Can this power supply be defective even though it provides correct voltages?

[tigrou]

My main computer died two weeks ago. While under use, I got a black screen (but computer continue running  ).
I shut it down. Then, upon start :

- Fans are spinning, motherboard LEDs are on.
- no USB power (eg: keyboard or mouse)
- no video output

I tear it down completely (including removing CPU) : same as above. The Asus QLED (which is supposed to report issues) does not light up at (it should, since there was no CPU installed)
I tested PSU manually by shorting PS on : it provide correct voltages on all pins (including "power OK" pin).
=> I concluded motherboard was dead.

I ordered new 2nd hand motherboard (because new ones are not sold anymore).
Upon start : it starts then stop immediately. Like there is a short circuit. I tested the pins of the motherboard for possible short circuit between ground and the rest but couldn't find wrong. After those events, PSU seems still OK and provide right voltages.
=> Either 2nd hand motherboard is dead or PSU is the culprit.

Is it possible that beside providing correct voltages PSU cannot handle any load anymore ?
What other tests can I make ? I don't have advanced equipment like an oscilloscope.
 I am planning buying a cheap (20 bucks) PSU on Amazon for testing with a brand new PSU.
The old PSU (Corsair RM 750x) is still under warranty.

[M0HZH]

Yes, the PSU can be malfunctioning but still provide normal voltages at no load.

Try to measure the main +5V when computer is on, if there's no USB power then that might be the culprit.

+12V must have some power if fans are spinning.

Also, as a matter of principle, you should test without HDDs/SSDs or any other peripherals plugged in.

[alm]

In addition I would also measure ripple with the computer turned on. Depending on the DMM, measuring across each rail on the AC(m)V range might give reasonable results.

The ATX spec specifies a maximum, and given the symptoms I'd expect ripple to be much higher for it to cause these problems.

[tigrou]

Try to measure the main +5V when computer is on, if there's no USB power then that might be the culprit.

measuring across each rail on the AC(m)V range might give reasonable results.

Thanks to both of you. What you say make sense.
Maybe old (supposedly dead) motherboard don't care and stay "on" despite unstable voltages or no +5V, while new one is more picky and shut itself off.

I have a Aneng AN8008 so it should not be a problem.
"Measuring AC(m)V" : I guess it should be done under load as well.

EDIT :
Here is expected voltage and ripple from ATX 12V standard :

[alm]

"Measuring AC(m)V" : I guess it should be done under load as well.

Yes, ripple normally gets worse under load.

[MathWizard]

Do you have an oscilloscope ? Or any 12V lightblubs, or something else that runs off 5/12V that you be willing to risk testing it with ? Computer PSU's do a pretty good job of detecting problems and shutting down. They look for under/over-voltage on the inputs and outputs, and they look for over-current and overheating.

If you had a scope and could probe a few pins on the controller chips, it would see if they are going into shutdown, if they are still working themselves.

I've fixed a few computer PSU's, and I think they all had different problems, and only 1 or 2 was electrolytic capacitor's wearing out. (some cheap brands/models tho are bad for it tho)

[tigrou]

I did some measurements using the old motherboard (which still turn on) and everything disconnected (no CPU, no RAM, no SSD, ...)
Here is the results :

Voltage: seems to be within specs (with motherboard ON or OFF). The 5V is there even with motherboard ON.
I did the measurements in V DC mode because mV always show "0L" regardless the range selected (auto or manual).

Ripple : measurement made with motherboard ON.

	(1) mV, auto range	(2) V, manual range, 2 digits
+5V	53 mV	00.10 V
+3.3V	71 mV	00.06 V
+5VSB	53 mV	00.10 V
+12V	20 mV	00.14 V
-12V	13 mV	00.13 V

1) it wasn't stable : it displayed 0 most of the time then shortly oscillate around a value (what is shown in the table), then it repeats.

only -12V was stable in autorange.  I also got a 60 mV stable between two ground pins (don't understand why).
mV manual range displayed 0 regardless what was selected.

2) stable.

Did i measure ripple correctly ? Why does it oscillate, is ripple to low for DMM to catch it ?
I did measurements several times and make sure the probes were properly connected.

[alm]

I wrote AC(m)V just in case the ripple was so large to overrange on the mV ranges. There's no point using the ACV mode to measure 20 mV, it is unlikely that the meter is accurate below say 1% of range. I don't know why the reading is fluctuating in the mV range, but either way those ripple figures are not indicating a problem to me, at least not at this low load.

If you try to measure 12V on the mV DC range, of course it will overload the meter. You have to switch to a range higher than 12V dc to measure 12 V dc. The AC ranges are AC-coupled, however, so the DC part is blocked and only the AC part (ripple/noise) is measured. That's the theory at least. The AN8008 has an AC bandwidth of 40 Hz - 1 kHz, so it's certainly not measuring all the noise.

[MathWizard]

Those ripple voltages look fine, for no load anyways. Yeah IDK what it would measure on a scope tho, but it sounds fine enough.

Assuming the problem is the PSU, and not the CPU or GPU had a short or something.

If you wanted to proceed to taking it apart, well there's safety issues of course, but you can measure a lot of stuff unplugged. Usually when I work on a PSU (just my own stuff), I just start at the plug at work my way in, just doing some resistance and diode checks. I learned the basic layout of computer PSU's that way. I've fixed a couple that way too, just finding the issue when unplugged.

You really want a scope tho.  Even 1 of the $20 mini-ones could let you see more than a DMM would, like a reset line going high or low.


The PSU's have a mini-PSU inside, that provides Vcc for the chips on the high voltage side, and 5V standby power on the low voltage side. A lot of stuff must be working on yours.

It still could be some simple, easy to find problem tho, but there's plenty of hard problems, and then desoldering some parts of these can be a nightmare.

[M0HZH]

Sounds like it's more likely that the CPU is dead.

[tigrou]

Sounds like it's more likely that the CPU is dead.

I agree. I phoned to a computer repair center yesterday. The guy on the phone his a veteran that has been repairing computers for almost 30 years. While he couldn't help me (he has no spare parts to be tested), he told me it's likely CPU to be dead.
I did the following test : I tried to power the new motherboard without CPU : it started for the first time (LEDs went on and it stayed like that until I shut it down). With CPU, it start then shutdown immediately.

I replaced the motherboard because I was confident that the Asus QLED would report something if CPU was dead (but it's not always the case, based on what I read on reddit and what repair shop told me). Also I did not expected a high end Intel CPU (9700K) to die like that. Usually, it's either PSU or motherboard which goes bad.

You really want a scope tho.  Even 1 of the $20 mini-ones could let you see more than a DMM would, like a reset line going high or low.

Yes, I agree. That will probably be the next purchase.
I had in mind to open PSU for inspection but since it still under warranty (until 2028) I will not attempt it.

[tigrou]

Something I don't understand related to AC ripple measurement :

In autorange mode I never got something stable. It always showed me 0 then oscillate around some value (eg: 50mV) then show 0 and such. Why ? Is ripple to low so DMM cannot detect correct range ?

The only way I got something stable is in AC V manual, with 2 digits.
For the +5V, I got 0.1V which is 100mV and also outside max allowed ripple.

Here is the measured ripple for that PSU : https://www.tomshardware.com/reviews/corsair-rm750x-v2-psu,5585-9.html

[Jwillis]

If the power supply is not at fault you should still be able to enter the bios? If the CPU is at fault the BIOS will tell you that it's not detecting it. Bios should still work with or without CPU . Remove all peripherals including drives and memory.  Try with a different power supply . Even a lower powered supply will be enough power to enter the bios . If theirs still a blank screen try a different graphics card. If you still get a blank screen its your mother board.

[M0HZH]

No.

The "BIOS" is really just a flash chip with a bootloader and some stored settings, you still need a CPU, RAM, a working graphics card and a working motherboard to run it.

[Jwillis]

No.

The "BIOS" is really just a flash chip with a bootloader and some stored settings, you still need a CPU, RAM, a working graphics card and a working motherboard to run it.

I've done it many times on mine. But it depend if your motherboard supports it .

[M0HZH]

No.

The "BIOS" is really just a flash chip with a bootloader and some stored settings, you still need a CPU, RAM, a working graphics card and a working motherboard to run it.

I've done it many times on mine. But it depend if your motherboard supports it .

I'd be curious to know what motherboard is that. Can I have the make and model?

It would need to have its own CPU+memory+graphics to actually display/take inputs from the keyboard and change settings, as far as i know not even server boards have those features.

[tigrou]

+1 BIOS is supposed to be only some code stored in a chip.

It looks like even Q-LED need working CPU to report status (eg : incompatible CPU).

EDIT : A colleague lent me one old PSU (but still working) for testing : it's same behavior than with my current one.
So CPU is dead (this is a rare case). There is also a possibility that it is motherboard that killed it.

[tigrou]

Some update :

The PSU is indeed fine because I installed it on a new build and it works well.
I am now planning to sell the old motherboard. There is a slight chance that it's the one who killed the CPU and the last thing I want is to sell a defective motherboard that kills CPUs.
Usually this happens when one of the VRMs fails short.

I used multimeter and AFAIK there is no short between Vcore and VDD / VSS.
This can be checked very easily by measuring contacts between phase coils and the 12V CPU power.

However : when the dead CPU is installed on the motherboard : there is a short between VSS and Vcore ! So it just failed catastrophically.
The only phase not being shorted by CPU is the last one, which is used for iGPU (VCCGT pin).
I measured about 4.5 ohms which is to be expected.

I confirmed those measurement by also measuring checking pins on the CPU.