indicator for the DC link voltage Image 3 show the smaller volume of the enclosure with the cRIO a metal box containing developed hardware and two power supplies On the left side of image 4 a card for the hall sensors can be seen Most of the developed hardware is nevertheless in the metal box and a close up of it can be seen in image 5 The large circuit board is for the current controller On the top right is a circuit board with level converters for the power electronics On the small prototype board is signal conditioning and the large prototype board beneath it is more electronics involved in calibration Since it is very tight on tight with space in the box there 1s another prototype board under the large prototype board the can be seen in the image This board is involved in something called Shake to wake and is explain in Shake to wake in 3 2 7 3 2 1 The usage of a CompactRIO About the CompactRIO We did not develop the CompactRIO system but to give the reader a fair overview of the control system here is a short description of the cRIO The cRIO consists of three parts an embedded computer a field programmable gate array FPGA and a number of input and output modules I O modules The embedded computer that is used is a NI cRIO 9022 and the used FPGA is a NI 9114 All together they form the cRIO into a compact industrial controller and measurement system It is up to the programmer which part of the code t
44. e fixed to 5 ms Seen from the right 78 The last measurement was made in a different way Triangular pulses with the same current magnitudes were used But this time the pulse time was altered in correlation to the current magnitude to obtain a constant slope of the triangular shape Hence the axis earlier used for number of pulses was used for current slope instead and only one current pulse was shot The current slope was varied from 1 5kA s to 101 5kA s in steps of SkA S The resulting surfaces six surface in presented below A piece of each of these surfaces is grayed to show were the requested current slope could not be maintained due to the request of current magnitude 500 400 300 200 100 Maximum magnetization G 100 200 300 150 Current A Slope kA s Surface 13 Peak magnetization with triangular shaped current pulses with varied pulse time to obtain current slopes according to the slope axis 79 SUE a termine li ar MM iv 400 300 200 100 Maximum magnetization G 100 200 300 0 20 40 60 80 100 120 Slope kA s Surface 14 Peak magnetization with triangular shaped current pulses with varied pulse time to obtain current slopes according to the slope axis Seen from the front Maximum magnetization G 0 50 100 150 Current A Surface 15 Peak magnetization with triangular shaped current pulses with varied pulse time to obtain current slopes according to the slope axis
45. e for rotation using the distance to travel Further on there is a button to manually enable the servo controller The servo controller can indicate errors from the servo controller electronics the most important of these is decoded and displayed on the Servo sub tab If there is an error in the servo communication there are also indicators of that Finally there is the possibility to save the servo communication to a log file There is a sub tab called Software that contains controls and indicators to debug the embedded computer program For example to see if the diagnostics of the supply voltages consumes too much CPU time it is possible to force the program to do or not to do the diagnostics For the same reason it is possible to force the program to ignore the controls on the front panel If an error occurs there is an indicator to show the error message The rest of the controls and indicator on the sub tab involves debugging of FIFOs These controls and indicators are placed there because they have been useful during the debugging of the software If there are more bugs the user may want to use these controls and see these indicators E Pulse magnetiser RT 2 0 vi on Pulsmag 2 0 lvpr File Edit View Project Operate Tools Window Help seni Elapsed time Omni Dis Serial communication Pause sequence o N 3 FIFO communication Resume sequence 359 94 Deg 565 39 mm Diagnostics Calibration of comparators Curre
46. e minimum time between switches is sent to the block in ticks to the Min ticks input The output signals from the two HF protection blocks are sent to the Digital out block 51 4 4 3 Generation of tolerance bands Min ticks n protection S1 HF protection S2 Digital input ee S Use both interpreter zero vectors zm Dec S2 TB Analog reference Outer band distance A levels Tolerance band size A Compensation values Digital output Block diagram 2 FPGA code In the block diagram above blocks involved in the analog reference levels has been added to Block diagram 1 To control the tolerance bands the FPGA program uses four analog voltage output signals called Analog reference levels and 1s seen to the right of the Analog out It 1s the rightmost block of the blue colored blocks The levels are calculated by the Reference calculator which is the leftmost of the blue colored blocks It uses the Outer band distance A Tolerance band size A and the current reference to calculate the ideal value for the reference levels These are then recalculated within the Reference calculator with the Compensation values from the embedded computer before they are sent to the Multiplexers The Multiplexers chooses which reference levels to send to the Analog output block So far there is only one block that generates reference levels namely the Reference calculator block 32 4 4 4 Comparator calibration
47. e move Set the Ticks until next positioning check Set check position to true Flowchart 1 Servo Change position There is very often something to do when the servo motor has reached the desired position Either a current pulse or sampling of the hall sensor should be done at the new position Therefore a task called Servo Schedule at new position 1s enqueued when the servo motor has reached the desired position This task checks what there 1s to do and enqueues Current Run to do a current pulse or Hall Start sampling to take samples of the hall sensors These tasks will in turn start a chain reaction that will enqueue the task that requested the new position 40 Positions calibration As described in 4 3 1 the calibration of the position of the hall sensors relative to the magnetization head is an important step to get a working machine Another state machine is implemented in the operating system to do this calibration As seen below it is a state machine with nine states Start Magnetization Position calibratior Magnetization Calibratior reset band Magnetization Calibration pulse Magnetization Calibratior scan init Magnetization Calibration scan 1 Magnetization Calibration scan Z Magnetization Calibration scan Magnetization Calibration scan 4 Magnetization Calibration plot State machine 2 Position calibration 41 On the front pane
48. e parallel while loop for sending data with the serial port To have a FIFO queue for communication between tasks in the operating system and the parallel while loop that reads the serial port is also a good idea But in this case it 1s called receive queue and received messages is enqueued in the while loop and dequeued in the operating system When a message is available in the receive queue a task called Decode 1s started If the decoding finds that requested data was received it starts a task that can use it If the message got corrupted on the way it will be enqueue again to the send queue To be able to know what the message was there is a third queue called sent queue This contains the messages that has been sent and they are enqueued in the parallel while loop that sends commands In the normal case the message is not corrupted but the element in the sent queue needs to be dequeued and thrown away Positioning the servo motor The servo motor is used to rotate the magnet band and thus changing the position that is magnetized To move the servo to a desired position a command is sent to the servo controller It will take some time for the servo to rotate and then the position of the servo motor is checked This is done to detect if the servo motor can not position itself correctly There could be different reasons for this It could be mechanical problems power supply failure or random errors To implement this a two state in the operating system
49. e rectifier the machine is equipped with an inrush current limiter consisting of two contactors one time relay and three power resistors of 100 ohms The function is very simple the first contactor connects the mains voltages to the rectifier via the power resistors The voltage drop over the resistors enables the DC link capacitors to be charged at a more suitable current After approximately four seconds the time relay closes the second contactor which short circuits the resistors however during these four seconds the DC link voltage has been able to rise to approximately 500 volts The first contactor and the time relay is activated by an external input for 230 volts Hall sensors During the early test of the performance of the machine it was provided with a hall sensor mounted next to the outside of the magnet band Unfortunately this sensor was damaged during our tests 3 2 Developed hardware im 0 m er e e ge ep gt e v Ve 2 BL ES Ha ww 1 D 9 t E d d SE E ep Mi Ss E i zi zf d ag ACHEN AA zi S i LF347N i i6 H LI PR a c Giai S v U GXRS3ORC M339N 46 X N N I C ox E EN BE H P Se iu e m x cf i e a 2 ee y Image 5 The inside of the metal box containing the developed elect FONICS The developed hardware can be seen in image 2 3 4 and 5 On the far right in image 2 electronics for calibration can be seen it 1s th
50. enqueues other tasks so that they can continue Hall sensor sampling To make the measurement of the hall sensors more accurate averaging 1s applied As described in 3 2 5 the supply voltage for the hall sensor 1s needed to calculate the magnetic flux The FPGA program sums up a number of samples from a sensor and does the same for the supply voltage This is then repeated for the other three hall sensors The task named Hall Start sampling makes the FPGA program start this measurement When the measurement is done the FPGA program sends an interrupt to the embedded computer The embedded computer then enqueues a task called Hall Read sample This task reads the values from the FPGA program and divides them with the number of samples that were taken It then calculates the magnetic flux Since this procedure is started by the Hall Start sampling there is always a task that 1s interested of the result Of course this is tasks that are scanning the band To let these continue on their work the task Hall Read sample enqueues the one that initiated the sampling Serial communication with the servo For the serial communication there are two while loops parallel to the one of the operating system One of them is used to send data and the other is used to receive data To write and read from the serial port the standard VIs in LabVIEW to do that is used To communicate with the controller for the servo motor a special protocol is used It is described
51. erances is a little too small There will be three graphs showing the result of each setting The first one is the current response as in the earlier graphs The second graph is showing the switch pattern and the high frequency protection activity during the shot The third graph is showing switch to switch time of each IGBT These results of the correctly tuned controller are shown in the three graphs below Current Sam iis Current A l I l l l l l l I l l 0 0005 0 001 0 0015 0 002 0 0025 0 003 0 0035 0 004 0 0045 0 005 0 0055 Time s Graph 9 Current response for a triangular reference with a peak value of 300 A correctly tuned controller 65 EM BEN oo um Switch pattern analyser 0 l l l l I 0 0 001 0 002 0 0053 0 004 0 005 0 006 Time 5 Graph 10 Switch pattern during the triangular pulse of 300 A correctly tuned controller Switch time monitor 0 00085 0 0005 0 00075 0 0007 0 00065 0 0006 0 00055 0 0005 0 00045 0 0004 0 00035 0 0003 0 00025 0 0002 24 0 00015 ae 0 0001 j SES 0 a 0 001 0 002 0 003 0 004 0 005 0 006 Time 5 Time between switches 5 Graph 11 Switch times during the triangular pulse of 300 A correctly tuned controller 66 These next three graphs shows the behavior of controller where the tolerances are a little too tight The interference of the high frequency protection a
52. es of the magnet called magnetic domains This is achieved by exposing the magnet to an magnetic field usually provided by an electromagnet Since different materials have different properties it is not unreasonable to assume that they should be magnetized in different ways If the desired remanent flux is large the flux applied by the electromagnet must also be large Nevertheless there 1s profit in exposing a magnet for a flux that is much larger than the possible amount of remanent flux The maximum applied flux is not the only important parameter The way the flux is applied is also important If the magnet 1s able to carry an electric current eddy currents will occur if the flux is applied to fast These will act like a magnetic shield which inhibits the magnetization since they will provide a flux that 1s opposite to the magnetizing flux As mentioned this magnetic property differs between different magnet materials Even the properties inside a piece of magnet differ from spot to spot This makes it difficult to manufacture magnets of a consistent quality since it seems like it is desirable to treat every spot of it individually To get around this problem LTH has produced a prototype with the aim to individually magnetize small spots of permanent magnets The magnet that is to be magnetized 1s placed in an air gap where the flux is concentrated to a small slit This is done dynamically with lenses made of copper where the eddy currents generate
53. ettings for the current with a start value delta and the number of times to increase Beyond that the Pulse tab has a setting for the pulse time and the Slope KA s tab has a setting for the number of pulses to do at every position Finally there is button to start the testing called Start surface measurement Here it 1s also good to know the total number of test positions and the number of rounds that 1s needed to do 34 4 3 2 Operating system Structure When making a program for the embedded computer it is important to have some sort of plan from the beginning Especially if the program will have as many features as this program have From the beginning we knew roughly what type features the structure of the program would need There is a need to run different parts of the program in parallel and that is easy to do with parallel loops in LabVIEW Related to parallel features is that the user wants to be able to give the machines various commands in arbitrary order Some of them might even need to run in the background after they have started In addition many of the tasks need to share variables To implement that kind of program with a state machines and parallel loops would probably be hard It would also grow in size and complexity considerably fast This would have a negative impact on other features of the structure of the program namely that it needs to be easy to develop the program and add more tasks with ease The solution used to sa
54. etween rounds s 1015 A 300 Use 1 Use gis Use 17 ao o in Use 25 Zo Number of positions to use s 4 J Ma Use 2 70 Use 10 4375 5 Use 18 Jo Use 267 o Width of teset stretch mm Use 3 J 100 Use 11 1400 Use 19 0 Use 276 Jo 430 Use4 136 opw Use 206 Jo 2 270 Pulse type ee pe Use 13 Use2ir pm Use 290 triangle v ws yo d 6stSurface Use 4 EES gss tag 520 Use 23 15 Use 31 ig Uses zen Use 16 7 A Use 24 Jo Use 32 Ze Pulsmag 2 0 lvproj cRIO Pulsmag 4 Front panel 7 Magnetization Surfaces 33 The last sub tab in the interface 1s called Surface and is seen above It is used for testing the performance of the magnetization device These tests always start with a band that is evenly magnetized this is done with the same reset method that was mentioned earlier The simplest test imaginable to then perform is to shoot a pulse and scan the affected stretch of the magnet An analysis of the result can then tell how much it was magnetized and how long section of the magnet that was affected The result will depend on a number of things and this sub tab is used to investigate them Things that have been tested are to vary the current pulse type number of pulses and the slope Again it should be noted that the calculation for the slope is only valid for the triangle pulse type The tests in this sub tab are designed to va
55. eved magnetization level with the reset is done in the Magnetization Surface 47 pulse reset scan analysis and decision where to go next is made based on the result When the reset 1s adequate it is stored in the file The flowchart for the Magnetization Surface pulse pulse is seen below Start run pulses gt max pulses No Ye S Look at next position Use this position Last position Yes Increase pulses Enqueue Current Generate Enqueue Current Run Enqueue Magnetization Surface pulse pulse pulses gt max pulses Yes pulses pulses 0 Increase Current with Delta Increase run pulses Current gt stop current Enqueue Magnetization Surface pulse pulse scan No Enqueue Current Generate Enqueue Servo Change position Increase reset pulses Set run pulses 1 Flowchart 5 Magnetization Surface pulse pulse 48 It is a more advanced state machine than previous once but it uses the same concept The called Magnetization Surface pulse first pulse will produce the first pulse then jump to the Magnetization Surface pulse pulse It is almost the same but with some minor differences since it is producing the first pulse When the tasks that implements these flowchart executes the machine will start at the start current and the starting number of pulses at the first test position Then it will increase the number of pulses with delta number of pulses a
56. fferent interrupts 26 4 3 The structure of the program on the embedded computer 4 3 1 User interface Current control snatyas Flux Free nun scan Hal srao position calbraton Magnet disgnosties Scan band Hal serzor 1 can outer 3 125 mp aae 9 Maas me 315 mvs an 92 ES 40 100 Diagnostics Calibration of comparators CUTenE CONT Magnettzation Food band 4111 Pube generator Get Res En ve an Pulse npe Amplitude eo Trance 2 Amplitude Gauss un m 5 B 5 5 9 o 8 Sepe kas Fute tms P am a 600 i Fute sme Pulse tine to ust bh ar Sereor 3 D Emm animer 13m Controller parameters 100 Torako Error before 3 bari e faj further nl 1 0 3 d o Ke r H 59 I x 3 F 100 E 190 Plot reference RUN l Plot inner bands o M Save data to fles Plot outer bands D W s 005 m Fusmag EI Front panel 1 Whole system The front panel seen above is the whole user interface Since it is a tab based interface not everything can be shown at the same time The idea with the tab based interface 1s to save space on the front panel and make it user friendly but at the same time advanced and powerful by only showing the necessary controls graphs etc to the user The interface utilizes the space of two screens and can be divided into three parts Some controls are always good to keep within reach and these are placed in the upper part of the left
57. h set of settings there are two different surfaces showing the peak magnetization 70 respectively the magnetization width and all surfaces is shown in three different angles To begin with two measurements with identical settings except for the pulse shape was made The first measurement is done with half sinusoidal pulses and the second is made with triangular pulses The current is starting a 0 A and is increasing by 5 up to 150 A The number of pulses is starting at pulse and is increasing by 5 up to 101 pulses The pulse time is fixed to 5 ms The resulting surfaces for the half sinusoidal current pulses are shown in the six surfaces below Am i 300 Jl lt lt c BEE e di e i 00 7 i Maximum magnetization G Ft 00T 150 Current A Number of pulses Surface 1 Peak magnetization with half sinusoidal shaped current pulses and with pulse time fixed to 5 ms 71 Maximum magnetization G 300 0 20 40 60 80 100 120 Number of pulses Surface 2 Peak magnetization with half sinusoidal shaped current pulses and with pulse time fixed to 5 ms Seen from the front 72 400 300 200 e e Maximum magnetization G 100 200 Current A Surface 3 Peak magnetization with half sinusoidal shaped current pulses and with pulse time fixed to 5 ms Seen from the right Current A Number of pulses Surface 4 Pulse width with half sinusoidal shaped current pu
58. has been added Servo Display rotor position and Servo Change position The first 1s enqueued when a message about the position of the servo motor is received The message is converted from the position format used by the servo controller into position in degrees and mm It also enqueues the Servo Change position if it was the one that requested the position The Servo Change position is enqueued when a task wants to change the position The tasks that request the change in position must provide the new position in the variable New position in mm in the variable cluster Servo Change position ensures that the servo motor will be at the desired position by looking at the present position and the desired position order the servo controller the move according to the difference in position This means that Servo Change position will be enqueued more than once to ensure that the servo motor will be at the desired position If it does not make it on the first try it tries again until it decides that it 1s not possible to reach the goal and alerts the user A flowchart for the logic of this task is presented below 39 Start Check position Order to read position ES position withir tolerance Abart sequence Number of positioning tries 307 Enqueue next task to do Clear tasks to schedule Set check position to false pe Calculate remaining distance in degrees order t
59. hat all of the values shown in the indicators are okay they could be downloaded and used in the FPGA with the Use button Finally there is a control called Voltage scale factor and it 1s ideally the inverse of the conversion ratio calculated in Voltage measuring in 3 2 2 Since it 1s a control it is hence possible to compensate if the real conversion ratio differs from ideal one Pulse magnetiser RT 2 0 vi Front Panel on Pulsmag 2 0 lvproj cRIO Pulsmag File Edit View Project Operate Tools Window Help tt 17pt Appication Font 82 gt Stop execution Elapsed time Omn 0 1 Pause sequence Resume sequence Serial communication FIFO communication 352 22 Deg 565 37 mm Diagnostics Calibration of comparators Current control Magnetization Fixed band t Reference aM Pulse generator Current hel Kl E Pulse type Amplitude 60 riangle D 450 eg Slope kA s Pulse time 4665 0 0015 2 Pulse time E 45 d time ae Joos SU 40 Controler parameters 35 Tolerance Error before 30 bandsize A further action 96 I lt 20 40 5 1200 299 Current A Ed EX 20 Ben NM 15 Plot reference RUN Iv Plot inner bands Save data to files Iv Plot outer bands Cursors X Y l pe P D 1 1 1 1 1 D 1 1 1 0 005 244 0 0 0005 0 001 0 0015 0 002 0 0025 0 003 0 0035 0 0
60. hat is to be executed in the embedded computer and which part that 1s to be executed in the FPGA and the code look similar but it 1s executed in very different ways The code executed in the embedded computer executes task by task since it 1s a single core processor nevertheless it 1s executed in a real time operating system The code that is downloaded into the FPGA is practically built into hardware This sets clear limitations in the complexity and the size of the code The big advantage is that everything that runs into the FPGA is executed simultaneously hence it is very powerful in controlling applications if it is used properly The FPGA is connected to both the I O modules and the embedded computer Hence controlling tasks can be handled by the I O modules and the FPGA only with extremely short response time while heavier less time critical calculations is delegated by the FPGA to the embedded computer The embedded computer is equipped with two Ethernet ports These are used for communication with the host PC that 1s used for programming and interfacing it for example Furthermore it is equipped with one RS232 port for miscellaneous use and an USB port to connect an extra storage volume 3 The RS232 port is in our case used for communication with the servo used for positioning of the permanent magnet Input and output modules The cRIO is very flexible in the way that you can choose I O modules that suits you task In this project five mod
61. he relays As mentioned the power electronics is supplied directly from rectified three phase mains The original design described earlier 1s used The servo uses two different power supplies one that 1s delivering power to the motor drive and another to supply the control electronics Hence it is possible to start the controller up to download safe controller parameters for example before turning on the power All power supplies are connected to the mains via a fuse recommended by the manufacturer except the supply for the servo power which is internally protected The power electronics is connected via three 16A fuses placed in the inrush current limiter box At last there 1s an extra power supply for the temporary installation of the variable inductance It provides 5 V and 24 V for maneuvering a relay and a contactor Hall sensors The magnetizing result is measured with a set of hall sensors Both the inside and the outside of the band is measured and since the maximum field strength of the band was unknown two different sensors was mounted to ensure both high accuracy as well as high field strength measure capability This means that totally four sensors are used They have a rated supply voltage of 5 volts which is available But since it is used to feed digital circuits it was not used for this purpose Instead the analog 15 volt was used a regulated down to 5 volts via a voltage regulator The output of this type of hall sensor is a
62. hey are only used in the embedded computer Related to the local variable is the global variable Instead of being located on the front panel of the program it is placed on the front panel of a separate VI This makes the global variable accessible from more than one VI If more than one global variable is needed more variables can be added to the separate VI with the first global variable The structure of the FPGA program makes global variables very useful to create variables that are used within the FPGA In the program for the embedded computer global variables are only used for communication between loops that run in parallel FIFO queues FIFO Queues First In First Out Queues are essential in both the program for the embedded computer and the transfer of large amounts of data from the FPGA program to the program in the embedded computer or vice versa A queue is a data structure that allows data to be produced and consumed at different times but still not lose the order of the data A FIFO queue dequeues the data in the same order as they are enqueued Interrupts The program in the FPGA can generate an interrupt to the CPU in the embedded computer An interrupt 1s meant to interrupt the normal execution of the program in the embedded computer and run a special part of the code called the interrupt routine The interrupt routine is then supposed to take care of the reason for the generation of the interrupt It is possible to generate up 32 di
63. how to implement something that can do one or a series of pulses in a row A flowchart of the state machine can be seen below Start First position mark pulses gt pulses to do Enqueue Current Generat Enqueue Current Generate Enqueue Magnetization Position calibration Enqueue Servo Change position Enqueue Current Run mark pulses 1 Increase reset pulses Flowchart 2 Magnetization Calibration pulse 42 The variable mark pulses is in the variable cluster and is initiated to zero in the Magnetization position calibration state The state machine utilizes the fact that both Current Run and Servo Change position will start a chain reaction that will enqueue task that implements this state machine Reset state machine A flowchart of the state Magnetization Calibration reset band can be seen in the flowchart below Start Calibration reset position gt stop position First position es Enqueue Magnetization Position calibration reset pulses gt pulsesjposition Enqueue Current Generat Calculate new position and update Calibration reset position Enqueue Servo Change position Engueue Current Run reset pulses 0 Increase reset pulses Flowchart 3 Magnetization Calibration reset band It looks similar to the previous flowchart and that is not a coincidence The method to utilize the chain reaction is
64. ht tolerances result in more current ripple than necessary Suggestions of improvements The overall current performance could be improved a little more if the tolerance bands where shaped progressively to always utilize the highest possible switching frequency However it would not improve the maximum ripple level of a given current pulse with a correctly tuned controller A more sophisticated solution of the variable inductance with a bigger variety of inductances could be implemented but faster power electronics would probably be a better solution 87 6 2 6 2 1 6 2 2 Another approach is to control the actual flux instead This could be done by for example a time integration of the voltage applied on the flux proving coils which will be proportional to the provided flux if the resistive voltage drop is subtracted Magnetizing results First of all when reasoning about these measurements the reader should keep in mind that the used magnet material 1s bonded ferrite This means that no significant eddy current should occur in the magnetic material Reset There is not much to say about the reseting results It was obtained by trial and error to find something to make further tests from The method is as one can see working well and it is in practical use very time effective The big variations at the beginnings and the ends of each measurements corresponds to the ends of the magnet band and the small variations corresponds to
65. iO SSalety Oz ve O 57 4 9 SWIICH Pattern Analyzer ee ee ee ee an eis 58 4 5 NEST TD Bro SEAS EN 58 SEN 60 Ob SEHEN COMMON ee Reli 60 3 2 Aere 69 Do ROSE U Dre ee ee 69 322 UAC NET URL TTE 70 SG MEE Ur 86 Discussion and Cone SiON E 87 DIE A OO e een es 87 6 1 1 The work of the Controller sale a netu td TO 87 6 1 2 4 SUSSCSUONS EIERE a ad 87 OZ ee 88 QUAM MEM T dT ee ee ea Des 88 0 2 2 Een Eeer 88 Referenten 89 PANT TIE T ee ee Dee 91 EE 9 Ee RE 9 ligi Nee IPIS SRI EE FT OT m 93 NDPALDEAJT COC EE EE 95 1 1 Introduction Background In a large range of electromechanical applications a magnetic force is used in the conversion between electrical power and mechanical power in either direction Many of these devices involve permanent magnets and in these devices there is often a strong correlation between the performance of the used permanent magnets and the performance of the final device For that reason it is desirable with permanent magnets whose properties are within tight tolerances Different materials have different magnetic properties 1 What is common for those material that are used for permanent magnets are the ability to remember a magnetic flux this is called remanence The properties of the remanence differ between different materials in several ways Generally the most important parameter is how large flux the magnet is able to store Magnets are magnetized by affecting the magnetic entiti
66. ication E 6 T FIFO communication D E 132 V 359 99 Deg 56548 mm Diagnostics l Calibration of comparators l Current control Magnetization Free run Hall sensor position calibration l Magnet diagnostics Scan band l Surfaces l index pos current hall Jo go Present index emm Note Thise indexes does NOT use sensor position correction Speed control Position control normal mode 11 195 987 Magnetize Scan Plot From index From index From index UT 40 3 Jo 1760 2 to to to BO index index index 100 SE Mo Jeso e OE ME OE A 20 n zm ali Kai e i 40 160 Bo pa 200 G 220 ERE l Create indexes in array Modify current values in array m Start index Pos offset Startindex Current 3 3 1 a 35 Stop index Pos increment Stop index e ZK RE SEN 28 Create indexes Modify current values 1240 260 mn mum 280 300 320 9i 340 um 4A z zm Ze um o TTT TT um EAGLE C zem TTT mm co Pulsmag 2 0 lvproj cRIO Pulsmag 4 Z Front panel 5 Magnetization Free run tab The first sub tab is Free run and is seen above It gives the opportunity to rather freely program an array with pairs of positions and currents The values in the array can then be used to magnetize 3 the magnet band scan the band or to plot a result of a scan in the Free run scan tab in the graph tab interface The
67. igital inputs and outputs and the sampling rates of the A D and D A converters In the FPGA it is possible to make the signal processing fast and achieve control loops running faster than one us when using the digital I O signals Tasks with a close relation to 25 4 2 4 2 1 4 2 2 4 2 3 the I O modules like fast control loops is also directed to the FPGA Tasks that involve input from the user or presenting data to the user is not that time critical and is thereby natural to implement in the embedded computer Tasks involving signal processing before presentation to the user could be implemented on the embedded computer or the FPGA Where it is implemented depends on demands on speed amount of data complexity of the algorithm and the integration with the rest of the program It is also possible to split it and do some processing in the FPGA and the rest in the embedded computer Communication between the different parts Everywhere in the programs both in the embedded computer and the FPGA there is a need to send and receive information from other parts of the program As expected there is more than one way to do this and the ones used in this project will be explained here Variables Sometimes a control or an indicator on the front panel is needed in more than one place in the program This is achieved with a so called local variable to the control or indicator Local variables can only be used in the same VI and in this project t
68. implements three timers where two of them were mentioned in 4 3 1 namely the elapsed timer and the cool of timer The third timer involves the servo motor and more specifically the time it takes for the servo to move to a new position It also now and then asks the servo controller if there are any errors and checks if the servo controller has sent any new messages Another interesting thing is how the state machine described above 1s started It is started by a task called nit which enqueues the dle task The nit task also initiates some variables and the communication with the servo The dle state in turn 1s enqueued before the while loop that is a part of the operating system In program like this the number of variables used within the program to store values and send values between tasks get fairly large To cope with this a cluster containing the variables 1s used The cluster is coupled to a shift register on the operating system while loop This means that the values stored in the cluster from one task can be read by the next task If all tasks pass on all the variables except the one that they change a value that 1s stored can be read by any task that later executes The FPGA program can generate three different interrupts To deal with this a while loop parallel to the one of the operating system 1s used When an interrupt occurs it will simply enqueue a task to the task queue of the operating system that handles the interrupt The FPGA p
69. it Current control Overview The current that is driven through the flux providing coils is regulated to a shape and magnitude decided by the user to allow maximum flexibility To ensure good control of the current ripple a tolerance band controller is used As mentioned earlier a hybrid between the analog and the digital tolerance band controller is used The desired tolerance bands in other words the current shape and magnitude are generated digitally But the comparisons ie what determines if the current is within tolerance or not is done analogy which gives almost immediate response To make this possible the generated bands DA converted with an analog output module and feed to the peripheral electronics The comparison signals 1s feed back to the FPGA where the choice of new signals to the IGBT drivers 1s made There are totally four tolerance bands two inner and two outer The purpose of the inner bands is to know if the current is within tolerance or not The purpose of the outer bands is to know if the chosen control action was sufficient or not The schematic of the peripheral electronics is found below and will be explained the following paragraphs 11 R24 R25 R20 100 R END R39 3 3k S i GND 551 i 3 um 150 6 LM338N DEND c 15 Serer L d Ka SE But 10k TOR A G D NE
70. l the user has the ability to decide which of the steps in the calibration procedure that should be done The Magnetization position calibration state supervises the whole calibration procedure A state machine with a supervising state makes it easy to skip one of the states and thus implementing the feature to decide which of the steps in the calibration procedure that should be done A supervising state is also a convenient way to initialize variables before entering the other states First the Magnetization Calibration reset band will be executed to reset the band then a pulse at a known position is shot with the Magnetization Calibration pulse state After that it is time to scan the band and first an array to put the measurements is created in Magnetization Calibration scan init the magnetization level is measured with and stored in the array with the four states named Magnetization Calibration scan x where x 1s a number from to 4 Finally the result is plotted with the state Magnetization Calibration plot This state also calculates the correction factors for the position When the USE button on the front panel is pressed the task that implements this state is enqueued This time the calculated correction values will be saved in the variable cluster and the plots will be updated and use the new correction values Pulse state machine Let s look a little bit closer on the Magnetization Calibration pulse state since it is a perfect example of
71. layed These samples are written to a FIFO queue called Analog samples The DC link voltage is also sampled during the current pulse It would be neat to put these samples in another FIFO queue However it is only possible to have three FIFO queues and the third 1s needed for another thing The solution is put both the current sample and the voltage sample in the same FIFO queue The samples are taken at the same time thus there will be equally many of them and they will be interleaved in the FIFO queue The software in the embedded computer needs to take care of this and to make the data useful in needs to separate it into two vectors 37 The third FIFO queue is called Digital samples and it transfers data about the switching instances of the power electronics to the embedded computer When a switching instance occurs an element is written to the FIFO queue This elements contains information about which switch was activated the boolean value after the switch and on which clock cycle it occurred The embedded computer can then reconstruct the entire switching pattern from the data in the FIFO queue When a current pulse has been played and the FIFOs Analog samples and Digital samples is filled with data the FPGA program generates an interrupt This means that the program can continue with the next step That could be analyzing the data or continue with a magnetization procedure Therefore the System Read FIFOs task that 1s enqueued by the interrupt
72. ls The level conversion 1s made in the same way as for the enable signal The capacitor in the transistor coupling shortens the fall time of the converted signal since the base current will be greater at the transition The capacitance 1s experimentally determined to make the fall time as close the rise time as possible It 1s feed to an inverter to buffer and make the signal to the lower IGBT and then inverted again to make the signal to the upper IGBT The last input is only buffered with to inverters to use for other purposes The level converter listens to the error output of the drivers Since it is negated it 1s inverted and then converted into a 5 V signal by two resistors and feed to a input on one of the NI 9401 modules The inverter is driving a transistor to with the purpose of providing the card with an extra error indicating signal output with the capability to driving a led or a relay for example The used logical gates were 744HCU04 13 TC4049 14 and HEF4081 15 16 3 2 4 Illustration 4 Level converter electronics Power supplies There are several different power supplies for different parts of the machine There are different reasons for using more than one power supply First of all many different voltages are used Secondly it is not wise to use the same supply voltage to feed for example both analog and digital circuits Thirdly if you use many different supplies you can start them up in an order that
73. lses and with pulse time fixed to 5 ms 73 Width mm Number of pulses Surface 5 Pulse width with half sinusoidal shaped current pulses and with pulse time fixed to 5 ms Seen from the front 74 Width mm Current A Surface 6 Pulse width with half sinusoidal shaped current pulses and with pulse time fixed to 5 ms Seen from the right 75 The resulting surfaces for the triangular current pulses are shown in the six surfaces below 400 300 200 100 100 Maximum magnetization G 200 300 150 Current A Number of pulses Surface 7 Peak magnetization with triangular shaped current pulses and with pulse time fixed to 5 ms Maximum magnetization G 300 0 20 40 60 80 100 120 Number of pulses Surface 6 Peak magnetization with triangular shaped current pulses and with pulse time fixed to 5 ms Seen from 76 Maximum magnetization G Current A Surface 9 Peak magnetization with triangular shaped current pulses and with pulse time fixed to 5 ms Seen from the right Current A Number of pulses Surface 10 Pulse width with triangular shaped current pulses and with pulse time fixed to 5 ms 71 Width mm 60 80 100 120 Number of pulses Surface 11 Pulse width with triangular shaped current pulses and with pulse time fixed to 5 ms Seen from the front Width mm Current A Surface 12 Pulse width with triangular shaped current pulses and with pulse tim
74. m selector Dec zero vectors IHNEN TB gt HF protection S2 Start playback Clear play FIFO Analog reference levels Reference calculator Outer band distance A Tolerance band size A Multi plors Analog out Embedded computer Measurments Compensation values Compo Compl Comp2 Comp3 Comparator cal start calibration Digital output Test current direction Block diagram 4 FPGA code To generate a current pulse the Sequence player is added to the block diagram and the resulting block diagram is seen above The current reference is changed by the Sequence player to generate a current pulse It gets the data points for the current pulse from a FIFO queue called Current to play which the embedded computer puts the data points into When the Start playback signal is activated the Sequence player starts to consume the data points at a constant rate and uses the value as the present current reference During a period of time this will create a current pulse The Seguence player is also responsible for changing the values of the D and the Zm signals with the C I D and C Zm signals before it starts to generate the current pulse 54 4 4 6 Error watcher Compo Min ticks S1 S1 TB HF protection S1 Digital input comparet vectortype Zero Use both g P t interpreter selector zero vectors S2 TB CC Comp HF protection S2 Start playback Sequence Clear play
75. mains voltage A close up image of the inside of the smaller volume is seen below Image 3 The smaller volume with cRIO metal box containing developed electronics and two power supplies 6 3 1 2 3 1 3 3 1 4 Magnetizing head re A E Image 4 The magnetizing head of the pulse magnetizer machine The magnetizing head is seen in the image above and consists of an adjustable yoke with an air eap containing the magnet that 1s supposed to be magnetized Next to the magnet the yoke ends up with two flux collection lenses that are concentrating the flux to a defined area This area 1s shaped as an upended rectangle with the height of the magnet and the width equal to the desired resolution of the magnetization Next to these lenses there are two coils providing the flux when an appropriate current is driven through them These coils are coupled in series Magnet holder The magnet holder is a prototype made for evaluation of the rest of the machine It is simply two pieces of magnet band glued to the cutting surface of a piece of plastic pipe The plastic pipe is in turn attached to a rotatable base consisting of an aluminum plate and a servo motor This means that you can choose which part of the magnet band that 1s to be exposed to the concentrated flux of the magnetizing head by sending a position reference to the servo Power electronics The old power electronics was used as well however a new controller was a part
76. n a scanning sequence of the magnet band or a magnetization sequence of the magnet band 1s started a timer starts The timer is used to show the time in minutes and seconds since the sequence was started and 1s located in the upper part of the interface In the same area there are buttons to pause the execution of a sequence and resume it later These buttons are called Pause sequence and Resume sequence When the servo can not go to the desired position it will also pause the sequence in the same way as the pause button does the sequence can then be resumed 27 using the resume button when the user has solved the problem The upper part of the user interface also contains a dial and an indicator of the DC link voltage and a dial and indicator of the rotor position in degrees Finally there is two boolean indicators one that show when there is serial communication with the servo controller and the other shows when the RT uses the FIFOs for sampled current and voltage data Underneath this is the main tab interface with four different tabs diagnostics calibration of comparators current control and magnetization Three of these tabs have in turn a tab interface into them these tab interfaces is referred to as sub tabs E Pulse magnetiser RT 2 0 vi on Pulsm File Edit View Project Operate Tools Window Help Elapsed time O mn Ols Serial communication Pause sequence E x FIFO communication l Resume sequence A 359 96
77. nd ylaoel tturrent zlabel Maximum magnetization G if slope 5 Slope mx x 1 15 x end 99 length x 4 Hy yl 22 02 y end 7 else Pulse mx x 1 2 x end my y 1 2 y end end nxi myi meshgrid mx my Zi interp2 x y zl mx my linear r gqure rlgure Start SULT Mey Z1L if slope xlabel Slope kA s else xlabel Number of pulses end ylabel Current A zlabel Maximum magnetization G clear n clear nbr of index clear divide index Clear resetname clear dataname clear datanumbername 100 PeakNDeltFinderOneRound m function result invalid result current out pulses out PeakNDeltaFinderOneRound reset data pulse data usedpes t ons 1Lera tions Scaningres pulse width current delta current pulses delta number of pulses number of pulses max number of pulses invalid current invalid pulse result zeros iterations 14 invalid result NaN iterations 14 half width floor pulse width scaningres 2 for n l iterations index usedpositions n scaningres result n current pulses result n 3 14 pd pulse data index halft widthzindexthalt width 2 rd reset data index half width indexthalf width 2 ir current gt 1 PeakValue RelPeakValue FirstIndex SecondIndex DeltaX PeakNDeltaFinder pd rd else mid floor pulse data length pd 2 PeakValue pd mid RelPeakValue pd mid rd
78. nd shoot the resulting number of pulses at the next position to use The number of pulses is increased until they reach the limit of the testing interval Then the current is increased by delta current and the number of pulses starts over again at the starting number of pulses The task will continue to do this until there is no more positions to use for tests or there is no more tests to do The Magnetization Surface slope pulse works in the same way but the number of pulses 1s fixed and instead the slope After that once again a scan of the magnet band is performed and the result 1s then saved If the machine is finished it alerts the user about that else it continues in the loop and runs more test pulses To be able to perform the analysis of the results in MATLAB the test conditions needs to be known This is achieved with the file name The format of the file name and an example can be seen in table 1 and table 2 below For the tests with number of pulses number of pulses is used instead of slope in the file name The Pulse time is irrelevant for the test with slope but a value is used anyway to be consistent with the file name Table 1 File name format l 2 3 4 5 6 7 8 9 10 11 Date Io A Deltai A pulsesO Delta Max Pulse withWaveform Pulse timeUser xls pulses pulses mm ms string Table 2 File name example 11 08 02 0 0 5 0 1 5 101 30 0 Triangle 0 005 Final xls
79. netization relative to the magnetization for the reset at the z axis The third surface plot has the width of the magnetized segment on the z axis The script is called Data divider m and can be found in the attachment MATLAB code The function PeakNDeltFinderOneRound mentioned earlier uses the reset data and result data from one round along with conditions of the tests It goes through every tested position and analyzes the data with a MATLAB function called PeakNDeltaFinder for every sensor When it 1s finished it returns the result It also returns matrix with the invalid results that 1s used to make the erayed area in the surface plots The code for PeakNDeltFinderOneRound can be found in the attachment MATLAB code The PeakNDeltaFinder function takes two vectors one containing data with a peak in and one containing the data from the reset scan It then finds the peak value in the data and calculates the width of the peak It also returns the peak value the data relative to the reset The code for PeakNDeltaFinder can be found in the attachment MATLAB code If the user wants to study the pulse that analyzed to get the data point in a surface there 1s a m script that does that When an interesting data point has been found the x and y values is entered in the script The script finds the round containing that pulse and plots it It also tells which of the pulse that is the interesting in the command window in MATLAB The script is called peakCurren
80. nt control Magnetization Calibrate Calibrating l Scale factors for analog output Scale factors for analog input Comp OK 0 0211182 Comp O M 0 0137329 Comp 1K 0 0209961 Comp 1M 0 0140381 Comp 2K 0 0211487 Comp 2M 0 0088501 Comp 3K 0 0211487 Comp 3M 0 0105286 Current sensor K 47 5625 Current sensor M 0 0105591 Voltage scale factor 101 v Pulsmag 2 0 lvproj cRIO Pulsmag 4 S Front panel 3 Calibration of comparators As mentioned in Calibrating electronics in 3 2 2 analog electronics have uncertainties that is 29 measured and taken care of in software To perform this calibration the sub tab Calibration of comparators that is seen above is used It is easy to perform a calibration it is just to press the Calibrate button After the calibration is done the scale factors that was calculated from the measurements is presented to the user in indicators in the sub tab The K factor for each channel 1s the calculated gain for each channel It should be compared to the conversion ratio 0 02 V A that was calculated in Current measuring in 3 2 2 The M factor is an calculated offset and for the comparator channels it will compensate for hysteresis in the comparator and offsets in amplifiers The fact that they compensate for the hysteresis means that they should not be zero but sill rather small Comparing the M factor between the channels there should be two groups channel 0 and 1 and channel 2 and 3 If user thinks t
81. o variables The user interface for this script is used in the same way as for the previous script The script is called MATLAB file loader m and can be found in the attachment MATLAB code The analysis 1s done with the third m script First of all the user needs to enter what type of test that is analyzed if it is a performance test with different currents and number of pulses or with different currents and slopes The user can also specify which of the sensor to use First the script extracts the used positions from the data Then it starts to divide the rest of the loaded array into smaller arrays that each one contains the data from one scanning round of the band In this process the reset data 1s denoised The number of test positions that has been used in every one of the scanning rounds is also extracted and saved in variables Note that there are two scans for each test round one from the reset scan and one from the pulse or slope scan The scanned rounds are then iterated through and analyzed by a MATLAB function that has been created it is called PeakNDeltFinderOneRound The results from PeakNDeltFinderOneRound are added up to a resulting matrix After this the script takes the data corresponding to the chosen sensor from the matrix and plots three different plots The first 1s a surface plot with the test conditions on the x and y axis and the maximum resulting magnetization on the z axis The second surface plot has the maximum resulting mag
82. of the thesis The power electronics has been used almost as it was except for some smaller changes and repairs It is 7 3 1 5 3 1 6 built like a module with IGBT s with suitable drivers DC link capacitors rectifier and inductors for the rectifier mounted on a metal plate The IGBT s and the rectifier are further more mounted on a water cooling block however this is not connected since the losses does not seem to cause any significant temperature rise It is supplied directly from rectified 400 volts three phase mains There is no transformer isolating this power supply from the mains hence it is completely isolated from the rest of the construction This means that there is a lot of power available to the rectifier since there is no stray inductance of resistance as in regular power supply with a transformer To make it possible for the rectifier to survive a sudden drop of the DC voltage there are two inductors of 0 3 mH between the rectifier and the DC link capacitors with the aim to smooth current peaks Between the rectifier and the inrush current limiter there is a mains filter with the aim to prevent the machine from disturbing other equipment Inrush current limiter When energizing the DC link the voltage across the DC link capacitors are usually zero This means that the current through them will huge if the rated voltage of the machine is applied at once To prevent this inrush current from breaking the mains fuses or even th
83. positions to magnetize scan or plot can be made arbitrary in the array To make that possible in this tab all references are made to the array indexes and not the position values To manually enter many values in the array is time consuming so there is two toolboxes to do that The first creates equally spaced position values and the second fills them with current values The servo motor can run in two different control modes speed control and position control Normally the position control is used because it is the position that is interesting to control The speed control is used to make the servo and thereby also the magnet band spin at a constant speed This could be used to check mechanical variations and other testing To change between the two control modes the servo needs to be disabled before the change and enabled after by the Enable servo button P Pulse magnetiser RT 2 0 vi on Pulsmag 2 0 vproj cR10 Pulsmag File Edit View Project Operate Tools Window Help Stop execution N Elapsed time Ready mus 1 Serial communication Pause sequence A FIFO communication 6 h e 132 V 359 96 Deg 565 42 mm Diagnostics Calibration of comparators Current control Magnetization Free run Hall sensor position calibration Magnet diagnostics Scan band Surfaces Sensor 1 Sensor 2 Sensor 3 Sensor 4 Start calibration Do reset Do pulse Doscan fe e fe Current 2
84. preter Tolerance band size A Comparator values OK Multi plexers Mode Embedded Measurments Compensation values computer Compo Compl Comp2 Comp3 Reference levels Comparator cal start ka Comparator calibration Clear driver error Emergency stop Comparator cal in progress ke hall sampl TEC uisi dii rS tput Take hall sample Analog input Disable drivers ha nable drivers outpu Driver error Measurments Digital output Fmhedded computer Voltages OK Operational Shake to wake Activate Emergency stop Block diagram 7 FPGA code In the block diagram above the Safety features Shake to wake and Comparator error blocks has been added These two first 1s responsible for turning on the DC link voltage when the Activate signal is activated and everything is okay The Comparator values OK signal should indicate that they are okay the software controlled Emergency stop should not be activated and there should not be any Driver error The Shake to wake block will provide a 1 kHz signal to the Shake to wake electronics when the Operational signal 1s true The Comparator error block is also added in block diagram 5 It looks at the comparator signals to detect 1f there 1s any illegal combination and stores it 1f one found An illegal combination could be that the comparators indicates that the current is both higher and lower than the reference at the same time This 1
85. re shown by the two lower lines in the switch pattern graph Current EB jg i Real v Current A I I l l I l 0 0005 0 001 0 0015 0 002 0 0025 0 003 Time 5 l I l l l 0 0035 0 004 0 0045 0 005 0 0055 Graph 12 Current response for a triangular reference with a peak value of 300 A incorrectly tuned controller 67 ei Enc Eo gt mM Switch pattern analyser p T l l 0 001 0 002 0 003 0 004 0 005 0 006 Time 5 Graph 13 Switch pattern during the triangular pulse of 300 A incorrectly tuned controller Sibch time monitor 0 0007 D 0006S 0 0006 1 00055 1 0005 O 00045 0 0004 1 000 35 0 0003 O 00025 0 0002 Time between switches 5 a 00015 0 0001 SE 5 R n 0 l l T 0 001 0 002 0 003 0 004 0 005 0 006 Time s Graph 14 Switch times during the triangular pulse of 300 A incorrectly tuned controller 68 At last a graph showing the DC link voltage during a 300 A triangular current shot is shown in the graph below DC link voltage HEA s sotan o d E dd d rh ele Vi h L I I l l I I 0 001 0 002 0 003 0 004 0 005 0 006 Time s Graph 15 DC link voltage during the triangular pulse of 300 A 5 2 Magnetizing performance As mentioned the magnetizing performance of the machine was tested These tests were focused on the ability to magneti
86. rmData end FirstHalf NormData 1 1 Moonshine FirstIndex min abs FirstHalf PeakValue ResetMean 2 ones length FirstHalf 1 SecondHalf NormData I 1 NbrOfElements Moonshine SecondIndex min abs SecondHalf PeakValue ResetMean 2 ones length SecondHalf 1 SecondIndex SecondIndex length FirstHalf DeltaX SecondIndex FirstIndex end 103 peakCurrentPlotter m current current0 pulses number of pulses current goal 190 X in 2 Suri we plor pulse goal 965 X in a surface plot test pot nbr 1 round nbr d St Je current lt current goal if pulses gt max number of pulses pulses number of pulses current current delta current else pulses end gest Po ne IDE test polme nbr i datanumbername genvarname workname eval temp datanumbername if test point nbr temp test Point nbr 1 round nor round nor 17 end ii current gt Max current disp This was bad 1 break end end Aa datanumbername genvarname workname eval temp datanumbername while pulses pulse goal pulses test point nbr test point nbr 17 ir test polnt nbr gt temp teot point be e 13 round nor round nbr 1 datanumbername num2str round nbr eval temp end D datanumbername end dataname genvarname workname eval temp data dataname if pulses max number of pulses disp This was bad 2 t else interres
87. rogram generates interrupts when the calibration for the comparators is done when a magnetization pulse is done and when a sampling of the hall sensors with averaging 1s done Current pulse generation To shoot a current pulse the pulse needs to be generated in software first The method used in this project is to generate the pulse in the software of the embedded computer and then load the data points to FPGA The FPGA then consumes the data points at a constant rate and uses the value as the present current reference During a period of time this will create a current pulse To send the generated data from the embedded computer to the FPGA a FIFO queue called Current to play 1s used Using a FIFO queue is a good idea since that is are made to send large amounts of data from the embedded computer to the FPGA or vice versa The task Current Generate generates the current reference described above and writes the data to the Current to play FIFO To get the desired current the current controller in the FPGA needs the correct tolerance bands The Current Generate task therefore also sends the correct values for the tolerance bands to the FPGA When the data is loaded the only thing remaining to do is to start the playback of the data This is done by the Current Run task After a current has been played the user may want to see how the resulting current was Therefore the FPGA program samples the real current at the same rate that current reference is p
88. rrent and voltage measuring electronics X4 3 Current and voltage measuring electronics X4 5 Current and voltage measuring electronics X4 4 Current and voltage measuring electronics X4 5 NC Screening box ground Current and voltage measuring electronics X1 5 NC Current and voltage measuring electronics X1 5 Current and voltage measuring electronics X1 5 NC Current and voltage measuring electronics X1 5 Calibrating electronics X1 1 NC Calibrating electronics X1 1 NC NC NC NC Current and voltage measuring electronics X1 4 NC Current and voltage measuring electronics X1 3 Current and voltage measuring electronics X1 1 02 NC NC DIO3 Current and voltage measuring electronics X1 2 DIO4 Calibrating electronics X1 3 NC NC DIOS Calibrating electronics X1 4 DIO6 NC NC NC DIO7 NC NI 9401 2 COM Level converter electronics X4 2 NC NC COM Level converter electronics X4 2 COM Level converter electronics X4 2 NC NC COM Level converter electronics X3 6 COM Level converter electronics X3 6 NC NC COM Level converter electronics X3 6 COM Level converter electronics X3 6 NC NC COM Level converter electronics X3 6 COM NC DIOO Level converter electronics X4 2 NC NC DIOI Level converter electronics X4 2 DIO2 Level converter electronics X4 2 NC NC DIO3 Level converter electronics X3 1 DIO4 Level converter electronics X3 5 NC NC DIOS Level converter electronics X3 4 DIO6 Level conve
89. rrent response for a triangular reference with a peak value of 200 A 61 Current Current 4 I l l I l l l D 0 0005 0 001 0 0015 0 002 0 0025 0 003 0 0055 Time 5 Graph 4 Current response for a triangular reference with a peak value of 100 A Current Current A l l I l l I I 0 0005 0 001 0 0015 0 002 0 0025 1 003 0 0035 Time 3 Graph 5 Current response for a triangular reference with a peak value of 50 A 62 l 0 004 l 0 004 I a 0045 l 0 0045 l 0 005 I 0 005 l 0 0055 l 0 0055 Current au Current A l l l l l l l l l l I Dn 0 0005 0 001 0 0015 0 002 0 0025 0 003 0 0035 0 004 0 0045 0 005 0 0055 Time 3 Graph 6 Current response for a triangular reference with a peak value of 20 A Current E ei l ZS nu l Current A l l I l l l l I l I l 0 0005 0 001 0 0015 0 002 0 0025 0 005 0 0035 0 004 0 0045 0 005 0 0055 Time 3 Graph 7 Current response for a triangular reference with a peak value of 10 A 63 Current Spam Real a ek Selm Current 4 na an I I l l I l l l l l l I D 0 0005 0 001 0 0015 0 002 0 0025 01 005 1 0035 0 004 0 0045 0 005 0 0055 Time s Graph 8 Current response for a triangular reference with a peak value of 5 A 64 To show the work of the high frequency protection a current shot of 300 A is used first with a correctly tuned controller and then with a controller where the tol
90. rter electronics X3 3 NC NC DIO7 Level converter electronics X3 2 Interconnections Level converter electronics X6 2 Shake to wake electronics X1 1 Level converter electronics X1 1 IGBT driver supply voltage Level converter electronics X1 2 IGBT driver ground Level converter electronics X1 3 IGBT driver S1 lower Level converter electronics X1 4 IGBT driver S1 upper Level converter electronics X1 5 IGBT driver S2 lower Level converter electronics X1 6 IGBT driver S2 upper IGBT driver error out Level converter electronics X1 7 93 Calibration electronics X3 1 Diagnostics electronics X1 1 Calibration electronics X3 2 Diagnostics electronics X1 2 Analog supply voltage 15 Diagnostics electronics X4 1 Analog supply voltage 15 Diagnostics electronics X4 2 Digital supply voltage 5 Diagnostics electronics X4 3 Digital supply voltage 4 15 Diagnostics electronics X4 4 Shake to wake electronics X2 1 Control panel Shake to wake electronics Shake to wake electronics X2 2 Control panel Shake to wake electronics 94 MATLAB code cRIO_result_files_getter m CRIO Eep li e b cd cRIO Surface results fileslist dir cRIO for n 1 length fileslist disp strcat num2str n fileslist n name end file 1 while file gt 0 file input Select file enter 0 to close if file gt 0 fileformat fileslist file
91. ry two of these parameters the current and the number of pulses or the current and the slope This will result in a number of test configurations To cover these configurations the test starts at a point on the magnet band The servo then moves the magnet band so that a new test point comes up and a second test is performed It is important that these test points do not interfere with each other This procedure is done over and over again until all the test points have been used Now it 1s time to scan the whole magnet band for the analysis This procedure can then be started over again to cover more test points Of course the magnet band needs a reset before each round with set of test pulses The test points are specified in the Surface sub tab and it is possible to specify up to 32 points These points should be points on the magnet band where the band 1s healthy It is also important that the different points behave in a similar way to the same test conditions since this contributes to the uncertainty of the measurement To check the condition of the magnet band the Magnet diagnostics sub tab is recommended to use To perform a test it is also important to state the number of positions to be used and the width that should be used to examine the result The scanned result is saved into a file for further analysis and that requires a file name The file name contains information about the test conditions and the word entered in the User string The pulse t
92. s In the settings for the pulse generator the pulse type amplitude and pulse time can be configured The pulse time is either entered directly or calculated from the values of the slope and amplitude controls It should be noted that this calculation is only valid for the triangle pulse type In the settings for the controller parameters there 1s a control for size of the tolerance band this value 1s used for the inner bands The value of the outer bands is given as a percentage of the inner bands When small currents are generated inductance 1s added in series with the flux providing coils for better performance This means that there are two identical sets of controls for the controller parameters In fact there is a third set of controls since the inductance changes when the inductor starts to saturate Depending on the peak current the system uses one of these three settings for the current controller Finally it 1s possible to plot the reference inner and outer bands in the same graph as the resulting current Last but not least is the Magnetization tab in the main interface It is used for magnetization and measurements of the resulting magnetization There is a sub tab system in the Magnetization tab to separate different magnetization related activities P Pulse magnetiser RT 2 0 vi on Pulsmag 2 0 lvproj cRIO Pulsmag File Edit View Project Operate Tools Window Help P Stop execution N Elapsed time Re mns T Serial commun
93. s a tool that could come in handy to detect 1f there 1s problem with noise in the analog electronics 57 4 4 9 Switch pattern analyzer Comparator error Comparator error Min ticks Comparators at erro Compo SI TB HF protection S1 Use both zero vectors 52 TB HF protection S2 H Analog reference levels Inc YD m an ye omparator ector type Digital input TT compa zm SE Tee Driver error A 4 Start playback Sequence Clear play FIFO player Outer band distance A Reference Tolerance band size A calculator Analog out Embedded Measurments Compensation values computer Compo Compl Comp2 Comp3 Reference levels Comparator cal start Comparator calibration Clear driver error Emergency stop omparator cal in progress ie hall san p en BESSER ubl om tput Take hall sample Analog input Disable drivers ha nable drivers outpu Enable test current f Digital output Test current direction 3 P Safety Embedded features o operational computer Voltages OK Comparator values OK Multi plexers Driver error Measurments Shake to wake Shake to wake Activate Emergency stop Block diagram 8 FPGA code In the block diagram above the Switch pattern analyzer block is added This is the final block in the FPGA code and is used to analyze the switch pattern of the IGBTs but also the desired switching instances This block
94. s it possible to set parameters etc to both the cRIO and the servo before powering up anything else The second switch turns on the power supplies to the peripheral electronics 1 e the 15 volts to the analog electronics and the 5 and 15 volts to the digital electronics including the drivers for the IGBT s This makes it possible to make the calibration of the analog electronics or troubleshoot in a safe way if needed It is also used to the power supply for the variable inductance and as a main switch for the cooling fan to the flux providing coils The third switch turns on the power supply to the servo Finally the fourth switch turns on the DC link voltage by applying 230 volts to the enable input on the inrush current limiter The order the different switches are listed in above is the order they are supposed to be started in to For safety they are coupled in a way which means that a higher order switch becomes ineffective if not all of the lower order switches are set to on The switches for the servo power and the DC link are furthermore only functioning when the machine 1s in Operational mode The condition for Operational 1s that none of the two emergency switches has been tripped One of them is located on the control panel and the other one is represented by the Shake to wake which will be explained later in this chapter Every power supply except for the DC link is equipped with a pilot lamp that indicates if it is supplied with
95. screen the rest of the left screen contains the main tab interface The right screen contains a tab interface controlling which graph to show and these tabs correspond to tabs in the main tab interface These three areas will be described separately in the following text It is natural to start describing the controls that always are within reach This part of the user interface contains a system for activating and deactivating the machine It contains an emergency stop button which disables the servo power and the DC link if they are turned on This is done by turning the shake to wake signal off Next to the emergency stop button is a button named Stop execution when pressed it will stop the execution of the RT and FPGA program if the DC link voltage 1s below 20 V The machine goes into the operational mode and a boolean indicator called Operational on the front panel turns green when the Activate button is pressed However there are some conditions for the Activate button to turn the machine into the operational state These conditions are no error from the IGBT drivers may occur the supply voltages has to be within limits and the compensation values for the comparators has to be okay This is indicated by a boolean indicator that is located above the Activate button When the conditions are not met the indicator 1s red and contains the text Check diagnostics and when the conditions are met the indicator is green and contains the text Ready Whe
96. sed workload of the FPGA will not decrease the performance It is also possible to make FPGA programs with very low latency The limitations of the FPGA is its size and thereby the size of the program that can be implemented on it Other disadvantages is that compared to the embedded computer the compile time for the FPGA is many times longer and the programs are thereby harder to debug This makes the development time for the FPGA programs longer To not reach the hardware recourse limit of the FPGA and have a to long development time it 1s essential to not use the FPGA unnecessarily The programs for the embedded computer and the FPGA are developed in National Instruments development environment LabVIEW Except for hardware restrictions the programs for the embedded computer and the FPGA is the same and the development tool takes care of the integration in the real time operating system on the embedded computer and with the FPGA The programmer programs a task for the embedded computer the FPGA or a combination of both This choice can later be changed and the code can be reused as long as it meets the hardware restrictions of the intended target Sometimes the choice between the embedded computer and the FPGA is easy and sometimes it 1s harder The FPGA is the link between the embedded computer and the I O modules and a custom FPGA program has been developed to provide communication between them A custom FPGA program provides maximum control of the d
97. set of the band m one direction and then a scan of the band This is repeated but with a reset in the other direction The result is then plotted in the Magnet diagnostics in the graph tabs It is possible to plot the raw data from the scans and or the difference between the two scans If the absolute value of the magnetization level is the same for the reset with positive current and negative current the difference will be zero To do a quick scan of a segment of the whole band the Scan band sub tab can be used The result is plotted in the Scan band tab in the graph tabs The data can be saved to a file using the controls in this sub tab E Pulse magnetiser RT 2 0 vi on Pulsmag 2 0 lvproj cR File Edit View Project Operate Tools Window Help I Stop execution f 3 Elapsed time NN e es 7 Serial communication D 5 FIFO communication n ZI m Resume sequence Operational 12V 55557 Deg mm Diagnostics Calibration of comparators Current control Magnetization Free run Hall sensor position calibration Magnet diagnostics Scan band Surfaces Band reset settings Current settings Pulse Slope hA s Final check Ireset E 250 Slope O Number of pulses v N CN EN d 15 P Reset resolution 7 a Delta slope Total number of test positions T H Y a Fr DU 551 Reset pulse time 7 e Sy Slope res Rounds to do LV ng 1 y SES Reset tolerance G 7 S 8 Slope max Cool off time b
98. show a rather interesting phenomenon which in a way explains the look on the surfaces at higher currents The widening of the pulse is continuing while flux density is decreasing at the location where it should be peaking A proposal to an explanation of why is that it is due to unexpected eddy currents in the air gap where the magnet is located This theory is supported by surface 13 where one can see that the extremely flat slopes which of course results in less eddy current does not suffer from the same peak flux reduction at high currents This is further supported by pulse 4 which is done with a small current slope and a big current Of course it will result in an extreme pulse width to 88 References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Joseph J Stupak Jr Methods magnetizing permanent magnets Oersted Technology corporation pp 1 8 October 2000 Mats Alak la and Per Karlsson Power Electronics Devices Converters Control and Applications Lund University Department of Industrial Electrical Engineering and Automation pp 33 106 National Instruments Corporation Operating instructions and specifications ComapctRIO NI cRIO 9022 pp 5 17 National Instruments Corp Operating instructions and specifications NI 9205 pp 1 235 National Instruments Corp Operating instructions and specific
99. sive vector 1 e Zero output voltage These signals are used by Use both zero vectors It translates them into signals to the bridge where Inc will be translated into 7 TB true and S2 TB false Dec will be translated into 7 TB false and 52 TB true and Zero will be translated into either 7 TB false and S2 TB false or SI TB true and S2 TB true depending which of them that were used the last time When passive vectors are used it changes the used passive vector every second time Since it 1s a tolerance band controller it can introduce undesirable high switching frequencies when it is not properly tuned Therefor there is a high switching frequency protection implemented the HF protection as seen in the upper right corner of block diagram 1 in the attachments There are two identical blocks for this purpose one for each leg in the bridge The only thing that differs is the 7 and the 5S2 in the names The desired leg position comes from Use both zero vectors and if it 1s okay to switch it is directly forwarded to 7 The protection uses a timer that starts counting when a switching occur to prevent that the same transistor 1s turned on again in a certain time This means for example that the transistor could be turned on and immediately off again but then the transistor can not be turned on again before the timer has counted to the desired limit that defines the highest switch frequency The same thing applies for next turn off Th
100. stration 3 Calibratine electronics The supply voltage is 5 V and the resulting resistance is around 1 ohm which gives a resulting current on 5 A The current runs through the current transducer 20 times which makes the control system think that the actual current is 100 A All switches are actually relays which can be maneuvered from the cRIO and to minimize the risk for disturbances there are opto couplers between the cRIO and the relay driver The upper most relay closes the coil which thus are disconnected during normal operation The two other relays make it possible to make the control system experience both 100 A and 100 A The current through the coil 1s measured as a voltage with the 0 1 ohm shunt by the cRIO with one of the channels on the NI 9205 module during the calibration The execution of the calibration will be described in the software chapter of this report 15 3 2 3 Voltage measuring The DC link voltage is measured only for monitoring To ensure galvanic isolation between the DC link and the control system a voltage transducer from LEM is used It is a LV 25 P SP2 and it is approved for measuring up to 4 1kV 9 Actually it is a current to current converter with galvanic insulation and a conversion ratio of 2500 1000 The resistors that are used are5 15 10 75 kOhm for the primary side and 3 100 300 Om for the secondary side mE 2500 1000 J This gives a total conversion ratio of 399 7s000 0 01V V
101. t hardware servo or software The Hardware tab shows information about the supply voltages in the control electronics These voltages are continuously controlled to be within the tolerance limits The nominal values and tolerance for the supply voltages are set 1n this sub tab Since resistors are dividing the voltages before they are read by the A D converter this sub tab contains controls for these scale factors Here is also information about eventual comparator and driver errors Finally here are settings for sampling of the hall sensors The hall sensors are continuously sampled just like the supply voltages But when they are used for taking a reading of a specific position of the magnet the A D digital converter goes to a special mode that takes an average on a number of samples This 28 number is controlled in this sub tab There are also settings for the sensitivity and a possibility to trim the offset error of the sensor Settings involving the servo and monitoring of the serial communication can be done in the Servo sub tab When the RT program wants to rotate the magnet band it orders the servo to go to the new position and after an appropriate amount of time it ensures that the servo has reached the new destination In the sub tab there are settings for the tolerance of check at the new destination diameter of the magnet band for conversion between degrees and position There 1s also a setting for a variable used to estimate the tim
102. t mode a contactor makes the current flow through the flux providing coils only In low current mode this contactor is opened and another contactor is closed which makes the current flow through an extra inductance of 15 8mH To improve the signal to noise ratio in the current measurement 9 extra turns around the current transducer is added in the same way when choosing low current mode this result in a new conversion ratio of 0 2A4 V Of course this 1s taken care of in software It should be mentioned that during the development of the current controller it was unknown that there was a need for good current quality at this low currents Hence these parts of the hardware are implemented in a more temporary way Calibrating electronics The current controller is partly an analog construction with all that it implies and as mentioned the comparators have a small hysteresis that must be taken care of The amplifiers and the current transducer are assumed to behave linearly but with an uncertainty in amplification and offset All of the comparators have a certain position when the current is within tolerance and the other position tells the control system when it is not This means that the comparator hysteresis can be treated as an offset as well 14 To be able to compensate for both offset and amplification uncertainties a test current is needed This 1s in principle achieved with voltage source and a resistor Illu
103. tPlotter m and can be found in the attachment MATLAB code 59 5 Results 5 1 Current control One of the objectives in this thesis was to develop a current controller for the flux providing coils In order to facilitate the tuning of the controller the current is sampled by the cRIO and plotted on the front panel during the test shots The same plot tool is used to show these results Several different currents were tested to show the performance The results will begin with tests at the rated current for the machine 315 A and then gradually decrease to 5 A The tests at the rated current will be shown for both a half sinusoidal pulse and a triangular pulse while the rest of the currents only will be shown for triangular pulses These results are shown in following eight eraphs Current Beam l Peal nau Current A l l l l I l I l l l l 0 0005 0 001 0 0015 0 002 0 0025 1 003 0 0035 0 004 0 0045 0 005 0 0055 Time Graph 1 Current response for a sinusoidal reference with a peak value of 315 A 60 Current a l Pez aM 3 130 Current A T eT I 140 l I l l l I l l l I l 0 0005 0 001 1 0015 0 002 0 0025 0 003 1 0035 0 004 0 0045 0 005 0 0055 Time 3 Graph 2 Current response for a triangular reference with a peak value of 315 A Current Current A l l I l l I l l l I 0 0005 0 001 0 0015 0 002 0 0025 0 003 0 0035 0 004 0 0045 0 005 0 0055 Time 3 Graph 3 Cu
104. tal output Block diagram 1 FPGA code 50 The logic of the current controller 1s seen in the block diagram above The block Digital input in the upper left corner provides the four signals from the comparison electronics to the logic of the current controller that is implemented in the FPGA The logic of the controller is implemented with the three subsequent blocks Comparator interpreter Vector type selector and Use both zero vectors The Comparator interpreter interprets the signals from the comparators and creates two new signals D and Zm I D tell the controller if it is supposed to increase the current or decrease the current It is not always certain if a passive voltage vector zero output voltage increases or decreases the current and since it 1s desirable to use passive voltage vector every second switch the controller needs to be informed about what the passive vector achieves This is done with the signal Zm This application is a special case since it is known if the pulse is positive or negative before it occurs Therefor there is an ideal value for the signals D and Zm to start at and they can be forced to change by setting the C I D or C Zm input high These two signals D and Zm 1s sent to the Vector type selector which only translate them into Inc Zero and Dec Inc corresponds to an active vector with a positive output voltage Dec corresponds to an active vector with a negative output voltage and Zero corresponds to pas
105. te with the servo and the time it takes for the servo to rotate The time of the communication differs from time to time depending on if the data is intact at arrival or not To turn rotate the magnet band 180 which 1s the worst case it takes barely 6 s To scan the whole magnet band with the highest resolution that the equipment allows it takes barely 30 min to perform the scan The highest resolution that the equipment allows is 0 25 mm and with that resolution there is more than 2000 points on the magnet band to scan This means that is takes less than s to move the magnet band 0 25 mm and to take a measurement with the hall sensors 86 6 6 1 6 1 1 6 1 2 Discussion and conclusions Current control The work of the controller As one can see the controller is keeping the current within tolerances when it is possible At graph 1 the current is falling behind in the beginning of the half period This is because the voltage is not high enough in comparison with the inductance to obtain a sufficient current slope the controller could not possibly do more For the triangular pulse of the same magnitude in graph 2 is the current slope on the way up no problem In Graphs 3 5 one can clearly see that on the way down zero output voltage is not enough to make the current fall in the desired way The current crosses the outer upper band and the controller changes negative output voltage to make the current fall faster Furthermore the
106. ting pulse end test Point nor 104 genvarname workname pulses delta number of pulses datanumber num2str round nbr datanumber num2str round nbr pulses delta number of pulses datanumber ITI gata num2str round nbr FLoure figure starti4 plot temp datai ER temp data ltsensor 3 xlabel Position mm ylabel Magnetization G center 220 axis center 15 center 15 300 500 105
107. tisfy the needs was to develop a structure that could be seen as a very simple form of operating system The first building block of the operating system is a FIFO queue called the task queue It contains elements that refer to a task to do so when there is a new task to do the corresponding element is enqueued in the task queue The second part of the operating system is a while loop that starts to dequeue one element from the task queue That element is then used to execute a piece of code that implements that task Each task is implemented in a case is a case structure and the case structure would then contains one case for each task The case structure is the third and last part of the operating system These three building blocks are shown in LabVIEW code below ts EL System Init ACA 1 Task queue T 2 While loop 3 Case structure LabVIEW code 1 The three main parts of the operating system in the embedded computer 35 Implementing a state machine This simple operating system can be used to implement a state machine Imagine a simple state machine with two states that enqueues each other like in the one below Start State machine 1 Example of state machine To implement the state machine the case structure in the operating system would contain two tasks one for each state The program in each task would simply enqueue the other task When this state machine is started 1t would continue forever Imagine that t
108. ts cRIO NI 9205 ATO All AD AIB Al4 Al5 AI6 AT AIS AI9 ATIO AIII ATI2 ATI3 All4 AII5 All6 AlI7 ATIS ATI9 AI20 ADI AD2 AD3 ADA AI25 AI26 AI27 ADS AI29 AI30 ABI COM DOO Diagnostics X1 1 Diagnostics X3 1 Diagnostics X3 2 Diagnostics X3 3 Diagnostics X3 4 NC NC NC Diagnostics X1 2 NC NC NC Hall sensor supply X2 1 Hall sensor 1 Hall sensor 2 Hall sensor 3 Hall sensor 4 NC NC NC NC NC NC NC NC NC NC NC Screening box ground variable inductance electronics Variable inductance electronics AISENSE Diagnostics circuit X5 1 91 PFIO NI9215 AIO AIO All All AD AR A AB NC COM NI 9263 AOO COM AOI COM AO2 COM AO3 COM NC COM NI 9401 1 COM NC COM COM NC COM COM NC COM COM NC COM COM DIOO NC DIOI DIO2 NC Current and voltage measuring electronics X3 1 Current and voltage measuring electronics X3 2 Current and voltage measuring electronics X3 3 Current and voltage measuring electronics X3 2 Current and voltage measuring electronics X3 2 Current and voltage measuring electronics X3 2 Current and voltage measuring electronics X3 2 Current and voltage measuring electronics X3 2 NC Screening box ground Current and voltage measuring electronics X4 2 Current and voltage measuring electronics X4 5 Current and voltage measuring electronics X4 1 Current and voltage measuring electronics X4 5 Cu
109. ules are used where two of them are the same The used modules are NI 9205 that 1s an analog input module with 32 channel and 16 bits resolution 4 The sampling rate is 250 kS s distributed over all used inputs It 1s also provided with a digital output In our case it is used for non time critical data sampling as hall sensor data and diagnostics of the machine NI 9215 that is an analog input module with four inputs and 16 bits resolution 5 The sampling rate is 250 kS s however this module is capable of sampling all inputs simultaneously at a sample rate of 100 kS s NI 9263 that 1s an analog output module with four output and 16 bits resolution 6 The sample rate is 333 kS s if one channel is used and 105 kS s if all four is used NI 9401 that is a TTL I O module with 8 channels 7 The channels can be configured as in or outputs in groups of four and the performance varies with the number of channels used Two of these modules are used 10 3 2 2 Interface The interface to the user consists of two parts the front panel on the host PC and the control panel located on the machine During normal operation the user only uses the front panel for programming and monitoring of the machine The control panel is there for both practical and safety reasons It makes it possible to shut parts of the machine down during troubleshooting for example and if something in the control system hangs up the user can safely stop the machine from
110. was okay or not The signal from the emergency stop on the front panel is also sent to the FPGA The task reads a signal from the FPGA that indicates if the machine is ready to be started changed into the operational state and a signal indicating if it 1s in operational state 4 4 TheFPGA program 4 4 1 Overall structure The FPGA program is as far as possible implemented in a logical block structure The thoughts behind the making of a program for the FPGA are fundamentally different from the ones used in the embedded computer in this project In the program for the embedded computer the simple operating system makes it obvious that the execution is sequential To get a program to work in the operating system the parallelism of the software would need to be adapted to that To utilize the parallelism of the FPGA the program needs to be truly parallel and this makes the code harder to understand straight away To describe the FPGA program and make it easier to understand it is explained with block diagrams The block diagrams are as far as possible made to correspond to the actual code The blocks are representing VIs and are all placed in a VI called Pulse magnetizer 1 0 vi This makes them run in parallel The signal lines between blocks are implemented with global variables 4 4 2 Current control logic Digital input Comparator EES Min ticks n protection S1 HF protection S2 nc Lise both zero vectors 52 TB Digi
111. when generatong the 4 plots Oo create the variable used positions eval usedpositions workname 1 32 1 2 Clear unused positions from usedpositions n 1 while n lt length usedpositions if usedpositions n 1 0 usedpos t nsin I else n mili end end usedpositions usedpositions 2 O create a vairiable containing the scanning resoulution eval scaningres workname 34 1 workname 33 1 nbr of index round 180 pi scaningres divides the loaded datafile inso smaller parts that contain one scan of the band divide index 35 n 0 eval datalength length workname while divide index lt datalength n n t 1 resetname genvarname workname reset numZstr n l dataname genvarname workname data num2str n datanumbername genvarname workname datanumber num2str n eval resetname j workname tidivide 1ndsx divide indgextunbr Gf press SS TL divide index divide index nbr of index 1 eval dataname f workname aLv de ingcex divide indextubr or index 1 7 7 eval datanumbername workname divide indextnbr of index 1 2 eval resetname Yt ene wden resetname Bye bi gqrwolog a one A eaund t st s eval resetname E421 wden resetname Vd E pad s C ele tretti beyt d t es iti t ys eval resetname E ta wden resetname Ets yola at nett d Teymi eval resetname EE
112. wo of these state machines are implemented and that they are independent of each other Now this operating system starts to be useful since it is easy to see that this could be expanded to more than two state machines It is also possible to see that this could be used to implement more advanced state machines In fact each state in a state machine could be rather advanced in itself When the programmer understands the structure of the operating system well it 1s a very powerful tool It will be easy to develop a state machine and change it during the development without the need to be concerned about the other state machines in the operating system This operating system has one drawback nevertheless When a task executes it executes and it will only stop when it is finished and hands the control to the operating system This implies that a task can not make a busy wait because that will stop all other tasks to execute A busy wait is seldom a good idea anyhow More important is that the execution time of a task can get so long that it will interfere with other tasks The programmer thereby needs to ensure that a task does not take too long time Of course it 1s possible to create a loop next to the operating system loop to deal with time critical tasks These will then use the RTOS on the embedded computer take care of the parallelism of the loops Luckily it is possible to enqueue elements to a queue in one loop and dequeue the elements in another parallel
113. ype is determined by the setting in the Current control sub tab but since it 1s a part of the file name it also needs to be put in a string in the Surface sub tab To ensure that the magnet band has been reseted properly a test 1s done to ensure that the magnetization level is within a certain tolerance entered as a reset setting Since the magnetization head and in particular the coils get hot when running large currents there is a fan to cool off the magnetization head This fan could potentially disturb the measurements Therefore there is a timer that starts after the reset and pauses the testing until the timer runs out This time is also entered as a reset setting Next of 1s the settings for the current Since the goal of these measurements is to investigate what happens at different currents there needs to be a way to get different currents The way it is done is by three controls The first is the starting current Jo A typical value for the starting current is zero The current is then increased in steps with amount of Delta I this 1s then repeated the number of times entered in res These three settings will result in a maximum current that is displayed in the indicator max The current is one of two parameters to vary The second is the number of pulses or the slope In the sub tab there 1s another tab interface with two tabs where each contains settings for the number of pulses or the slope Both of these sets of settings are similar to the s
114. ze small spots of the magnet band The studied parameters were the peak magnetization and the spread around the magnetization peak called magnetization width These parameters were studied as a function of different current magnitudes shapes and different pulse times There will also be some graphs on some of the accomplished magnetizing pulses 5 2 1 Reset results To make these tests possible the band was restored to fairly uniform magnetization every time all positions of the band were used The method of this restoration called reset in the program was found by trial and error and the result is shown in reset result below 69 009 055 00S I I I AN p J0suac 009 055 005 l I Zu z 1osuag OS 00t l as en OS 00 e ens 0SE I ose ww uorisod DU ww uorisog ODE osz i 052 U 002 002 OST i 007 SAW T 4auut UEIS p JOSUAS JEH jaw SZT uu URIS z 40suas eH L osz 55ne5 aprgyduy s55ne5 apnyydwy 009 oss oos AN 4JOSUaS 009 oss 005 IW 40suaG USt D k as RH Dk 00 e ens ww uorisod 0SE Ose UU I ww uormsog 00E l sz osz 002 002 U OST I OST I 00T OS 0 s55ne5 apnyyduy jaw g g 48jno ues c 10suas eH 001 0S 0 ssned apnyyduiy SAW SZT ano uens p JOSUAS EH Reset result 5 2 2 Surfaces For eac

Further development of a pulse magnetizer - IEA

Contents

Download Pdf Manuals

Related Search

Related Contents