Beginner's question about PDB and/or PSU fault/replacement - PT-825 to replace PT-826?

[jkn_sth]

Hi all
I am trying to diagnose a power problem on my Supermicro Home server. It's an old build but new-ish to me.

I think I am mostly there but would like to check a few points. This is my only home server and I have learned a lot by browsing this and other forums, but I still have a few questions.

The server is a Supermicro CM826 Chassis with a X9DRH-iTF motherboard. There are two PWS-920P-SQ power supplies, although I have only used one at a time. There is a PDB PT-826-8824 between the PSUs and the motherboard.

I have previously had the server running fine, but left it off for some months (too expensive to run!). When I powered it up again after that gap, things started to go erratically. The server would not power up (no fans or BIOS screen etc.) from time to time. After a few cycles things stopped totally.

I did a bit of exploring as to what could be wrong, and probably went down a few rabbit holes. IIRC there was a fried-looking insect on the motherboard and I feared that that might have caused major damage. I also stripped down the board (no CPU or memory etc.), with no joy.

Eventually I got a bit smarter and tried plugging in a different power supply - a 750W standard ATX PSU. That at least showed that the BMC was working. And then to my surprise I was able to start the server properly with this PSU, via the front panel switch.

[aside - I had guessed that because there is an I2C connection between the motherboard and the PDB, that that is what was used to power up the PSU. I think I now realise that that is not the case - the I2C is mostly used to get sensor information from the PSU/PDB, and that the power supplies are true on via the 'usual' pin16 low mechanism. See more on this later]

So having refitted CPU memory it looks like my motherboard is working properly. I am now trying to work out if it is the PSUs or the PDB which is faulty, and looking for a bit of certainty on this.

Some points:
- I tried to take pin 16 of the 24-pin ATX connector low with the SMC PSUs/PDB - no effect (ie. the PSUs did not start up)
- The LED at the rear of the PSUs are both amber when plugged in at the mains.

I am currently suspecting the PDB is the problem, but I would like to narrow this down a bit further. The PDB I have is PT-826-8244. These seem very difficult to obtain in Europe (I am in the UK). Having looked a bit further, I realise that my PSUs are the narrow 19-pin variety, and my PT-826 PDB is the wide 23pin variety.

So my Question #1 is: am I right in thinking that I could fit a PT-825-8244 PDU to this chassis and PSU instead of my PT-826?

Another point:. As mentioned above I now realise that the PSUs are turned on in a similar way to standard ATX supplies, by taking a standby pin low. I tried to do this with the output signals from the PDB to the motherboard without success... but if the PDB is the problem then that might not prove much.

If I were to be confident of the correct equivalent pin on the edge connector of the SMC PSU ... would it be a good test to (carefully!) take that low, and see if it springs to life? I am trying to be sure that the PSUs are OK, and hence the PDB is faulty, before orderin a PT-825 PDB as a replacement for my PT-826 PDB.

Oh, one more thing - I don't suppose there are any well-known and easy to fix PDB faults that might account for my trouble?

Thanks for any words of wisdom...

Jon N

 

[CyklonDX]

from what i've read i would replace the pdu it might be a bad cap.
But check whole mobo for any bulged caps, and replace the cmos battery - then check again.

 

[jkn_sth]

Hi
Yeah, perhaps I should have said that when I was suspicious of the motherboard (removing DRAM, CPUSs etc) I looked it over for bulging caps. I also cleared the CMOS memory, and replaced the CMOS Battery.

I have tried to perform a visual inspection of the PDB (I have removed it from the chassis but not disassembled it) and can't see anything obvious, sadly.

 

[CyklonDX]

you can check your rails and components if you have multimeter.

 

[jkn_sth]

you can check your rails and components if you have multimeter.

Well, this is my question really ...
There are no volts at the output of the PDU when trying to use the SMC PSUs. I cannot remember right now if the BMC LED is active in that setup.
What I would like to get confirmation of is whether I can test a PSU, not connected to the PDU, by taking pin ... 6 IIRC ... of the finger connector Low?

I think that should cause the 12V supply to start, and maybe the PSU fan to run.

If that works, I could perhaps try to plug the PSU into the PDU, take the equivalent pin on the ATX supply low, and see if that gets back to the PSU.

 

[CyklonDX]

first try to short 2 pins on pdb - confirm its not powering.

i recall there used to be a pin breakdown for sc psu's i just can't find it anymore - and likely its not as easy to power it on in first place.

 

[jkn_sth]

Yes, I have already tried this (I wrote above) - no joy. But that only tests (PSU + PDB), it is the PSU on its own I want to try the equivalent test on.

 

[CyklonDX]

I think you would have to apply 220-300ohm resistance onto pin5 and ground to power it on.



buuut... if psu was at fault you should have amber light as i recall on the front of the psu when plugged in. *(green blinking when its good - and not powered on - solid when powered on.) I don't recall them having a breaker - but double check if they have a breaker.

 

[jkn_sth]

Thanks for the image. I have not been able to find a definitive pinout for the PSUs I have (PWS-920P-SQ); there is a generic ASCII-art one here:

Supermicro PSU pinout

which states pin 5, not pin6, for /POn:

                          +12V                          DCG SMPF /POn  SDA  SCL  A0   A1   IS
-------------------------------------------------|---- ---- ---- ---- ---- ---- ---- ---- ----
 19   18   17   16   15   14   13   12   11   10    9    8    7    6    5    4    3    2    1
 38   37   36   35   34   33   32   31   30   29   28   27   26   25   24   23   22   21   20
-------------------------------------------------|--------- --------- ---- ---- ---- ---- ----
                             Gnd                             +5V SB    Gnd  -5V  GS  +12S  GS



                               +12V                               DCG  SMA /POn  SDA  SCL  A2   A1   IS    +12V
-----------------------------------------------------------|---- ---- ---- ---- ---- ---- ---- ---- ---- ---------
 23   22   21   20   19   18   17   16   15   14   13   12   11   10    9    8    7    6    5    4    3    2    1
 46   45   44   43   42   41   40   39   38   37   36   35   34   33   32   31   30   29   28   27   26   25   24
-----------------------------------------------------------|--------- --------- ---- ---- ---- ---- ---- ---------
                                Gnd                                     +5V SB   Gnd  Fan  GS    -   GS     Gnd

buuut... if psu was at fault you should have amber light as i recall on the front of the psu when plugged in.

erm ... the PSUs do have an amber light on (the back), I state this in the original post. But I think that means 'PSU in standby', not PSU fault?...

 

[CyklonDX]

oh yeah, thats a psu fault. but both of them?

 

[CyklonDX]

just tested how psu's look like when plugged but powered off on 1u server
*6018U-TR4+ | 1U | SuperServers | Products | Super Micro Computer, Inc.
PWS-1K02A-1R - blinking green light.
// it was amber for first 2 sec after plugging it in, then started blinking green - when powered on it was solid green.

I have 2 other sc boxes, but i can't test them until weekend.

// but your psu is like 25 usd +-, so it shouldn't be too expensive to get replacement.

 

[jkn_sth]

Hmm...

My Chassis manual has this to say about the LEDs:

3-5
Power Supply LEDs
On the rear of the power supply module, an LED displays the status.
• Solid Green: When illuminated, indicates that the power supply is on.
• Solid Amber: When illuminated, indicates the power supply is plugged in and turned off, or the system is off but in an abnormal state.
• Blinking Amber: When blinking, this system power supply temperature has
reached 63°C. The system will automatically power-down when the power supply
temperature reaches 70°C and restarts when the power supply temperature
goes below 60°C.

So it seems to me that steady amber indicates that the PSUs are in 'standby' (plugged in but not turned on)?

 

[jkn_sth]

// but your psu is like 25 usd +-, so it shouldn't be too expensive to get replacement.

The prices are higher in the UK ... but I am trying to work out if it is worth buying a PDB first, or a PSU. Or if my PDU is broken to the extent of frying PSUs connected to it...

 

[jkn_sth]

I have just checked one supply, and confirmed that the LED is steady amber when 'bare' (not connected to PDB)

I think I am going to try to short pin 6 (not pin 5 as per your image) to GND (on the *underside* fingers - I think your image is wrong there too) and see what happens. I am a bit nervous about it though, it was the feasibility of doing this I have been trying to establish....

 

[luckylinux]

The prices are higher in the UK ... but I am trying to work out if it is worth buying a PDB first, or a PSU. Or if my PDU is broken to the extent of frying PSUs connected to it...

Funny because the Supermicro PSUs I bought at very low Prices were from the UK. IIRC there are at least 2 Suppliers that are good, possibly 3. But you have to send them an Offer . Not sure you'll get the same Prices I got though, I bought >= 5 Pieces at a Time though.

EDIT 1: since my preferred Supplier does NOT have this exact Model in Stock right now, the Cheapest I found by quick Search are around 25 GBP

Supermicro PWS-920P-SQ Power Module 110V AC, 220V AC | eBay.de

Supermicro Power Module 110V AC, 220V AC. Modular & Fixed Port Switches. Hard Drive Caddies. Power Options. Power Cables. Need help?. RoHS EN IEC 63000:2018. Safety EN 62368-1:2014 EN 62368-1:2014+A11:2017.

www.ebay.de

Bitte entschuldigen Sie die Störung...

Note that I CANNOT recommend these Sellers as I never bought from them in the Past.

Other Results:

Bitte entschuldigen Sie die Störung...

Other Articles from one of my favorite Sellers in the UK (exact Model you look for is NOT in Stock ATM):

Bitte entschuldigen Sie die Störung...

 

[CyklonDX]

I have just checked one supply, and confirmed that the LED is steady amber when 'bare' (not connected to PDB)

I think I am going to try to short pin 6 (not pin 5 as per your image) to GND (on the *underside* fingers - I think your image is wrong there too) and see what happens. I am a bit nervous about it though, it was the feasibility of doing this I have been trying to establish....

careful - make sure you are not using 12V rail - one side is ground other is 12V

 

[seadog2441]

From my experienxe those PDB (ie 825 and 826) are interchangable, the main different in them should be the output plug, some got more PCI 6pins or PCI 8pin, could not remember which is which.
I was in the same situation as you when running various 825 and 826 chassis(s), I love those, they are so versatile and interchangable. Being old chassis, eventually I have to replace PSU, PDB, Backplane... one after another and keeping spare. The PDB 825 is the best option in my part of the world, cheap, easy to find if I dont need to run anything special (ie high power GPU). At bare minimum they are doing fine for 826 chassis.
The PSU you got, 920SQ, on another hand is one of best out there for SM chassis included the mighty 846/847, and they are not cheap but worth every penny.

 

[jkn_sth]

^^^ thanks for the perspective on PDB 825 and 826s. I think the final part of the part no (-8824 in my case) indicate the output wiring. There is an SMC link which lists what each one has. In my case I can get hold of PT-825-8824, ie. with the same output wires as the PT-826-8824 I have. I only need to fit a Molex to SATA power adapter for my booting SSD. I must have had one of these adapter cables fitted to my PT-826-8824 but it seems to have fallen into a wormhole, I cannot find it anywhere. There is no easy way of powering a SATA SSD from the motherboard, I think?

Anyway ... I did the experiment I mentioned above on both of my PSUs; running 'bare', I carefully connected pin 6 (/POn) to Ground. In both cases the PSU fans started up.

I also checked that with the SMC PSUs + PDB connected to the motherboard:

• 5V SB was on
• The BMC light was flashing (I couldn't remember if this was te case
• the /Pon pin on the ATX connector (pin 16) was high
• and, when I press the fron panel switch, the /POn pin goes low (but the PSUs fans do not start)

So given all of this I am now fairly confident that it is indeed the PDB that is the issue. I shall order a PT-825-8824 and hopefully this will fix my problem.

 

[jkn_sth]

well, good news for me ... I received my PT825-8824 and quickly tried it. It looks like things are back up and running!

curiously the I2C header does not have as many wires (3 not 4) as my PT826, but I think the missing signal will not make any difference. Apart from that it looks like (as expected from the '-8824') the wiring is compatible with what I had. So hopefully I am back in business...