Dave has changed a few things in the Rev B schematic of his power supply design. Talk on opamp specifications, I2C I/O expansion and AVR microcontroller pin usability.
I first watched this build series back in 2013, and at least once a year, I come back and watch the entire series again. Why? When I feel like I am in a slump, this series reminds me why I started doing this. Attention to detail, on the fly problem solving, and pushing the limits of what can be done on a budget. Still love this series on the PSU. Happy Thanksgiving 2018 from the U.S.A..
I want to thank you very much for encouraging us, teaching us, give us what you know. The words of gratitude that we use in this way are not enough to thank you a thousand thanks¡¡¡¡¡
I don't get why you want to bitbang SPI rather than use the hardware SPI port. The chip specifically handles sharing SPI and ISP because ISP is only active when the chip is in RESET mode. It's quite safe to mix the two on the same pins.
Very nice revisions and sheet layouts Dave! I think this is a good project example because you're interfacing I2C, using PWM etc. and yes, you could use a higher pin-count MCU or a PIC or Vortex 32 but it wouldn't be something that your average electronics hobbyist would even consider learning a lower level language nor having to buy programming hardware to modify and experiment with. Also I think it's keeping more of the 'vibe' with open source. Nice work again, and thanks for the detail!
@logicmindtube Yes, it will be time lapse of some sort, as I have several hours of layout footage captured that I plan to do a voice-over too. No idea how well that's going to turn out though...
12 ปีที่แล้ว +1
I love it how you use a screwdriver as a pointer! How engineering of you! (but it still gives me the creeks when it scratches the paper...)
@keenantims Yes, you are right, with hindsight I should have used the bigger device. I missed it. I had used the 8bit one before so was kinda caught up on that and stopped searching. Damn. The hardware SPI is left free so an Ethernet shield or some other external serial device can use it.
The image in the TLC272 datasheet @22:51 is a copy of the photomask for the metal on the IC (black=metal). It looks like this chip only has one layer of metal. Metal connects to the sources and drains of the transistors; polysilicon is used for the transistor gates and short connections underneath the metal. The pads with notched corners in the LL and LR corners are not bonded out, apparently. (They're probably only used at wafer test--"multiprobe" in TI jargon). Pin 4 (GND) is bonded to the pad at top center; pin 8 (VDD) is bonded to the pad at bottom center. Pins/pads are numbered clockwise.
@ahhuhtal When I remember correctly (did it once), you can still program it, but it will need high voltage programming (rasing the reset line to 12V). This can no longer be done with a standard ISP programmer and the pins you use for programming are then different, too (because many AVRs don't support serial HV programming).
@TerminalJack505 Usually the way to go for more pins is to use 74xx595 which is dirt cheap 8-bit serial in, parallel out register with latch, SPI-like input. These I2C exanders have more logic inside, are bidirectional, have interrupt and maybe few more things. Downside is slower I2C access, perhaps not critical in this application. I think Microchip have also the same with SPI bus, it just takes few more pins.
@electrodacus A $2 ARM cortex M3 is pointless if you need a DIP package. There are *countless* better/cheaper/faster/integrated microcontrollers I could have chosen from a dozen brands, but I didn't, for good reasons.
Typically an Arduino compatible header for serial programming has reset just after the RX/TX so it can be triggered by DTR handshake line from the RS232 adapter. This makes the ATMEGA run the bootloader and talk to the PC. You can press the reset button by hand and let go at the right time but it's a bit of a pain. Also the "C" version didn't have a grade letter, they all had "C" at the end it stands for commercial temperature range. Good stuff, looking forward to the next one.
Dave, since you're going to use an external resonator you might consider using a more UART-friendly frequency--like 7.3728 MHz. Also, you might be able to use an analog MUX to reclaim your SPI pins. You could use something like the 74HC4053 and use the RESET signal as the selector and route the SCK, MISO and MOSI pins to either the ISP header or your SPI device(s) based on whether the programmer is asserting the RESET signal or not. I've never tried this but it *should* work.
I understand the use of Atmega368 for full arduino compatibility, but I have loaded the arduino bootloader in the Atmega644 with minimal modifications (just because I used a different clock freq), you get all the pins you need, more RAM and flash and you can use the arduino IDE without any problems!!
@electrodacus I think the problem with this is that the only larger Arduino-compatible chip is *not* through-hole, and that was one of the restrictions of this project.
Watch about these TLC272's I've spend a bit over 200 now and about 2% just burns out with no apparent reason :/, and in the circuit (so it is not ESD)... If you want to pull it down to 0V for sure, load the output a little (resistor to ground) but you will test that. In my experience they were all under a mV offset on the output in my circuit (the cheapest ver.) so that is really not needed. I used them because of really low input bias current that i needed, and single supply operation.
@21:40 Dave goes on about the odd order of the A, B and C version. It's not A, B and C. It's " ", A, and B. Notice the CD is tacked on the end of all 3 of them. the odd one out at the bottom of the list doesn't have a letter.
@Vlakpage The 74HC4053 is a 30 cent part and is bi-directional. You won't get the functionality that I'm suggesting from a serial-in, parallel-out part. The point of using the 74HC4053 or similar was to allow the RESET signal--which is held low by the programmer during programming--to select between the ISP header and the SPI bus and let Dave use hardware SPI instead of having to bit-bang to the SPI devices.
Hi Dave, The AVRs may have interrupt capability on all the pins, but I don't think the arduino IDE implements interrupts on any of the pins except PD2 & PD3 (interrupts 0 and 1). Your code may not end up being "arduino compatible"
Regarding the TLC277 variants. The C version are rejects from the A and B versions. If the part fails specification testing but passes functional testing then brand it with a C and still sell it.
@electrodacus Of course if you want it cheap for some 10k run, you'll design accordingly. But keep in mind, this kit is made to be simple to build, use and hack around. That's why it uses mostly DIP parts (there are no widely available ARMs in DIP) and it's compatible with Arduino so you can take advantage of existing IDE and code/libraries already developped by the community. I highly doubt there is simple way to get multiplatform STM32 toolchain with IDE and upload capability with $5 flasher.
Hardware SPI is like two lines of code. Just make sure that the peripherals have their /CS pins idle during programming. Hardware I2C doesn't like glitches on the bus. Once I've ended up resetting I2C interface before every communication cycle (noise+long wires). The software implementation is quite a bit of code, unless you are using a library. Add a resistor between C24 and reset pin, leave ISP reset tied directly (as is). Some programmers will not be able to reload that cap fast enough.
Been watching this from #1, talk about feature creep.. ;) Well done though, you sir should change profession to teaching professionally, you are quite effective at presenting information in a way people can follow.. Or maybe make a few DVD's teaching electronics for hobbyists to buy that can be more complete and structured than a blog..
For those looking, Dave has made the schematic available on his website. Dave, don't you just love your project being designed by committee? LOL. Keep up the good work and don't get discouraged!
Dave, why the 8-bit GPIO when 16-bit are only pennies more? MCP23017 seems to make more sense here - simpler & cheaper. No reason not to use the hardware SPI either, it's less code & performs better, isolate it from the ISP with some resistors so programmer can override if necessary. @electrodacus While I agree in principle, for a hobbyist kit the whole Arduino compatible thing is a pretty big draw I think, and will encourage experimentation. Worth it for pennies worth of I/O chips IMO.
i think you should have no trouble sharing the ISP and SPI pins, ISP is simply SPI with RESET acting as a chip-select just take care to make sure the programmer floats the master out when its idle, and the other chip-select lines are inactive when programming
@electrodacus Besides keeping it simple Arduino-like, you have option to go with some 40 pin DIP AVR or you can take some Arduino Mega compatible chip which has more than enough I/O but is significantly more expensive. Of course there can be some variants with say PIC32 and built-in Ethernet or whatever but then we're getting into some different platforms which may not have all the codebase and tools simple and ready to use and finally program the chip.
I guess this design has a big weakness. You have the current sensing with those 4 resistors. They equal 1 Ohm. So if you drive your Power Supply with 1A load the output voltage does not really equal what you have set it for. Because of the voltage drop caused by the current shunt. So in worst case (max load 1 A), there will be U= R * I = 1V difference. Pretty much if you ask me. Output Voltage = R_load * I + R_shunt* I
+7enderThomas The voltage regulator is after the current regulator, so it compensates for that loss. You will however lose 1V of your supply voltage. Not that big a problem as you can just use a higher voltage supply.
@7enderThomas You missed the explanation for this part of design, those four resistors only used for very low current measurements, all other time they are bypassed by software controlled Q4 (see RANGE pin) with very low resistance. besides that, your formula is incorrect: R_load * I does not make any sense because this is constant voltage power supply as long as you do not exceed the current limit set point. The voltage on the load will be the set point voltage.
why you did not used a very cheap(2$) stm32f103c mcu it has a lot of 12bit adc and 16bit pwm and it is an arduino ide compatible and it has a lot of good Features in you project
@electrodacus Why go on about this? It is pointless. I wanted a DIP package. I wanted Arduino compatibility. You get neither of those with those other more powerful devices. There are countless other bigger chips I could have chosen STM32 is only one of a dozen. I didn't for good reasons.
@ 00:20 true engineering skill: throw something away and it lands on the desk like someone put it there. :D
haha, this made my day
I first watched this build series back in 2013, and at least once a year, I come back and watch the entire series again. Why? When I feel like I am in a slump, this series reminds me why I started doing this. Attention to detail, on the fly problem solving, and pushing the limits of what can be done on a budget. Still love this series on the PSU. Happy Thanksgiving 2018 from the U.S.A..
I want to thank you very much for encouraging us, teaching us, give us what you know. The words of gratitude that we use in this way are not enough to thank you a thousand thanks¡¡¡¡¡
I don't get why you want to bitbang SPI rather than use the hardware SPI port. The chip specifically handles sharing SPI and ISP because ISP is only active when the chip is in RESET mode. It's quite safe to mix the two on the same pins.
@MrP00f So it's available on the I/O connector if required for some other more important task.
@keenantims The hardware SPI is now available to the external I/O connector.
Very nice revisions and sheet layouts Dave!
I think this is a good project example because you're interfacing I2C, using PWM etc.
and yes, you could use a higher pin-count MCU or a PIC or Vortex 32 but it wouldn't
be something that your average electronics hobbyist would even consider learning a lower level language nor having to buy programming hardware to modify
and experiment with.
Also I think it's keeping more of the 'vibe' with open source.
Nice work again, and thanks for the detail!
@logicmindtube Yes, it will be time lapse of some sort, as I have several hours of layout footage captured that I plan to do a voice-over too. No idea how well that's going to turn out though...
I love it how you use a screwdriver as a pointer! How engineering of you!
(but it still gives me the creeks when it scratches the paper...)
I cant wait to get one, maybe four. I will build the others for my 3 boys. This will be like Heathkit. Great way to learn Thanks DAVE
@exor674 Correct. And it's a good example to show the use of I2C I/O expansion devices.
Great! How about some hardwire testing now, with the extra IC's, LCD and the arduino?
@envisionelec Yeah, that's why I added them. Gives people something new to experiment with too.
@keenantims Yes, you are right, with hindsight I should have used the bigger device. I missed it. I had used the 8bit one before so was kinda caught up on that and stopped searching. Damn.
The hardware SPI is left free so an Ethernet shield or some other external serial device can use it.
The image in the TLC272 datasheet @22:51 is a copy of the photomask for the metal on the IC (black=metal). It looks like this chip only has one layer of metal. Metal connects to the sources and drains of the transistors; polysilicon is used for the transistor gates and short connections underneath the metal.
The pads with notched corners in the LL and LR corners are not bonded out, apparently. (They're probably only used at wafer test--"multiprobe" in TI jargon). Pin 4 (GND) is bonded to the pad at top center; pin 8 (VDD) is bonded to the pad at bottom center. Pins/pads are numbered clockwise.
@ahhuhtal
When I remember correctly (did it once), you can still program it, but it will need high voltage programming (rasing the reset line to 12V). This can no longer be done with a standard ISP programmer and the pins you use for programming are then different, too (because many AVRs don't support serial HV programming).
@TerminalJack505 Usually the way to go for more pins is to use 74xx595 which is dirt cheap 8-bit serial in, parallel out register with latch, SPI-like input. These I2C exanders have more logic inside, are bidirectional, have interrupt and maybe few more things. Downside is slower I2C access, perhaps not critical in this application. I think Microchip have also the same with SPI bus, it just takes few more pins.
@jmpattillo I didn't even look into if it was possible on the Ardunio. In any case, some people may want to write their own code.
@electrodacus A $2 ARM cortex M3 is pointless if you need a DIP package. There are *countless* better/cheaper/faster/integrated microcontrollers I could have chosen from a dozen brands, but I didn't, for good reasons.
Great update Dave!!
Is Rev B available yet?
Typically an Arduino compatible header for serial programming has reset just after the RX/TX so it can be triggered by DTR handshake line from the RS232 adapter. This makes the ATMEGA run the bootloader and talk to the PC. You can press the reset button by hand and let go at the right time but it's a bit of a pain. Also the "C" version didn't have a grade letter, they all had "C" at the end it stands for commercial temperature range. Good stuff, looking forward to the next one.
@electrodacus You missed the whole point. That's not a DIP part, which I am trying to use. Nor is it Arduino compatible.
Dave, since you're going to use an external resonator you might consider using a more UART-friendly frequency--like 7.3728 MHz.
Also, you might be able to use an analog MUX to reclaim your SPI pins. You could use something like the 74HC4053 and use the RESET signal as the selector and route the SCK, MISO and MOSI pins to either the ISP header or your SPI device(s) based on whether the programmer is asserting the RESET signal or not. I've never tried this but it *should* work.
I understand the use of Atmega368 for full arduino compatibility, but I have loaded the arduino bootloader in the Atmega644 with minimal modifications (just because I used a different clock freq), you get all the pins you need, more RAM and flash and you can use the arduino IDE without any problems!!
@electrodacus I think the problem with this is that the only larger Arduino-compatible chip is *not* through-hole, and that was one of the restrictions of this project.
what approach would you adopt if you wanted a cc/cv power supply to go up to 100v and 10 amp or more? looking for ideas please.
Beauty!
Watch about these TLC272's I've spend a bit over 200 now and about 2% just burns out with no apparent reason :/, and in the circuit (so it is not ESD)...
If you want to pull it down to 0V for sure, load the output a little (resistor to ground) but you will test that.
In my experience they were all under a mV offset on the output in my circuit (the cheapest ver.) so that is really not needed.
I used them because of really low input bias current that i needed, and single supply operation.
can I get the full schematic with the code for the IC :) would be really helpful
I²C expansion ICs - Love 'em.
@21:40 Dave goes on about the odd order of the A, B and C version. It's not A, B and C. It's " ", A, and B. Notice the CD is tacked on the end of all 3 of them. the odd one out at the bottom of the list doesn't have a letter.
what's the software you use sir to plot your circuit & convert it to a pdf file?
thanks in advance
@Vlakpage The 74HC4053 is a 30 cent part and is bi-directional. You won't get the functionality that I'm suggesting from a serial-in, parallel-out part.
The point of using the 74HC4053 or similar was to allow the RESET signal--which is held low by the programmer during programming--to select between the ISP header and the SPI bus and let Dave use hardware SPI instead of having to bit-bang to the SPI devices.
@horfittunge They are not available in DIP for Arduino compatibility.
I'm curious to why you decided not to use a 16 pin I2C expander instead of the 2 8-pin I2C port expander. Was it a board layout decision?
@electrodacus I'm very well of the countless alternatives available, thanks.
Hi Dave, The AVRs may have interrupt capability on all the pins, but I don't think the arduino IDE implements interrupts on any of the pins except PD2 & PD3 (interrupts 0 and 1). Your code may not end up being "arduino compatible"
Hi Dave!
Why did you decide against using hardware SPI?
Where is the link to the schematic? not in part#8 or this one.
@jpelczar It's recommended in the AVR literature for "noise reduction". I don't think it's needed, but people bitched when I left it out!
@jmpattillo Correction: It looks like there is at least one library (PinChangeInt) out there that adds pin change interrupts to arduino. PinChangeInt
Regarding the TLC277 variants. The C version are rejects from the A and B versions. If the part fails specification testing but passes functional testing then brand it with a C and still sell it.
"Everything you're doing is wrong. And even if it wasn't wrong, we'd say it was." Love, The Interwebz.
Why use two MCP23008 and not one MCP23017?
Why would you use external IO expansion chips instead of just using MCU with higher pin count?
@electrodacus Of course if you want it cheap for some 10k run, you'll design accordingly. But keep in mind, this kit is made to be simple to build, use and hack around. That's why it uses mostly DIP parts (there are no widely available ARMs in DIP) and it's compatible with Arduino so you can take advantage of existing IDE and code/libraries already developped by the community.
I highly doubt there is simple way to get multiplatform STM32 toolchain with IDE and upload capability with $5 flasher.
Hardware SPI is like two lines of code. Just make sure that the peripherals have their /CS pins idle during programming.
Hardware I2C doesn't like glitches on the bus. Once I've ended up resetting I2C interface before every communication cycle (noise+long wires).
The software implementation is quite a bit of code, unless you are using a library.
Add a resistor between C24 and reset pin, leave ISP reset tied directly (as is). Some programmers will not be able to reload that cap fast enough.
Is this schematic available for download?
can you have the Altium Design original file of this schematic?
@Gameboygenius Is that an SI unit? :->
Hey Dave, how come you didn't just use a larger micro instead of adding I/O-chips?
Been watching this from #1, talk about feature creep.. ;) Well done though, you sir should change profession to teaching professionally, you are quite effective at presenting information in a way people can follow.. Or maybe make a few DVD's teaching electronics for hobbyists to buy that can be more complete and structured than a blog..
i would be cool if the screen followed the current.Going from green (low current) to red(high current) :P
Right, so the change rate of the µCurrent range switch is 1 BM^-1. Interesting...
sweet!
Dave, where's the pdf schematic? can't find on eevblog.com
For those looking, Dave has made the schematic available on his website.
Dave, don't you just love your project being designed by committee? LOL. Keep up the good work and don't get discouraged!
Dave, why the 8-bit GPIO when 16-bit are only pennies more? MCP23017 seems to make more sense here - simpler & cheaper.
No reason not to use the hardware SPI either, it's less code & performs better, isolate it from the ISP with some resistors so programmer can override if necessary.
@electrodacus While I agree in principle, for a hobbyist kit the whole Arduino compatible thing is a pretty big draw I think, and will encourage experimentation. Worth it for pennies worth of I/O chips IMO.
How on earth can you make a half-hour video of a piece of paper -- the same piece of paper --, and make sure the video is not boring?
Once i figure out how to beef this up to give me a great deal more current i think ill build me one of these and try it out against my current unit
i think you should have no trouble sharing the ISP and SPI pins, ISP is simply SPI with RESET acting as a chip-select
just take care to make sure the programmer floats the master out when its idle, and the other chip-select lines are inactive when programming
The naming scheme is TLC272xCD
Where x is:
blank - low grade
A - medium grade
B - high grade
Makes perfect sense :P
@electrodacus Besides keeping it simple Arduino-like, you have option to go with some 40 pin DIP AVR or you can take some Arduino Mega compatible chip which has more than enough I/O but is significantly more expensive.
Of course there can be some variants with say PIC32 and built-in Ethernet or whatever but then we're getting into some different platforms which may not have all the codebase and tools simple and ready to use and finally program the chip.
I guess this design has a big weakness. You have the current sensing with those 4 resistors. They equal 1 Ohm. So if you drive your Power Supply with 1A load the output voltage does not really equal what you have set it for. Because of the voltage drop caused by the current shunt. So in worst case (max load 1 A), there will be U= R * I = 1V difference. Pretty much if you ask me.
Output Voltage = R_load * I + R_shunt* I
+7enderThomas The voltage regulator is after the current regulator, so it compensates for that loss. You will however lose 1V of your supply voltage. Not that big a problem as you can just use a higher voltage supply.
@7enderThomas
You missed the explanation for this part of design, those four resistors only used for very low current measurements, all other time they are bypassed by software controlled Q4 (see RANGE pin) with very low resistance. besides that, your formula is incorrect: R_load * I does not make any sense because this is constant voltage power supply as long as you do not exceed the current limit set point. The voltage on the load will be the set point voltage.
@electrodacus same goes with DAC and ADC
@implor It's got a power switch :->
@EEVblog The unit is 1 (Blue Moon)^-1. ;)
@eevblog
its all blue color today, LOL. it took me some time to distinguish you from your table mat :D
Chameleon mode lolz
Hate to be pedantic, but it's not an arduino. An arduino is the board, not just the Atmega 48/88/168/328 ;)
@philbx1 Someone gets it. It's just "the vibe"! (Australian joke...)
@PetrFM 5V is no good when you need >10V :->
why you did not used a very cheap(2$) stm32f103c mcu it has a lot of 12bit adc and 16bit pwm and it is an arduino ide compatible and it has a lot of good Features in you project
@crenn6977 @EEVblog Oh, I see that you've already responded to keenantims about that.
@electrodacus Why go on about this? It is pointless. I wanted a DIP package. I wanted Arduino compatibility. You get neither of those with those other more powerful devices. There are countless other bigger chips I could have chosen STM32 is only one of a dozen. I didn't for good reasons.
@heroineworshipper It's not a contest to have a better micro.
What you call C is CD without a prefix. Worst: CD. Less bad: "A"CD. Better "B"CD. Best 277CD (This one has a new number code.)
I don't understand anything.