PSU issue.

[atlanticjim]

So I get this machine (Dell Dimension 4600i) in with the symptom of being hard to start for a month and a half , he has to press the power button a few times before it will start and then finally would not start. Off the top of my head I say. . power button or PSU. Pull the power button and test it and it is just fine.
Used my PSU tester and the PSU reads ok on all outputs.
I noticed that there are the "1-2-3-4" light on the back panel but none light up at all, so I swap the PSU for one laying on the bench and the machine boots without a problem.
Of course I call the client, get the ok for the job, replace the PSU and cmos battery and all is good.

My question is this. Why did the PSU read ok on all outputs but not work in that machine? I bench tested it again and got the same good readings.

 

[normbarb]

PSU Testers

I have had the same issue with PSUs testing good but not working. I believe that a lot of these testers are doing open circuit test and the unit is failing under load.

Norm

 

[Styxbound]

Psu

When you say you bench tested was that with a multi-meter or psu tester, Jim?

 

[Feraxks]

I have had the same issue with PSUs testing good but not working. I believe that a lot of these testers are doing open circuit test and the unit is failing under load.

Norm

This is quite common whether it be a PSU for a computer or a PSU for a piece of military hardware -- saw it a lot during my time in the Navy.

 

[Cambridge PC Support]

and that's why I never use a PSU tester

much easier to swap in a known good PSU

 

[joydivision]

The power supply is working fine under a very limited load, as soon as you put a load on it it fails. This is why it 'tests' ok but dosn't work in the real world. I've seen a few car batteries which read 13v but yet the engine still won't turn over. Its because the battery cannot deliver the ampage.

 

[atlanticjim]

Thanks for the replies.


I guess that I will treat it like I do in medicine. The psu tester has a high false negative. (will often test as ok when failed) but low false positive (if the PSU is not putting out on a side, it will show it on the tester).

So then if negative test on the PSU tester, I move along to swapping PSUs.

Thanks again for the input.


btw: using psu tester, not multimeter. If I am going to go to multimeter, it is just as efficient to swap a known good psu since I am not going to repair a psu.

 

[Styxbound]

Agreed power supplies are not really practical to repair. I've spent a lot of time pulling them apart usually to find it's some part not easily replacable. I did succeed with one a couple of weeks ago when it was only one bad cap, though.

A very pertinent point which has been made here for testing all sorts of things_even cmos batteries_ is that voltage is only a half-test; one must also test amperage to know anything. I understand that even a horse-radish can show a voltage, butof course no amperage.

 

[anonymous Mac Tech]

The PSU showing voltage just means there is trickle voltage coming to the board and that's about it. I always do a swap with a known good PSU just to be sure in these situations.

 

[hondablaster]

Out of curiosity can you take the bad psu and put it in a healthy computer and get the same symptom?

Just for my own personal wisdom........ I know it seems pointless but I like knowing these little things..... LOL It would be interesting if it worked fine.

 

[Styxbound]

Theoretically if you put a good psu and it then works, then it was a healthy machine, but___

There's something I'm still wrestling with about all this. Numerous times I've had boards on the bench which wouldn't start with SOME known good supplies but would with another; I've been playing with one like that today. It seems to be somehow related to the last power state, though I haven't been able to pin it down. For example I've taken a psu from a troublesome machine which wouldn't start, put it in one working well, then go back to the first and have it turn on. This is one problem I feel like I'm close to solving, I just need to get a little more methodical about testing.