Home

PowerSocket - The Media Computing Group

1. for the range 1200 to 1400 the JNDs are 140 to 150 for the lower part and 150 to 160 for the higher part 130 to 150 are the JNDs of the range 1000 to 1200 4 Evaluation 110 120 and 130 are barely recognized detection rate is lower then 24 in the range 1000 1200 which would sug gest that they belong to a prior range This is strengthened by the fact that the JND 120 is somewhat recognized 38 in the first part of the 1200 1400 range but in the higher part the JNDs 130 and 140 are barely recognized both de tected by 15 Therefore the JNDs 110 to 130 should be tested in the 800 to 1000 range As 130 and 140 are not rec ognized very well a detection rate of 38 only in the first part of the 1400 to 1600 range and 150 is not recognized at all these JNDs will probably still be too low for this range In the next class 1600 to 1800 rather low JNDs are recog nized This can be explained that at 1600 the animation changes color to red which was a very bright color We can see that 170 is recognized with 54 as well I assume that this value is a more correct value as it would fit to the fact that the JND of the next class is between 190 and 200 de tection rate of 62 and 53 and the JND of the last class is over 200 With this information I can now interpolate the JNDs of the ranges from 1000 to 2000 For the range 1800 to 2000 the JND will probably be as already mentioned between 190 and 200 From this I ca
2. ings of the 8th International Conference on Mobile and Ubig uitous Multimedia page 15 ACM 2009 81 83 Index abbrv see abbreviation adapter of the MCU 25 adjustments to the EM773 smartmeter 17 advantages of the EM773 16 appliance level metering 8 9 availability of metering data 35 36 bar chart of the relative detection rate 43 benefits of the prototype 32 benefits of the software design 41 42 box containing the mains circuit 29 box containing the PCBs 29 31 box of the prototype 29 31 breakdown of the power supply 17 communicating with the LED driver 37 costs of a prototype 7 deleted parts from the analog front end 18 deleted periphery around the MCU 19 20 discussion of the result of the user study 46 48 drivers 33 38 ease of installation 10 electrical safety 28 evaluation 43 49 future work 52 53 GPIO driver 34 35 household level metering 9 immediate and detailed feedback 1 implementation of the rotating animation 39 40 initialisation the LED driver 37 integrating an opto coupler 21 LED driver 36 38 84 Index LED PCB 27 main board 24 25 mains circuit of the prototype 29 mechanical safety 31 methodology of the user study 43 45 metrology engine 35 36 modular design of the prototype 24 27 modules from the EM773 SDK 33 35 modules I have constructed myself 35 41 overview of the energy consumption 11 package ANIMATIONS 38 power supply 26 27
3. problems with Watt and kWh 1 2 programming and debugging the prototype 41 properties for prototypes used in large scale studies 6 7 prototypes avoiding Watt and kWh 2 replacement of the power supply 17 reproducing a prototype 6 ring buffer of the LED driver 36 37 safety by using a fuse 28 strength and weaknesses of appliance and household level me tering 10 summary 51 52 the choice for an appliance level meter 11 the resulting circuit around the MCU 21 22 the rotating animation 39 UART driver 34 unmeasurable appliances 10 usability of a prototype 7 user studies conducted with energy measuring prototypes 5 using the LED driver 37 38 using the metrology engine 35 using the rotating animation 40 41 Typeset October 27 2013
4. the ports that were makes the switch 3 1 in figure 3 5 and the ports to the relay connected to the 3 2 in figure 3 5 superfluous Consequently they have not relay are freed and a been integrated in the prototype introduced in this work switch controlling the Another part I deem unnecessary was the LED that signals relay removed which current measurement channel is in use This was due to that this information could be printed on the UART lastly an LED interface Hence this LED was left out of the prototype I signalling the constructed As mentioned in the beginning of this chapter measurement signal the wireless transmitter unit was left out this means that is removed the connections denoted by 4 in figure 3 5 are not present in my prototype 3 1 Hardware design DT REEUEREREED 21 Figure 3 6 The MCU with some ports and connections depicted is the breakdown of this part of the circuit 1 Power supply for the MCU 2 XTAL for the MCU unit 3 1 UART port isolated by an opto coupler 3 2 Switch to get the MCU in programming mode 4 I O block of the MCU circuit The result of the component removal can be found in figure C 3 In contrast to removed parts one crucial part has been added This is an opto coupler Adding this part enables debugging over UART even when the power sup ply of the circuit has different voltage levels then the device that debugs programs it The metering unit requires th
5. DIE bs ely hd ex 1x3 A6 EENHJELIWNS ED SSA Ons EAE TODA fer a wap o Ti Lo zo Co a zu OOO zr ENAEr NLS a a SAO Y ATOMS ence EN u 038 End Hyeras Vid FF e E gt AA CA ta OLY OTL MODSI ATODI gt E 0105 N N Era cn Fayoz SVETON E Dos e ns a vos 2 N38 go s4001 fm EEH Ener Ener a Fold EODH f DT a TAYHTOITE LI OXL L TOId 044I7099113 01355 2 70018 CO r Nu TEN AR TYH OREL LOM Tend fF an ar O YO TED Syb Told ovis JOY ITA ANDI ELVA TEE Td p TL OMS 4 z HEIDIT Por MaM TWISTED E TOIS a de NOYCHEIHT IL TIEL Y Z Todd Be TA YHTE TTH ooldrjesay fF 158 yms BPS zo Loze 60 T za is EAE AQ 3 10 501 S13 ENEr ru oO Figure B 3 The EM773 microcontroller with some periphery attached 61 16 000 MHz Es c100 C101 TP OL2381AHN COB eu Mm an zj Fla c128 C129 JA ojs Ch Dp z 3V3 u7 ttnH a 6 a VIF o c133 C134 L13 C123 L15 Lis 15nH smiH C126 c 25 PF a swi UPDS713TK c132 IFI tn 1 Y BOS mm JP11 On PCB Figure B 4 The wireless transmitter of the EM773 smartmeter 63 Appendix C Schematics of the prototype C 1 Schematics In this part of the Appendix the schematic of the proto type introduced in this work can be seen Figure C 1 de picts the power supply of the prototype It is based on the viper22al which is a switching mode power supply As a
6. can be seen from figure D 1 the EM773 SDK contains the drivers for UART and the drivers for the ports of the MCU The drivers for the ports of the MCU are provided in a sub package called GPIO As the LED animations stem from PowerSocket the EM773 SDK did not have support for an LED driver Furthermore there was no driver available for the TLC5947 I have used Therefore I have constructed the driver for the LED driver myself Details concerning this driver of the LED driver will follow later 33 As basis for the animation the animation of PowerSocket is used modules from the EM773 SDK are used the cortex MO system modules are used from the SDK ittakes care of the clock settings of the MCU the SDK contains the drivers for UART and the GPI O ports 34 the UART driver is used by calling an init function first and calling the corresponding functions for sending and receiving two modules are used to control the GPI O pins from both modules the init method should be called with the module pcb the function of a pin can be set the module gpio takes care of handling the GPI O pins 3 Creating a small scale version of PowerSocket When the UART is used it first has to be configured This is done by calling Ser_Init config In this function config is a parameter that is a struct of type uart_config_t How config looks like this can be seen in figure D 1 The parameter config passes the UART configur
7. toggle the clock pin the next one sends a bit or applies a configuration to the LED driver The driver of the LED driver keeps sending one bit at a time until the last bit of the cell that was lastly written has been send If the last bit of the last of the ring buffer is sent the then received configuration at the LED driver is applied After the timer is activated the initial content of the ring buffer is sent to the LED driver and applied This sets all brightness levels of the LEDs to zero The core part of the LED driver is sending single bits this is done automatically as already mentioned by using a timer interrupt The LED driver keeps sending until the send pointer reaches the write pointer of the ring buffer When the write pointer is reached the sending is paused as there is no more data available When the last bit of the last cell of the ring buffer is sent which means that a full config uration is received at the LED driver the configuration is latched and put in use This is done by first disabling the LEDs by setting the corresponding pin to high which reduces flickering in the animation Afterwards the pin that accepts the received configuration latch is set to high Then both pins are set to low in reversed order The reverse order ensures that the LEDs are activated after the received configuration is put to use As can be concluded from the last paragraph the driver deals with all necessary communicat
8. 2000 4 8 2000 2200 3 1 Table F 2 The result of the first part of the user study Bibliography Wokje Abrahamse Linda Steg Charles Vlek and Talib Rothengatter A review of intervention studies aimed at household energy conservation Journal of Environmental Psychology 25 3 273 291 2005 Arduino URL http arduino cc en Main arduinoBoardDuemilanove Blue Line Innovations URL http www bluelineinnovations com Loove Broms Cecilia Katzeff Magnus B ng Asa Nyblom Sara Ilstedt Hjelm and Karin Ehrnberger Coffee maker patterns and the design of energy feedback artefacts In Proceedings of the 8th ACM Conference on Designing Inter active Systems pages 93 102 ACM 2010 Marshini Chetty David Tran and Rebecca E Grinter Get ting to green understanding resource consumption in the home In Proceedings of the 10th international confer ence on Ubiquitous computing pages 242 251 ACM 2008 Sarah Darby Making it obvious designing feedback into energy consumption In Energy efficiency in household ap pliances and lighting pages 685 696 Springer 2001 Sarah Darby The effectiveness of feedback on energy con sumption A Review for DEFRA of the Literature on Meter ing Billing and direct Displays 486 2006 2006 DIY Kyoto URL http www diykyoto cor Energy Inc URL http www theenergydetective com 79 80 Bibliography FabLab Aachen URL http hci
9. Kyoto The Kill A Watt is an on outlet measuring device whereas the Wattson measures the energy use of the complete household One thing most commercial systems have in common is that the unit used for energy usage which is kWh or Watt like the Reichelt measurement unit shown in figure 1 1 In Broms et al 2010 there is mentioned that the concept of kWh or Watt is hard to understand for users as it is just a number This is confirmed by the literature studies from Darby 2006 and Abrahamse et al 2005 In Parker et al 2008 it is stated that the energy usage of a household is indirect feedback is not sufficient detailed and immediate feedback is needed commercials systems use Watt and kWh Watt and kWh are hard to understand ___ er w an u _ 1 Introduction ess uss la e a Figure 1 1 A Reichelt measurement unit displaying Watt usage right and used kWh left indication should be understood prototypes avoid the use of Watt and kWh have been constructed directly influenced by the behaviour of the inhabitants of the household If inhabitants cannot understand indicators for energy usage negative energy consumption behaviour will not be recognised and changed This problem is addressed in the HCI community Mostly by introducing prototypes that try to visualize the energy consumption avoiding the usage of Watt and kWh as indicator The Power Aware Cord Gustafsson and Gyllensw rd 2005 visu
10. Namely PowerSocket is an appliance level measuring device which I have chosen to construct This choice is motivated in chapter 2 How ever PowerSocket in its current state is too large to conduct a large scale user study with so it should be downsized Ideally the complete prototype should fit into the night lamp from IKEA see figure 3 1 If this succeeds the device will then meet some of the properties discussed in chapter 2 Namely it is easily installable as it can be plugged be tween an outlet and an appliance The size of the prototype is not too obtrusive The prototype is easily constructed as the exterior can be bought at low cost at IKEA 3 1 Hardware design uma TX Foo FX i Rtn mm Duemllanove lt pe ea ety Ne Sp oe io ens Figure 3 2 The Arduino Duemilanove prototyping board which is too large and expensive to be integrated in a small scale prototype To reduce costs more and because of advantages I will men tion later the EM773 from NXP is used in the prototype I constructed The current implementation of PowerSocket uses an Arduino Duemilanove which is more expensive then an integrated solution based on the EM773 One fur ther advantage is the integrated metrology engine of the EM773 Only a small analog circuit is needed for two channel measurement Another important aspect of this design choice is that it reduces the size of the PCB as the size of the Arduino Duemilanova is already quite lar
11. The main idea was to combine PowerSocket Heller and Borchers 2011 an appliance level measuring device with the EM773 smartmeter in a device from Ikea seen in figure 3 1 In this work only a prototype is introduced that tests the function ality of this design with exception from the energy mea surement circuit which is not functional during the time of 51 52 5 Summary and future work this work This prototype when the energy measurement circuit works can be used as a debug platform for future appliance level measurement prototypes It is particularly suited as debug platform because of the modular design of the prototype The visualization unit can be easily re placed the power supply replaced or extended and other periphery can be coupled on the ports of the MCU that are still free The software which is extended in context of this work annotates the prototype as debug platform As the programmer only has to write a driver for his or her ex tra attached periphery visualization unit and application animation itself The data from the measurement unit is made available and refreshed in constant time intervals so the programmer does not have to worry about the measure ment part Furthermore drivers for UART GPI O ports a wireless transmitter and I C are already available Lastly a user study is conducted in this work that assesses the performance of the rotating animation of PowerSocket This is done by first find
12. analog front end seen in figure 3 4 indi cated with 1 is removed This is done to decrease the size of the prototype It is necessary to evaluate if problems oc cur after removing the shield with the part of this circuit which was protected by the shield The relay of the ana log front end that can be seen in figure B 1 is a rather big part There is no additional need to switch the appliance connected to the meter off Users can plug out the appli ance from the meter therefore the relay has been removed The resistor and capacitor denoted by 3 in figure 3 4 are rather big components as well Their function is to relieve the removed relay when it switched This makes both com ponents superfluous and which is why they have been re moved In the end the result is the schematic which can be seen in figure C 2 Although the analog front end used in the prototype introduced in this work is a copy from the original schematic from the EM773 smartmeter it did not work in the prototype proposed in this work 3 1 Hardware design 19 Figure 3 4 The parts removed from the analog front end of the EM773 smartmeter 1 Shield that was put around the operational amplifier 2 Relay with which the measurement can be switch of The left 2 is the circuit of the relay the right 2 is the integration of the relay into the circuit of the front end 3 Capacitor and resistor integrated to relieve the relay when it is switched In
13. and TED 5000 Energy Inc Prototypes which can be clas sified in this class include 7000 oaks and counting Holmes 2007 Wattbot Petersen et al 2009 a systems that gives feedback using a mobile phone Weiss et al 2009 the Energy Aware Clock Broms et al 2010 and Dehems Sundramoorthy et al 2010 10 2 Related work I will now list the strengths and weaknesses of both classes summarized in table 2 1 First of all is the ease of installa an appliance level tion The advantage of application level energy measuring meter is easier to is that it is easy to install The meter only needs to be placed install between the outlet and the application as shown in figure 2 1 In contrast for domestic level energy measuring alter ation or additions around the main fuse need to be made when no smartmeter is available According to Weiss and Guinard 2010 this leads to high adaptation barriers Appliance level Domestic level Easy to install Overview of household energy consumption Detailed per device measurement Every device is measured Not every device can be measured Not so easy to install No overview of household consumption No detailed usage of specific devices Table 2 1 A comparison of strengths and weaknesses of both classes A weakness of appliance level measurement is that some devices cannot be measured as they are not plugged in into a energy outlet In Kim et al 2009b this proble
14. ence schematic have to be used and which one could be removed As the goal is to construct a small scale pro totype only the necessarly parts remain As such the wireless transmitter in the reference schematic which can be seen in figure B 4 has been removed Furthermore the power supply which the EM773 smartmeter uses de picted in figure B 2 was too weak to power the LEDs used for the animation introduced in PowerSocket Hence this power supply has been replaced The schematic of the cir cuit that replaced the power supply can be seen in figure C 1 The power supply is based on the SMPS switching mode power supply Viper22A E and the linear regulator LD1117533 The datasheets of both devices can be found in Appendix C 1 Furthermore as reference design for the SMPS circuit I used the STEVAL ISA035V1 The appli cation notes of this reference design can be found in Ap pendix C 1 as well The power supply consists of three parts as depicted in figure 3 3 The first part labelled with 1 in figure 3 3 transforms the voltage from the mains 230VAC to 12VDC The transformation from 230VAC to 12VDC is done by an SMPS The SMPS circuit of the power supply is a copy from the reference design of the STEVAL ISA035V1 The current Unneeded parts of the EM773 smartmeter are removed the wireless transmitter has been removed the power supply was too weak and has been replaced 18 the power supply can be extended by adding an
15. in figure 3 12 on the left Pressing the button has only effect when a running user study is ended During a study the button press is ignored When a user study is ended the animation is turned off It can be started 45 from this broad classification the user study was designed it consisted of a sequence of speed changes in the sequence speed jumps where encoded which represented a JND the sequence was randomized and speed jumps not representing a JND were mixed in the task of the participants was to recognize the speed jumps this resulted in how often a JND was recognized ina range 46 JNDs 2 5 belong to the range 0 to 10 JNDs 10 to 15 belong to the range 10to 25 JNDs 25 to 35 belong to the first half of the range 100 to 200 and 35 to 40 in the last part of the range 4 Evaluation again in the reverse order by pressing the red program but ton The participants are unable to pause the test Thirteen users participated in the user study Twelve of them were male and one was female The age of the par ticipants ranged from 22 to 58 The mean of the age of the participants was 31 None of the participants were color blind The results of the user study can be found in table F 2 Figure 4 1 shows a bar chart of the relative detection rate of the JNDs In the range from 0 to 25 Watt almost no JND is recognized Both JNDs recognized in this range have a detection rate of 8 This is probably ca
16. oaa The bill of material for the circuit of the LED PCB of the proposed prototype The coarse energy consumption classes I have found in initial testing The result of the first part of the user study 78 xi xiii Abstract A lot of thought has been put in how effective feedback is on energy consump tion There is a broad consensus that providing direct and immediate feedback has a positive effect on energy consumption behavior The type of feedback does not seem to be the only problem with feedback on energy consumption The way energy consumption is indicated seems to be too vague for a lot of people Most meters only showed kWh or Watt Therefore other visualization that try to avoid using kWh or Watt have been introduced Although some user studies are done with the prototype that contain alternative visualization none of the conducted user studies was powerful enough to prove that the introduced visualization led to more awareness then just displaying Watt or kWh One of the reasons such a user study was hardly realizable could be the way the prototypes used during the studies were constructed In this work I present a prototype that encompasses some advantageous proper ties for a large scale user study The circuit of the prototype is based on the EM773 smartmeter and the animation is based on PowerSocket Two versions of the proto type are envisioned The first is used to test functionality To m
17. of the main board and the MCU adapter using a socket 3 1 Connection for the Line L of mains 3 2 Connection for Neutral N of mains 3 3 Connection to the Neutral N of the appliance 4 Connection between the power supply and the main board using a socket 5 1 Connection using a socket between the LED PCB and the main board 5 2 Free I O block of the MCU 6 1 UART port 6 2 Pinheaders to put jumpers on when running the prototype on UART power 6 3 Pins to put a jumper on to set MCU in programming mode 7 Analog front end on main board of the mains 3 2 in figure 3 8 is needed As can be seen in figure 3 8 the line L and neutral N are redundantly connected namely to the power supply PCB and the main board With this redundant connection the power supply can be tested without being connected to the main board eliminating the risk to damage the main board during the construction and tests of the power supply The fourth and last PCB is the LED PCB depicted in figure 3 8 with the number 1 4 This board needs a connection to the MCU using the I O block of the MCU on the main 3 1 Hardware design 27 Figure 3 9 A HVQFN33 MCU bonded on an adapter board depicted by 5 1 in figure 3 8 The connection be tween the main board and the LED PCB is established by the pin headers on the LED PCB 5 2 in figure 3 8 Further more connections to the 3 3V wire of the power supply and ground are needed Bot
18. of the prototype is still possible It is advised when experimenting with 29 the mains circuit of the prototype is isolated by a box a ground fault interrupter is used as another layer of protection against electric shock The PCBs of the prototype are isolated by another box the second box is mounted on top of the first 30 3 Creating a small scale version of PowerSocket Figure 3 11 The box containing the circuit of the mains of the prototype Left the front of the box with the fuse and the socket for the application Center the top of the box the cables which connect to the PCBs of the prototype can be seen Right the back of the box there the plug white breaks out of the box The plug is connected to a ground fault interrupter A UART port and button is placed on the outside of the box which enables programming the prototype the PCBs to use the device seen in figure 3 11 on the right Figure 3 12 in the middle shows the box with its cover and figure 3 12 without the cover In the cover there is a window which enables watching the animation of the prototype although the cover is in place In the bottom plate there is a hole for the cables seen in figure 3 11 in the middle Furthermore on the outside of the box there needs to be a button that can switch the MCU in programming mode and a port for UART Having these on the outside of the box has the advantage that the prototype can be programm
19. pin of the MCU To connect the adapter to the main board I used two rows of double pin headers as seen in figure 3 9 on the right An advantage of a removable MCU is that if other parts of the circuit are tested the MCU is safe if it is removed Furthermore it is of advantage that if the MCU is broken the main board does not suffer from the attempt to solder a new one with bonding Circuit paths on a PCB tend to suffer when a joint on a patch is desoldered The third PCB is the PCB of the power supply 1 3 in fig ure 3 8 This PCB is connected to the main board with pin headers 4 in figure 3 8 Connections to the main board are the 3 3V wire and ground As the main purpose of the power supply is to transform 230VAC of the mains to 3 3VDC on the power supply PCB a connector for the line L of the mains 3 1 in figure 3 8 and the neutral N 25 the analog front end is situated on the main board the second PCB is an adapter for the MCU the MCU is bonded to the adapter the adapter can be removed from the main board for testing purposes the third PCB is the power supply which transforms 230VAC to 3 3VDC 26 3 Creating a small scale version of PowerSocket Figure 3 8 The modular arrangement of the first version of the prototype A dashed line symbolizes parts of the prototype that are occluded by other parts 1 1 Main board 1 2 Adapter for the EM773 MCU 1 3 Power supply PCB 1 4 LED PCB 2 Connection
20. prototype introduced in this work We have seen that the basis for this design are the EM773 smartmeter MCU and measurement and PowerSocket Heller and Borchers 2011 animation requirements for LED PCB I will now continue with the first version of the prototype I constructed 23 the constant current of the LED driver is configured using resistors 24 the prototype can be debugged even when the power supply is used the prototype is build modularly it consists of four PCBs a main board which connects all other PCBs the prototype can be powered with the power supply and the UART port 3 Creating a small scale version of PowerSocket 3 1 3 First version of the prototype As already mentioned at the start of section 3 1 2 the first version of the prototype is focused on functionality regard ing both the hardware and the software Therefore this ver sion of the prototype is a debug platform For software this is realized by providing a UART port which can be used even when the prototype is coupled on the mains As already mentioned this is done by an opto coupler The schematics for this part of the prototype is denoted 3 1 of figure 3 6 For programming when power is provided by mains both jumpers in the circuit of 3 1 in figure 3 6 need to be cleared When power is provided by the UART port both jumpers have to be set When the mains is coupled the jumpers may never be set For the second version of
21. update metering data is available the init function of this module takes care ofthe complete initialization of the metrology engine 36 After calling the init method constantly updated metering data is available in a variable had to write the driver of the LED driver myself the driver uses a ring buffer the write pointer moves with bigger steps through the buffer as the send pointer 3 Creating a small scale version of PowerSocket write pointer biti bit12 Figure 3 13 In this figure the ring buffer of the driver of the LED driver is shown Noticeable is the difference of movement between the two pointers Fmains metrology set_ranges ranges and metrology_start from the module metrology Fur thermore the initialization function calls the function that sets up the timer After the Initialise_Metering function is called the variable metering_result can be used to access the up to date energy measurement values Another module I have as already mentioned constructed myself is the driver for the LED driver This driver uses a ring buffer to store the data to be sent This ring buffer has the size of one configuration of the LED driver One configuration means the information brightness of the LEDs for every port of the LED driver that drives an LED This has the advantage that a configuration cannot be distorted before it is being used by the LED driver and data for a complete configuration
22. variation by changing the number of phases The main idea is that one LED in the LED ring has a full brightness level This specific LED rotates around the LED ring The configurations activated between the brightest level being at a specific LED and moves to the next LED are the phases I define the movement of the brightest LED to the next LED as a period At low RPM rotations per minute more phases are needed then a high RPM because a lower RPM has a longer period then a higher RPM This helps making the animation smoother At higher RPM less phases are used but this is not noticeable as the rotation speed is higher The color of the animation changes at certain thresholds set by defines The animation reads out the energy usage from the meter ing module and decides which color to use depending on the thresholds Furthermore the module then calculates which RPM corresponds to the energy usage at hand This is done with the formula rpm rpm_Per Watt x metering_result P rpm Per Watt depends on defines and is defined as const float rpm_Per Watt float MAXRPM MINRPM float POWERMAXTRESHOLD POWEROFFSET Coarse the formula looks like aa was Wat When the RPM has been calculated the phases needed per period to achieve this RPM can be calculated This is done by phases int float UPDATES _PER SECONDx60 float rpm LED_OUTPUTS OUPUTS_PER_LED 39 the animation shown by the prototy
23. 5 recognized in this range which would suggest they should rather be tested in the last half of the 100 to 200 range and the 100 to 200 range should be split As the JND 35 and 40 would rather fit in the last half of the range 100 200 the JND belonging to the first half of the range 200 300 will be bigger than 40 The JNDs 50 and 55 are tested in the beginning of the range 300 to 400 and are rather hardly recognized detection rate of 46 and 38 Thisis supported by the fact that the JNDs 60 65 and 70 which are tested in the last part of the range are hardly recognized detection rate is not higher than 23 too This suggests that the JNDs 50 and 55 rather belong to the last part of the range 200 to 300 In a next study the JNDs 45 to 50 should be tested in the first half of the range 200 to 300 and JNDs 50 and 55 in the last half of this range As already mentioned in the last paragraph the JNDs tested in the range 300 to 400 would probably be too low According to the last paragraph JNDs of over 55 should be tested for the first half of this range In the range 400 to 600 the JNDs 70 and 80 are already hardly recognizable de tected by 15 and 38 in the first and middle part of this range This would suggest that they are JNDs which be long to the last part of the range 300 400 For the range 300 to 400 we now have a list of JNDs to be tested which start with 55 and goes up to 80 The JND 80 in range 400 600 is slightly reco
24. 5C AFK476M50X16T F C7 0 14F 1KV X7R GRM55DR73A104KWO1L C8 100vF 16V X7R CC0805KRX7R7BB104 C9 10nF 6 3V JMK212B7106KG T Dz 12V Zener BZT52C12 7 F Dz1 16V Zener BZT52C16 7 F D1 D6 D8 1KV 1A S1M 13 F D5 600V 1A Ultrafast STTH1RO6A LO 4704H 140mA ISat 5300 33 RC L1 1mH 300mA ISat RCH895NP 102K RO 100 1W 5 PNP1WVJT 52 10R R1 1KQ 1W 10 MCMR12X102 JTL U1 Viper22A E VIPER22ADIP E U2 LD1117533 LD1117S33CTR 69 Table C 1 The bill of material for the power supply of the proposed prototype 70 C Schematics of the prototype C 2 2 Mainboard Component number s Required Part Placed part Cl 15nF 16V 10 C0402C153K4RACTU C2 1 2nF 25V 5 08053A122JAT2A C3 56pF 50V 5 GRM2165C1H560JZ01D C4 C6 220nF 10V 10 0603ZC224KAT2A C5 104F 6 3V 10 08056D106KAT2A C7 C8 47 nF 10V 10 Tantal T491B476K010AT R1 R2 R3 R4 R5 R6 270KQ 0 125Watt 0 1 ERA6AEB274V R7 R14 5 6KQ 0 0625Watt 0 1 CPF0603B5K6E R8 68KQ 0 0625Watt 0 1 CPF0603B68KE R9 1 2M9 0 0625Watt 0 1 1614959 9 R10 100KQ 0 0625Watt 0 1 CPF0603B100KE R11 R12 22KQ 0 0625Watt 0 1 CPF0603B22KE R13 4 7KQ 0 0625Watt 0 1 CPF0603B4K7E R25 5mQ 2Watt 1 CRF2512 FX ROO5ELF IC1 Quad operational amplifier NE5234D 01 512 Table C 2 The bill of material for the analog front end of the proposed prototype Component
25. EGW CA Ver 16A pdf C 1 Schematics 65 Y LS D 4 DI 30 Dt 4 Figure C 1 The schematic of the power supply of the prototype introduced in this work 66 C Schematics of the prototype Figure C 2 The schematic of the analog front end of the prototype introduced in this work C 1 Schematics 67 Figure C 3 The schematic of the MCU of the prototype introduced in this work and some periphery around it C Schematics of the prototype 68 DRIVERCON 3V3 THERMAL TLC5947 P30LEDBR PSI9LEDSG PSISLED7R PSIILEDSG PEIZLEDAR 3V3 3V3 CG CG ED6R LED6G LEDIG The LED PCB I designed Figure C 4 C 2 Bill of material C 2 Bill of material The following conventions have been used for all bills of material in this appendix The component number refer ences to the component number in the schematic The re quired part describes the properties the part needs The placed part is the manufacturer code of the part I placed on the PCB of the prototype proposed in this work C 2 1 Power supply Component number s Required Part Placed part Cx 0 022 4FE 50V X7R C0805C223K5RACTU C1 C2 104F 400V 105C ECA 2GHG100 C3 4 7 uF 25V X7R C3216X7R1E475K160 AC C4 0 47uE 25V X7R C2012X7R1E474K125 AA C6 47 uF 50V 10
26. List of Figures 3 10 Fuse and wiring of the mains in the prototype 28 3 11 Box containing mains circuitry 30 3 12 Box containing PCBs of the prototype 31 3 13 The ring buffer of the driver of the LED driver 36 3 14 The rotating animation of PowerSocket 38 3 15 Animation of the prototype 40 4 1 Chart of the result of the user study 44 B 1 Analogue front end of EM773 smartmeter 58 B 2 Power supply of EM773 smartmeter 59 B 3 The EM773 microcontroller with some pe TIPMOLY td hdc i eat is ze 60 B 4 The wireless transmitter of EM773 smartmeter 61 C 1 Schematic of the power supply of the prototype 65 C 2 Schematic of analogue front end of the pro LOLY PO Gnas es Oe ek eet eR eel Sete nds 66 C 3 Schematic of MCU and the periphery around it of the prototype o o o o ooo oo 67 C 4 The LED PCB of the prototype 68 D 1 Module diagram of the software architecture 74 E 1 Setting for flash magic 76 E 2 Setting for flash magic 76 List of Tables 2 1 el C 2 C3 C4 Fl E2 A comparison of strengths and weaknesses of both classes 2 22 22 20 The bill of material for the power supply of the proposed prototype The bill of material for the analog front end of the proposed prototype The bill of material for the circuit contain ing the MCU and some periphery of the pro posed prototype
27. PowerSocket Implementation of a Small Scale Power Consumption Measurement Platform Bachelor s Thesis at the Media Computing Group Prof Dr Jan Borchers gt Computer Science Department RWTH Aachen University Hi Pierre Schoonbrood Thesis advisor Prof Dr Jan Borchers Second examiner Prof Dr Ing Stefan Kowalewski Registration date June 26 2013 Submission date October 28 2013 I hereby declare that I have created this work completely on my own and used no other sources or tools than the ones listed and that I have marked any citations accordingly Hiermit versichere ich dass ich die vorliegende Arbeit selbst ndig verfasst und keine anderen als die angegebe nen Quellen und Hilfsmittel benutzt sowie Zitate kenntlich gemacht habe Aachen October2013 Pierre Schoonbrood iii Contents Abstract xi Acknowledgements xiii Conventions xv 1 Introduction 1 2 Related work 5 3 Creating a small scale version of PowerSocket 13 3 1 Hardware design 13 3 1 1 Existing prototypes or solutions used 14 3 1 2 Hardware design of the prototype 16 3 1 3 First version of the prototype 24 3 1 4 Benefits of the first version of the pro EPE Eur ade 32 3 2 Software design 2 2 2222 32 3 2 1 Software architecture 33 Contents 3 2 2 Programming and debugging the prototype o ooo o 3 2 3 Benefits of the software design Evaluation Summary a
28. ake experimenting with functionality easier the first version of the prototype is designed modularly This allows to exchange the visualization unit With the modularity and because the experimenter does not need to care about the measurement unit the first ver sion of the prototype can be regarded as prototype platform for energy consump tion visualization The second version of the prototype will be a minimized version of the first In this work the first version is introduced Lastly an user study is performed to asses the performance of the animation of PowerSocket In this study suitable energy consumption classes by using the just noticeable distance of the animation are tried to be found The result is a classi fication which give some insight what JNDs are suitable for certain energy usage classes Acknowledgements Firstly I would like to thank Prof Jan Borchers for hosting excellent lectures De signing Interactive Systems I Designing Interactive Systems II and Current Topics in HCI Those lectures inspired me to perform my bachelor thesis in this field Secondly I want to thank those who sponsored me some materials and crafting tools with which the exterior of the prototype is constructed In this context I would like to thank Jo Meens of Electro Meens for sponsoring me a fuse socket and move able ground fault interrupter Mainly I would like to give thanks to my dad for all of his tools and material I was allowed to us
29. alizes energy consumption of an appliance by light patterns in an electrical cord see figure 1 2 Powersocket Heller and Borchers 2011 mimics the movement of an electrical meter on an electrical outlet The T n r Kim et al 2009a visualizes energy consumption by using the Tree of T n r as a metaphor When energy is consumed the tree starts to loose leaves and ultimately ends up as a statue This statue is the tree of T n r in its present form as the original tree died The Energy Aware Clock Broms et al 2010 which is seen in figure 1 2 uses the dial of a clock to indicate the energy usage of the household at that time With low energy consumption the dial is short For higher energy consumption the length of the dial increases In Holmes 2007 a system is introduced that shows the number of oaks needed for neutralizing the carbon oxide emitted during the production of the energy that is used in one day by a building Figure 1 2 Left the Power Aware Cord shows a glowing pattern which indicates a certain energy consumption of an appliance Right the Energy Aware Clock is showing a pattern of energy usage The length of the dial represents the household energy usage at a certain point in time Both avoid using kWh or Watt as an indicator User studies have been conducted with the above proto types but there is no study with any of the prototypes which statistically proves that the introduced indicator for e
30. all scale cheap prototype based on a literature study of publications in which user studies with energy measuring prototypes which have a different visualization then displaying kWh or Watt only conventional energy measuring I will argue about how the user studies were conducted and what kind of problems the prototypes used during the studies had during the realization This results in criteria for prototypes to be used in a large scale user study In the second part I will argue about the advantages and disadvantages of appliance level metering and residential level metering I conclude the second part by reasoning about the choice I made which is to construct an appliance level meter In this first part I will start by giving a broad overview of how user studies with prototypes are conducted and what kind of results are obtained Considering the first aspect of the user studies I will discuss the size and length of the user studies The size of the user studies conducted by Gustafsson and Gyllensward 2005 Broms et al 2010 Heller and Borchers 2012 range from 9 participants to 15 participants The duration of the mentioned user studies range from a day to three months From the duration of the studies we can see that long term affects have not been studied and with the number of participants the strength of the results are questionable the sample size of the user studies were too small the length of the user studies we
31. ation containing information about the baud rate data bits stop bits and parity bits Once the UART is initialized the functions Ser WrStr tx_str to send and Ser_RdStr rx_str count stop_char to receive can be used The parameter tx_str is a pointer to a uint8_t This is the string to be sent The first parameter for receiving rx_st r is of the same type The data received is stored in the variable that the pointer points to Data is being read from the UART port until count 1 characters are received or the stop character which is defined by the parameter stop_char is received To control the pins of the MCU two modules are available The module pcb allocates a pin and sets its function Before any allocation is done it is advised to run pcb_init which sets all pins to unallocated When initialization is done the function pcb_alloc portpin func mode can be called to allocate a port and set its function The parameter portpin is the pin of the MCU to be allocated func defines the function to be set and mode passes the pullup resistor mode To free a pin of a port pcb_free portpin can be called The pin denoted by portpin is then unallocated The module gpio enables pins of the MCU to be used as GPI O pins Again before any pin can be used as GPI O pin an initialization function needs to be called this is gpio_init Ifa pin should be used as GPI O pin it first needs to be allocated as such This is done by gpio_alloc
32. ber 4 in fig ure 3 6 depicts the I O block of the microcontroller Here all ports of the MCU that are not in use to fulfil its basic func tions are combined The basic functions consist of energy measuring and being able to program and debug the MCU The I O block is the interface to connect the MCU with for example the LED PCB used in the prototype introduced in this work or a display This concludes the part in which the adjustments to the schematic of the EM773 smartmeter are made In this section I will discuss the design of the LED PCB that will show the rotation animation of PowerSocket Heller and Borchers 2011 The resulting schematic of the LED PCB circuit can be found in figure C 4 The heart of this PCB is the LED driver TLC5947 from Texas instruments This is a 24 channel constant current sink LED driver The datasheet of this part can be found in Appendix C 1 The animation of PowerSocket Heller and Borchers 2011 uses the colors green orange red This means a combined LED with a red and green LED suffices to show the colors of the animation As the LED driver is a constant current sink driver the combined LED needs a common anode I used the L59EGW CA LED from Kingbright The datasheet of the LED can be found in Appendix C 1 as well As the TLC5947 is a 24 channel driver and the LEDs re quire two ports per combined LED one for red and one for green the LED PCB has 12 LEDs In figure 3 7 this part is denoted by 1 I
33. can be calculated in advance In figure 3 13 we can see how this ring buffer works The ring buffer consists of 24 cells which contain the twelve bits that set the brightness level of the LEDs that are connected to the LED driver The write pointer marks the cell that is written last This means the write pointer moves in steps of twelve bits through the ring buffer The main reason for this is that the driver of the LED driver has the brightness of one LED as input The send pointer in contrast to the write pointer moves only one bit per step This is because the connection to the 3 2 Software design LED driver is a serial connection and therefore only one bit can be sent at a time The last bit sent to the LED driver is denoted by the send pointer To start using the driver of the LED driver it has to be initialized Which is done by setting the define UPDATES_PER_SECOND which sets the number of complete new configurations the LED driver can achieved per second and calling the function Initialise_LED_Driver This function first sets the pins of the MCU accordingly Four total pins are needed One pin provides a clock signal one pin the data signal another pin disables the LEDs and the last pin accepts a received configuration After the pin allocation is done the initialization function clears the ring buffer by filling it with zeros Afterwards a timer that fires an interrupt on two matches is configured One interrupt is used to
34. cure the PCB in place When a nut touches an electrical circuit a ring of perspex is used to isolate the nut from the electrical circuit On the bottom plate another nut secures the position of the bolt in a way that the bolt cannot move down any more As can be seen from figure 3 10 the LED PCB is not screwed This part of except for the LED the prototype should be easily replaceable since other an PCB which should imations requiring a different PCB for example a display be easily replaceable should be able to be tested as well This concludes the part of the first version of the hardware prototype I will now continue with some benefits of this version of the proto type which are of advantage for future prototypes which use appliance level energy measuring 32 the modularity of the prototype is its biggest advantage dangerous voltages are covered 3 Creating a small scale version of PowerSocket 3 1 4 Benefits of the first version of the prototype In this section I describe other purposes the first version of my prototype can be used for and what benefits it has One first benefit comes from the modularity of the first version of the prototype The visualization unit can be replaced and there are a lot of ports of the MCU yet unused see figure C 3 Therefore this prototype provides a platform to try out different animations for example using a display This can be done without worrying about the energy measure ment u
35. discope design and imple mentation of a fine grained power monitoring system for homes In Proceedings of the 11th international conference on Ubiquitous computing pages 245 254 ACM 2009b NXP Em773 URL http www nxp com products microcontrollers application_specific EM773FHN33 html Bibliography P3 international Kill a watt URL http www p3international com Danny Parker David Hoak and Jamie Cummings Pilot evaluation of energy savings from residential energy de mand feedback devices FSEC Rpt FSEC CR 1742 08 2008 Dane Petersen Jay Steele and Joe Wilkerson Wattbot a residential electricity monitoring and feedback system In Proceedings of the 27th international conference extended abstracts on Human factors in computing systems pages 2847 2852 ACM 2009 Plugwise URL http www plugwise com SmartLabs Smartlinc URL http www smarthome com Vasughi Sundramoorthy Qi Liu Grahame Cooper Nigel Linge and Joshua Cooper Dehems A user driven do mestic energy monitoring system In Internet of Things IOT 2010 pages 1 8 IEEE 2010 Markus Weiss and Dominique Guinard Increasing energy awareness through web enabled power outlets In Pro ceedings of the 9th International Conference on Mobile and Ubiquitous Multimedia page 20 ACM 2010 Markus Weiss Friedemann Mattern Tobias Graml Thorsten Staake and Elgar Fleisch Handy feedback Connecting smart meters with mobile phones In Proceed
36. e Of course I would like to thank all of those who advised and helped me which brings me to my last but not least acknowledgement I would like to thank my supervisor Florian Heller for supporting me I am especially grateful for his will ingness to help me even when no meeting was scheduled xvii Conventions Throughout this thesis we use the following conventions Text conventions Definitions of technical terms or short excursus are set off in coloured boxes Definition Excursus Source code and implementation symbols are written in typewriter style text myClass The whole thesis is written in American English Download links are set off in coloured boxes Chapter 1 Introduction Electricity is a ubiquitous source of energy A lot of people do not think about the consequences of using an electric device After a month or a year they get their electricity bill and are upset because they have to pay a lot of money Studies show that this indirect type of feedback is not sufficient to draw the peoples attention towards their energy consumption Abrahamse et al 2005 Fischer 2008 Although indirect feedback is not successful it is stated by Darby 2006 that by providing detailed and immediate feedback between 5 and 15 of a households energy consumption can be saved There are commercial products available for direct feedback for example the Kill A Watt by P3 international and the Wattson by DIY
37. e power supply which should be non isolated for operation Therefore adding the opto coupler is especially important for debugging the metering unit In figure 3 6 a breakdown of the circuit around the MCU can be seen The core of this part of the circuit is the EM773 microcontroller A user manual and datasheet of the EM773 can be found in Appendix C 1 Furthermore the power supply 1 in figure 3 6 and the clock source 2 in fig ure 3 6 of the MCU can be seen Although the EM773 has an internal oscillator I used an external clock source be cause the accuracy of an external oscillator is much higher This leads to more precision in timing functions of the MCU an opto coupler is added for debugging purposes an external clock source is used to increase accuracy 22 the MCU is programmed and debugged with the UART port a 24 channel constant current sink LED driver is used to control the LEDs combined red green LEDs are used for the LED circle 3 Creating a small scale version of PowerSocket and the energy measurement As already mentioned the MCU is debugged and programmed with the UART port The UART port isolated by an opto coupler can be seen in figure 3 6 denoted with number 3 1 In figure 3 6 a switch that is used to bring the MCU in programming mode is de noted by 3 2 Getting the MCU in programming mode is done by making a connection between ground and pin 1 of port 0 of the MCU and resetting the MCU Num
38. e a lot or not so much energy Both Darby 2006 and Fischer 2008 argue that to adept behaviour more efficiently the user needs to be aware of the energy usage of different appliances Furthermore ac cording to Kim et al 2009b direct feedback reveals the link between actions and their impacts Because the infor mation provided by appliance level feedback is direct and application specific both arguments of this paragraph sup ports that appliance level metering increases awareness In contrast Froehlich et al 2009 state that appliance specific breakdown is not possible with domestic level measuring This is caused by the arrangement of circuits in a home To get an idea of how much a specific appliance uses the user needs to observe the changes the appliance causes to the overall household consumption This requires additional effort from the user The main idea wherefore the prototype presented in this work is constructed is a large scale user study to show that by providing a different more clear way of showing en ergy consumption then conventional meters using Watt as primary indication contributes to more awareness of en ergy consumption The easiness of installation of appliance level measuring directly complies with the properties for prototypes used for large scale user studies Darby 2006 Fischer 2008 and Chetty et al 2008 argue for the im portance of awareness of energy usage of different appli ca
39. e front with the fuse and the socket for the application From the appliances plugged in into the socket the energy consumption will be displayed In the middle of figure 3 11 the top side of the box can be seen The connections to be PCBs of the prototype are visible namely the line L brown the neutral N blue and one black wire which connects the neutral N of the socket to the PCBs of the prototype On the right side of figure 3 11 we can see the back of the box of the prototype The plug of the prototype is situated on the back white plug with white cable Furthermore another electrical security measure can be seen in this part of the picture namely a ground fault interrupter black device with two buttons This device is coupled between the prototype and the wall socket It offers another layer of protection when a person touches a wire of the mains in the prototype When more then 10mA leak for example when a person touches a part of the circuit this device will terminate the mains to the prototype As we can see from figure 3 11 and know from the above part of own work wires and circuits on a PCB containing high voltages can still be touched by persons In the next part I introduce a box that encloses the PCBs of the prototype This second box is mounted on the box containing the mains circuit of the prototype see figure 3 12 on the left The cover of this second box is detachable therefore experimentation with the PCBs
40. ed and debugged without the circuit being exposed To get the prototype into programming mode the red button in figure 3 12 has to be pressed first and then the MCU has to be reset This is done by turning the fuse off and on To return to normal operation this process has to bre repeated 3 1 Hardware design 31 Figure 3 12 The box containing the PCBs of the prototype Left the box containing the PCBs with the UART port and programming button mounted on the box containing the circuit of the mains Center the top of the box with its cover with the windows for the animation Right the box without the cover the PCBs of the protoype are visible The second box takes responsibility for the mechanical safety as well I define mechanical safety as freedom from damage caused by moving the prototype When the pro the PCBs are totype is moved it can happen that parts are damaged or secured with bolts on connections get loose If certain connections for example the bottom plate of the ground of the power supply do not work any more the the second box prototype can suffer serious damage when put into oper ation eventually rendering it unusable Mechanical safety is attained by securing the PCBs with bolts The circuits of the prototype should not touch the bottom plate Therefore some distance to the bottom plate should be kept This is done by using longer bolts and one nut on which the PCB rests on and another one to se
41. ement Another two connec tions are needed for measurement These are the connec tion to the neutral N of the application 3 3 in figure 3 8 and the connection to the Neutral of the mains 3 2 in figure 3 8 which is ground if mains is connected The I O blocks for the MCU are also located on the main board 5 1 and 5 2 in figure 3 8 depicts these Lastly the analog front end 7 in figure 3 8 is also situated on the main board The analog front end has not been put on a separate PCB as this cir cuit is the EM773 smartmeter circuit except for some parts which are deleted Therefore it should be possible to be put to work without the need to replace a complete PCB The second PCB of the first version of the prototype is the adapter of the MCU 1 2 in figure 3 8 I used an adapter as I soldered the first version of the prototype completely by hand As mentioned before this was due to that 1 needed to experiment with the hardware functionality first since it was not completely clear if the changes made to the origi nal circuit of the EM773 would work The hand soldering led to some problems with the MCU due to the package of the MCU namely HVOFN33 which has no leads As I did not have the possibility to reflow solder at that time 1 used the bonding methodology to connect the MCU to an adapter The result can be seen in figure 3 9 on the left An other advantage of this adapter is that it allows to measure whether a signal arrives at a
42. environment of the prototype The fourth link contains a download link to the program with which I programmed the EM773 MCU using UART Flashmagic The last link provides a download link to a program with which I debugged my software using UART Terminal http www flashmagictool com download html amp d FlashMagic exe https sites google com site terminalbpp Terminal20130116 zip attredirects 0 Appendix B Reference schematic of EM773 smartmeter In this Appendix the reference schematic of the EM773 smartmeter is shown Figure B 1 depicts the analog front end of the EM773 smartmeter used by the metrology en gine of the EM773 microcontroller In figure B 2 the power supply of the EM773 can be seen Figure B 3 shows the EM773 microcontroller with all the periphery of the EM773 smartmeter attached This includes switches and the con trol of the relay In figure B 4 we can see the wireless trans mitter the EM773 smartmeters use to transmit data to a central hub 57 58 B Reference schematic of EM773 smartmeter SHIELDING SH 33n 3 2 LLOWGAIN 400V J HIGHGAJN 3V3 El N x OPAMPREF T T PMGD290XN 290 FID3_4 POLS Figure B 1 The analogue front end of the EM773 smartmeter 59 3V3 SAs7000 33D TEA1320T c18 220n The power supply of the EM773 smartmeter Figure B 2 B Reference schematic of EM773 smartmeter 60 Sena vena tayn 9
43. ep of the user study provides positive results and the second version of the pro totype is constructed then a large scale user study can be conducted that tries to prove that energy meters that avoid using Watt and kWh lead to more awareness then those that only display Watt or kWh Afterwards some experimenta tion can be done by extending the prototype with a wire less transmitter This would result an a combination of the appliance level and residential level metering class 53 Appendix A The development environment of the prototype In this section I provide links to the development environ ment I have used to program my prototype The first link is a link to the program I constructed my schematics and board layout of the prototype with Eagle light edition The second link contains the SDK of the NXP EM773 In this SDK drivers for the EM773 can be found as well as an example project in IAR embedded workbench Documen tation for the EM773 is in the SDK too The hardware de sign for the EM773 based smartmeter is in the SDK as well EM773 SDK The third link is a link to download the IDE with which the EM773 can be programmed IAR EMbedded Workbench kickstart edition tp ftp cadsoft de eagle program 6 5 eagle win 6 5 0 exe http www nxp com documents other EM773 SDK_Setup exe 55 Shttp supp iar com Common ProtectedDownload asp key 7WIBVKVG60 2C4681686 amp protocol HTTP 56 A The development
44. figure B 3 we can see that a lot of periphery has been at unneeded periphery tached to the MCU Some of the periphery can be removed attached to the MCU as it was not needed in the prototype introduced this work has been removed An overview of the removed parts can be seen in figure 3 5 The SDK of the EM773 smartmeter only offered the option to program and debug the EM773 microcontroller over a se rial interface UART As the JTAG interface is therefore not the JTAG interface in use it can be removed The JTAG interface is denoted has been removed by 1 in figure 3 5 None of the periphery used by the proto types in this work used the IC bus so the interface for the EC bus marked with 2 in figure 3 5 is removed as well As the I C bus as well mentioned before the relay that can switch the appliance attached to the meter on or off has been removed This 20 3 Creating a small scale version of PowerSocket APL CELTS of PODA 4 LOT ACT MAD A a 0 8 PO VELLO TTIABA LAN cu 7 ne Y PRO ae sof PIDO MAGIA Y LEO Ts poh DONA 13 20 ACLARA MATT TALA gt EMTTIPHMIS Figure 3 5 The removed parts from the circuit around the MCU of the EM773 smartmeter 1 JTAG debug and program interface 2 PC bus 3 1 Switch which can turn the appliance connected to the meter off 3 2 Ports to the relay and an LED that shows which channel for measurement is in use 4 Ports used to communicate with the wireless transmitter
45. g the prototype File ISP Options Tools Help SH ABEVR gt YA OS EM773 Erase block 0 0x000000 0x000FFF Erase block 1 0x001000 0x001FFF Flash Bant Erase block 2 0002000 0 002FFF Erase block 3 0x003000 0 003F FF COM Port COM 4 Erase block 4 0x004000 0x004FFF Erase block 5 0x005000 0xD05FFF V Erase all Flash Code Rd Prot Interface None ISP vr al Erase blocks used by Hex File Oscillator MHz 12 Step 3 Hex File Hex File C Users Pierre Dropbox Bachelor Arbeit Fresh build Debug Exe E Modified Friday October 11 2013 17 44 01 more info Verify after programming i dil ed _ Gen block checksums _ Execute _ Activate Flash Bank On Line training classes for microcontrollers and embedded networking and Internetworking www esacademy com fag classes Figure E 1 Screen shot of flash magic it shows the settings to use for programming the prototype COM Pot Baud rate Dala bis Pordy Slop bts Herdibakng CO cm 50 rs Cre Ga nore coms Sir C em Cum A 6 C odd RATISICIS ea 28800 128000 Cem 18 C ON OFF EM amo e o raso U mak n AISICISADNADIT gue CHR CHNO Caste FS C mace 2 CATS on TX IT men Semings Be cate TE eB gee fis cs ooo wm lop T Aosta Sept RT meen te Sega cos con Reece CLEAR Bet TOER HEX De BR Bn G BSCR Hex Sid og Sesto Bares Peigenen Figure E 2 Screen shot of terminal it shows the settings to be used f
46. ge see figure 3 2 Size and costs are properties already mentioned in the first part of chapter 2 that influence the ability of a prototype to be used during a large scale user study men tioned 15 to reduce costs the EM773 is used in contrast to a prototyping board the EM773 has an integrated metrology engine by using the EM773 the size of PCB is reduced 16 the prototype can be constructed on one PCB with hard wired connections this reduces the risk of loose connections The EM773 smartmeter comes with an extensive SDK the prototype introduced focuses on functionality 3 Creating a small scale version of PowerSocket The EM773 is a one chip SMD solution As such the com plete prototype can be constructed on one PCB which has advantages regarding prototype size All the connection on the PCB are hard wired this is planned in the second ver sion of the prototype in the first version the circuit is sep arated on several PCBs which eliminates the risk of loose connections A smaller risk of loose connections improves the reliability of the prototype As the complete prototype can be constructed on one PCB production of large quan tities is enabled The PCB can be made by external compa nies which construct the complete PCB This reduces con struction time and cost Another big advantage is that the EM773 comes with an extensive SDK this will be used in the software section of this work Withi
47. gnizable 38 in the middle of the range this would suggest to test the JND range of 80 to 90 in the first half of this range In the range 600 to 800 the JNDs 80 and 90 are barely recognizable recognized by 30 and 15 in the first half of the range This strengthens the assumption that they belong to the first part of the 400 600 range For the last part of the 400 600 range JNDs from above 90 to 110 should be tested These JNDs are slightly recognized witha recognition rate of 15 in the last part of the 600 800 range As aJND of 80 is already somewhat recognized with 30 in the beginning of the range 600 800 JNDs beginning from 100 should be tested in the first half of this range The JNDs 100 and 110 are tested in the middle part of the 800 to 1000 range They are barely recognized detected by 23 and 15 in this part of the range This would suggest that these JNDs are candidates for the last part of the 600 to 800 range 47 JNDs 45 and 50 belong to the first part of the range 200 to 300 and 50 to 55 to the last part of the range the JNDs belonging to the range 300 to 400 are 55 to 80 JNDs belonging to the 400 600 range are 80 to 90 for the first half and 90 to 110 for the second half 48 the JNDs 110 to 130 belong to the 800 to 1000 range the JND of range 1800 to 2000 is between 190 and 200 170 to 190 are fitting as JNDs for the range 1600 to 1800 the JNDs for the range 1400 to 1600 are 160 to 170
48. h connections are also provided by the main board as the main board itself is connected to the 3 3V wire and the ground of the power supply This con cludes the part about the four PCBs of the modular design The PCBs need a mechanical and an electrical safe casing This is what the next section is about I will introduce the casing and illustrate how I ensured mechanical and elec trical safety which are both needed as a small user study will be conducted with the first version of the prototype Furthermore electrical safety is an important factor when I experiment with the PCBs of the prototype The last PCB is the LED PCB which will show the animation the PCBs of the prototype need a casing for safety reasons 28 3 Creating a small scale version of PowerSocket Figure 3 10 Integration of the fuse in the circuit and the connections of the wires of the mains Brown Line L of the mains Blue Neutral N of the mains Green Ground of the mains a fuse is integrated to prevent serious damage after a malfunction of the prototype I will start with discussing electrical safety With electrical safety I mean the measures taken to avoid consequen tial damage when a part of the circuit malfunctions or causes damage to persons during operation Damage to persons during operations can be caused by touching a part of the circuit that is dangerous for example touching the power supply PCB on parts on which the mains
49. ich energy consumption metering can be done 2 Related work Figure 2 1 The idea of appliance level energy measuring the energy measuring device is put directly between the ap pliance and the energy outlet For conducting a large scale user study with a prototype we now have some properties for a prototype to be used during such studies We have seen that a prototype has to be able to be reproduced efficiently should not cost too much and should be functional reliable and not to obtru sive in its use The prototypes I studied did not comply to these criteria probably leading to problems when conduct ing a large scale user study When conducting a large scale user study to answer the question of increased awareness by using a metaphor for energy consumption one impor tant choice is on which scale the energy is measured In the next section I distinguish two classes of energy measuring devices analog to Weiss et al 2009 Froehlich et al 2009 and give some examples of commercial systems and proto types which belong to a respective class I will argue about the strengths and weaknesses of both classes and use this argumentation to strengthen my choice for the class I have chosen for my prototype There are two classes of energy measuring devices avail able on the market and used in research The first one is feedback on device level These systems measure the en ergy usage for every attached device The energy meter
50. in metering data from the metrology engine the function metrol ogy read_data metrology_result has to be used This function writes the metering data into the variable pointed at by the pointer parameter met rology _result if meter ing data is available The function returns whether reading metering data was successful or not This concludes the part in which I describe the modules I used from the EM773 SDK In the next part I will talk about the modules I have added myself To refresh the data from the metrology engine in a certain time interval and to make it available I added the module metering to the package METROLOGY The functionality of the refreshing is based on a timer The used timer is configured to count milliseconds An interrupt is fired and the counter resets when the counter reaches the value of the define METERING_INTERRUPT FREQUENCY The corresponding ISR reads out the data using the func tion metrology_read_data amp metering_result The metering data is then available in the variable metering_result The function Initialise Metering handles the configuration of the complete metering system This includes calling the functions metrology_init AHBC1kFrequency 35 the module metrology can be used to control the metrology engine init functions must first be called before use after starting the metrology engine its data can be read with a read method wrote a module that makes sure
51. ing devices are usually situated between the energy outlet and the appliance as seen in figure 2 1 Commercial sys Figure 2 2 The main idea of measuring the energy consumption of a complete household A sensor is integrated near the main fuse The usage data is then col lected and accessible over a terminal or mobile device tems already available in this class include Kill A Watt P3 international SmartLinc SmartLabs and Plugwise Plug wise Several prototypes have been constructed as well the first one is which include the Power Aware cord Gustafsson and Gyl appliance level lensw rd 2005 the T n re Kim et al 2009a ViridiS metering cope Kim et al 2009b Web enabled Power Outlets Weiss and Guinard 2010 and PowerSocket Heller and Borchers 2011 In the second class are the devices that measure the en ergy which a complete household consumes These sys tems usually measure the energy usage at the main fuse and communicate this to a central hub Usage data is then accessible at a central access point like a PC or a mobile device like a smartphone The general idea of this class is demonstrated in figure 2 2 Another more detailed op the second one is tion is to measure each circuit secured by a fuse individ household level ually Numerous commercial products are already avail metering able on the market in this class These include the Wattson DIF Kyoto Power Cost Monitor Blue Line Innovations
52. ing feasible energy usage classes and then check whether participants can classify devices accurately into these energy usage classes The first part of this study is presented in this work Although a precise classification of energy usage classes was not possible with the results of this study the study strongly hinted to the ranges of the JNDs possible for an energy usage class 5 2 Future work The user study presented in this work is divided into two steps Only the first of the steps has been completed In this first step interpolations had to be made of JNDs of energy consumption classes Although this first step gives already a good indication what the JNDs are this still needs to be confirmed in a next study In the second step there will be checked whether users can classify appliance accurately in the energy usage classes identified in step one In this sec ond step it would be beneficiary when the energy metering unit of the prototype introduced in this work would work Getting the energy metering unit to work would therefore be the next logical step after this work When the circuits of 5 2 Future work the prototype are fully functional then the prototype still has to be minimized The minimization would result in the second version of the prototype Although minimizing of the prototype with the PowerSocket animation would only be sensible when the second step of the user study has positive results When the second st
53. ion to the LED driver Therefore the program mer does not need to worry about it He only needs to 37 an init method should be called to initialize the driver 38 the input of the driver are the bits that define a brightness level of one LED the module animations makes sure exactly one animation runs at a time 3 Creating a small scale version of PowerSocket CITE Figure 3 14 The rotating animation of PowerSocket which change color based on the energy consumption Left low energy consumption Middle a medium level of energy is consumed Right high energy consumption fill the buffer with new data Filling the buffer is done by calling the function Write _LED_To Buffer ledbits This function needs twelve bit as input which define the brightness of one LED The bits passed in the parameter of the function are then put in the next available slot and the write pointer is moved one cell twelve bits further The function returns a write fail if the function is called if the buffer is full When the writing of the bits into the buffer succeeded the function returns a write success A package I completely wrote myself is the package that deals with the animations This package consists of two parts One module is the interface of the animations The other part of this package are the modules that define the animations themselves The module which is the interface to the animations is called animations It
54. irst of all the prototype must be able to be built in a large quantity For a result to be significant a large number of samples must be available In the case of a user study these samples are participants Depending on the type of the study all or a part of the participants need a prototype The first factor greatly influencing the ability to reproduce the prototype is the construction time of the prototype The person or team conducting the study usually cannot spend the time needed to construct the prototype themselves So an external company should mass construct these prototypes partly The cord used in Gustafsson and Gyl lensw rd 2005 is probably not suitable to be reproduced effectively If the production of the prototype can be done by an external company this will reduce the costs to construct the prototype Costs of the prototype is another important factor to consider when conducting a user study with use of a prototype The version of PowerSocket introduced in Heller and Borchers 2011 uses a Arduino prototyping board which in contrast to a integrated solution costs almost four times as much In Petersen et al 2009 the user needed a wireless network and an iPod touch or iPhone this would lead to rejection when users do not have them at hand Obviously a prototype should be usable when used during a study Usable in a sense that it is functional reliable and is not too obtrusive in its use If a prototype is n
55. is active A malfunction of a circuit often leads to a short circuit Measures to protect other devices when a device is short circuiting is integrating a fuse This is done in the first version of the prototype When the functionality is tested by using the first version of the prototype and the system works reliable the fuse will not be necessary in the second version of the prototype Figure 3 10 shows how the fuse is integrated into the circuit The fuse is situated directly behind the plug of the prototype and cuts of the line L when switched off When the fuse is turned off the complete prototype is switched off Therefore the complete circuit behind the plug of the prototype is secured with the 3 1 Hardware design fuse Afterwards the line L branches to the PCBs of the prototype and the socket of the prototype The PCBs of the prototype need the line L for measurement and as an energy supply To isolate the circuit of the mains in the prototype I constructed a box in which the fuse socket and plug of the prototype are integrated In figure 3 10 the circuit of the mains in the box of the prototype can also be seen As can be derived from the figure the neutral N is cut of by the PCBs of the prototype This is where the current mea surement of the appliance connected to the socket of the prototype is done Figure 3 11 shows the box containing the mains circuit of the prototype On the left side of figure 3 11 you can see th
56. m is men tioned and ceiling lights heating and ventilation systems some devices are not are given as example In contrast domestic level measure plugged into an ment measures the complete household the devices that outlet cannot be measured by appliance level measurement are measured as well household level With residential level energy measuring the users have energy metering a complete overview of the energy consumption of the provides a complete household This has the advantage of identifying the im overview pact of one device on the energy consumption of the house hold This enables users to identify the biggest consumers in their homes by relating the energy consumption of one energy usage of a device to that of the complete household This can con device can be related tribute to determine whether a device has a low or high to this overview consumption Appliance level measurement in contrast can only show the difference in energy usage between sin gle devices A complete overview of the energy consump tion of the complete household is hard or not possible to achieve Appliance level measuring therefore lacks the abil ity to determine the amount of power consumption by com paring it to a overall household consumption Regarding energy awareness this gives rise to the question How to make users aware if a device uses not so much or a lot of energy An important part of this question is when does a device us
57. makes sure that an animation is running In the first version of the prototype this is realized by defines However it is thinkable to extend the prototype with a button or switch which can be used to switch animations at runtime This can be useful for a prototype used during a user study The Energy Aware Clock Broms et al 2010 had such a system to change the animation The interrupt that causes the change of animation by button press will be handled in the animations module Furthermore this module is then responsible to change the animation accordingly 3 2 Software design The other part of the package ANIMATIONS are the ani mations themselves For the first version of the prototype I have decided to mimic the animation of PowerSocket Heller and Borchers 2011 The animation of PowerSocket can be seen in figure 3 14 It shows the consumption at low medium and high consumption levels The animation is a rotating animation which increases speed once the energy consumption increases Furthermore it changes the color at certain consumption thresholds from green to orange to red to indicate whether low medium or high energy consumption is at hand In my implementation the update rate remains constant I define the update rate as the rate a complete configuration is activated per second at the LED driver In other words the update rate is the rate of which the brightness of all LEDs change per second I have achieved the speed
58. mation of Power Socket Heller and Borchers 2011 is used as a reference for the animation of this prototype I will start with describing what part of the software of the EM773 smartmeter I use This can be seen in figure D 1 which is a module diagram The modules that have been painted red belong to the EM773 smartmeter SDK The EM773 SDK can be found in Appendix A The modules from the EM773 SDK that I have used includes the system files for the EM773 The system files are contained in the packages CMSIS and SYSTEM CMSIS is a package that contains general defines and operations for a cortex MO core In SYSTEM the interrupt vector and EM773 specific defines are contained Using the system_EM7xxx module the clock of the MCU can be configured The most prominent properties of the clock are its source and its frequency As I have used an external oscillator this has to be set in a define in the system_EM7xxx module The clock frequency can be configured in the app config module which is situated in the APP package These clock settings can then be put to work by calling the SystemInit function of the system_EM773 module This sets up the basics of the MCU As we want some devices attached and to be able to use the ports of the MCU it would be convenient to have drivers for them The devices used for the prototype are the ports of the MCU UART and the LED driver The drivers for these devices are situated in the package DRIVERS As
59. me jumps in JNDs can be noted after a change in color these happened at 1000 and 1600 this would indicate that the change in color influenced the result This suggests disabling a color change for the next user study which should confirm the interpolations I made These interpolations can be used to construct energy consumption classes The span of these classes should be longer then the JND of the range the class is in This en sures that the class is recognizable 49 using the JNDs energy consumption classes can be constructed Chapter 5 Summary and future work 5 1 Summary and contributions We have seen that there were some approaches to elimi nate the Watt indicator from energy measurement devices What never have been addressed is the question whether these different approaches truly lead to more awareness of energy consumption in contrast to conventional energy measuring devices using Watt or kWh as an indicator In this work I argued that a large scale user study to address this question of awareness would have been hard with the prototypes introduced in those works that ipresent an al ternative visualization From this argumentation some properties for prototypes used during large scale user stud ies are discovered When a prototype encompasses these properties it will be more suited to conduct a large scale user study with it In this work I have proposed a proto type that tries to encompass these properties
60. n conclude that a JND between 170 and 190 is most fitting for the range 1600 1800 As 160 and 170 are recognized 23 and 54 in the last part of the range 1600 1800 these JND are candidates for the range 1400 to 1600 This is furthermore strengthened as 150 still seems to be on the low end for this range Although 150 should be recognizable at the beginning of the range 1400 to 1600 This makes a JND between 150 and 160 candidate for the higher part of the range 1200 to 1400 For the lower part a JND between 140 and 150 should be recognizable as 130 and 140 are already somewhat recognizable detection rate of 15 in the higher part of the 1200 to 1400 range and 120 is recognizable 38 although not very good in the first part of the range For the range 1000 to 1200 a JND be tween 130 and 150 should be tested This is in accordance with the results as 130 recognized by 23 is still too low for the higher part of this range and 110 recognized by 23 as well is too low for the lower part This concludes the analysis of the test result of the first step of the user study Although the results already hint at some JNDs for ranges the JND of some ranges can only be coarse determined By interpolating a better insight in possible JNDs for a range can be obtained These interpolations needs to be checked in a further user study During the user study al most all participants found the change of color in the ani mation distracting As so
61. n of PowerSocket By extending the prototype with a wireless unit a combi nation between the two classes mentioned in chapter 2 can be made The unit on the socket displays the current en ergy usage the data from the wireless unit and from the main fuse can be used to make a detailed breakdown of the energy usage of the devices Chapter 4 Evaluation In this section I will describe a user study that tries to asses the performance of the animation of PowerSocket With performance is meant how well a user is able to estimate energy consumption based on the animation This user study was intended to be conducted in two steps In the first step the just notable difference IND of several ranges of the prototype will be assessed In the second step I will create energy consumption classes based on the results in step one In the user study associated with the second step the user then has to classify in which class a certain energy consumption belongs to In this way the accuracy of the energy usage classes and the animation will be measured In this work only the result of the first step is presented As can be seen in chapter 3 2 1 the RPM of the animation cor relates linearly to the energy consumption in Watt There fore the energy consumption in Watt will be used to indi cate the speed of the animation of the prototype In the first step a broad classification of energy consump tion had to be made This was done by buildi
62. n this SDK a documentation for the EM773 microcontroller can be found as well as the refer ence schematic of the EM773 smartmeter pictured in figure 3 1 in the middle I took this reference schematic as a basis for the design of the circuit of the prototype presented in this work 3 1 2 Hardware design of the prototype First of all the prototype is designed in two iterations The first iteration is focused on getting the hardware and software to work This is necessary because the reference schematic provided by the SDK from the EM773 smart meter has to be modified The second iteration then focuses on getting the prototype as small as possible so that it even tually fits in the exterior of the night lamp from IKEA seen in figure 3 1 on the right This has the benefit that I can first focus on functionality and reliability of the prototype which as mentioned in chapter 2 are both properties a pro totype should have for conducting large scale user studies with it During this phase I mainly use tools available in the FabLab Aachen http www nxp com documents other EM773_SDK_ Setup exe 3 1 Hardware design 17 C Figure 3 3 The breakdown of the power supply of the prototype 1 SMPS circuit this part transforms 230VAC down to 12VDC 2 Linear regulator 12VDC is regulated to 3 3VDC 3 Input and output section of the power supply In the first step I analyzed which components of the refer
63. n this figure we can further see which port of the LED driver corresponds to which LED and which color of the LED The colors of a combined LED are cou pled on a consecutive port of the LED driver This makes programming the driver easier How the LED driver is pro grammed will be further explained in the software section Number 2 of figure 3 7 shows the connections to the rest of the PCB of the prototype This consists of the connection 3 1 Hardware design w J i Figure 3 7 The breakdown of the LED PCB 1 Twelve LEDs of the LED PCB 2 Connection to the MCU and the power supply 3 Power supply of the LED driver 4 Resistors to configure the constant current of the LED driver of the LED driver to the MCU and the connection to the power supply How the LED driver is coupled to the con nections to the power supply is denoted by 3 in figure 3 7 A capacitor is used to stabilize the power supply to the LED driver In figure 3 7 number 4 shows how the current of the LED driver is configured There are several options possi ble for the constant current value possible therefore a con figuration is needed The configuration consists of putting a certain amount of resistance on the IREF port of the LED driver The exact resistance value for a constant current can be looked up in the datasheet from the LED driver avail able in Appendix C 1 This concludes the broad overview of the design of the circuits of the
64. nd future work 5 1 Summary and contributions 52 Future work The development environment of the prototype Reference schematic of EM773 smartmeter Schematics of the prototype Gil Schematics 222 dee Sa ers C 2 Billofmaterial C 2 1 Powersupply C 2 2 Mainboard E23 Led PEB scsi 228 Software architecture of the prototype Programming and debugging the prototype Results of the first step of the user study Bibliography 41 43 51 51 52 55 57 63 63 69 69 70 71 73 75 77 79 Contents Index 83 vii List of Figures 1 1 1 2 2 1 2 2 3 1 3 2 3 3 3 4 3 5 3 6 3 7 3 8 3 9 The Reichelt on socket meter The Power Aware Cord and Energy Aware Clock rea hanana a eed wat belt Appliance level energy measuring Domestic level energy measuring Used prototypes and commercial products The Arduino Duemilanove Breakdown of power supply The deleted parts from the EM773 smart meter analogue frontend The removed components around the micro controller of the EM773 smartmeter Breakdown of the circuit around the MCU Breakdown of the LED PCB of the prototype Modular arrangement of the first prototype MCU bonded to an adapter ix 14 15 17 19 20 21 23 26
65. nergy usage increases the awareness of energy consump tion Taking a closer look at the prototypes which are used during user studies some flaws can be discovered in the design of the prototype These flaws could make a large scale user study needed to prove increased energy aware ness much harder In this work I propose a prototype for conducting a large scale user study to address the question if a different indicator for energy consumption in contrast to the conventional indicators kWh and Watt leads to more energy awareness I will start to do a review of studies conducted with prototypes that use an alternative indicator for visualizing energy consumption Then I will hint to some flaws of these prototypes regarding the usefulness during large scale user studies From these flaws some properties for a prototype used during large scale user studies can be constructed From these properties I then choose whether I use a measuring device which measures only one appliance or the complete household Afterwards I introduce the prototype itself The hardware and software design will be discussed Afterwards a user test assessing the performance of the chosen animation is discussed there is no user study which proves increased awareness by using such prototypes will introduce a prototype that should enable such a study 1 Introduction Chapter 2 Related work In this section I will first give a motivation why I built a sm
66. ng classes with a span of 200 Watt through the complete range 0 Watt to 2200 Watt and then broadly determining the JND of these classes At 2200 Watt most household appliances do not use more than 2200 Watt the animation reaches its highest speed therefore no change in speed will be notice able for values beyond 2200 Watt At 2200 Watt the anima tion has a speed of 300 RPM and at 1 Watt starting thresh 43 a user study is conducted to find the just noticeable difference of ranges in the animation first a broad classification based on a pilot study was made 4 Evaluation 44 sassepo uoydummsuod 313u9 y Ul AN P3489 9y JO IPLI UONP P P AYL T F IMBI pp 7 i i F old the animation speed is 0 15 RPM When the JND of a consecutive class was much higher more then double the former class is splitted into two classes and the JND of the higher range class is determined The broad classification with their JN Ds is displayed in table F 1 From this broad classification the user study of the first step could be designed In the user study presented in this work a sequence of animations running at a certain speed is shown The sequence is randomly programmed but changes speed accordingly to JNDs of all the classes of the broad classification The sequence is random in a sense that JNDs of a certain class are rarely tested con secutively and the class from which the JNDs are tested is randomi
67. nit Because the power supply is on a separate PCB the design of this PCB can easily be altered without influ encing the rest of the circuit too much When a display for an animation uses 5V this wire can easily be added to the design Another advantage of the first version of the pro totype is that most parts of the circuit on which dangerous voltages can be found are covered and do not need to be altered except for the power supply which might need to be extended or replaced The biggest advantage is the use of SMD parts This enabled to create a small low cost PCB of the complete circuit of the prototype The small form fac tor and the low price are both properties that are of benefit for large scale user studies This was the last section of the hardware design in the next section I will talk about the software design 3 2 Software design In this part of own work I will describe the software design of the prototype The software between the two intended versions of the prototype does not vary In this section 1 first describe the software architecture and what it is based on Afterwards I explain how to configure programs to pro gram and debug the prototype This includes how to set up the programming environment Lastly I will indicate some advantages this software architecture has 3 2 Software design 3 2 1 Software architecture As basis for the software architecture the software on the EM773 smartmeter is used and the ani
68. number s Required Part Placed part C19 C20 12pF 50V 2 06035J120GBSTR C21 C22 220nF 10V 10 0603ZC224KAT2A R26 R27 3309 0 0625Watt 0 1 CPF0603B330RE L3 L4 6002 500mA BLM18AG601SN1D Q1 12Mhz quartz NX3225SA IC2 on adapter Energy metering Cortex MO EM773FHN33 551 IC5 opto coupler UART ADUM1281ARZ Table C 3 The bill of material for the circuit containing the MCU and some periph ery of the proposed prototype C 2 Bill of material 71 C 2 3 Led PCB Component number s Required Part Placed part Cl 2207F 10V 10 0603ZC224KAT2A R1 R48 Together 3 6KQ LED1 12 LED green red common Anode L59EGW CA IC1 PWM LED driver TLC5947 Table C 4 The bill of material for the circuit of the LED PCB of the proposed pro totype 73 Appendix D Software architecture of the prototype In this Appendix the module diagram of the software ar chitecture can be seen Itis depicted in figure D 1 74 D Software architecture of the prototype Figure D 1 The module diagram of the software architecture An arrow in this diagram means a module is used by another module The arrow points to the module that uses the module A green module means that this module is provided by the EM773 smartmeter SDK The module painted red are fully constructed by myself Appendix E Programming and debugging the prototype 76 E Programming and debuggin
69. opening the work space file in the root folder of the project This file is Energy meter eww For debugging Terminal is used A screen shot containing the setting to configure ter minal to communicate with the prototype can be found in figure E 2 ReScan can be used to discover the COM UART ports of the computer This COM port should correspond to the COM port assigned to the UART cable Before anything can be received the connect button needs to be pressed 3 2 3 Benefits of the software design Like during the hardware design the programmer does not as drivers are need to worry about the metering part as the metering re already available sults that become available in the metering module and get only an application in updated automatically can be used Hence the program most of the cases mer only has to worry about the app in this case usually needs to be written an animation the programmer wants to program Further more as the hardware design is modular the prototype can be extended by a wireless unit The design of the PCB for a wireless unit is already available from the EM773 SDK The driver for the wireless unit is available in the SDK too http www flashmagictool com download html amp d FlashMagic exe Shttp supp iar com Common ProtectedDownload asp key 7WIBVKVG60 2C4681686 amp protocol HTTP https sites google com site terminalbpp Terminal20130116 zip attredirects 0 42 3 Creating a small scale versio
70. or communi cating with the prototype for debugging purposes Appendix F Results of the first step of the user study Range Watt JND 0 25 2 5 25 50 5 10 50 100 15 20 100 200 20 30 200 300 40 45 300 400 55 65 400 600 70 80 1400 1600 140 600 800 90 100 1600 1800 140 160 800 1000 100 110 1800 2000 190 200 1000 1200 120 2000 2200 200 1200 1400 120 130 77 Table F 1 The coarse energy consumption classes I have found in initial testing F Results of the first step of the user study Table F 2 represents the results of the first part of the user study On the left you see the ranges from which a JND is tried to be found On top you see the JNDs tested in this user study The entries in this table represent how often users recognised a JND in a certain range Range JND 2 3 5 10 15 20 25 30 35 40 45 50 0 25 1 1 25 50 117 14 50 100 4 716 100 200 215 4 2 2 200 300 4 12 2 172 Range JND 50 55 60 65 70 80 90 100 110 120 300 400 6 5 1 3 2 400 600 a 600 800 4 2 2 2 800 1000 1 3 2 1 Range JND 110 120 130 140 150 160 170 180 190 200 1000 1200 3 2 3 1200 1400 5 2 2 1400 1600 5 5 1600 1800 6 7 6 3 7 1800
71. ot reliable for example the Power Aware Cord Gustafsson and Gyllensw rd 2005 had loose contacts during the user study the users will have a hard time using the prototype This can result in that the users will not use the prototype at all This is also true for the size of an prototype for an on outlet power measuring device the dimensions of PowerSocket Heller and Borchers 2011 was large A prototype which is too big often prototypes are bigger then is needed because this makes prototyping easier probably leads to a higher usage barrier Nobody probably would like a box standing in the room for a while just to measure energy consumption Furthermore depending on the location of the wall socket a big box can get in the way of other appliances Another aspect of obtrusiveness is how easy is a prototype installable which is mentioned by Kim et al 2009b too The Energy Aware Clock Broms et al 2010 obtains the energy usage data from the main fuse When such a system needs to be installed in an older house then some alterations need to be made to the circuit around the main fuse Alternations to existing circuits should be avoided as it brings in additional risk and possible rejection of the prototype by the potential user the costs of the prototype should not be too high a prototype should be reliable it should not be bigger than needed the prototype should be easily installable there are two scales on wh
72. other linear regulator from the analog front end the shield is removed as well as a relay that could switch off the appliance 3 Creating a small scale version of PowerSocket that can be drawn from the SMPS circuit is about 350mA and the voltage is about 12VDC The second part denoted with 2 in figure 3 3 transforms 12VDC to the voltage that is needed by the circuit of the prototype In the case of the prototype introduced in this work this is 3 3V This trans formation is done by using a linear regulator that is cou pled on the output of the circuit of the SMPS circuit When another voltage is needed for a prototype the power sup ply can easily be extended by coupling another linear regu lator on the output of the SMPS circuit on which the linear regulator depicted with 2 in figure 3 3 is attached The third part in figure 3 3 denoted by 3 is the input output part of the power supply The input is the neutral N and the line L of the mains Output is the ground and the out put of the linear regulator in case of the prototype pro posed here 3 3V When as mentioned earlier another lin ear regulator is coupled for an additional voltage one more output is necessary In the analog front end of the EM773 smartmeter which can be seen in figure B 1 some parts can be found that are not needed in the prototype The parts that I deem superfluous can be seen in figure 3 4 The shield around the operational amplifier of the
73. pe is an imitation of the animation of PowerSocket in every iteration of the main loop of the animation the RPM is determined depending on the current energy consumption in Watt furthermore the color of the animation changes depending on energy consumption thresholds 40 the rotating animation can be configured by setting defines by calling the Meter ing Animation_Rotation function the animation gets in its main loop 3 Creating a small scale version of PowerSocket Figure 3 15 The animation of the prototype mimicking the animation of PowerSocket shown in figure 3 14 The defines used in this formula are from the mod ule led driver This means a phase is defined as e De nn Afterwards the brightness level of every LED used in the animation can be calculated for a phase Figure 3 15 shows the result of the prototype mimicking the animation of PowerSocket shown in figure 3 14 Before using the rotating animation it can be configured This is done by setting certain defines MINRPM and MAXRMP determine the RPM range by setting its mini mum and maximum value The define POWEROFFSET determines from which energy usage the animation starts This define can be used to make sure no animation is shown when no appliance is plugged in The defines POWERMEDIUMTRESHOLD and POWERHIGHTRESHOLD can be used to determine on which energy usage level the animation switches its color from green to orange and from o
74. portpin mode dir The pin to be allocated is defined by portpin mode sets the pullup resistor mode and dir sets the pin to be either an input or output By calling gpio chdir portpin dir the function of the pin denoted by portpin can either be switched to input or an output If a pin is an output gpio_set portpin can be used to set the pin to a logical one and gpio_clr portpin sets it to a logical zero The function gpio_getval portpin can be used to read the value at the pin expressed by portpin When 3 2 Software design a GPI O pin is not needed anymore it can be unallocated with the function gpio_free portpin To make use of the metrology engine of the EM773 the EM773 SDK contains a module that controls this engine Before using the metrology engine it first needs to be configured This is done by calling metrology_init AHBC1kFrequency Fmains which needs the clock frequency AHBC1kFrequency and the frequency of the mains Fmains If Fmains is set to EM_AUTO the mains frequency is detected auto matically and returned as the result of the initialization function The second step of the configuration is to call met rology_set_ranges ranges How the parameter ranges is used is extensively explained in the user manual of the EM773 Before any metering data becomes available the metrology engine has to be started This is done with metrology_start To stop the metering metrology stop can be called To obta
75. range to red POWERMAXTRESHOLD defines the energy usage level on which the animation shows the highest consumption More energy usage than defined by this threshold will not result in a change of the animation The animation is started after configuration by calling Metering_Animation_Rotation which should be done in the module animations This function starts the main method of the animation In this main method the RPM phases and brightness of the LEDs are calculated as described in the previous paragraph When the brightness of A LED is calculated it is send to the driver of the LED driver by the main method of the animation With the description of the animation the end of the software 3 2 Software design 41 architecture part of this work is reached 3 2 2 Programming and debugging the prototype I will now give a short guide on how to program and de bug the prototype First of all an UART cable has to be connected to the prototype and the computer which will program the prototype The ground of the UART cable is mounted on the left side of the UART port of the proto type visible in figure 3 12 on the left Afterwards Flash Magic has to be downloaded and run Figure E 1 shows a screen shot on how to configure flash magic for program ming the prototype In step 3 a path to a hex file should be defined The hex file is found in project debug exe Open ing the program in IAR EMbedded Workbench is done by
76. re too short proving the increased awareness has been avoided a reason could be the prototype a lot of replicas are needed a prototype should be able to be built fast 2 Related work During the conducted user studies the question Does this prototype lead too more energy awareness then conventional systems has never been addressed In Gustafsson and Gyllensw rd 2005 Heller and Borchers 2011 and Petersen et al 2009 animations and or interfaces are tested on how appealing they are to the user and how well they are understood In the study from Heller and Borchers 2012 the prototype is passed for a week to users to experiment with it In Holmes 2007 Kim et al 2009a and Weiss et al 2009 prototypes are constructed but no user study has been done at all Broms et al 2010 draw a similar conclusion as in Darby 2001 and Darby 2006 that providing direct feedback leads to an increased awareness of energy consumption Apparently the conducted user studies were too small and too short and did not address the issue that the proposed concept leads to an improved awareness in contrast to conventional systems It is interesting why this issue has never been addressed accordingly One of the reasons could be how the prototype has been built In the next section I will describe problems with prototypes that are constructed and what kind of prototype is needed to perform a large scale user study F
77. reference board the STEVAL ISA035V1 is used To get the required 3 3V a linear regulator is used coupled on the output of the SMPS circuit The linear regulator used is the LD1117S33CTR The bill of material for the design of this power supply can be found in Table C 1 In Figure C 2 the schematic of the analog front end of the prototype presented in this work can be seen Table C 2 shows the bill of materials used to construct this part of the circuit In figure C 3 the schematic with the MCU EM773 and some periphery around it can be seen The EM773 microcon troller used in the schematic has a user manual and a http www st com web en resource technical document datasheet CD00087939 pdf http www st com web en resource technical document application_ note CD00159053 pdf Shttp www st com st web ui static active en resource technical document datasheet CD00000544 pdf http www nxp com documents user_manual UM10415 pdf 64 C Schematics of the prototype datasheet Table C 3 shows the bill of materials for these parts of the circuit Figure C 4 shows the LED PCB The main part of this board is the LED driver TLC5947 from Texas Instruments I have used the L59EGW CA LED from Kingbright for the LED circle of the LED PCB 5 http www nxp com documents data_sheet EM773 pdf Shttp www ti com lit ds symlink tlc5947 pdf http www kingbright com manager upload pdf 1366958416 L 59
78. rwth aachen de fablab Corinna Fischer Feedback on household electricity con sumption a tool for saving energy Energy efficiency 1 1 79 104 2008 Jon Froehlich Kate Everitt James Fogarty Shwetak Patel and James Landay Sensing opportunities for person alized feedback technology to reduce consumption In Proc CHI Workshop on Defining the Role of HCI in the Chal lenge of Sustainability 2009 Anton Gustafsson and Magnus Gyllensw rd The power aware cord energy awareness through ambient informa tion display In CHI 05 extended abstracts on Human factors in computing systems pages 1423 1426 ACM 2005 Florian Heller and Jan Borchers Powersocket towards on outlet power consumption visualization In CHI 11 Ex tended Abstracts on Human Factors in Computing Systems pages 1981 1986 ACM 2011 Florian Heller and Jan Borchers Physical prototyping of an on outlet power consumption display interactions 19 1 14 17 2012 Tiffany Grace Holmes Eco visualization combining art and technology to reduce energy consumption In Pro ceedings of the 6th ACM SIGCHI conference on Creativity s cognition pages 153 162 ACM 2007 Ju Whan Kim Yun Kyung Kim and Tek Jin Nam The t n r design for supporting energy conservation be haviors In CHI 09 Extended Abstracts on Human Factors in Computing Systems pages 2643 2646 ACM 2009a Younghun Kim Thomas Schmid Zainul M Charbiwala and Mani B Srivastava Viri
79. the prototype the opto coupler and the jumpers will be re moved This removes the possibility to debug the proto type when mains is coupled The possibility to experiment with the functionality of the hardware is mainly provided by the modular arrangement of the first version of the prototype This modular arrange ment was made because if an error occurred in a part of the circuit then only that part for example the power supply needs to be reconstructed The modules of this arrange ment can be seen in figure 3 8 The first version of the prototype consists of four PCBs The first one is the main board 1 1 in figure 3 8 which connects all the other PCBs First of all the main board contains the socket for the adapter of the MCU 2 in figure 3 8 Fur thermore the programming and debug interface parts de noted by 6 in figure 3 8 is located on the main board This includes connections for the button to put the MCU into programming mode 6 3 in figure 3 8 and the jumpers to enable programming the MCU with power over the UART port 6 2 in figure 3 8 This results in two options to power the prototype The first one is to use the power of the pro gramming device via the UART port 6 in figure 3 8 The second option is to connect a power supply PCB to the main board using the socket for a power supply 4 in figure 3 8 3 1 in figure 3 8 is a connection to the Line L of the mains 3 1 Hardware design which is needed for measur
80. tions This would suggest that it is sufficient to increase the awareness per device to increase the awareness overall As the awareness per device is a strength of appliance level measuring as well as the easy of installation I have chosen to construct an appliance level meter In the next section 1 will introduce a prototype that tries to encompass all of the above mentioned properties for a large scale user study 11 appliance level metering lacks an overview to which consumption can be related to awareness of energy usage of different appliances is important household level metering cannot be application specific appliance level metering shows energy consumption of different appliances and is easily installable have chosen to construct an appliance level meter Chapter 3 Creating a small scale version of PowerSocket In this section 1 will introduce a prototype which has the purpose to be used during a large scale user study As men tioned before I will therefore try to encompass the proper ties mentioned in chapter 2 into the prototype I will start with illustrating the hardware design In the following the software architecture will be explained in detail 3 1 Hardware design In this section the hardware design is explained I will start with clarifying why I used existing prototypes and solu tions and explain the prototyping phases I had in mind Af terwards the broad hardware design of the circ
81. uits of the prototype and the first version of the prototype will be in troduced The second version the minimalistic prototype is not introduced in this work as I was not able to construct it before this work was written I will conclude this part by bringing up other possibilities the introduced prototype has 13 14 3 Creating a small scale version of PowerSocket Figure 3 1 The prototypes and products used to construct the prototype Left the Powersocket prototype in its present state a rather large prototype Center the EM773 smartmeter from NXP This device has no power indication Right an IKEA night light It is very small and has a transparent ring around the socket PowerSocket the EM773 smartmeter and a night light from IKEA are used the circuits of the prototype are based on the EM773 smartmeter PowerSocket is too large ideally the prototype fit into the IKEA night lamps 3 1 1 Existing prototypes or solutions used The initial idea was to combine the animation of the Pow erSocket prototype 3 1 left with the circuits of the smart meter prototype from NXP based on the EM773 figure 3 1 center and a night lamp from IKEA figure 3 1 on the right Powersocket serves as the basis for the anima tion of the power consumption a rotating animation which changes speed and color depending on the energy con sumption Another aspect of PowerSocket influences the prototype I constructed as well
82. used by the JNDs which are tested in the middle of the range and the JND of 10 on the end of the range To examine the JND of this range more pre cise it is advised to split the range As a JND from 2 10 will probably be recognized in a range from 0 to 10 Regarding the range from 25 to 50 we can see from the results that the JND is probably between 10 and 15 According to the results the JND of the range 50 to 100 is between 15 and 25 In this case 15 and 20 are values in the start of the range and 25 is a value on the end of the range This would sug gest that the JND of the range 50 100 is about 15 to 20 and from the range 75 100 about 20 to 25 These JNDs needs to be verified in a next user study In the ranges from 100 to 200 and 200 to 300 the tested JNDs are probably too low for the area for which they are tested as the detection rate for the JNDs of both ranges is not higher than 38 The JNDs 20 to 25 of range 100 to 200 is tested in the middle of the range The JNDs of its prede cessor class rather suggest that the JND is above 25 for the start of the range 100 to 200 The JNDs 30 to 35 are tested in the last part of this range which would explain why they are missed rather often they only have a detection rate of 15 For a next user study the JNDs 25 to 35 should be tested in the start of the range In the range 200 300 the JNDs 35 and 40 are tested in the first half of the range They are hardly recognizable 30 and 1
83. zed Speed changes that do not contain any useful speed differences are integrated as well to reduce the pre dictability Furthermore the intervals between the speed differences of the sequence are randomized as well This is done to cancel out a learning effect To furthermore make the user less aware of an incoming change the sequence is hard coded on the MCU As soon as the observer initiates a speed change this could hint at an oncoming change By the time the user observes a speed change he has to tell the observer whether it was a speed increase or decrease The observer is acknowledged of the speed change by the MCU as it sends a message that the speed of the animation has changed to the computer of the observer Furthermore the number of the actual speed change is send to the observer This information is not visible to the user The observer has to make notes whether the user recognized the speed changes correctly This is done on a form that should not be visible for the user Otherwise this could influence the user as he can see whether he or she already missed some speed changes To cancel a learning effect a user gets the sequence in its original order or backwards It will be di vided such that user n gets the original sequence and user n 1 the sequence backwards and so forth This way the original order and the backwards order is altered for con secutive users This is done at runtime by pressing the red program button seen

PowerSocket - The Media Computing Group

Contents

Download Pdf Manuals

Related Search

Related Contents