Troubleshooting of redundant F750E-S0 Dell PowerEdge Server

[omega]

Hello all,
as a beginner electronics hobbyist, after a few years I would like to place another post on this Forum section, given that I did not succeed in finding any useful advice among the other posts.

Over the last years, I have been using a PowerEdge Dell Server with two redundant PSUs, namely the 750W F750E-S0 ones (aka 06W2PW). Unfortunately, last summer one of them failed (perhaps owing to an overheating), and the server had for working to take into account the other one only. Of course I opened and tried to troubleshoot the failed PSU, but each cap I tested seemed to be fine both as regards its onboard capacitance and ESR values. Also other components seem to be fine, but I did not test all components extensively.

Moreover, since I had no luck to find any schematics of F750E-S0 in the web, I had to reverse engineer this to me rather complex PSU (much more than a standard ATX) and discovered myself some interesting features, such as its active PFC section (with its switch-off relay) and two separated secondary rails (one of which lies in a soldered, extra-miniboard difficult to access and troubleshoot). On the other hand, I did not investigate the failure any further, and put the pieces of the PSU in a box.

Now that the other PSU is also failing, my server has gone eventually off, and so I need to find a solution for both the redundant PSUs.

Anyway, I observed a strange behaviour in the last failing PSU (frankly, I don't know whether it was also present in the former failed one), through which the expert eyes of the community could hopefully address me to a solution to the trouble: when I pull off the PSU, then I put it back to its housing and press the server on button, apparently nothing happened and the PSU troubleshoot LED is off, but after say 5-8 mins (and some occasional and random noises and clicks), the server "magically" starts and the LED is green (the action also occurs in the PSU only, therefore it is not related to the server motherboard or its peripherals).

Of course, that behaviour (which I guess will only last for a short time) makes anyway the server unstable and I have to fix it, but I guess that the some-mins-delay in powering on might be suggestive of a failure of (bad)caps capacitance. But which one(s)? Have I to investigate in the primary or secondary circuit to find the culprit?

Your attention and possible help will be greatly appreciated. and I also think that your hints may be useful to other redundant PSU servicers.

Thanks

[CapLeaker]

Delayed start problem is in the primary, usually a small starter cap. Having unstable server after start, the problem gets trickier. You don't really know if the problem is in the secondary caps or the main filter cap. Depending on how many hours that thing already ran and how much of a reliability you need, you change all the caps, or just a few. Up to you.

[omega]

Thanks for your answers!

Having unstable server after start, the problem gets trickier.

I did not make myself understood well. Once the PSU is started, the LED is green and the server wokrs fine.

Anyway, I will look for that "small starter cap", but if I remember properly, besides the security Y and X caps, the first caps in the primary circuit are the two big caps (at the centre of the picture), respectively 330uF 450V and 47uF 500V, topologically connected through a diode. I did not even understand that odd topology and the duplication of the big caps.

[CapLeaker]

Ah, so it only doesn't start after a power outage or after a scheduled turn off…
if that's your only problem, it's the start up cap that's bad. Borrow wifey's hair dryer and blast some heat in the PSU on the back for like 30 seconds. The server should instantly start.

[omega]

if that's your only problem, it's the start up cap that's bad.

Thanks again for your hint! Actually, I have found that almost hidden cap beside the big one. Its specifications are 100 nF, 450 V. But I have tested its capacitance and it seems to be fine (140 nF), so as the big cap. Is it really the culprit of this delay? By the way: I would be happy if i could measure the high-voltage loss of these two caps (I have tested both caps through a standard capacimeter). Which is the right tool to accomplish these measurements?

[CapLeaker]

This is the wrong cap you looking at. You are looking for a small electrolytic cap in the primary. Maybe my bad for not clarifying this earlier.

[omega]

This is the wrong cap you looking at. You are looking for a small electrolytic cap in the primary. Maybe my bad for not clarifying this earlier.

Oh I see. I can confirm you that in the primary side there is NO (little) electrolytic capacitor besides the two big ones, which however at the voltage of my testers seem to be both fine as regards both capacitance and ESR. In the primary section, there are just two red polypropylene and a few SMD ones (in the other side of PCB). Maybe that (component crowded) PSU does not follow the ATX standard. At this point I do not know what to check.

[sam_sam_sam]

What small value capacitors are on the daughter boards because sometimes they are hiding on these boards

[omega]

What small value capacitors are on the daughter boards because sometimes they are hiding on these boards

Thanks for your comment. In my case, in the daughter board (the standby voltage rail?) there is no electrolytic cap before the mosfet and the stepdown transformer. After that, there are one or two concealed Capxon 270 uF electrolytic caps, virtually impossible to reach & replace without desoldering the daughter board. Anyway I think they might have little to do as regards the huge startup-delay trouble because they work to rectify the transformer wave.

[CapLeaker]

Any pictures?

[omega]

Any pictures?

Do you mean the picture of the hidden cap in the SB rail?

[CapLeaker]

Maybe… any cap near a heat sink or any heat source is also suspect. It should be a cap near a PWM / PFC IC.

[omega]

Maybe… any cap near a heat sink or any heat source is also suspect. It should be a cap near a PWM / PFC IC.

as I wrote, and as you can see yourself in the first pic, this redundant PSU is a masterpiece of crowded component. I believe it might get hot with use. Anyway, the Capxon cap(s) I showed you is(are) buried in the secondary side of the daughter board which is difficult to extract and inspect. I think I would need low melt solder (which I don't have) to remove their many solderings.

[CapLeaker]

omega
Yeah... its got to be in the primary. Here is a lil cheat sheet on things: Google "P5104" and download that pdf. there is a little example schematic on how this thing works. There is something called VBoot. See how it works.

[omega]

omega
Yeah... its got to be in the primary. Here is a lil cheat sheet on things: Google "P5104" and download that pdf. there is a little example schematic on how this thing works. There is something called VBoot. See how it works.

Actually, on the mainboard of the PSU there are two NCP5104 ICs. Only one of them (U1) has a cap (C14) between the pin 8 (VBOOT) and 6 (BRIDGE), but it is an SMD one and seems to be fine (about 200 nF). Also the diode on the right in the picture seems to be fine (0.6 V drop).

[CapLeaker]

omega
Look if Vcc pin 1 reads within spec during startup. Did you try heat to see if the PSU will start better when warmed up?

[omega]

omega
Look if Vcc pin 1 reads within spec during startup. Did you try heat to see if the PSU will start better when warmed up?

Sorry, at present I can not switch on the PSU because, as you can see in a picture of mine (comment #6), I have removed the two big filter caps. I can only test the board turned off.