Home

Circuit Analysis and Improvements ot the

Contents

1. Vswi Vsw2 Vv Vv PULSE O 110 PULSE 10 05 tran 0 1000 panini 107010 model VCSW SW Ron 0 001 Roff 10000Meg Vt 5 Vh 0 tran de 110 15 Figure 16 Schematic of the modified 28V regulator This is the circuit which was used for SPICE simulations in part 7 of the document The DC voltage source V1 at the top of the schematic mimics the transformer and rectifier bridge for DC sweeps i e simulations of the V I characteristics In this case the output current driven is presented at the right by a variable load I For AC transient simulation at power switch off the AC voltage source and the rectifier bridge shown in the diagram were connected instead of the DC source and the variable load I was replaced by a 51 85Q resistor corresponding to a 0 54A current both in series with voltage controlled switches VCSW in order to mimic toggling of the power switch by means of a pulse of the control voltages Vswl and Vsw2 of these switches The results of the SPICE simulations of the modified regulator are presented in the part 8 below A comparison to measurements results in my rig as well as to the simulation of the original circuit will be presented too 8 Measurement results on the modified PSU and comparison to simulation results The Figures 17 and 18 show the simulated V I characteristic of the modified circuit The figure 18 also includes results from measuremen
2. 2SC1815 BC639 max Collector Emitter Voltage Vcxo 50Vdc 80Vdc max Collector Base Voltage Vcg 60Vdc 80Vdc max Emitter Base Voltage VEg 5Vdc 5Vdc max Collector Current continuous Ic 150mAdc 500mAdc max Total Power Dissipation Pp T 25 C 400mW 625mW maximum Junction temperature Ty 125 C 150 C Case TO 92 TO 92 Table 4 Characteristics of the original transistor Q and the proposed replacement n a not available calculated from data sheet Modification of the power switch There is another design weakness left the last but one I found We saw in Figure 1 that the power switch provides the regulated 28V to the antenna tuner the LPF unit the meter lights and to the driver of the 21 7V pass transistor We also know that the current across the switch is 0 54A lights dimmed as long as we don t operate the antenna tuner or switch the filters The wires which connect the PSU to the switch and then distribute the switched voltage to these loads have an inductivity in the order of 1wH When toggling the switch we abruptly cut this current Consequently at the contact cut off by the switch a negative voltage spike will be induced according to Faraday s law V Le I t EBSAGV also used the same transistor BC639 in his earlier TS930S PSU repair notes 15 The inductivity calculated for a strait wire 50cm long 1mm in diameter is about 0 76uH 16 At the contact wired to the PSU point 28B before the mo
3. If regulation starts at load currents higher than 110mA we simply have to add an equivalent load permanently by means of a 259Q resistance from the output to ground Indeed this would work too But such a resistor would dissipate more than 3W of heat We would have less waste of power if instead to the PSU output we connect the permanent load between the bases of the pass transistors and ground We can do this because the pass transistors are current amplifiers The minimum current needed to achieve the same effect would for the same point of operation and with some simplifications of calculation only be 110mA divided by the current gain of the pass transistors This is what in fact we did with the resistor we calculated above 110mA 0 7mA 41kQ 259Q Just another way of thinking In order to find out for which value of resistance the potentially dangerous voltage at worst case conditions disappears I subsequently ran several simulations similar to Figure 9 I found out that this happens at about 23kQ Therefore in practice the resistor inserted should not exceed 22kQ For such a value even if D were dead the junctions of the pass transistors are at 125 C and your power network delivers at voltage 20 above the nominal value the output voltage of the PSU is limited to 29 3V at points 28A 28B By modifying the value of the resistor we can also reduce the influence of the non linear Vgg 1p characteristic of the pass transistors For a giv
4. 0 0004A N series with the AC load 0 0006A j eaea The transient was simulated for a load current of 0 54A corresponding to the current surged in RX mode which is represented by a 53Q resistor switched off at t 1s The 0 00004 L o 0002A supply voltage a rectified AC source was also switched off at t 1s The gray trace in Figure 5 shows the voltage 0 0004A across the two 22000uF capacitors Cg and Co following the rectifier bridge see Figure 1 From t 0s to 1s this j 0 00064 voltage shows the 50Hz ripple after the rectifier not resolved at this scale From t 1s on this voltage decays as the two capacitors start to discharge The black trace maaan shows the transient of the voltage at the DC lead to the Os 5s 10s 15s 20s 25s 30s PA which is not switched off Until t ls the PSU provides the regulated target voltage 0 0008A Figure 5 Transient after power on t 0s and off t 1s Left scale 28 5V supply to the PA unit black rectified e After switching of the load and for a time period of input voltage gray more than 15s the output voltage to the PA black Right scale Irn blue Ip7 red Ico turquoise curve raises above the target voltage compound base current to pass transistors pink The excess voltage transient in between t 1s and t 15s is consistent with my experimental observation and demonstrates that there is indeed a possibility to approach the dangerous Vceo l
5. 1 2W5 b7 sv 1 0w 37 0V 0 8W 36 5 0 6W 36 0 0 4Ww 35 5V 0 2W 35 0 0 0W pa ods ots obs 12s tbs 20s 28s 20s 325 3bs 40s OAN os 06s 12s nbs 24s ads abs Ads ajs Sus Figure 6 Output voltage transient following power Figure 7 Power dissipated in zener diode D1 of the switch off for different junction temperatures from PA unit following power switch off if D7 PSU 25 C black to 125 C pink in 25 C steps considering broken at different junction temperatures in 25 C steps a broken zener diode D7 Ip 0 from 25 C black to 125 C pink For the driver transistors there is still one the last protection left The zener diode D1 on the PA unit But this is a 35V zener the same voltage as the maximum rating for the MRF485 driver transistors safety margin zero And if D is already dead this diode will be exposed to the excess voltage transients each time you switch off your rig Figure 7 shows the power this diode has to dissipate It turns out that for almost 1s after switching off the rig the power rating of the diode is exceeded During some tens of ms a long time for a semiconductor the excess power is 40 above the rating and even higher when the pass transistors are hot Most semiconductors do not appreciate repetitive thermal stress This may affect the component sealing and the surface passivation of the chip or trigger the creation of dislocations in the crystal which at a longer term d
6. Location Reason Importance 1 1 1kQ resistor from collector of Q to PSU board Avoid unregulated operation at Mandatory ground low load 2 Replacement of the pass transistors by Heat Sink Increased Vcg margin Recommended MJ15003 Increased thermal margin 3 0 1Q emitter resistors for the pass Heat Sink VBg current voltage counter Mandatory transistors reaction Safe power handling 4 Connection of the anode of Dg to the PSU board Include R 4 into the regulation Optional output end of Ry4 loop eliminate the 0 5V drop of the PA supply at 10A 5 Additional 10Q resistor in parallel to Ry4 PSU board Safe operation in case of ICR Mandatory if mod 4 wire break is done 6 Connection of the switched 28V line Heat Sink Avoid meter lights flicker Optional small yellow wire to the output side of slightly with the PA current not applicable if mod Rus 4 is not done 7 Replacement of D by a 1N5364BG PSU board Higher Pp and T rating Recommended increased protection in case of another component failure 8 Replacement of Q by a BD708 PSU board Higher Icmax rating better Recommended protection against current spike at power on 9 Replacement of Q by a BC639 PSU board Higher Voz rating better Recommended protection against applied Vcr voltage at power on 10 lF 10nF in parallel to the power switch Power switch Avoid spikes in the switched Recommended 28V line line when switching off 11 15V zener diode from VR to ground PSU bo
7. e D is dead e The rig and in particular the pass transistors are hot and by error the rig is switched off under transmission at full power output e The rectified input voltage varies by 20 with respect to its nominal value due to an equivalent variation of the mains power voltage The justifications for these assumptions are as follows We want to know what happens if D7 our emergency break fails for any reason The thermal junction to case resistance of the 2N5885 pass transistors is 0jc 0 875 C W At full TX power output each transistor dissipates about 63W Consequently the junction temperature is about 55 C higher than the case Considering the temperature the case reaches in practice we can estimate that a junction temperature of 125 C is realistic In general we won t switch off the load via the power switch of the rig right after transmitting the pass transistors therefore would generally have the time to cool to lower temperatures before we switch off I agree but nevertheless this possibility exists e g by error you could switch off during an RTTY PSK31 or SSTV transmission Normally we should encounter a 20 excess voltage only if there is a substantial failure in the electrical power net but we don t have control on this In terms of robustness and security such a risk might it be unlikely should be mitigated Figure 9 shows the results of a simulation with D7 cut stepping the component temper
8. moving the point where the regulator senses voltage to the output side of R 4 Indeed R is still connected to the point 28B Why that Remember how the comparator Q works The emitter is provided with the full output variation whereas the base is delivered with only about a half of this variation page 2 of this paper We saw that consequently the base emitter voltage of Q varies out of phase with any output voltage variation and by about half of the amplitude This provides another opportunity for 17 Tn both there is a series 100uH coil at the 28V line as well as the 0 47uF capacitor C66 to ground in the LPF unit and the 10nF capacitor C36 in the AT unit 18 using a 1 10 probe and a trigger threshold superior to 2 8V AC coupled corresponding to the switched 28V 19 A real life transistor Qs won t be completely off due to leakage in off state Consequently there is a chance that the negative spike appears at the emitter of the 21 7V pass transistor Q on the heat sink too As the collector of the pass transistor is at a potential of 32V it requires a spike of only about 50V to exceed the maximum Vcro 80V this transistor 2SD843 is rated for Of course you could connect these capacitors from point 1 of connector 7 to ground in the PSU but this is curing the symptom not remediation of the cause the 5I t term 16 improvement If we set the base potential at a constant level we will have 100 of the output voltage variation
9. regrets I looked for an alternative solution to the VK4AMZ modification and therefore investigated in detail the operation of the original design as detailed now 2 General analysis of the 28 5 V regulator The schematic of the PSU and adjacent circuits is shown il an2 i 2ngees in Figure 1 ak The power supply unit delivers 28 5V to the PA antenna i tuner LPF unit meter bulbs and the 21 7 V pass transistor It also provides 21 7V used to generate other voltages for the rest of the rig The pass transistors Q and Q of the 28 5V supply are visible in the upper right part of figure 1 Q above the power switch is the pass transistor delivering 21 7V Q is the driver for the 21 7V pass transistor The voltage regulator for the 28 5V supply to the PA etc is visible in the upper part of the PSU unit constituted by Q Q and adjacent components The PA is permanently connected to the right end of the Soc Rare Ey 0 05Q resistor Ry4 the voltage drop across Rj4 is OR Oa epaizgemo ope ivo oo iures provided to the signal unit via connector 6 of the PSU for Eom Se iienaa o3 iute the display of the PA collector current ICA and JCB u ms ait The PSU voltage at point 28B before R44 is supplied to IEE SW UNITIX41 1410 000 1 the signal unit via JCA for display of VC at the meter Q3 Fe eee a to Qs serve f
10. stays below the zener voltage of Dy Therefore no significant current flows in this diode and consequently the base current to the pass transistors is controlled by Q1 only As a consequence a collector current flows if a load even weak is connected In other words the point of regulation onset was shifted from hoas 110mA see Figure 4 to Ipoaq 0 this completely suppressed the regimes A and B Looking at the low current region below 1A output current we see that the V I characteristic is bent in contrast to the range of almost constant slope observed at higher currents This is due to the base emitter characteristic of the pass transistors Effectively the difference between the gray trace potential at node C_Q and the black one voltage at points 28A 28B corresponds to Vgg if we neglect the voltage drop in resistors R4 and Rs We therefore can expect an influence of the temperature of the components in particular the junction temperature of the pass transistors A good design is one where external influences are eliminated within reasonable limits where component tolerances in particular of semiconductors non linear characteristics and temperature drift of active elements are compensated by the fact that passive linear elements govern the characteristics of the circuit To strengthen the robustness of the PSU I investigated what happens under worst case conditions I defined the worst case as follows
11. the circuit under an angle of view of potential danger If in practice this danger really exists may it be only by the unfortunate conjunction of particular circumstances is of secondary importance Up to you to decide if you adhere to this philosophy and which modification proposed you want to implement The easiest way for doing so is to unsolder Ds and to reconnect it at the copper side of the board together with the 10Q resistor and the little wire to point ICR 12 Due to the 1 3A bias current 28 435V 13 We will see in part 5 that in the signal unit there is a design error in the circuit for the measurement of the PA voltage VC position of the meter and also how this error can be cured 13 Replacement of D I profited from the fact that I took the board out of the rig to do some other preventive modifications First I replaced the zener diode D RD33FBD by another one a 1N5364 with higher rating of power and maximum junction temperature to provide better longer protection in case of another component failure The wires of the 1N5364 are of larger diameter but still fit in the holes of the board if you remove carefully the residual solder Table 2 summarizes relevant characteristics of both zener diodes Replacement of D RD33FBD 1N5364 max Total Power Dissipation Pp Ta 25 C 1W SW maximum Junction temperature Ty 175 C 200 C 30 35V min 31 35V min Zener Voltage Vz 31 97V max 34 65
12. the simple addition of a resistor from the collector of Q to ground This modification e eliminates the unregulated regime at low load currents e inhibits excess voltage transients at power off e provides protection to D in the PA unit e mitigates potential danger as it increases by one the number of failures necessary to come into a critical situation for the driver transistors in the PA unit The next part of the document deals with further means of improvement of the robustness and with opportunities to enhance the performances of the 28 5V regulator 4 Further robustness improvements Now that we have done the most essential modification protect the PA unit and other loads against overvoltage we can continue our analysis of the circuit from a standpoint of safety margin for the components We will also reveal some simple means to improve the performance of the regulator Replacement of the pass transistors recommended The original pass transistors Q and Q on the heat sink are 2N5885 types These are 60 V 25 A 200 W transistors Though I never had a problem with the original transistors I replaced these transistors by MJ15003 transistors ON Semiconductor formerly Motorola with substantial higher Vcgo 140 V by about identical maximum limits Ic 20A and Pp 250 W Table 1 summarizes relevant characteristics of both transisors Pass transitor replacement 2N5885 MJ15003 max Collector Emitter Vo
13. to connect to the mains AC power contacts a 10nF ceramic capacitor and a 1uF chemical one towards the PSU towards the load As these capacitors need to be charged at the moment we switch off they absorb current and therefore reduce the slope I t After addition of these capacitors my scope didn t trigger any longer Some people reporting broken 21 7V pass transistors might have been victims of the voltage induced by switching off the 0 54A current in the 28V line I didn t have a problem with this transistor but anyway it costs just two capacitors to feel better 5 Further improvements of the regulator performance The important most improvement of the differential resistance of the PSU was achieved by including the resistor R44 on the heat sink into the regulation loop see part 4 Also the proper choice of the resistor which inhibits unregulated operation also allowed improving the linearity of the differential resistance see part 3 Why do I care about a low and linear differential resistance A linear differential resistance should be aimed for the following reason As the small signal equivalent circuit of the PA includes this resistance non linearity of the source series resistance would lead to non linearity of the transfer function of the amplifier stages too Of course in the TS930S this will happen only to a little extent for there is the low pass filter constituted of L13 C23 and C34 at the supply lead in the P
14. to remove the bridge rectifier from the sink and the screws which fix the heat sink to the chassis Remove the screw of the rectifier but leave the wires connected Don t reassemble right away read the next two steps first Insertion of emitter resistors for the pass transistors mandatory Together with the replacement of the pass transistors I inserted 0 1 emitter resistors for both transistors It always strikes me why there are no emitter transistors in the original design of the PSU Power supplies with pass transistors definitely need them The characteristics of the actual transistors aren t ever exactly equal Thus connecting pass transistors in parallel inevitably results in unequal currents and hence different temperature rise As the current gain hpg of bipolar junction transistors increases with temperature the divergence of the currents in both transistors increases too This can lead to a situation where one transistor drives almost all of the current Fortunately one single transistor 2N5885 by its own can handle the around 10 A when transmitting at full HF output though it is driven hard with respect to the thermal power it has to dissipate The 2 2 Q base series resistors R4 and Rs on the PSU board effective reduce disequilibrium in base currents but they do not compensate inevitable hpg drift Emitter resistors in contrary provide a counter reaction by reducing Vgg with increasing Ip Emitter series resistors contribute to t
15. translated into Vpr still out of phase of course The easiest and largely sufficient means to do so is to connect a 15V 0 5W zener diode between the hot end of VR cathode and ground anode The resistor R would now play the role of the series resistor which fixes the operation point of the zener diode The potential at the base of Q is now despite a little positive influence we will see a little bit later independent of the output voltage it is defined by the zener voltage and the divider consisting of VR and Rg This indeed increases the gain of the error amplifier by a factor of 2 which also means that the differential resistance of regulator is divided by 2 The range for voltage setting by means of VR still is right For ease of adjustment I profited from my modifications to replace the original trimmer resistor by a 500Q linear multiple turn Cermet trimmer As we achieve an almost constant base potential by means of the zener diode it simply doesn t matter in this context if we pick up the voltage before Rj or after it Nevertheless I spoke about a little positive influence on the base potential of Q2 What about this Saying that the 15V zener keeps the voltage at the hot end of VR constant is true in a first order approach It would be the case for an ideal zener diode but not for a real one where the voltage drop varies very little with the current and thus is also toa very little extent a fun
16. two 22000uF capacitors become discharged output voltage non regulated As said before here too we should neither stick to the precise values of the currents nor precisely to the time points calculated the tolerances of the real component characteristics in the rig will influence the speed by which the PSU undergoes the different regimes So far we could say what does it matter the voltage at the output approaches but never reaches the critical limit of 35V for the driver transistors But what if D were dead for some reason be it aging Remind that at low output current the base potential of the pass transistors and therefore the output voltage of the PSU is limited to about 32 5 V only by means of the zener current which counterbalances the current of Ry Simulation without D indeed shows that in this case the peak of output voltage reaches 38 4 V This is above the 35V Vceo rating of the driver transistors MRF4835 already at room temperature and the situation is even worse if we consider heating of the pass transistor junctions and hence higher current gain as can be seen in Figure 6 At 25 C the value calculated is 38 4V At higher junction temperatures the junctions of the pass transistors can reach up to 125 C the peak voltage exceeds 39 7V V output_to_pa Output_to_PA I D1_PA Unit 40 5V 2 2W 40 0V 2 0W ba sv 1 8w4 39 0 1 6W5 38 5 1 4W 38 0
17. 2A and 10A cursor positions The slope approx 60mV A gives the differential resistance 60mQ of the original PSU in this range The differential resistance will be discussed in the forth part when we speak about improvements of the regulator performance The collector voltage red trace of Q i e the signal from the error amplifier of the regulator shows the same overvoltage behavior for low load currents Beyond about 110mA of load current the collector potential of Q increases as expected such as the base current of the pass transistors does light green trace right scale Let us have a closer look At the node C_Q the collector of Q1 the sum of currents must be equal to 0 Iga and Ips are the base currents in the pass transistors and are considered equal for the purpose of this analysis By convention currents are counted positively towards the node currents effectively flowing away from the node therefore have a negative value Consequently according to Kirchhoffs law we have Tc q1 Irz Ip 2 Tr 0 jo REA ii ae The situation at node C_Q is shown in Figure 4 a ie simulation zoomed at the strange region of low loads 33 2V 32 8V gt LT SPICE defines the directions of the currents according to its own rules At selection of the currents Ip7 and Ips were effectively counted towards the node but have a negative value For Irz and Ic 9 LT SPICE determines the values of currents flowing awa
18. A unit The capacitors will buffer voltage variations supplying extra current when the PA surges a high current and therefore virtually reduce the non linear behavior of the source Nevertheless it is always a good choice to remedy to the cause rather than using filters to limit its impact in particular when this can be done by simple means A low differential resistance keeps the collector emitter voltage constant Of course the linearity of the RF signal of the driver and the PA is mostly dependent on a stable bias current The supply voltage also is set to define together with the load impedance a point of operation where Ic Vcg of the RF transistors is linear If the voltage becomes a function of the current not only because of the Ic V cpg characteristics of the transistors but also because of the internal differential resistance of the power supply non linearity can increase Again the low pass filters as mentioned above will reduce this effect but we can help by limiting voltage variations through a low differential resistance of the power supply The following modification is certainly not an essential one I never had complaints on the bands concerning spectral widening due to non linearity and if using a good microphone TS930S owners always get excellent modulation reports So you might accuse me tending to perfectionism and I admit you aren t completely wrong Nevertheless you noticed in part 4 that we only spoke about Dg when
19. Circuit Analysis and Improvements of the TS930S Power Supply Unit Switching off your rig can kill the driver transistors F5VIM DF7NT Klaus SCHOHE December 2011 1 Introduction In this document I present an analysis of the TS930S power supply unit PSU and propose some modifications efficient and easy to implement which strengthen the robustness of the 28 5V power supply to the PA The PSU as we know is the perhaps only weak point of this great radio which still is a reference in the high end class of HF transceivers A lot of posts on internet report problems with the original PSU sometimes resulting in killed driver transistors In my rig bought a couple of years ago from a fellow ham I too observed some intermittent failures sometimes exhibiting excess voltage which fortunately didn t lead to destruction of the driver transistors expensive and difficult to find In order to cure the failure and eliminate the intrinsic weaknesses of the original design I investigated the PSU using traditional circuit analysis SPICE simulation and measurements on my rig Along with this analysis presented step by step simple modifications will be proposed which enhance the performance characteristics of the PSU In what follows the important most point is the one which reveals an intrinsically design related regulation failure and exposes related possible failure mechanism This is the subject of the rather intriguing statement I added to t
20. V Iz 10mA Iz 40mA Table 2 Characteristics of the original zener diode D and the proposed replacement In normal operation the slightly higher zener voltage is of minor importance the reverse current can be neglected for now we normally are far below the zener voltage due to proper regulation at all load conditions Remplacement of Q I examined the voltages currents and power dissipation of Q at different load conditions None of the limit ratings was exceeded or insufficient safety margin was detected However transient simulation showed an important current pulse at power on of the rig Figure 12 shows a simulation of the power on transient of the output voltage black trace and of the collector current of Q red trace This reveals that about 78ms after switching on the rig the collector current undergoes a spike of almost 1A zy war wet 12h Though the simulated spike is very short and normally 4 should not lead to destruction the maximum allowed 24ay4 i ee H 11A i at continuous current of the original transistor is two times Srp E E E L I E Canas SE E DEE ETT bigger I replaced the original transistor by a BD708 transistor ST Mircoelectronics for the following 2 p om reasons 18V A 0 8A i First The precision of transient simulation more than 16v j oza other modeling is subject to component tolerances and parasitic elements difficult to include in the model The current spike in fact i
21. _Igo9 is approx 0 7mA see cursor e andthe node potential at this point is about 29 1V Care should be taken interpreting figures from this diagram First of all as indicated above some components differ from the real components used And even if we have a model for the component in the circuit characteristics of real components are subject to dispersion within tolerance ranges in particular the current gain hpg of the transistors Consequently the steep linear range Regime B may extend to currents higher than expected from the simulation shown in Figures 3 and 4 Secondly as we know the current gain of bipolar transistors increases with temperature and consequently the threshold for regulation onset shifts to higher output currents If the point of operation is close to the limit between Regimes B and C overvoltage may occur at higher temperatures while it doesn t at lower ones This I guess is the reason for my observation during testing with dummy DC loads that excess voltage occurred occasionally after warm up of the PSU while for the same low current there was no overvoltage when it was cold Before we come to the remedy let us investigate my other very strange observation of an overvoltage transient right after switching off the rig Let us also think about the potential danger and failure modes triggered by this non regulated regime at low load conditions e This will show that unlucky but realistic circumstances ma
22. ard Increase of the comparator gain Optional by 2 reduction the differential resistance 12 Readjustment of VR to 28V PSU board Because the operation point if Mandatory still adjust to 28 5V if mod 4 is not done the regulator slightly modified 13 Readjustment of IC meter 0 set point Signal Unit The additional current 0 5A Mandatory if mod 6 check of IC meter calibration in Ry4 modifies OA point of the is done in all cases PA current meter display recommended 14 Connection of cathode D77 in the signal Signal Unit Correction for true VC Recommended unit to the ICB potential measurement Connector 6 pes aa Figure 15 Photography of the modified PSU board The modifications 1 4 5 and are shown on the photography of Figure 15 The three 1 1kQ resistors modification 1 are visible at the upper part of the picture Below these resistors the modifications 4 and 5 are visible Note that the diode Dg is now at the solder side of the board with the cathode soldered at its original point of connection The anode is now connected to the plot ICR of connector 6 The 10Q resistor connects in between the anode of Dg the copper plane of point 28B i e the original point of connection of the anode The 10Q resistor and the anode of the diode as well as the wire to point ICR aren t soldered anywhere on the board they are soldered together in the air The diameter of the diode ensures th
23. at this point cannot touch anything The 15V 0 5W zener diode VR to ground according to modification 11 is visible in the upper left part of the photography 19 The schematic of the modified 28V regulator of the PSU with the adjacent components on the heat sink are shown in Figure 16 The components indexed h are located on the heat sink 1 e the pass transistors the emitter resistors added mod 3 and the resistor R44 for the PA current measurement The components Rg 1 1kQ Rg 10Ohm and D 15V zener are the other components added This is the diagram used for simulation As I didn t dispose of the SPICE models for BD708 1N5364 and 1N1555 other component models where used For these components the designations of the components used in reality are indicated in blue According to the data sheet the varistor SV 03Y is modeled by a constant voltage source of 2 1V with an internal resistance of 16Q vi ser 0 74 39 837 Vv Zs A et a c4 220004 22000 p k C J 1 0 01 h bt 3x 3k3 0 5 W parallel KH Rs1 D im T 1K1 D8 Varistor ne FLZ33 1N5364 we Rser 16 D7 21 MJ15003 N F 1N1555 D6 c5 c3 ce H at tt Th TEzo0 001 1N4148 0 014 u ka 02 SINE O 42500000 B639 vcsw a BZX84C15L je ZXB4C12L MTZ12JC F Oc si s hd a 1k2
24. ature used in the calculation in the range 25 C 125 C and sweeping and the input voltage from 32V to 48V VNominai 20 As can be seen in Figure 9 there is an almost 0 5V increase in output voltage if the junction of the pass transistors heat up from 25 C to 100 C In the case where the junction temperature of the pass transistors is set to 125 C the output voltage exceeds the target voltage already at the nominal rectified AC voltage and reaches a critical level if for some reason the mains power rises by 20 corresponding to 48V after the rectifier Consequently cutting R22 out is not a satisfying solution and we have to search for an alternative means to eliminate the unregulated range of operation at low output currents Let us take an analytical approach In Figure 4 we can see that the point of regulation onset is characterized by a current Ig of approximately 0 7mA denoted by the cursor and that the node potential there is about 29 1V Without surprise the value of this potential equals the output voltage 28 5V plus Vgg of the pass transistors At no output current of the PSU the base current of the pass transistors will almost be equal to zero too Under this conditions if we manage to obtain a node potential of 29 1V the current in Roz of 0 7mA must be deviated somehow Somehow because at this potential there is no possibility of a significant current in the zener diode D7 also Q is at the edge but still in the
25. be reduced to 2 7kQ in order to return to a positive differential resistance But here I didn t do further fine tuning I kept the original 3 3kQ resistor R and pragmatically went to real life measurements It turned out that despite that I didn t lower the value of R the output voltage still dropped see part 6 by 15mV when driving the current from 0A to 10A instead of rising I attribute the discrepancy between reality and simulation to the fact that most passive components in particular resistors are described by idealized ones that I didn t have the model for all active components effectively used that all components do not heat the same way the pass transitors will at least be warm if not frankly hot where as Q and the zeners won t heat up significantly by themselves and stay close to the temperature in the rig while I didn t make attempts to attribute individual temperatures to each component during simulation that other parasitic elements exist such as resistance of the leads which I didn t include Modeling always has limitations even if there is a model for each component used if we go beyond this limits we are speculating not simulating Up to you to decide based on the measurements in your rig if you modify R with in order to minimize further the drop of the voltage supplied to the PA unit If you fine tune R7 you also have to readjust VR We might consider also use of a fixed voltage regulat
26. c see the curser window in Figure 3 indicates a differential resistance of about 60mQ within the regulated range This is the equivalent internal resistance of the voltage source which represents the regulator In figure 11 we saw that a differential resistance of 5 7mQ 8 1mQ is achieved at point 28A 28B and after the modification to cure the overvoltage issue We can profit from this low differential resistance if we move the point where the voltage is sensed from the point 28B after the fuse to the lead of the resistor R 4 where the PA unit is connected This includes the resistor used for the measurement and display of the PA current in the regulation loop and therefore eliminates the voltage drop of OV 0 5V for 0A 10A current supplied to the PA Logically the differential resistance is reduced by the value of this resistor i e 50 mQ The easiest way to implement this modification is to connect the anode of Dg to the plot ICR of connector 6 For this purpose you could unsolder the anode leg and run a short wire to this plot The regulator would then be supplied with the voltage at the other end of the resistor via the cable connected connector 6 However before doing so consider following Imagine that for some reason the wire ICR is broken or you simply forgot to reconnect the connector 6 after some maintenance In this case no voltage is sensed and the regulator provides maximum current to the bases of the pass transi
27. circuit to start providing polarization to the pass transistors We will see later that indeed Rz has an influence but definitely not the way we want This resistor without other modification might even cause the destruction of the driver transistors in point 3 we will analyze this potential dangerous effect in detail As far as start up of the circuit is concerned again I didn t see any difference in simulation with SPICE both with and without Ry connected What about Dy At a first glance we might attribute to this 33V zener diode a exclusively protective role i e a security function which consists in limiting the base potential of the pass transistors in case of a component break down in the error amplifier e g a Q short And this is what I first thought too But investigating further it turned out that even without any failure of whatever component D may not prevents us from disaster We will see again in point 3 what this is due to and how we can give D it s exclusive role of an emergency break as it should be The base series resistors of the pass transistors R4 and R partially compensate disequilibrium in current and power distribution between both pass transistors We ll see in the fourth part of this article that we can do better Some of you may already see a design weakness here but this is not crucial for the security of operation We will come back to this observation later when we speak about improvements
28. ction of the voltage at point 28B where R is connected to An increase at this point leads to a little rise of the base potential and if compared to a constant potential at the emitter of Q this increases the collector current of the transistor Such an increase happens if the voltage between point 28B and the connection point of D the output of the regulator rises due to the output current flowing in R 4 Remember it is now the point after the resistor Rj which has a constant voltage In other words by connecting R before and Dg after the resistor Rj4 we introduce a small component of current voltage reaction We could even push this effect to a situation where the voltage rises together with the current i e we would have a negative differential resistance Of course this is not our aim negative differential resistances can lead to instabilities and we even use them to built oscillator in circuits we call negative impedance converters This effect evidently is governed by the operation point of the zener diode we added from R to ground which means by choosing the value of R we can enhance or reduce the effect of further reduction of the differential resistance The Figure 17 shows that indeed with the original value of 3 3kQ of R we should expect a slightly negative differential resistance because the output voltage increases by about 32mV if we vary the output current from 0A to 10A I found by simulation that R needed to
29. dification the spike is positive This spike is buffered by C6 and C7 of the PSU 15 We see that due to the second term of the equation virtually there is no upper limit for the amplitude of this voltage but oxidized contacts will help This spike won t harm the LPF and the AT units The meter lights and the relay for the remote connector to an external PA won t care either However the spike may be delivered to Qe in the PSU the driver of the 21 7V pass transistor The capacitor C on the PSU unit 22uF won t be of much help for a short spike because of the parasitic series inductivity of such a chemical capacitor and the series resistor for stabilization of the base potential of Q6 R23 470Q further reduces the capability of C 3 to buffer the spike With a scope I indirectly observed the presence of a very short spike at the switch contacts It triggered the scan of the beam but I couldn t measure the amplitude and duration The measurement of a sharp single spike requires a probe and a memory scope with high bandwidths If we don t see a spike it may be due simply to limited measurement capabilities we cannot conclude with certainty to the absence of such a spike In similar cases we would find means to switch softly in order to prevent an induced voltage which eventually might harm but this obviously was omitted in the original design e I added two capacitors in parallel to the contacts of the switch be careful not
30. e current in the zener diode D red and in Rz blue only In both ranges overvoltage occurs The absence of collector current of Q for Tjoaa lt 110mA provides evidence that the error amplifier doesn t control anything in this range of operation e To my knowledge this intrinsic effect due to the design of unregulated operation of the TS930S PSU wasn t reported yet and published investigations most often suspect observed excess voltage to be due to some potential or real component failure The three regimes of operation can be characterized as follows Regime A Iz controlled In the range up to 60mA the current in the zener diode varies with the potential of the node while the current in resistor Ry is almost constant The current in the pass transistors and the output voltage essentially follow the I V characteristic of the zener diode Dy Regime B Irz controlled In the range 60mA 110mA the potential of the node about 0 5 V higher than the output voltage is below the knee of the zener I V charactersitic the current in Ry varies while the current in the zener diminishes Only R32 governs the V I characteristic of the PSU Therefore in this range the V I characteristic is linear Regime C I g controlled regulated operation The threshold of the regulated regime i e the onset of Ic q occurs at a load current of approx 110mA while the current in Rz becomes constant At the point of onset of the regulation e
31. e of the ae A E12 series lower than the calculated value The 29 0v L 100mA characteristics in Figures 8 and 10 are almost identical a excepted that in Figure 10 a small current in Raz is plotted P8 sv a p 8Oma which of course doesn t exists when we cut this resistor ay out Figure 8 28 8 5 H 60mA Lara ee Lic Irz2 not revolved due to the scale in Figure 10 is equal to the voltage drop across the resistor divided by the eee resistance it varies according to the node potential from 730pA at Iroa OA to 704A at Ipoag 12A The current Oma Ir22 is compensated by the current flowing to ground i e away from the node in the resistor we added Remember this is how we calculated the value of the Bs See resistor 28 3 5 40mA 28 2 5 Dh H 60mA bev BN ae L 80ma Figure 10 Simulation with R in place and 39kQ H 20mA resistor from the node C_Q to ground pa ov lt j Left scale Voltage at points 28A 28B black deliverd to egal gee the PA unit green and potential of the node C_Q grey i Right scale currents at the node Irz blue Ip red Ic q Siia e o a a a a a turquoise and compound base current to the pass OA 1A 2A 3A 4A 5A 6A TA 8A 9A 110A 11A 12A transitors pink Figure 10 proves that our approach insertion of a resistor from the collector of Q to ground remediates the design error Another way of thinking to remediate the problem of the non regulated regime is to say
32. e rhythm of my CW code so I moved this wire to the same point as the heavy red lead of the PA unit i e to the output side of Ryy This has do be done when the heat still sink isn t mounted back into the rig As additional current the 0 54A provided to the antenna tuner the LPF unit the 21 7V pass transistors and the meter bulbs now flows across the current sense resistor a readjustment of item 4 in the adjustment procedure of the service manual IC meter 0 point is necessary The adjustment range of VR g in the signal unit is sufficient to compensate this extra current Check item 6 IC meter of the TX adjustment procedure page 70 in the service manual too though the Ic meter should give the correct PA current right away after readjustment of the IC meter 0 point You may have noticed that we only moved D to the other side of R 4 not the point where R is connected This is not an omission we will speak about this in part 5 Cold solder points replacement of D Q and Q modification of the power switch Working on the PSU I finally noticed that some of the intermittent failure was due to a cold solder point the R1 lead towards connector 3 I found some other suspect solder points obviously without consequences which I reheated for security Have a look too we know that for example the digital unit also has a reputation to sometimes suffer from bad soldering In the discussion which follows you will see that I analyzed
33. egrade the excess carrier profile of the pn junction It might only be a question of time until this ultimate protective means also fails Once D in the PA unit is killed too the next time you run the rig may be the last time for the driver transitors When you switch off again they might get shot Of course with what I am saying here I do not consider that a the problems reported with broken driver transistors are due to these excess voltage transients there are multiple other reasons which may be the cause But if besides the MRF485 you don t find other broken components except D7 PSU and D PA unit this cause route is a good candidate Anyway once a potential risk is identified our aim must be to eliminate the cause and mitigate the risk This is the subject of what follows The remedy How can we prevent the excess voltage transient effect how can we eliminate the unregulated regime at low output current As we saw the overvoltage is due to the current supplied by R33 at no or low load As this resistor isn t necessary for the PSU to start we could simply cut this guilty component out And in fact it would work under normal operation conditions but not under worst case conditions At room temperature and at nominal AC mains power voltage a simulation without R22 indeed shows that the voltage at no load stays limited to 28 7V as shown in Figure 8 My advice If you observe high voltage with the VC meter or if yo
34. egulator To obtain a ImV resolution for the measurement with a digital voltmeter the voltage was measured against a stabilized power supply of 28V considered to be stable during the short period of time necessary for the measurement 7 I didn t attempt to refine this model in order to simulate the V I characteristics more precisely for load currents exceeding 10A for this is of minor importance in my situation I am running my rig on a non resonant loop remotely tuned at the antenna feed point This leads to a negligible load reflexion coefficient at the PA output on all bands Consequently I do not encounter PA currents superior to 9 4A 9 Conclusion As presented in part 3 the original design of the TS930S power supply unit bears a potentially dangerous design error At low load conditions the output voltage exceeds the nominal 28 5V This is due to a non regulated regime which occurs for output currents lower than about 100mA Each time the rig is switched off the PA unit is exposed to excess voltage because during the time the two 22000uF capacitors following the rectifier are still energized the PSU is in this non regulated regime Under unfortunate circumstances voltage in excess to the rated maximum collector emitter voltage can be applied to the driver transistors This might explain some failures of the TS930S driver transistors reported A simple modification is proposed to overcome this design failure and completely eliminate the vo
35. ematic are real life SPICE models i e including parasitic elements series inductance parallel conductance The simulation includes the 4 7kQ resistor across Cs and Co in the rig but not shown in the TS930S main schematic Figure 1 The load is simulated by a variable load I varied from 0 to 12A in ImA steps as indicated by the SPICE directive The source in reality a rectified AC source is modeled with a constant DC source because LT SPICE doesn t allow time dependant sources i e AC in DC sweeps The compound current of the two pass transistors is calculated simply by 2 x Ips The compound base current of the pass transistors has a negative value for by convention of the current node analysis we will perform later the current is counted positively in the sense flowing toward the node As can be seen from the black and blue curves which respectively show the voltage current characteristics at 28A 28B and after the 0 05Q resistor there is indeed a range up to a load current of some hundreds of mA where the voltage exceeds the 28 5V target voltage e This simulation confirms that by design there is a threshold of load currents below which the regulation doesn t work Beyond this point as expected both voltages drop as the current increases the blue curve i e the voltage the PA sees showing the additional voltage drop in the 0 05Q resistor The popup window in Figure 3 indicates the voltages supplied to the PA unit at the
36. emperature compensation first because of the electrical counter reaction secondly also because their resistivity increases with temperature For this reason they should follow the temperature increase of the transistors they should be bolt on the heat sink as close as possible to the respective transistor As a rule of thumb emitter series resistors should drop about 0 5V at the maximum current of the pass transistors Consequently R 0 5V 5A 0 1Q At steady state and full power TX the resistors dissipate a power of 2 5 W each e I used 35 W TO 220 package resistors MHP350R100F from BI Technologies bolt on the heat sink To bolt these resistors no need to drill additional holes in the heat sink The pass transistors are flanged with 4 screws The two screws at mid height of the heat sink provide the collector connection for the transistors The two other screws the upper and lower most ones on the heat sink can be used to fix the emitter resistors To do this you need to move the plastic insulator originally underneath the washer and nut i e on the inner side of the heat sink to the other side under the head of the screws This electrically isolates the screws from the collectors The flanges of the resistors are isolated from the resistor Consequently you can bold bolt the resistors directly onto the heat sink without an insulator with respect to the screws and without mika or teflon sheets But of course put some heat conducting co
37. en node potential set by VR a further increase of this current by reducing the value of the resistor from the node to ground cannot be equilibrated by the current of Ry The error amplifier keeps the node potential constant by injecting additional current into the node In other words by reducing the value of the resistor from the node to ground we also reduce the influence the Vpp Ip characteristic of the pass transistors We therefore can expect an improvement of linearity of the differential resistance by choosing an appropriate value e In practice for good linearity without exceeding reasonable additional dissipated power I used a resistance of 1 1kQ I soldered 3 resistors of 3 3kQ 1W each in parallel from the node at collector of Q to ground The power dissipated in these resistors is about 740mW Further reduction of the resistance would still enhance linearity if the V I characteristic of the PSU because it increases the base current of the pass transistors but we must consider power dissipation in the resistor added The result of my choice is given in Figure 11 i et a ae apea Figure 11 demonstrates that a collector current of Q now oan also flows at no load at all differently to the case of the baa ee net H sitan original design Figure 4 the case were R22 is cut m Figure 8 and the case with R22 kept in place and a po oy oee L 180mal 39kQ resistor connected to ground Figure 10 ko sv f 150mA Due to the prese
38. he title of this document Before going into detail I want to say a few words on two of the other modifications published David s modification W6NL mainly is not related to the 28 5V supply Nevertheless for me implementing David s modification is an absolute must in order to get rid of the ugly original stabilization of 7 5V and 15V voltages by means of series resistors and zener diodes in the fan cabinet Some years ago I implemented David s improvement and am very happy with it Implement David s modification if not done yet you won t regret Concerning the 28 5V supply the modified PSU circuit proposed by Michael VK4AMZ without any doubt would have been my choice for it s elegance and intrinsic reliability It provides state of the art regulation together with overcurrent and overvoltage survey associated to protecting thyristors However this board probably intended to early TS930 transceivers doesn t provide the 21 7V regulator which Kenwood implemented in later series My transceiver 5M serial would need an additional board if replacing the original PSU unit by Michael s one in order to pilot the pass transistor of the 21 7V regulator There is not a lot of space around the PSU and the transformer and it would have been difficult to associate an additional board replicating the 21 7V regulation Qs Dio C13 R23 C12 and in particular the space consuming chemical capacitor C3 and the 6 pin connector 7 of the original PSU board With
39. his point went up to again about 33V for a couple of seconds after switching off the rig First I suspected a component failure but whatever point I measured everything seemed to be ok in steady state I therefore simulated the circuit with LT Spice in order to check the operating points of the components more precisely against the corresponding values calculated by simulation and also in order to verify if perhaps a component was driven to close to its rating suspecting that perhaps some overheating could lead to the failure I observed Again everything was ok so far if at least about 1A of load current flew at the output However at low load or no output current at all the calculated operation points confirmed my experimental observation of excess voltage at the PSU output The Figures 2 and 3 show the simulated circuit and I V characteristics of the PSU simulated with SPICE for the net labels indicated in red on the schematic V 28a_28b Wfoutput_to pal Vic_at 2I R5 33 5 OmA Copie Article TS930PSU Original raw X Cursor 1 33 0 M 10mA Vioutput_to_pa Horz 24 Vert 28 4444V Cursor 2 32 5 i Vfoutput_to_pa M 20mA Horz 104 Vert 27 9649V Diff Cursor2 Cursor 2 0 i Hore 84 Vert 479 465mV 30mA i Slope 0 0599332 f B1 5 40mA 5 ah aoe fanss rn 31 0v 4 L 50mA va TR E Eata Tu am ti
40. imit of the driver transistors of 35 V when switching off the rig Between t 11s to 15s the output voltage decays towards the target voltage From t 15s to 17s the output voltage again is equal to the regulated one and after t 17s finally decays according to the voltage transient of the two capacitors Considering the currents flowing to the node at the collector of Q as in the preceding simulation we can see that effectively the PSU again undergoes all three regimes discussed above before decaying to zero zls Regime C regulated Iz p7 is almost zero the error amplifier delivers a current Ic g of about 5mA out of scale in Figure 5 Ir22 corresponds to the potential difference of the node and the input voltage ls 11s Regime A non regulated the collector current of Q is close to zero zener current in D limiting the base potential of the pass transistors the current in R decays according to the discharge of the 22000uF capacitors lls 15s Regime B non regulated the collector current of Q is close to zero no zener current because the potential of the node is now below the zener voltage the base current of the pass transistors is supplied via Ra only 15s 17s Regime C regulated again by the collector current of Q the current in Ry now corresponds to the new potential difference between the node and the residual voltage at the input which continues to decay all other currents are insignificant 17s Decay to zero as the
41. lbag OA 12A Left scale Vou black V rectifier characteristic of the modified 28V regulator blue Vc qi red Right Scale Compound base For comparison with Figures 3 and 17 note the zoomed currents of the pass transistors green Ic q pink voltage scale Figure 18 indicates that by simulation black trace an about 35mV an increase of the voltage supplied to the PA can be expected for load currents ranging from 0A to 10A However the voltage measured in the rig blue measurement points after the modifications with R kept at 3 3kQ as discussed on page 17 drops by 15mV see dotted interpolation curve between no PA current i e loaa 0 54A and full TX output 9 1A PA current plus 0 54A for the rest of the rig This voltage drop corresponds to a differential resistance of about 1 6mQ The difference in V I characteristics between the simulation and real life measurements is attributed to limitations of modeling of the bridge rectifier with ideal diodes and series resistances This rudimentary model is sufficient for simulation of the transient behavior of the regulator but evidently cannot account for the non linear I V characteristics of the real bridge rectifier The measurement was carried out using the TX on a dummy load as a variable load The current was measured at the PA supply lead taking into account the additional current of 0 54A supplied to the LPF unit the AT unit the meter lights and the 21 7V r
42. ltage The Figure 13 shows how IC and VC are measured in the signal unit The Figure 14 shows the corresponding layout in the signal unit The voltage used for VC measurement in the original design is supplied to the signal unit via the ICA lead of connector 9 point 1 In the signal unit the zener voltage of D77 XZ 200 20 6V 500mW is subtracted in conjunction with R274 47kQ and the resulting voltage is supplied to the meter via VR VC set point 470 KQ see Figure 13a In the original design the voltage displayed on the meter is the voltage sensed before R44 i e when you run full TX output 10A PA current the PA voltage will drop to 28 0V while the meter still displays 28 5V In other words what ever happens the meter in VC position displays how the voltage regulator of the PSU reacts to current variations but not the voltage effectively supplied to the PA This should not be the intent of the VC measurement On the other hand if you implement the modification including Rj in the regulation loop the voltage supplied to the PA will always be 28 0V but now the voltage measured at ICA will rise by OV 0 5V according to the PA current Of course we know that now the voltage supplied to PA at any current is what we measure at TX on with no modulation CW or key up This is better but we still have a display error at full TW output in the reverse sense now x o o o oO gt 9 1K Figure 13 IC a
43. ltage Vox 60Vdc 140Vdc max Collector Base Voltage Vcg 60Vdc 140Vdc max Emitter Base Voltage VEg 5Vdc 5Vdc max Collector Current continuous Ic 25Adc 20Adc max Total Power Dissipation Pp Tc 25 C 200W 250W Pp derating above 25 C 1 15W C 1 43 W C maximum Junction temperature Ty 200 C 200 C Thermal junction case resistance 0jc 0 875 C W 0 7 C W Case TO 204AA TO 3 TO 204AA TO 3 Table 1 Characteristics of the original pass transistors and the proposed replacement 8 Also proposed by Michael VK4AMZ in his PSU modification 11 The intent of this replacement is to ensure protection of the driver and PA transistors through increased safety margin with respect to the guaranteed maximum collector emitter voltage Vcego Also due to lower thermal junction case resistance 9 c and higher maximum dissipated power of the MJ15003 the replacement increases the margin for thermally safe operation At 63W DC power per transistor corresponding to 10 5A output current of the PSU and a case temperature of 70 C the junction temperature is 125 1 C for the 2N5885 and 114 1 C for the MJ15003 At 70 C case temperature the thermal power margin with respect to a DC power of 63W taking into account the maximum power at 25 C and the power derating is 85 25W for the 2N5885 and 122 65W for the MJ15003 i e the replacement leads to a 44 increase of thermal power margin To change the pass transistors you have
44. ltage spike when the rig is switched off Further on simple modifications are proposed to enhance the robustness of the PSU and to improve the performance of the PSU With these modifications the output voltage now stands solid as a rock The variation of the 28 0V supplied to the PA is only 15mW as compared to the about 0 5V in the original design for a load variation from RX to full power TX Again as said in the introduction for rig of early series without the 21 7V regulator I would recommend the replacement of the PSU with the VK4AMZ design Finally without any relation with the subject of this article I equipped my TS930S with the Inrad roofing filter as well as 250Hz CW filters on both the 8 830MHz and the 455kHz IF The result is amazing With the roofing filter I can work in SSB without any problem very closely to AM stations we still find from time to time in the upper part of the 40m band and even when the 40m or 80m band is crowded there is no more need to use the attenuator With the IF filters in conjunction with the variable bandpass tuning you can reduce the bandwidth to some tens of Hertz even in a contest the band becomes quiet For me providing the TS930S with these steep filters is definitely worth the price Have fun with your TS930S 73 Klaus FSVIM 22
45. mpound underneath the flanges of the resistors Insert an appropriate washer between the isolators and the nuts or heads of the screws Tighten the bolts firmly to ensure good thermal contact with the heat sink and secure the bolts with a drop of nail varnish on each nut One word to those who might consider the heat sink temperature as a concern The power dissipated by both resistors 2 5W per resistor at full power TX does not lead to a temperature increase of the heat sink The power dissipated by each pass transistor is reduced by the same amount because of the reduction of Vcg as mentioned above Improvement of the regulation loop optional The two joint modifications presented in the following do not have an impact on the robustness or safety of operation They allow improving substantially the differential internal resistance of the regulator They are presented here rather than in part 5 because one of this modifications is to be implemented on the heat sink before you mount it back into the rig ie 10A at full power TX plus 0 54A for the other loads of the PSU OTF you look for other resistors than the type indicated don t select on the basis of low temperature coefficient The lower the resistivity drifts with temperature the less you profit from the thermally induced counter reaction Select with respect of power dissipation limit and type of the case 12 For the original PSU the calculated slope of the V I characteristi
46. nce of the 1 1kQ resistance to ground the error amplifier delivers about 26mA also at no load pean Thoaa OA see the light blue trace of Icqu 28 745 H 90mA e The presence of this current indicates that the output ba ev 4 Ree voltage is controlled by the error amplifier for all load conditions 28 5 5 t 30mA e The design weakness of unregulated operation at om low load is completely eliminated nm As can be seen by the black trace in Figure 11 the output voltage characteristic is fairly linear The simulated differential resistance at point 28A 28B i e the slope of atma the black trace varies only very slightly from 8 1mQ at Thoad 1A to 5 7mQ at Thoad 10A 60mA 28 0 5 120mA ae For comparison in the regulated range the differential E aA ii resistance at point 28A 28B of the original design varies from 35 0mQ at Iroad 1A to 6 8mQ at Iroad 10A 27 8V T T t T f t T T t 1 T T 180mA 0A 1A 2A 3A 4A 5A 6A TA BA 9A 10A 11A 112A The voltage at points 28A and 28B drops by about 68mV Figure 11 Simulation with a 1 1kQ resistance if the current rises from 0A to 10A full power TX Left scale Voltage at points 28A 28B black delivered to the PA unit green and potential of the node C_Q The voltage supplied to the PA of course undergoes an grey Right scale currents at the node Ip red Ic a1 additional drop of 500mV due to the 50mQ series resistor turquoise and compound ba
47. ncrease of the base potential and hence a 50mV decrease of Vpgg of Q2 In other words while we provide the regulator with the full output voltage variation at the emitter of Qn half of this variation is lost for Vpg by the voltage divider for the regulator set point The NPN transistor Q is the first stage of the error amplifier in the regulation loop its role is to compare the sample of the output voltage to the set point which defines the target voltage As we saw Vgg of Q varies out of phase by about half of the output voltage variation Consequently the collector current Ic of Q varies also out of phase If the output voltage increases with respect to the voltage set point defined by adjustment of VRj the collector current Ic q2 decreases The PNP transistor Q4 the second error amplifier stage is the driver for the pass transistors it only amplifies the error signal provided by Q The transistor Q is polarized with a negative Vgg voltage corresponding to the voltage drop across R a 47kQ resistor Therefore Vgg of Q is proportional to the current flowing in this resistor A decrease of the collector current Ic q2 therefore also reduces the collector current of Q and hence leads to less current injection in the pass transistors Consequently Vcg of the pass transistors increases This limits the output voltage variation to the regulator error voltage which is function of the total loop gain the higher this gain the more stable is the o
48. nd VC measurement in the Signal unit Figure 14 Board Layout Signal unit In order to measure the real voltage supplied to the PA the correction is quite simple but you need patience If you have the courage to remove dozens of connectors and screws from the signal unit board you can connect the cathode of D7 half way between connector 32 and the IF crystal filters to the ICB point instead of ICA on the nearby connector 9 see Figure 13 An easy way to do this is to unsolder the diode D7 and to move the diode to the foil side of the board This provides true measurement of the PA voltage The modification doesn t impact the IC measurement derived from the ICB potential through R272 R273 and VRjx and from the ICA potential through R275 R277 and R376 in the signal unit In part 7 we will summarize all the modifications proposed and present the schematic of the modified 28V PSU It only neglects the voltage drop in the heavy red 28V lead from the PSU to the PA The resistance of this lead is less than 2mQ and consequently the measurement error is less than 20mV at full TX output a difference we anyway won t see on the meter 18 7 Summary of the modifications schematic of the modified 28V regulator Table 5 lists the proposed modifications the reason for modification as well as my subjective appreciation whether the modification is essential or not Modification
49. of the PSU performance LT Spice can be downloaded as freeware at http www linear com designtools software or run a Google search gt However there is a little influence of C on the ripple rejection Also C may serve to prevent other effects e g oscillation which perhaps were observed with real world components during the test phase of the design I didn t investigate this point further 3 Regulation limitation and resulting potential danger for the driver transistors In this part I will come back to my observation of occasional overvoltage of the 28 5V output of the PSU As said in the introduction I observed some strange behavior of the PSU Sometimes the meter lights shone brighter or flickerd somewhat randomly At the same time the Vc meter of the TS930S showed voltage in excess to 30 V Of course the first thing I did was to open the rig and disconnect the heavy red 28 5V lead to the PA I then tested the PSU with dummy DC loads I had on hand high wattage resistors and also 12V car bulbs in series I noticed with low load currents that an overvoltage could occur after some time of heat up while it was absent when the PSU was cold Further on I observed with these tests that the PSU behaves as expected for a wide load range butat low load or no load at all the output voltage raised to almost 33V Also when I ran the PSU with the small yellow wire still connected to point 28B I observed that the voltage at t
50. off state Both conditions 29 1V potential at the node current Ir22 compensated by an additional current flowing away from the node otherwise than through the components already connected can be met with an additional resistor from the node to ground The value of the transistor is easily calculated according to Ohm s law R 29 1V 0 7mA 41 6kQ The 41 6kQ value is the upper limit for the resistor we have to insert In case we use a lower resistance the potential will drop below 29 1V This leads to a reduction of the output voltage which will be detected by the error amplifier This in turn will cause the collector of Q1 inject current into the node in order to establish the node potential at a new slightly lower equilibrium point of operation All you need to do is readjust by means of VR this equilibrium point of operation in order to fit with the output voltage with the 28 5V target This means that within limits we can choose the value of the resistor in order to assure correct regulation also under the worst case conditions listed above But before doing so let us check the validity of our analysis let us see what happens if we use a resistor close to this calculated value weal Weal Vioupat_to_pa aA Figure 10 shows a simulation of the regulator with R22 D7 Ic Q1 271 R5 140mA 3 kept in place and with a 39 kQ resistor connected from EA ert sama the node C_Q to ground the closest resistor valu
51. ogize for a too partial investigation i Simulation with SPICE not shown here confirmed the shift of the threshold to higher currents if a higher hrg was used for modeling Simulation with stepped temperature also confirmed the shift to higher output currents of the threshold for regulated operation Though we focus here on what happens at power switch off this simulation also includes the transient at power switch on This was done in order to investigate the start up as discussed in part 2 of this document It allowed also checking if a critical value could be reached during the time of settling No particular effect was revealed at highly time resolved scanning of the switch on transient See eee an en For simulation of the transient behavior the model 0 00164 r y schematic was modified The constant voltage source was replaced by a circuit consisting of an AC source and the bridge rectifier each diode including a 0 36Q series 0 001 4A ll agiza resistor corresponding to the resistance of the diode itself and half of the internal resistance of the transformer as L 0 00104 determined by the voltage drop under load measured in my rig In order not to introduce an artifact into the f 9 00084 transient simulation the variable load which in fact is a swept current source was replaced by a fixed resistor in series with a time controlled switch in series in order to mimic the power switch A similar switch was placed in
52. or 7815 instead of resistor R in order to provide even higher stability or better ripple suppression to the base potential However this solution bears a potential risk in that sense that when switching on the rig the emitter potential of Q2 may built up earlier than the base potential This might lead to reverse base emitter polarization which destroys Q if its maximum emitter base voltage rating is exceeded 2 The power dissipation of R rises slightly it is 62mA at a maximum full power TX You don t need to change for a higher power rating 3 The more current the zener diode drives the more stable the voltage will be a By reducing the value of R we increase the current in the zener diode and consequently the operation point is shifted to a point where the I V characteristic is steeper i e Vz varies less with the voltage at point 28B 17 6 Design error of the PA voltage measurement VC meter The last design error I found concerns the measurement of the PA voltage with the meter VC position As can be seen from Figure 1 the two potentials before and after R44 are routed to the signal unit across the connections ICA and ICB of the connector 6 of the PSU board The potential difference allows IC measurement For the VC measurement by error the potential ICA is used This voltage doesn t account for the PA current dependant voltage drop in Ry4 The voltage displayed by the meter in the VC position therefore is not the real PA vo
53. or control of the PSU cooling fan H or Y rn R ume wt KH i yya In the following we will focus on the 28 5V regulator Figure 1 Schematic of the TS930S power supply The 28 5V regulator operates as follows e The output voltage is sensed at point 28B via R and Dg The forward biased diode Dg the varistor De and the reverse biased zener diode D are subject to a current of about 10mA In conjunction with R they define the emitter potential of Qo to Vea Vout approx 15 3V e The potential difference of 15 3V subtracted from the output voltage is the sum of the forward voltage in Dg of an about 2 1V drop in the varistor Dg at 10mA and of the zener voltage of D3 Consequently any change of the output voltage induces an equal change in the emitter potential of Q2 e On the other hand the base potential of Q is defined by the output voltage and the voltage divider constituted of R4 VR and Rg This potential is about 13 7V so roughly half of the output voltage thus providing about 0 5V of base emitter voltage Vpgg to Q2 A variation of the output voltage will affect the base potential by only kalf this value If the output voltage decreases by for example 100mV due to increase of the load current or by a drop in the main power supply voltage the base potential of Q drops by about 50mV Likewise a 100mV increase of the output voltage e g due to increased current gain of the pass transistors resulting from heating leads to a 50mV i
54. roposed replacement Replacement of Q 2SB861C BD708 max Collector Emitter Voltage Vcxo 150Vdc 60Vdc max Collector Base Voltage Vcg 200Vdc 60Vdc max Emitter Base Voltage VEg 6Vdc 5Vdc max Collector Current continuous Ic 2Adc 12Adc max Total Power Dissipation Pp T 25 C 1 8W 1 81W Pp derating above 25 C 14 4mW C 14 5mW C maximum Junction temperature Ty 150 C 150 C Thermal junction ambient resistance ja n a 70 C W Case TO 220 TO 220 Table 3 Characteristics of the original transistor Q and the proposed replacement n a data not available calculated from data sheet Remplacement of Q Similar to Q I examined the voltages currents and power dissipation also of Q at different load conditions For this transistor we need to consider that at the very first instant after switch on the collector emitter voltage is close to the input voltage of the PSU i e about 40V The original transistor a 2SC1815 has a Vceo rating of 50V In order to gain safety margin I decided to replace this transistor by a BC639 4 which has a maximum allowable Vcxo of 80V Also I had a SPICE model for this transistor replacing the transistor allowed me to model the behavior of the comparator stage in the rig with better precision The table 4 gives the principal characteristics of the original transistor and the proposed replacement for Q2 Replacement of Q
55. s related to the slope with which the input voltage builds up The steeper this slope the 12V4 more current Q has to inject to make the output voltage a rise towards the target voltage i For steady state simulation I was able to determine the ev 1 combined internal resistance used in this simulation too of the transformer and of the rectifier bridge but this w b oIa allows only limited extrapolation to the dynamic case _ ___i__ Second The current delivered by Q is the current vi m U 0 1A necessary to sustain the output current Consequently if dms Bms 72ms 7ims 8 ms S4ms BBms 92ms Jms 100ms 2 i for reasons of tolerances the current gain of the real life pass transistors is lower than what is simulated the amplitude of the spike is higher This is also the case when the rig is particularly cold Figure 12 Zoom at the power on transient 64ms 100ms after switch on Left scale output voltage black Right scale Collector current of Q1 red Therefore I decided to gain safety margin with respect to the collector current by replacing Q The BD708 has a 6 times higher continuous collector current limit than the original 2SB861 The Veg and Vcg ratings of the BD708 are lower but still sufficient for the given case Q is exposed to about 12V DC The other maximum ratings are similar for both transistors 14 Table 3 summarizes relevant characteristics of the original transistor Q and the p
56. se current to the pass R14 which increases the differential resistance seen by transistors pink the PA by the same amount We observe that with this little modification the node voltage stays well below the zener voltage and therefore the current in D is almost zero its value is 0 6uA not resolved in Figure 11 due to the scale 10 Consequently D7 now has an exclusively protecting role in case of other failure e g collector emitter short in Q but no longer has a voltage limiting function even in operation without component failure as it was in the original design Therefore we gain one level in terms of robustness While in the original circuit a failure of D7 alone was sufficient to cause voltage in excess able to potentially damage D in the PA unit after this modification breakdown of D7 may result in overvoltage only in conjunction with an additional component failure As aconclusion of this part of the analysis we can state The original 28 5 V regulator bears a potential dangerous design weakness e At low load current the voltage is unregulated e At each time the rig is switched off the PA unit undergoes an excess voltage transient the voltage of which is only limited by Dy e If D is broken the voltage transient reaches the Vceo rating of the driver transistors of 35V provided D in the PA unit is still operational and exceeds the rating in the contrary case if D is defective The design error can be cured by
57. stors Consequently the output voltage moves up to the 35V zener voltage limit defined by D minus the base emitter potential difference in the pass transistors the voltage drop in the emitter resistors and the voltage drop in R 4 Depending on the output current the output voltage would be somewhat 33 34 V in this case In order to avoid dangerous overvoltage in case both D and the ICR lead are broken it is wise to connect a 10Q resistor 1 4 W is sufficient in between the anode of Dg and the point where it was originally soldered And you have to run a short wire from the anode to point ICR on the connector as said above Under normal circumstances output voltage sense will be provided by this short wire In case the point ICR on connector 6 is open the voltage sense will be provided by the 10Q resistor This limits the output voltage of the regulator also in this case The voltage supplied to the PA before this modification dropped from at bit less than 28 5V to about 28 0V at full power output e VRI should now be set to 28 0V i e the same voltage as before at full TX power As due to this modification the lead to the PA is regulated at 28 0V point 28B in turn will now rise by up to 0 5V when we drive a high current TX mode This doesn t harm the little yellow wire from 28B to the second contact pair of the power switch already ran 28 5V in the original design I didn t want to see the meter light slightly flicker in th
58. ts on the modified circuit The diagram of Figure 17 can be compared to Figure 3 corresponding to the original design As can be seen the excess voltage in the load range below 110mA completely disappeared Moreover the 60mV drop of point 28A 28B visible black trace in Figure 3 is reduced Even more evident the 0 48V drop at of the PA supply voltage at 10A blue trace and cursor box in Figure 3 is eliminated As the internal resistance of the transformer and of the rectifier bridge determined experimentally was included in this simulation the voltage supplied by the rectifier drops with increasing load current blue trace As a consequence beyond about 11 8A the PSU output voltage black trace also drops because the voltage at the node of collector Q1 red trace starts to drop at this maximum load current From this point on the output voltage is no longer regulated this is visible on the pink trace of the collector current of Q in Figure 17 which is bent and becomes a linear function of the output current in the range above 11 8A Measurements with various loads showed that in the modified circuit in my rig the transition from regulated to non regulated operation occurs at approximately 11 4A which is fairly close to the limit determined by SPICE simulation 20 v_out V v_rectifier Figure 17 Simulation of the modified 28V regulator Figure 18 Simulated black and measured blue V I for
59. u see the meter bulbs shine brighter than usual do not switch off the rig Pull the plug of mains power cord from the socket This should prevent the PSU undergoing the excess voltage transient as it will see a sufficient load during the first seconds of the voltage decay V 28a_28b 07 V c_q1 lc Q1 output_to_pa 2 1 R5 R22 29 1 29 0 28 9 28 8 5 28 75 I28 5V 28 4 28 3 5 28 2 28 1 28 0 27 95 140mA 120mA 100mA 80mA 60mA 40mA M 20mA 27 8 OmA 20mA 40mA 60mA 80mA 100mA 120mA 140mA T j T T 6A 7A BA SA t t 10A 11A 124 Figure 8 Simulation with R removed Left scale Voltage at points 28A 28B black deliverd to the PA unit green and potential of the node C_Q grey Right scale currents at the node Ip red Ic qi turquoise and compound base current to the pass transitors pink 28a_28b 36 4 35 745 35 05 34 3 5 33 645 32 9 5 32 2 4 31 5 5 130 8V 7 130 1 F al sel lll 28 745 I28 0 f t f t t 32V 34 36 38 40V 42yv t Aay i AGY 48y Figure 9 Simulation of Vout VS Vin in case of failure of D at 5 different temperatures in 25 C steps from 25 C black to 125 C pink In Figure 8 we can see that without resistor R22 the potential of the node
60. utput voltage with respect to load current or input voltage variations So far for the steady state operation principle What do C and R3 serve for and how does the regulator start Right after switch on the emitter and base potentials are zero and therefore we might think that C is intended to give a kick to the base of Q2 I am not sure about this First the potential at C4 increases asymptotically to the value of the supply voltage with a time constant defined by C4 and R and moreover the supply voltage itself also builds up as the two capacitors Cg and Co together 44000uF get charged by the rectified AC voltage taking into account the series resistances of the transformer and of the bridge rectifier this process is quite slow Consequently we cannot consider that Cs provides a significant potential spike to the base of Q2 Secondly to my understanding both from conventional circuit analysis and from SPICE simulation the circuit starts through the residual off state current in Q gt Of course in absence of a polarization Vgg an ideal NPN transistor would not allow any current to flow as we apply about 40V to its collector But a real transistor will Due to the high value of R a quite small off state current in Q allows polarization of Q and therefore will start regulation I simulated the circuit without C and didn t see any difference in the switch on transient behavior The same manner we might think that Rz allows the
61. v a B0 5 4 60mA ine her W3 ate 10 1 00000 0 041 2 a im b e oon mu v B0 0v H 70mA Le Besse Ta WVR tb en ho ba IG c jpe le p9 5v 80mA pr at ale ia i Je Ime nii ia 2200p Ks 2200p fo Fv fm E X y Po i 29 0 5 28 5 5 28 0 5 7 5v t T T f T T j r r i 7 120mA 0A 1A 2A 3A AA 5A 6A 7A BA 9A 10A TA 12A Figure 2 Design used for LT SPICE simulation of the Figure 3 Simulated V I characteristics and the 28 5V regulator compound base currents of the pass transistors of the See the text for explanations regarding the models used original 28 5Vregulator for loaa 0A 12A I concluded that something must be wrong by design Therefore I ran a simulation with a sweep of the load current from 0A to 12A and it confirmed my experimental observation e The PSU by design cannot regulate at low load conditions Here is the reason why Remarks concerning the modeling Simulation with SPICE is often hampered by the unavailability of SPICE models of the components For this simulation I used the models of components as in the original design excepted for Q BD140 model Q BC639 model Ds simulated by a 2 1V voltage source with internal resistance of 16Q read from I V characteristics in the datasheet of the SV 03Y varistor Ds 1N4148 model and D BZX84C12L model The chemical capacitors C4 Cs and the two 22mF capacitors denoted Cs and Co in the sch
62. y from the node 32 4 32 0 31 6 I31 2 The signs for the labels of the currents at the top of the diagram were set to respect the rule of counting Article TS930PSUOriginal raw X positively currents flowing towards the node 30 8 725 078pA As can be seen from the output voltage characteristic at point 28A 28B black line at this level of zoom there are 3 distinct V I dependencies 30 4V 4 Diff Cursor2 Cursor1 30 0 e anon linear V I dependence for OmA to 60mA output 29 6v current 29 2 4 e a linear range from 60mA to 110mA with a steep slope 28 8 e beyond 100mA an also linear V I dependence but with a much smaller slope 128 4 1 i 1 1 1 2 1mA OmA 50mA 100mA 150mA 200mA 250mA 300mA The latter corresponds to the normal and expected Figure 4 Node at Collector Q1 load range 0 300 mA Left scale Output voltage black node potential grey Right scale Node currents Ir blue Ip7 red Ic qi turquoise and sum of the base currents to the pass transistors pink For cursor 1 see text regime of the regulator we already saw in Figure 3 Looking at the currents we notice that in the lowest two ranges i e up to 110mA no significant collector current is supplied from Q light green trace In this range of low output current the current flowing to the pass transistors pink trace is balanced by th
63. y lead to destruction of the driver transistors in the PA unit when switching off the rig Analyzing the schematic of the rig we notice that the PA is connected permanently to the current sensing resistor Ry4 However the 28 5V to the antenna tuner LPF unit meter bulbs as well as the 21 7V pass transistor are delivered through a contact pair of the power switch in this switched lead I measured a current of 0 54A with meter lights and display dimmed It is due to this switch why we observe an overvoltage at 28A 28B as well as on the lead to the PA Right after switching the rig off the capacitors Cg and Cy 22000 uF each near the rectifier in Figure 1 are still energized to about 40V I measured 39 8V across these capacitors in my rig in RX mode When switching off all loads are disconnected from the PSU excepted the residual current in the PA I measured 31 2uA at no polarization you won t switch of your rig during TX would you Consequently we encounter a situation where the operation point moves from the regulated range C to the non regulated range A when switching off This first results in overvoltage according to Figure 4 Then after some seconds the output voltage decreases following the decay of the voltage at Cg and Co To illustrate this analysis I performed a simulation of the transient behavior of the 28 5V regulator Figure 5 shows this time transient simulated with LT SPICE Tf ever it was in advance I apol

Download Pdf Manuals

Related Search

Related Contents

- OoCities  Service Manual: HS2P series (G407/G408/G410/G411), IS450S  DEPOT EXPRESS - mode d`emploi - E  À NE PAS FAIRE À FAIRE  Guía del usuario  le collaborateur  Étude des pratiques de remédiation et de pédagogie différenciée  

Copyright © All rights reserved. Failed to retrieve file