DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)

[T3sl4co1l]

Ethernet has a 1nF 1.5kV from cable (center tap on the transformer) to GND for ESD protection.  Note that that's a consistent value (i.e., figure 8kV contact discharge from 300pF), but only for single shots (there's no leakage to drain accumulated charge!).

Tim

[VEGETA]

Just a quick question, can you achieve isolation of 2 channels with only one AC-DC module (or any DC source)? I thought about it and my answer was no. I just wanted to read your opinion.

However, there is the floating regulator design so does it allow for example connecting the 2 channels in series or parallel assuming both of them has only one power source (a battery pack for example).

[prasimix]

As far as I know the answer to the first question is no, and yes to the second one as you suggested.

[VEGETA]

As far as I know the answer to the first question is no, and yes to the second one as you suggested.

Good, looks like I got it correctly.

However, for the floating regulator.. I kinda have a problem of understanding how it would be 2 channels in series while both of them returns to battery negative terminal.

I will ask someone here since he know a lot about them.

[prasimix]

For two channels in series you cannot expect that both of them provide a full range but just half, e.g. 2 x 20 V for single power input that can provide up to 40 V on output. You still have to make a "floating" control logic for at least "upper" channel.

[VEGETA]

For two channels in series you cannot expect that both of them provide a full range but just half, e.g. 2 x 20 V for single power input that can provide up to 40 V on output. You still have to make a "floating" control logic for at least "upper" channel.

If both 20v 2A then two in series will be 40v 2A and in parallel 20v 4A, I know that already since it is what all PSUs are.

but what is the floating control logic for the upper channel? In my design, we used an isolated DC-DC module to get the control logic working and so on as I remember. Now we have to get 1 per channel so we have to get 2? what if I plan to use a microcontroller with external ADC + DAC to control both channels...? how will all that fit?

[prasimix]

If both 20v 2A then two in series will be 40v 2A and in parallel 20v 4A, I know that already since it is what all PSUs are.

Yes, but in parallel you could go up to 40V if their single power input provide enough voltage and power.

but what is the floating control logic for the upper channel? In my design, we used an isolated DC-DC module to get the control logic working and so on as I remember. Now we have to get 1 per channel so we have to get 2? what if I plan to use a microcontroller with external ADC + DAC to control both channels...? how will all that fit?

Yes, you should deploy two isolated bias power supply for control logic each referenced to positive power channel output. If MCU/digital control is required you have to isolate it on digital (before DAC/ADC looking from digital to analog section) or analog side (after the DAC/ADC section). Digital isolation is easier and cheaper to achieve without compromising linearity/precision. SiLabs has great isolators for attractive price.
Some sort of isolation between MCU and more then one power channels is required anyway if you'd like to have two separated floating power output.

[VEGETA]

Yes, but in parallel you could go up to 40V if their single power input provide enough voltage and power.

Yes, assuming the source is powerful.

A side question: can this be achieved by 4 parallel 18650 li-ion batteries (using buck-boost converter)?

If MCU/digital control is required you have to isolate it on digital (before DAC/ADC looking from digital to analog section) or analog side (after the DAC/ADC section)

You mean powering the arduino by an isolated dc-dc converter? as well as DAC and ADC?

SiLabs has great isolators for attractive price

I only buy stuff from Aliexpress, thus I searched and found this one: ADUM1201 and ADUM1400 which is 4 channels (2 for each supply channel). Looks like it has this one from Silabs too: SI8660BA.

 Looks like it is just up to 5v but I guess this is suitable since ADC\DAC won't go above. So now I can just power the MCU via any means then get the signal (SPI,I2C) from it to this IC then to DAC. The same for ADC too. Is this method correct?

However, I guess you could use one IC per channel since you don't want to mix the voltages between them, is it correct?

Some sort of isolation between MCU and more then one power channels is required anyway if you'd like to have two separated floating power output.

Isn't this achieved already by using digital isolators between ADC\DAC and MCU for each channel? while MCU itself could be powered directly from the battery pack.

I have a design that is working in LTSPICE here in the forum, with cooperation with kelinstein if you would like to take a look.


_________________

EDIT: I have attached the design image quickly.

[prasimix]

A side question: can this be achieved by 4 parallel 18650 li-ion batteries (using buck-boost converter)?

I have no experience with 18650 batteries. Possibly yes to some extent since such batteries are used in Dave's uSupply.

If MCU/digital control is required you have to isolate it on digital (before DAC/ADC looking from digital to analog section) or analog side (after the DAC/ADC section)

You mean powering the arduino by an isolated dc-dc converter? as well as DAC and ADC?

Yes, your digital control (MCU+peripherals) need separated power rail, that can be floating or e.g. PE referenced.

I only buy stuff from Aliexpress, thus I searched and found this one: ADUM1201 and ADUM1400 which is 4 channels (2 for each supply channel). Looks like it has this one from Silabs too: SI8660BA.

 Looks like it is just up to 5v but I guess this is suitable since ADC\DAC won't go above. So now I can just power the MCU via any means then get the signal (SPI,I2C) from it to this IC then to DAC. The same for ADC too. Is this method correct?

However, I guess you could use one IC per channel since you don't want to mix the voltages between them, is it correct?

That's correct.

Some sort of isolation between MCU and more then one power channels is required anyway if you'd like to have two separated floating power output.

Isn't this achieved already by using digital isolators between ADC\DAC and MCU for each channel? while MCU itself could be powered directly from the battery pack.

I have a design that is working in LTSPICE here in the forum, with cooperation with kelinstein if you would like to take a look.

You have to provide isolation in all sections: analog/linear and digital. No common ground is allowed if you want two separated isolated channels or their coupling in series. That doesn't matter for parallel coupling.

[morris6]

It's a bit silent here.. let's give this thread a bump. The EEZ H24005 is a great project FWIW and I have build three pieces so far.


@prasimix:

I feel a bit like the sorcerers apprentice digging into this. The possibility to open the psu-firmware project file was not presented and discussed yet, nevertheless.. 

It all started with trying to correct some typo's in the screen texts. It looked like this could be done using EEZ-Studio after opening the psu.eez-project file. I felt a bit uncertain about what I was doing, but it worked out fine. After correcting some typo's in the texts of the pages just "Build" and "Save". Several files are updated auto-magically... Then start the Arduino environment, compile and load the corrected firmware. Nice!

Next was cleaning up the small text font characters, pixel editing character by character. Same procedure.. Nice!

And then.. the degree symbol. It shows on screen as "oC" while "°C" is intended. Of course this is a "minor" issue, but a device wherein so many hours of development are invested deserves better, I think. For this a simple text correction didn't work, for several different reasons I think. Ascii code page problems, 'char' type signed or unsigned, It proved too difficult for me, I am only apprentice remember..!
So I copied a trick that is used elsewhere in the psu firmware. Also icons are addressed as text item. I created a new icon "°C" at position 151 in the small fonts section of the psu.eez-project file. Most probably you consider this as a definite "NoNo", editing this file by hand. But it worked. After changing the text string in value.cpp from "oC" to "\u0097" the new icon "°C" shows up on the display, after compile and load of course.

I took the liberty to open an issue on Github about this and uploaded the altered files there.

Some pictures of the display "Before" and "After" are posted here to get an idea of the cleaned fonts and the "°C" symbol.

Maybe these alterations can be taken along in the next firmware update.

[prasimix]

Glad to see that someone had a time and patience to dive into the GUI project without waiting that I eventually find a time to describe a whole process (and what I've promised a few times in the past ). Anyway, it seems that you successfully managed what you'd like to change what is a sort of proof that working with EEZ Studio ESP module is pretty intuitive. I presume that you are already find out that M2 version of EEZ Studio is needed to edit EEZ H24005 project, and this is a note for other who is willing to follow your steps.
Thanks for adding your changes into GitHub. We'll try to include that possibly in first next firmware release.

[morris6]

Yes, I was using the M2 version. You mean M3 version cannot be used yet or anymore for this purpose?

[prasimix]

Unfortunately M3 cannot be used for H24005 project, it will support only a new STM32 MCU board. But, if any issue arise with using M2 feel free to report it on GitHub and we'll try to take care of it.

[mcdanlj]

I was one of the crowd supply backers, so I didn't build my H24005 from components. Now I have an... opportunity to learn more!

One way or another, with the channels coupled in parallel, I fried ZD7 and ZD10. I was set to 4.2V 10A because I was using it to top-balance cells in a Li-ion battery. I did not turn power off while moving clips from one set to the next and don't know whether I had a transient or brushed the clips against inverted voltage, but in any case both TVSs gave up the ghost with visible smoke and damage; one looks like it caught fire; of course it's ZD7, the one that's between the relays and the header, so it will be fun to get to. (I noticed that the KiCad layout moved ZD7 to the other side of the header; maybe I'm not the only one?)

Before I order replacement SMAJ48A-TR from Mouser, I'm curious what else might have died and need replacement, so I can ship all the likely-needed parts at once... I checked the R65 sense resistors on the power boards and they are still good, so I'm hoping the FETs behind them are OK. Would it make sense to remove the cooked TVSs and try powering the unit back up with no load and check the output voltage, then try a dummy load, or are there other things worth testing first?

I'd appreciate suggestions. Thanks in advance for any help here!

[prasimix]

Hm, I guess it is better chance that was happened due to inverted voltage then transient that has to be enormous (48 V and more). I'm aware of stupid placing of ZD7 on crowdfunding version of the board. The "easiest" way is to remove K_SER relay first and then ZD7. That issue is rectified in more recent revision.

It makes sense to remove ZD7 and ZD10, start the unit with both channel off, set currents to small value (i.e. below 100 mA), try to enable outputs of one channel at time and see what will happen.

[Helix70]

I have the crowd supply version and I find the encoder is unusable to adjust current and voltage. I have compiled and updated firmware, have I missed a setting, or is the encoder operation just not used?

I also don't like the colours used, can I change them?

[prasimix]

I have the crowd supply version and I find the encoder is unusable to adjust current and voltage. I have compiled and updated firmware, have I missed a setting, or is the encoder operation just not used?

That's strange, that can be frustrating. Could you please give us more info, in what way it is unusable? Perhaps a short video of "unusability in action"? Does your encoder push button works? It's possible that you have mechanical failure, too.

I also don't like the colours used, can I change them?

We still didn't manage to provide a simple way of color changing (e.g. by providing various color scheme including a user defined one). It theory is possible to change it by editing psu.eez-project file using the EEZ Studio M2 (see below). Currently that is a substantial work, there is many pages that has to be checked (despite the fact the we are tried to use style "inheritance"). Once modified it has to be build and new firmware has to be compiled and uploaded on the unit (using the Arduino IDE). There are many points where you can failed. Good thing is that you can always goes back to the latest working firmware by simply download it from the GitHub.

[Helix70]

I will do a video when I get a chance. Basically when you turn the knob, sometimes it increments in the correct way, sometimes it goes backwards, or jumps around without moving. The button works correctly. I have tried to find the correct method of turning it, holding my tongue at the right angle, slow or fast, but it is a bit hit and miss.

[Helix70]

It seems to be more of a problem clockwise. See video attached.

https://youtu.be/-LBV2er1DRk

Thanks.

[prasimix]

Many thanks for additional info. My first impression is that something is wrong with encoder. Before trying to replace encoder, I suggest to probe encoder outputs with 2-ch scope. It shouldn't be problem to access encoder pins when top cover is removed. Encoder should generate a two square signal with shifted in phase.
I presume that you're using the latest firmware, too.

[Helix70]

Yeah, latest firmware. Yes, ok, I will test the encoder, just wanted to know it it was strange, sounds like it is.

Thanks.

[prasimix]

Perhaps applying some cleaning agent such as Kontakt 60 could resolve mechanical issue.

[mcdanlj]

Hm, I guess it is better chance that was happened due to inverted voltage then transient that has to be enormous (48 V and more).

That was probably the problem. This is a salvaged chevy volt battery with tabs sticking out every 4.2V cell and I expect I brushed the wrong one. It would be easy to do, and easy to not notice doing it.

It makes sense to remove ZD7 and ZD10, start the unit with both channel off, set currents to small value (i.e. below 100 mA), try to enable outputs of one channel at time and see what will happen.

ZD7 was a pile of char, so I cleaned it up so it wouldn't conduct. ZD10 looked undamaged but disconnected at one end and sticking up at an angle, and looked a bit like a failed cold solder joint. I have not tested it, merely removed it.

I tested it without ZD7 and ZD10. In the 0-1V range, floating with no load, calibration is off by .001V on channel 1 and  .005V on channel 2, and with a 2Ohm load by about .02V in that range. I had not run calibration on it recently so I don't know whether that's even a change.

Looks like it survived.    I'll order new TVS diodes.

Is the remote voltage sense tolerant of inverted voltages? It has occurred to me that I could build charging cables with a diode embedded inline to protect against another inadvertent touch, with separate remote sense wires in order to get precise control of voltage on the other end of the diode. Does that make any sense?

Thank you so much for the help!

[prasimix]

Is the remote voltage sense tolerant of inverted voltages? It has occurred to me that I could build charging cables with a diode embedded inline to protect against another inadvertent touch, with separate remote sense wires in order to get precise control of voltage on the other end of the diode. Does that make any sense?

Yes, and it has remote sense reverse polarity protection (Q18, Q20) that in case of inverted voltages immediately disable power output (via Q17) and notify MCU of error condition (Q19). Please note that such protection works for output voltage of approx. 1 V and above.

[mcdanlj]

Yes, and it has remote sense reverse polarity protection (Q18, Q20) that in case of inverted voltages immediately disable power output (via Q17) and notify MCU of error condition (Q19). Please note that such protection works for output voltage of approx. 1 V and above.

That's awesome! I ordered some high-current schottky diodes to build these cables and they'll arrive with the replacement TVSs. Very much appreciated!