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

[mrpackethead]

Nice project. I'd be keen to take it with pretyt much everything done.   

[VEGETA]

I wanted to ask you about how you made your design stable? and most importantly, is it stable and working nice with low currents like 10mA or maybe less (in CC mode of course)? I've made a circuit that does good stuff but still has these major issues.

One big difference between our circuits is this:

1- you use opamps with negative voltage rail not single supply.

2- your CV loop is somehow weird, I mean R37 and R62 (consolidated pdf) followed by the traditional diff amp. I would say why you put these 2 resistances since you are gonna use a diff amp anyway. I couldn't figure the purpose of C47+R48 as well as D16 too.

picking compensation caps and resistances values is still foggy for me as I am learning, perhaps it is all in the ac analysis rather than blind trial and error.

You went for low-side current sensing which is ok as well as my circuit too. However, what is the benefit of getting this instead of a simpler high-side current sensing using something like LT6106? Here you may get away with just a voltage divider to get your V_monitor instead of the diff amp right?


Thanks for your efforts. I hope I have enough money (and courage ) to get this when it is available.

[prasimix]

I'd like to emphasize once more time that post-regulator section is not mine but overly simplified Liv's design.

Yes, the R37 and R62 possibly looks weird since one part of the circuit is "missing" on Sheet #3 and can be found on the Sheet #8 when you can see that by default R37 and R62 are connected to power output (Out+, Out-) using small signal relays (K_S1 for Ch1 and K_S2 for Ch2). They as used as "protection" in remote sensing mode of operation when sensing wires are disconnected.

C47+R48 improve over- and under-shooting performance. D16 (same as D18) is used against integrator wind up.

I spent some time trying to find out proper high-side current monitor but without success. You can find interesting posts of this theme on void_error's PSU project thread. There is a few issues here: huge imprecision on small current, min. working voltage (in general cannot works with Vout below ~3 V) and/or very limited bandwidth. The latest is not an issue if it is used just as current monitor, but if you need to add it into CC control loop then it become a problem. For min. working voltage issues you could try to put current monitor before series/pass element but if you're using BJT then you'll also its measuring bias current, and if you have down-programmer (as in this project) then situation/error is even higher. It seems to me that most of the current monitoring ICs are designed for monitoring/supervising (not to actively regulating) well defined load: e.g. DC motor, solenoid, battery, etc.
Therefore low-side monitoring is used and I tested if with both LTC2057 and OP27. Precision is good and can be improved further by adding e.g. two ranges: 500 mA / 5 A and range selection can be MCU controlled.

Diff amp is not used just for monitoring but is an active part of CV loop. Using voltage divider could be used as simple/cheap voltage monitor, but even in that case it should be buffered: that means that you need to use one op amp anyway.

[prasimix]

The firmware M3 is released today. We had some interesting time when for the first time a more then one interrupt is used and spent almost a month trying everything to make operation stable and predictable. Finally we found that Arduino's SPI transactions doesn't masks interrupts properly and we have to see what Arduino team will say. Another interesting thing was related to coupling channels outputs in series and parallel. It's not so straightforward as one can think because this power supply has down-programmer circuit that can make some problem when channels are coupled ni series and enters CC mode or when coupled in parallel and enters CV mode. I'll prepare a video to show what could be happen as we solved that successfully.
With this firmware we covered almost everything on hardware level (only digital input, and load/batter NTC are not addressed). Please find below the main features of this release. Shown screenshots are from software simulator. All yellow painted text represents enabled/"clickable" options (in a "follow the yellow brick road" manner).

Channel coupling
Allows doubling of max. output voltage (up to 80 V) or current (up to 10 A). Outputs are wired internally thanks to two MCU managed power relays. Presentation and programming of output values and protections are unified/centralized.




Tracking mode
Presentation and programming of output values and protections are centralized. When any protection tripped on one channel it will shut off both channel. Output enable/disable function is also unified.



Additional system protections
This is initated by Kleinstein's suggestion and we added two more options:



AUX temperature sensor settings



Lock local console (touch screen)
When locked only unlock icon is enabled. Lock can be protected with system password.



Sound control

[VEGETA]

Good job doing the display! BTW what is the display used (price)? You mentioned that you will release the software to design these GUIs, is there any intention left to do it?

thanks again.

[prasimix]

Display is possibly the cheapest 3.2" touchscreen that can be found on the eBay (example). Model name is TFT_320QVT.

The EEZ Studio that we are using to editing menus and pages is still not ready for public release. We are working on it and hopefully be ready sometime in 2017. We'll try to add also support for RIVERDI displays based on more powerful controller (FT800/FT801) like 3.5" resistive and capacitive one. 

[prasimix]

I tested today how fast is remote control using SCPI commands via serial (USB port). It's initiated with question got by email from guy who would like to know is it possible to generate square wave signal faster then 10 Hz. He has another brand power supply that cannot offer more then 5 Hz.

Without any optimization and with selecting most demanding viewing mode (bar graphs) that we are refreshing as soon as any change is detected (but not faster then ~300 samples/s) I got the following:

Display is ON (that means display will be frequently refreshed):
18 Hz, if connected load does not cause changing of mode operation (or CV only)
12 to 14 Hz depending of the selected viewing mode, and when mode operation is changing (CV-CC)

Display is OFF (option DISPLAY_OPTION=0 in conf.h file):
36 Hz regardless if the mode of operation is changing or not!

For testing I used simple Python script that you can find in attachment. Here is a screenshot for 12 Hz SCPI VOLT 0.1 - VOLT 5.1 changed in the loop on the both channel (PSU is in the TRACking mode of operation):



In firmware M4 we are planned to add SCPI LIST subsystem when you'll be able to load waveform shape that will be internally executed in the loop. Our goal is to reach 120 Hz (sine wave generation) in that way, but we'll see.

Apart to digital (SCPI) remote programming it's already possible to achieve much higher speed using so-called remote voltage programming (VOLT:PROG). In this case the PSU can easily survive much faster changes, e.g. 500 Hz as shown in next couple of screenshots (square and sine, yellow trace is output, cyan is controlling analog signal produced by signal generator):





 

[mrpackethead]

Can i  buy one yet>

[prasimix]

Can i  buy one yet>

Hi, you cannot buy one but you'll be able to build one if group buy will be successful (I opened a separate thread for that topic). In best case (Level D) you shouldn't need more then a hour or two to have ready to run unit. In the mean time all firmware features you can check right now using Software simulator (Windows download is here, Mac/Linux instructions are here).

[VEGETA]

@parasimix

what is the tool or IDE you used to create the software simulator? and how to update the firmware of a running unit in the future?

[prasimix]

what is the tool or IDE you used to create the software simulator?

As you can see on the firmware home page on the GitHub under section "Simulator GUI Front Panel", It is based on SDL, SDL Image and SDL TTF.

and how to update the firmware of a running unit in the future?

... like the first time by simply connect the PSU using USB to your PC that runs Arduino IDE as described here.

[jondaddio]

Looks like a nice power supply!

I see the title of this thread says "DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)", but I noticed the Crowdfund says "0 – 40 V @ 0 – 5 A with a 155 Watt limit (with selected AC/DC modules, otherwise up to 200 W)" in the spec section.

1. Could you clarify what the voltage range is?  I assume it is 0-40V based on the power spec.
2. Not being very familiar with your design, what does "with selected AC/DC modules" mean?
3. It appears the current is up to 5A (vs. the original 3A) and power is up to 200W, depending on what "with selected AC/DC modules" means.

Thanks

[prasimix]

Looks like a nice power supply!

I see the title of this thread says "DIY SCPI programmable dual channel bench PSU 0-50V/3A (now EEZ H24005)", but I noticed the Crowdfund says "0 – 40 V @ 0 – 5 A with a 155 Watt limit (with selected AC/DC modules, otherwise up to 200 W)" in the spec section.

1. Could you clarify what the voltage range is?  I assume it is 0-40V based on the power spec.
2. Not being very familiar with your design, what does "with selected AC/DC modules" mean?
3. It appears the current is up to 5A (vs. the original 3A) and power is up to 200W, depending on what "with selected AC/DC modules" means.

Thanks

Yes, it was a long journey that end up as a 40 V / 5 A max. 155 W model for crowdfunding based group buy. Selected model is Mean Well LRS-150-48 and two can be housed into enclosure. But, you are free to use existing Power boards with other modules to reach max. 200 W per channel. Also Power board can be powered from AC input (i.e. mains transformer). In that case with bigger transformer, and by adding few components on the Power board you can go again up to 200 W per channel.
Two channels can be also internally coupled in series or parallel and firmware will works with them as one "big" channel capable of delivering up to 80 V or up to 10 A (max. 310 W with above mentioned modules).

[jondaddio]

Yes, it was a long journey that end up as a 40 V / 5 A max. 155 W model for crowdfunding based group buy. Selected model is Mean Well LRS-150-48 and two can be housed into enclosure. But, you are free to use existing Power boards with other modules to reach max. 200 W per channel. Also Power board can be powered from AC input (i.e. mains transformer). In that case with bigger transformer, and by adding few components on the Power board you can go again up to 200 W per channel.
Two channels can be also internally coupled in series or parallel and firmware will works with them as one "big" channel capable of delivering up to 80 V or up to 10 A (max. 310 W with above mentioned modules).

Thank you for the clarification, and the quick response!

[2bluesc]

Surprised this hasn't been mentioned yet, but it appears that the Crowd Supply Group Order page was put live on Jan 6th with a complete price of $400.  I'm tempted. It looks awesome and love open source, but I don't reallly need another power supply.   

[prasimix]

A new video about channel tracking mode:



... and about channel coupling (in series and parallel) is now available:



I'd also like to thanks everyone who decided to join us on the group buy that is started a few days ago.

[prasimix]

Surprised this hasn't been mentioned yet, but it appears that the Crowd Supply Group Order page was put live on Jan 6th with a complete price of $400.  I'm tempted. It looks awesome and love open source, but I don't reallly need another power supply.   

Thanks. I hope that over the time will be clearly obvious that important part of this project (software/firmware "suite") goes well beyond implementation in the programmable power supply. Therefore maybe in the near future will be something here for you too .

[prasimix]

Presentation of a few additional protection mechanisms:

• Disabling outputs when Auto-recall user profile option is selected
• Shutdown when any of protection tripped
• Switching off all outputs when any of protection tripped
• Disable all outputs on start up

External/remote voltage programming when PSU output follows analog signal (2.5 V for full scale, input protected and backed with software OVP). Thanks to this you can e.g. use it as a pre-regulator when testing/designing another variable power supply .

[0xPIT]

Are there plans to have stock in the EU?
I'm not interested in paying import duties for a European product...

[prasimix]

Hard to say as long as campaign set goal is not reached. Any pledge could help in this moment. Please note that is a "group buy" organized for interested backers and I'm not (ab)using crowdfunding principle as a cheap marketing vehicle for my business. I got power supply in accordance to my wishes, learned a lot, had a great opportunity to met people on this and couple of other forums to discuss various topics related to this project.
I could risk to some extend an amount of money for limited stock, but limited means higher production cost. Therefore you cannot expect that order of e.g. 10-20 can cost equal to order of 100+ units.
I have to repeat once again even in case of "EU stock" please note that you cannot avoid VAT (Croatian 25%) if you cannot provide a valid VAT number.

[jondaddio]

I just ordered a complete kit which happened to be the 100th complete kit (site now says "100 claimed").  Approaching the goal fairly quickly.

[minion]

The goal has been reached!

Congratulations Prasimix! 

link: https://www.crowdsupply.com/envox/eez-h24005

[prasimix]

Just to repeat here what is posted on other part of the forum: I wish to thank you so much for your support :hug: (sorry, no icon associated). Stay tuned and I'll inform you here and via campaign subscriptions about manufacturing details that should start in coming days.

I really do believe that backers will like what they'll get and continue to get without paying for unlocking firmware nor spending nights trying to hack it or e.g. wondering what some IC with identification markings removed by laser could be . All of you who pledged just for bare PCBs keep in mind that I'll try to assist you as much as I can, and don't be afraid at least two people so far succeed in that small adventure.

[prasimix]

We still have over a month left in the campaign, so there’s plenty of time to exceed our initial goal and hit some stretch goals. Please take our survey to help determine our stretch goals.
Many thanks in advance.

[prasimix]

A new feature that comes with M4 firmware (work in progress) is YT view, a fourth way of monitoring up to two output values like voltage, current and power (delivered energy in Wh and Ah will be added into future):



it can be used for monitoring e.g. battery charging process and more importantly to display output changes when arbitrary waveform generator will be added. Next logical step will be to add support for storing and retrieving data on local SD card that will give more autonomy when no PC or similar devices are near by for data logging using SCPI remote control.

This is also the first intensive GUI task that requires TFT display with better controller that has built-in functions like zoom, scroll, pan, etc. That's a reason why a new display is mentioned in survey as one of possible stretch goal on CrowdSupply campaign.