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

[prasimix]

Hm, now I'm confused. If R179 is removed then you shouldn't have correct reading since nothing is connected to non-inverting input of IC7. also did you try to slightly decrease resistance of R63 to e.g. 80 mOhm (you can manage that also with soldering another 33 Ohm resistor in parallel).

[morris6]

Sorry, I didn't remove R179, I shorted it to find a possible effect. None found.

Putting 33 ohm parallel to R63 is exactly what I did earlier in post #942 when I suspected R63 high. That started me to have a look at the trace resistance.

[prasimix]

Ok, did you try to go in opposite direction and increase R63 a bit, and see what reading you can get on the full scale after calibration?
I do believe that I'll find time and parts soon to complete r5B13 that you sent to me and check on my side what is possible with existing design.

[morris6]

@prasimix

This time I increased R63 with one of 0.01 in series as you proposed. Full scale, after calibration, was 435 mA.
Next I did was 33 ohm parallel to R63, it reached 497 mA.
I then tried 12 ohm parallel, resulting in indeed 500mA, full range in 0.5 a range.
I also did R63 in series with 0.01 and a wire to R65, but that reaches 498 mA max.

It seems you can already see whether 500 mA will be reached, the last calibration step will have a higher current than the target of 480 mA. Am I right with this assumption?

When I parallel the trace from R63 to R65 with a piece of wire, the 480 mA calibration step gives 525 mA actual current, and 500 mA full scale is no problem.

As a last one I did R63 in series with a 0.01 - 0.03 parallel combination, followed by a wire to R65, bypassing the trace. This gives for the R63 combination calculated resistance of 0.082 + 0.0075, so just below 0.09 ohm. Calibration current at target 480 is now 490 mA. And, of course, 500 mA full scale is OK. 

[prasimix]

It seems you can already see whether 500 mA will be reached, the last calibration step will have a higher current than the target of 480 mA. Am I right with this assumption?

Yes, that is right assumption. Rsense should be selected in a way that when uncalibrated max. current goes beyond full range (i.e. more then 500 mA in this case). Anyway, it seems that R65 on r5B13 version should go below initially proposed 82 mOhm. Checking what is available on e.g. Mouser site, we can find the following values: 81, 80.6 or 80 mOhm.

[morris6]

Well, in stead of decreasing the value of R63 isn't it a better choice to decrease the resistance of the trace between R63 and R65, N$70 in the design files? See post #942.

[prasimix]

You're right, that should be a simpler to change in next revision then looking for a resistor with "exotic" value

[morris6]

Yeah, those low TC resistors are difficult to find.
For further trials I just ordered some 15 ppm/K ones: https://www.buerklin.com/en/smd-precision-metal-film-resistor/p/16e866?an=20000000.20100000.20101000

They also have 0.01, 0.04 and 0.05 in this series.

[prasimix]

Dear all, the firmware v1.1 is now released. But, this time it does not comes alone but accompanied with first milestone of EEZ studio. A first introduction how to use one of its modules is now available in separate thread here. More specific information how to use it to access EEZ H24005 I'll continue to post in this thread.
Two main feature of v1.1 are added file management for built-in SD card and internal data logger. In this version both of them are accessible only using SCPI commands, but don't worry a decent support to automate interaction with SD card and data logger is included in mentioned EEZ Studio.

[jfitter]

I have a new Astec LPS355 (48V/350W) single ended supply left over from a previous job. Am I missing something or will this circuit only work with two independent supplies? I see that the -ve rails are separated from the power ground in the post regulators so does that still mean the outputs can be stacked? Power supply design is somewhat of a new area for me. Any help appreciated.

[prasimix]

Here is a video (again ) that show how two EEZ H24005 channels outputs can be combined in series and parallel:



If your Astec outputs is not earthed (i.e. its floating) you can use it as power source for Power board. Please note that you can use max. 200 W of its 350 W.

[jfitter]

Thanks for your reply, however I am still unsure. I have seen the video but it shows only how to configure the software.
The SMPS I have is this http://www.farnell.com/datasheets/55283.pdf. I have only one of them. It is a single output, 48V and is isolated, so there is +48V and 0V.

My concern is that if I use this SMPS then both post-regulator boards will share a common +48V and 0V. The supply is rated at 350W which is enough to drive both regulators to their rated limit (almost) but the common 0V worries me. How does the power supply stack one output on top of the other to get 80V and can this be done using just a single SMPS?

It is a rather nice SMPS and it would be a shame to waste it.

[prasimix]

Ok, now I understand what are you asking. You have to use one AC/DC module with floating output per channel, that's the only way to keep them isolated from each other and to couple their outputs in series for total output voltage of up to 80 V.

[prasimix]

Preview of two main features added to firmware v1.1: SD card file management and internal data logging using SCPI commands. The EEZ Studio version M1 (its ESW module) is used for interaction with the EEZ H24005 (i.e. initiate data logging and exchange files).

[Pjoms]

First, really nice work with EEZ Studio and the data logging!

With FW 1.1 I have noticed a strange behaviour that never seen before on this power supply.
I have measured the time that takes for the voltage to recovery then a load is connected.

The setup:
Volt: 5V
Current: 3A
Load: Resistor 4,7 ohm

The load is connected from plus to the anode on a BTW68-600 thyristor to get a clean coupling without any bouncing.
 
I have used this setup before and always got a recovery time about 2 to 8 ms.
With FW 1.1 I often gets two dips in the voltage! Why???

CH1 (yellow): Output voltage measured on the terminals of the psu.
CH2 (cyan): Gate voltage on the thyristor, and trigger.
CH3 (magenta): Voltage over anode and cathode on the thyristor.

[prasimix]

Please let me know what you have under Current ranges settings (Current range and Autoranging)?

[Pjoms]

Current range: Best (default)
Autoranging: On

[prasimix]

Please try it with Autoranging disabled.

[Pjoms]

Current range: Best (default)
Autoranging: On

And if current rage is fixed on 5A there is of course no problems... [emoji849][emoji4]

But a little bit strange that I didn't see two dips before the official fw 1.1?

[prasimix]

Yes that's strange since as far as I know nothing is changed in that part of the code. Anyway, I'll try to check that in coming days on my side, too.

[prasimix]

You actually touched here the question of autoranging performance. With MCU that is busy with so many other tasks it require considerable time (up to 8 ms) to change from lower to higher current range. With adding one more range (0-50 mA) in revision r5B13 it makes it even worser for current that belongs to the highest current range (i.e. above 500 mA). Maybe we should start to think here about adding extra analog circuits to accomplish autoranging more efficiently. Perhaps with two comparators (in case of three range version) to detect "out of range" and switch range to next (higher) one. Any suggestions from your side are welcome, and I'll also try to think about it.

[Tazz]

Do you still have free interrupt lines to hook the comparators in a very simple (and interrupt latency efficient) manner ?

[prasimix]

Thanks Tazz for your input. We gave up for now from interrupts since we never got reliable interrupt handling on Arduino Due (it was reported to Arduino but no action there so far). Therefore a lots of attention was put into making polling (main loop) as fast as possible. But, as we can see it can lasts up to 8 ms, that is way to slow in comparison with analog part capability.
My first thought is to use comparator with enable function and in case of autoranging it will become enabled and take care about activating accompanying current range MOSFET (i.e. one of Q13, Q14 or Q34 on this schematics). When autoranging is activated, output pins of I/O expander are not used neither firmware should take care about current range switching.
If that makes sense I should start a quest for suitable, or better to say, affordable comparator.

EDIT: Second thought is to use general purpose comparator that is continuously monitoring "out or range" condition and when autoranging is enabled it's output state will be forwarded to the MOSFET using some simple logic gate (e.g. '125, '08, etc.).

[Tazz]

Hooo my .... I remember why I hate arduino now. Interrupts broken on a µC env, I will forbid my son to touch such thing until this whole ecosystem is more sane otherwise he will only learn how to think/do bullshit.
It is a miracle what you accomplish for EEZ H24005.
It would even be better with proper µC support as completely relying on full HW/analog solution to every encountered problems will rapidly become a dead end.

Yes simple (good) comparators and output gating by the µC would be a very good solution for this case.

[prasimix]

I cannot say what's going on with AVR and some other Arduino boards, but Due doesn't work and Due is unfortunately "dead meat" so we can expect any progress there.
But, this is a good opportunity to mention something else. We're spent some time searching for Arduino alternative. First one was XMOS xCORE200 announced in other thread, and I even made an evaluation board for that MCU (with multiple mistakes and wrong presumptions, the latest corrections are on GitHub). XMOS is so far put on hold primarily due to lack of needed information and support. Therefore we started to explore STM32, and I can say that it's progressing very well. Here is a screenshot (board is STM32F429I-DISC1). Porting GUI/HMI was gone almost seamlessly in a couple of days. I'll keep you informed about further progress and some other activity regarding the next PSU revision .