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

[prasimix]

Please try the following scenario: instead of using built-in coupling is series, use tracking mode, set both channel to max. voltage, use external wiring and by changing current check if you can get channel fault detected error.

[Pjoms]

Please try the following scenario: instead of using built-in coupling is series, use tracking mode, set both channel to max. voltage, use external wiring and by changing current check if you can get channel fault detected error.

Not able to set 5A.
No shut down, but:
**ERROR: 500,"Down-programmer on CH1 switched off"
**ERROR: -350,"Queue overflow"
**ERROR: -245,"CH1 output fault detected"
**ERROR: -350,"Queue overflow"

[prasimix]

Good, what was load and set current when you got output fault detected?

[Pjoms]

Good, what was load and set current when you got output fault detected?

Vset: 80V
Iset: changed from 3A to 5A by keypad
Load: 10 ohm

[prasimix]

I also reproduced latest issue. It's generated by situation that power limitation was intended to block: that power on any of AC/DC module do not exceed 155 W. If that happens even for very short time, the PWRGOOD signal will failed causing that firmware report e.g. -242,"CH1 fault detected" and enters standby.

A new firmware build is now available in master branch. Please download (or git pull) it and check what you'll get now.

[Pjoms]

A new firmware build is now available in master branch. Please download (or git pull) it and check what you'll get now.

It looks like I have some smoking hot resistors here now, and a stable behaviour from the PSU and the power supervisor.
Good work!   

[Helix70]

Does the latest firmware need any hardware modifications? Are the hardware modifications optional or required for correct operation?

[prasimix]

No hardware modifications is needed, it was purely firmware issue taking care of existing hardware limits (i.e. used power modules that can deliver max. 155 W of possible 200 W).

[prasimix]

I've missed this for a long time, but now it's added: channel's output control using "third" hand or foot pedal, when nobody is around for assistance, etc.

[Romain]

Hello prasimix

I have an issue with my recently assembled power supply (crowdsupply buy, I know it's been a while!  ).
It did work fine when I finished assembling it, however, I did not go through the calibration at that time. It rebooted once with no apparent reason but it was already late and I thought I would come back to it later.

Today it just won't boot. It keeps resetting after the "Ethernet initialization" message and boot loops...
Disconnecting the 48V PSUs and USB/Ethernet cables didn't help.
The wiring has been checked and seems ok, I followed the assembly instructions here: http://www.envox.hr/eez/bench-power-supply/group-buy-kit-assembly.html

Thinking that it could be a firmware issue, I updated the firmware to the git head this morning: 0fd1a68, without success.
I probed the 5V which seems stable on the Arduino at the time of the reset.

See the video of the boot loop here, hope that helps !


Thanks!

[prasimix]

Please establish serial connection (e.g. Arduino Serial monitor or any other serial terminal app) and let me know what is displayed there.

[prasimix]

I started few days ago searching for possible alternative for AC/DC two outputs module VTX-214-005-0512. It's price at the beginning of this year was under 6 GBP on Farnell, and since then it jumped up to 14 GBP! Why is that, I don't know. For crowdfunding campaign I managed to buy it directly from manufacturer for 5 GBP.
So, we need small and cheap flyback and two most problematic components are controller and transformer. I don't like an idea of winding my own transformer that should withstand AC mains and it is small enough. Fortunately I found for me a new source of magnetics and that is Feryster from Poland. They have a plenty of ready-made transformers and various chokes used for switching applications. In addition it seems that they took extra time to support Power Integrations controllers such as TNY266.
Feryster has two outputs transformer TI-EE16-1534, that is available thru TME (link) and Remagas. I've contacted Feryster and they offer me even bare PCB for TNY266 evaluation that is published in Power Integrations EPR-9.



As a first step I decide to use component values as specified in schematics from page 4 of EPR-9:



As we can see only zener diode is used for regulating output. That is just fine in our case since that 5 V is used for driving relays and LEDs on Arduino Shield, and it's further decreased to 3.3 V with an LDO. TNY266 operates on 132 kHz that is jittered to reduce EMI. It's not audible at all regardless of connected loads that I tried.
Interestingly it starts to work with only 40 Vdc, therefore for first start I've used successfully EEZ H24005 outputs coupled in series for total of 80 V .

You can see below few measurements on primary and secondary side (all measurements on primary side was conducted with isolated scope!).
First screenshot is TNY266 Drain (pin 5) where Vin=230 Vac and without any load (LED only):



Same pin, but this time with 8R2 power resistors connected to main output (+5V):



As we can see peaks goes up to 540 V, despite the used clamping (200 V TVS + schottky). I don't know how it looks to you, maybe there is some room for improvement here and I'll try to use RCD snubber instead. From the other side the PCB layout is really not representative (despite the fact that is designed by manufacturer not Feryster who sold it to me). Actually it looks like they didn't follow even their own recommendations for good layout.

Next screenshot is output ripple on +5V output without load (LED only):



Again, but this time with 8R2 load connected:



Finally we can see how output ripple looks like on second output (+12V) with 66R load connected, and 8R2 on main output. It's expectedly much higher since it lack LC filter as in case of main output. But even as is it should be fine for powering cooling fan.




 

[chris_leyson]

Hi prasimix, haven't looked at the Power Integrations TinySwitch designs but it's good news you've got a transformer and PCB sorted out from Feryster. TNY266 is rated for 700V so a 200V clamp should be good but an RCD clamp might be worth a try. I've been looking at LinkSwitch designs with primary side regulation around the 10W to 12W power level. For the most part it does what it says on the data sheet but sometimes the jitter modulation seems to do unexpected things at certain load currents. I'm still looking into that one.

[prasimix]

Thanks for this info. Please keep me informed if you got any new insights about it. In my case consumption is much lower it's between 3.5 and 4.5 W. But as you said, if that depends of load it could happen even on smaller load. I have to conduct more testing on my side too.
In the meantime I tried RCD snubber with R=100K and C=2n2. It looks better now, for example with Vin=230 Vac and connected load of 8R2 on main output (+5V) I got peaks that is more then 100 V lower:

[Romain]

I couldn't get anything on the serial connection. I noticed that the USB cable was not plugged into the right connector in the video. I changed it to the programming connector (lower connector), but still couldn't get anything at 9600 baud.
I tried other speeds, but couldn't get anything either.
After some tinkering, I managed to get the PSU not to boot anymore 

I had to reflash the Arduino firmware to fix the issue.
By doing that I noticed that plugging the Arduino directly to my laptop (without passing through the Aux power board) allowed it to boot completely! I suspected a power rail dropping and took the following readings on the Arduino shield IDC-10 connector X16.
People can access the pinout here (I so love open source!): https://github.com/eez-open/psu-hw/blob/master/Consolidated/EEZ%20PSU%20consolidated%20r5B13a%20Sheet%209of12.png
IDC-4: PWR_SSTART
IDC-5: FAN_SENSE
IDC-6: FAN_PWM
IDC-7: DOUT2
IDC-10: +5V

The 5V is the interesting signal here, as it drops to about 3.8V when the Arduino resets.

[prasimix]

The USB cable was connected on the right port (i.e. Native) that is set as default in firmware for version 1.01 and newer (that is mentioned in building instructions). Did you try to establish serial connection with USB cable plugged in directly into Due?

[Helix70]

Hi guys,

My power supply is going great. One question, is there a known issue with the USB? I can't connect to my power supply or the TechSpace's power supply with either of two HP Windows 10 laptops. My Home PC works and finds the Arduino Due port, so does my Mac laptop, but not the Windows 10 HP laptops. I just get a "Composite Device" with an exclamation mark.

When I try a bare Arduino Due, no problem with either laptop, comes straight up.

I am connected to the native port on both power supplies. I thought I read something about the isolator chip, but can't find it, not sure if it was relevant.

Thanks

Josh

[prasimix]

It's reported as issue #53.

Unfortunately it's related to AdUM3160 isolator chip. It has speed selection and it was hard-wired to Hi-speed on r5B12 revision (used in crowdfunding campaign). It seems that some PCs don't like it and I'm not sure is that a software (operating system) issue or hardware (USB ports).

Some people reports that if you connect it to Programming instead of Native port on Due it works fine. If you are going to make that changes for further working with firmware add change before compilation as stated in post #750.

[Helix70]

Thanks. Seems not OS related, my desktop (i5-3570K based Asrock USB3) and laptop both running latest 1709 Windows 10. Both laptops are intel based (one Core i7, one Core i5). My 2009 macbook works (also intel, but a much older Core 2 Duo).

[prasimix]

Maybe AdUM3160 is too "dumb" or new PC chipsets USB support is too "smart". It's pity that I didn't noticed that during development. I'm using five years old Gigabyte motherboard with Ubuntu and it works just fine.

[Helix70]

Yeah, weird. I also tried a USB 2.0 hub and a USB 3.0 hub with the offending laptops, to no avail. At least I know I am not crazy (well, not more crazy than usual).

Thanks, and what a great project this is. I am still waiting for the GUI designer to be released so I can change the colours, as the angle I use it the yellow/brown is hard to see.

[Romain]

Hi, I tried to establish a connection with the Native USB port through the AUX board but it doesn't enumerate as a serial port (W10 Pro 1709).
As you suggested, I connected directly my laptop to the Native USB port, and this allows the Arduino to complete the boot process. Here is the trace:

Code: [Select]

**TRACE: Ethernet not connected!
**TRACE [2017-12-08 23:55:57]: PWR_DIRECT -> HIGH
**ERROR [2017-12-08 23:55:57]: 630,"Fan test failed"

Let me know if you need me to probe anything on the Shield or the AUX board.
Thanks

[prasimix]

That's all what you have? If that is a case then something is wrong on AUX PS board. Fan error is expected since you have not connection with AUX PS board. Try to power up AUX PS board only and check output of TR1. Maybe you have the same issue as reported in post #762. Another usual suspect is ADuM3160 (IC17).

[Romain]

About the serial, yes that's all I have, I start Putty as soon as the port shows up in the device manager. I just managed to get a few lines more 

Code: [Select]

**TRACE: Touch screen calibration points: 3533, 447, 439, 3520, 415, 480
**TRACE: Touch modification ENABLED
**TRACE: Ethernet initialization started...
**TRACE: Ethernet not connected!
**TRACE [2017-12-09 00:59:31]: PWR_DIRECT -> HIGH
**ERROR [2017-12-09 00:59:32]: 630,"Fan test failed"

Output of TR1 is as follow:
Voltage across C77 (not mounted) is 17.57V when the standby switch is ON, or 10.44V when it's OFF  .
Voltage across X5-1 and X5-10 on AUX board is 4.99V.

I do not understand the fan error since the ribbon cable is plugged between the Shield board and the AUX board.
I can make pictures of the assembly if that can help diagnose the issue.
The self-test also indicates "SD card failed".

[prasimix]

Can you add a small load on +12V rail (e.g. 220 Ohm resistor) and measure voltage once again?