Defective power supply?

Topic

New Reply

Hi,

Apologies in advance for the lengthy post. Anyone patient enough to read through all this has my sincere thanks!

In short, I have a problem with my custom built PC that I think may be caused by a defective power supply unit (PSU) and I’d like your input as to whether you believe my hunch that the PSU is the culprit seems correct and if so whether I should replace the PSU sooner rather than later.

Here’s where this starts to get lengthy: the affected computer was assembled a few years ago and I have it connected to a wall socket (along with monitor, speakers and printer) through a power strip. The power strip has its own on/off switch and when I’m done using the computer for the day I switch it off (of course after having shut down the computer).

The problem I’m seeing – which happened since the very day I finished assembling the computer – is the following: I turn on the switch on the power strip and then push the power-on button on the computer, but the system does not boot up, it sits there with a blank screen, fans spinning and discs working without ever reaching the POST screen nor does Windows load. When this happens the power-on led remains dark and a led on the motherboard lights up indicating a “CPU fail”.

When this happens, I press the power-on button for a few seconds until the system powers down and then press that button again: the power-on led is lit, the screen lights up with the POST screen, Windows loads properly, no warning leds on the motherboard are lit and the whole system works correctly, i.e. it is perfectly stable and I can put it through as much stress as I like without trouble. The few BSoDs or freezes I’ve had were always traceable to a software problem and I never noticed any tell-tale signal of hardware failure.

After several tests across the years (see list at the bottom), I suspect the culprit of this odd behavior may be the PSU: maybe it’s defective and its over-current protection kicks in unnecessarily when the system is “cold” and I’m powering it up the first time after I switch on the power strip. I’m assuming the “CPU fail” led on the motherboard simply lights up because the over-current protection prevents current from reaching the CPU to protect it.

Do you think the above problem may be caused by a defective PSU? If so, would you recommend that I replace the PSU right way?

I didn’t use a “cheapo no brand” PSU for this computer, it’s a 650W Antec TruePower Classic PSU (TP-650C) which is more than enough for this system (at most my system pulls about 420W), but we all know that even branded products sometimes can be defective.

BTW, the problem seems to be increasingly frequent since a few days ago. That’s when I did something different after the system failed to start correctly. In fact, after shutting off the system I didn’t press the power-on button right away, instead I switched off the power strip to disconnect an unneeded battery charger that was plugged in, switched on the power strip and when I pressed the computer’s power-on button nothing happened at all: the computer seemed to be dead with no fans or disks working, no led lit on the motherboard. I assumed either the power supply or the motherboard had died, but then after a few minutes I could power up the computer again and everything was working just fine all over again.

Here’s a list of things I tried to trace the source of the problem:
– replaced the CPU with a different one;
– replaced the video card with a different one;
– replaced the memory modules;
– ran the system without a video card plugged-in and using the on-board GPU;
– ran the system with a single memory module;
– put memory modules in different memory slot(s);
– used a different power strip;
– used a different wall socket at home as well as a wall socket in an entirely different building;
– updated the motherboard’s BIOS at least twice to use the latest version;
– disabled CPU C-states in BIOS;
– disabled use of XMP profile for the memory modules;
– checked for failed or close-to-failure capacitors on the motherboard and found none;
– checked for damage on PSU cables and found none;
– checked that all cables are properly plugged-in onto the motherboard;
– connected the reset switch to the power-on pins on the motherboard to test whether the power-on switch is defective.

The only things I didn’t try so far are replacing the PSU and replacing the motherboard, but because the problem generally happens only on starting up “cold” and the system is otherwise stable, I’m thinking it’s more likely that the issue is with the PSU rather than the motherboard.

Any helpful input is appreciated, thanks!

Reply | Quote

Replies

I had a PC that did something very similar to this.  This was several years ago, so the details are a little fuzzy, but when I took a little peek inside the PSU (took the cover off), several of the capacitors were bulged, and replacing the PSU eliminated the problem.

There’s a good chance that’s what is happening. How old is the PSU?

When the computer is first turned on, the ATX specification requires that the PSU do a self-test, and if it finds the voltages to be stable and within limits, it sends a power-good signal to the motherboard, telling the motherboard to go ahead and boot. In the absence of power-good, the motherboard sends a reset signal to the CPU… and your motherboard was saying CPU failure, which just seems like too big a coincidence if it isn’t the PSU.

There’s no guarantee it’s not the motherboard, but I’d be thinking PSU just as you are if this were my PC. You might be able to use a digital multimeter to test the power-good pin on the power connector (back probing from the side of the connector facing you) on the motherboard when the failure is happening. A 5v signal indicates that power-good is being sent by the power supply, and 0v means the PSU has failed its self-test.  Here is a link that shows the pinout for the connector.

If you get no voltage on pin 8 (PWR_OK), that would support the idea that the PSU is bad. I suppose it is possible (and this is just a guess) that something in the motherboard could be messing with the voltage (short circuit, etc.) and causing the PSU to pull the PWR_OK signal low, but I would tend to suspect the PSU itself first.

 

Dell XPS 13/9310, i5-1135G7/16GB, KDE Neon 6.2
XPG Xenia 15, i7-9750H/32GB & GTX1660ti, Kubuntu 24.04
Acer Swift Go 14, i5-1335U/16GB, Kubuntu 24.04 (and Win 11)

1 user thanked author for this post.

Berserker79

Reply | Quote

Thank you for you reply Ascaris. It really sounds like your experience was very similar to what I’m seeing here now.

I bought this PSU in January 2016. Can’t see whether there is a “manufacturing date” somewhere on the unit now, but the box has a 2013 copyright notice, so I guess that the PSU could be as far as 7 years old counting from the potential mfg. date and at least 4.5 years old from when I began to use it. Not sure age is the root of the problem given that I had this kind of issues since day 1 (indeed, I remember a failed power-up as I did the first test, but back then I assumed that the blank screen meant I hadn’t properly plugged-in the video card).

For the time being I’m not keen on taking the cover off the PSU to check for bulged capacitors, since I still have about 1 year worth of warranty and maybe I can still RMA the unit to get a replacement.

Testing with a multimeter sounds like a good idea, but I need to get my hands on a multimeter first. I don’t have one at home and while I could probably borrow one at work this is unlikely to happen anytime soon (company still closed after the lockdown). BTW, just to be on the safe side, can you please confirm that the following are the correct steps for the multimeter test you suggested?

1) Wait for next power-on failure;
2) Turn off PSU using its own switch;
3) Unplug the ATX power connector and test the power-good pin

Reply | Quote

Just realized my own sillyness! Since the suggestion is to test the power-good pin on the power connector by “back probing from the side of the connector facing [me]” of course I should not turn off the PSU… ^^;

BTW, finally managed to find a multimeter, so I can actually test the PSU.

Reply | Quote

What happens if you shut down but leave power applied (ie: don’t turn off the strip)? Will it power up normally after sitting for some time? If so, I’d guess you have a cap leaking down in either the psu or the mb.

1 user thanked author for this post.

Berserker79

Reply | Quote

Thanks for your reply. If I don’t turn off the power strip the system powers up normally after just a few seconds (at least that is what happened so far).

I visually inspected the mb for bulging or leaking caps and could not spot any. This leads me to think the problem is the PSU, but for the time being I can’t check its caps, since taking its cover off would void what little is left of the PSU warranty.

Reply | Quote

“but then after a few minutes I could power up the computer again and everything was working just fine all over again.”

FAILS – when time interval from Strip ON to Computer ON is short.
OK – when interval is (unspecified) longer.

Can you test with varying intervals, and find a ‘minimum’ delay interval required to power on?
This ‘sounds-like’ a ‘control voltage’ isn’t coming up fast enough. Bad capacitor?, surge resistor increased resistance?, open diode?, … ?

I there a minimum Strip OFF time interval that induces this scenario?

1 user thanked author for this post.

Berserker79

Reply | Quote

Thanks for the reply Paul.

I suppose you refer to the scenario where strip ON->computer ON results in the “power on, but does not boot” event, rather than that one-time scenario I mentioned where strip OFF->STRIP ON->computer ON resulted in the “complete fail” event.

If so, I tried varying lengths of time of about 1 min, about 5 mins, about 15 mins and about 30 mins between strip ON and computer ON, but in all cases I’ve got the “power on, but does not boot” event at least once. I suppose that I could do more tests, also with longer time intervals, now that the issue has become more frequent.

For the time being I can’t visually inspect the insides of the PSU (still got some months worth of warranty), so I can’t rule out a bad capacitor or an open diode. I’ve been thinking that surge resistor increased resistance could be the root cause, seeing that I’ve had this problem ever since installing the PSU.

Reply | Quote

See if there are settings in the UEFI setup (BIOS) that affect what happens on a cold start up.

Recently, I installed an Nvidia 1660 Super in our Dell 3668 desktop by adding connectors I made and selecting the PSU rail that was less utilized, the non-cpu rail, I believe.  Using temp monitoring software, Gaming at 1080p with high frame rates had very little effect on temps or fan speeds even though adding the card supposedly overdrives the 240 w psu.  Video editing software works fine, also.  (The machine has 16 gigs ram and an i7 processor, mechanical HD)

Surprised me, I was expecting a psu upgrade would be needed, a pain since the case is small and a new psu would likely have to be external.   The 3668 is 2017 vintage and used often; I guess my point is good psu’s tend to be underrated, so I’d be surprised if yours is bad, although that’s possible.

While experimenting with the Nvidia card, the computer was cycled on and off many times.  Sometimes it started normally, other times, it would start but not boot, a few times I had to remove the power cord for it to boot on start up; many combinations of start up issues.

Forward a few weeks and I’m installing a Dell T30 server (‘cuz the price was great!) as a file/media server in our little home network with similar experimentation learning how to mount drives at boot in Linux using the terminal.  Again, lots of weird startup behavior as I fiddled.  T30’s setup (bios) has a number of power on options, I suppose because you want a server to behave predictably after a power failure.  I had to change default power settings so the machine wouldn’t auto-restart after power down.

Both computers work perfectly day to day now; they’re turned on/off with their power switches, remain plugged into a simple power strip.  One thing I noticed experimenting was start up after the power cord on either was disconnected took considerably longer than if just the power switch was used and sometimes had to be repeated with the power switch for proper boot.

I’m definitely not an expert on what computers do with the quiescent current they draw after shutdown but what you’re experiencing isn’t surprising.  Some of my older machines would boot perfectly every time with power strips turned off; the two I mentioned above won’t.

If mains isolation after shutdown is why you switch off the power strip, maybe consider a small UPS, some decent ones with short durations are very inexpensive now.

1 user thanked author for this post.

Berserker79

Reply | Quote

Why not just try a new power supply?? That should be the one component that could be re-purposed for years and could likely be used on your next build.

Just because you don't know where you are going doesn't mean any road will get you there.

Reply | Quote

Have you tried buying a PSU tester? I have found this tool to be very invaluable since it tests whether or not the PSU reaches its appropriate voltage as well as test the delay of how long the Power Good signal takes to generate. I believe the ATX standard mandates that a P.G. signal should take about 100 to 500 ms. So it sounds like that your PSU when started from cold the circuitry is not able to to get from 0 V to 5V in the right amount of time. In fact I just returned a PSU because it failed that test. The beauty of a PSU tester is that you can use the same 24 PIN cables to do the test and no need to fuss with the Volt meter etc. They are somewhat cheapish to get but a great tool to trouble shoot. Let me know how it works out for you.

Reply | Quote

Thank you for the suggestion, anonymous. While I have managed to get my hands on a voltmeter since starting this thread, using a PSU tester sounds like a much more practical and simple approach compared to back probing the power cable: I’m going to look for one of these testers and let you know if that works.

Reply | Quote

Berserker79 wrote:

The power strip has its own on/off switch and when I’m done using the computer for the day I switch it off (of course after having shut down the computer).

Berserker79 wrote:

If I don’t turn off the power strip the system powers up normally after just a few seconds (at least that is what happened so far).

I’ve read through this entire thread.  Your PSU is OK, your MB is OK.  Turning off the power strip results in bleeding all stored current from all capacitors, in both the PSU and the MB.  The function the capacitors serve is to supply “rush” current to a circuit on startup.

A capacitor is similar in its characteristics to a battery.  A capacitor will hold a charge when it is powered off, but not for very long.  Just as a rechargeable battery takes some time to recharge, a capacitor takes a little time to reach full charge.  When you perform a normal shutdown of your PC and power it off with the power button on the front of the cabinet, the MB is still receiving a trickle of power from the PSU.

That trickle of power is keeping the charge in the various capacitors on the MB topped off.  Similarly, leaving the PSU plugged in keeps the capacitors in the PSU topped off with a trickle charge.

Startup power goes through the capacitors.  If they are at full charge capacity, the circuit simply switches on at full power; there is no surge and no power drop.  If the capacitor has been depleted, startup power still goes through the capacitor, but it doesn’t reach the circuit downstream from the capacitor at full power requirement—some power is being drained by the capacitor being recharged.

In effect, when you switch on the power strip and then try to power on the PC, there’s a bunch of capacitors in both the PSU and the MB being recharged before they can fulfill their intended purpose.  By the time you restart your PC after the first failure, all the capacitors have been fully charged, and everything proceeds normally.

Don’t turn off the power strip.

Always create a fresh drive image before making system changes/Windows updates; you may need to start over!

We all have our own reasons for doing the things that we do with our systems; we don't need anyone's approval, and we don't all have to do the same things.

We were all once "Average Users".

Computer Specs

1 user thanked author for this post.

Berserker79

Reply | Quote

Thanks for your thoughts on this issue bbearren. This problem has affected my current build since I originally put it together and still puzzles me.

If the problem lies in that when I switch on the power strip and then try to power on the PC some capacitors are not at full charge, then shouldn’t switching on the power strip and then waiting some time before I power on the PC give time to those capacitors to recharge? I’m asking because sometimes power up of the PC failed despite waiting (a few minutes or even half an hour) after switching on the power strip.

Reply | Quote

bbearren wrote:

Your PSU is OK, your MB is OK

I agree that both are OK, but the PSU does not start up as it should from cold, as @bbearren says.

You can either live with the problem, buy a new PSU or, assuming it works, leave the power strip on, your choice.

cheers, Paul

Reply | Quote

Thanks Paul. I intend to test the PSU as soon as I can, in the meantime the plan is to live with the problem given that it does not happen on a too frequent basis and there are no other signs suggesting some hardware failure might be around the corner. But I will likely buy a new PSU the moment I decide to build a new PC to replace this one.

Reply | Quote