Hya,

Situation: One of my Debian penguin occasionally goes wrong, -dead mouse, hang while shutdown, too slow mouse click response and random troubles. Usually, I do not worry much, and replace logic board, memory, disk or whatever.

Peculiar situation With this power supply and case, there was a problem about 2 years ago. New logic board, memory and HD are now in trouble. With the replaced logic board, memory and HD, I put them into some other case to pinpoint where trouble was. For some reason, no trouble for more than year.

Interpretation Power supply is one of the common components with troubled setups. I do not think that case and power cable can cause trouble.

Question Is it good idea replace components again? Can power supply cause trouble? Has anybody had similar situation?

Any suggestion will be appreciated. Thanks in advance.

cheers

Yes, power supply can cause problems like this. You ought to try another one (probably a stronger one).

Seriously ?.

Everything can cause trouble. When I dismantle a case I keep everything I can extract as spares - cable and PSU included.

Hi,

Of course you can have issues with any mechanical connection. PSU can cause issues with different loads that may be the limit of the PSU. You can test the PSU with a DVM/VOM (Digital Volt Meter/Voltage Ohm Meter which is a analog system) when the system loads the PSU.
You need to be careful when measuring the signals and not short anything with the probes. Shot gunning components is one way to test the system but this will not necessarily show the real reasons for failure.

Trouble shooting a system is a art and should not be performed by a novice who could cause additional issues if not performed properly.

You can size all the component loads so you will know if the PSU meets those requirements.

Hope this helps.
Have fun & enjoy!

The power supply is one of the components. But just as important as the mainboard or any other component. If you want to do troubleshooting you should have sufficient known good components at hand.

Making measurements on the PSU output voltages is about the only thing you can do in terms of measuring. Usually the BIOS has a diagnostic page showing the PSU outputs. That is, mainboard inputs.

IMHO a Digital voltmeter is next to useless. What you see on the display is an indication of what voltage could be. But it does not have to be that due to a variety of reasons. Use a DVM to measure your house outlets or car battery and that is very its reliability stops. Unless you have 25 years experience as an EE and then you might know when DVM values are reliable.

An analog meter is a somewhat better. If a layman can read it.

I am afraid the best diagnostics tools nowadays is swapping components one by one for knows good ones until it is completely obvious that something is failing.

jlinkels

One gets themselves a simitruck then puts a six cylinder engine in it in other to drive it. Yes he is going to have complications with it. Because it doesn’t have enough power to drive it.

Hya

Thanks for replies.

The power readings of 12, 5 and 3.3.volt by bios are satisfactory. I do not have an external volt meter, unfortunately.
Yes, the power (calculated) needed is within specification of power unit. I have another setup with identical power unit, which has more components, and it has been healthy.
This time, I will replace power cable also.

I cannot figure out why computer stays healthy for a while before it starts go south.

I will update.

cheers.

This is the usual way, how things work (or die).

that says nothing.

and than you have also the ripple and other stuff on the line.

The voltage in bios, is the voltage when the box does nothing, which is not under load.

--

you need an oscilloscope to see some characterisctics. and than you need a currenct clamp to check every line. and htan you have the issue that it is not always the same load.

also power readings (implies you are talking about watts) and than you say voltage (You aware of power = voltage multiplied with ampere // PW = Ueff · Ieff · cosϕ )

don t mix up things.

--

what i read on other computer forum: 550 watt psu, ~80 euro, should fit for general purpose when you do not use more than one gpu. also use a proper power supply. anything with 200-300 watt will be too underpowered. and than you have the issue with different watts (ampere on different voltage rails) and other stuff. so better get a 550 watt or higher psu

Hya,

Thanks for replies.

Unfortunately, I have nothing like oscilloscope. "Ripple, surge etc" may cause trouble like this, I guess. But if this is the case, new one should run into trouble right away, not two years healthy interval.

Yes, stuffs die after healthy period. I guess. I forgot if clause -if power supply is to blame,...

cheers.

So, you switched to a different chassis and the same thing happened as with the other chassis? Very mysterious! I use a PSU checker to check PSUs. It's always been correct so far. They cost about 15USD.

The chassis can cause problems. I had a PC that was always flaky. It got worse when it was hot. But it was never enough to really bother me. Then one day the machine failed. The power caps on the mobo were bulged. So, I took everything apart very carefully, and there was an extra standoff behind the motherboard, that had been intermittently shorting the board to the grounded chassis.

It was a slow death for the mobo. But after I got a replacement board of the same model and removed the extra standoff, everything has work perfectly for years. Now I always count the standoffs and motherboard mounting holes before I install the board!

I've also had weird responses running a system without the I/O cover plate in place.

Hya,

Thanks for your reply.

Your standoff story is something like an old transistor age story, yes, it can happen.

PSU checker and DVM might be useful, if I had good experience with them. I will replace components again and see how it goes.

Now I know anything (more than I think) goes.

cheers

I'n not so sure that your symptoms show PSU failure. A classic result of inadequate power is the computer shutting down without any log record: the CPU, typically under heavy load, decides the power is not enough for reliable operation and it stops, just as if it were overheating.

There's a limit to what you can do in diagnostics unless you're a trained technician with a load of equipment. That's what computer repair shops are for!

As far as getting a new PSU, there are web sites where you can enter details of your components and get a suggestion of the wattage (is there such a word?) required.

The scientifically correct word is power rating. Which makes it strange that American English doesn't seem to have the word "Wattage" whereas, for example, Dutch does have it.
OTOH American English has the word "Ampacity" which gives me stomach pains.

Seriously now, processor power consumption varies wildly between 10 Watts and > 110 Watts. I think virtually all power supplies can handle 200 Watts. But for an 80 Watts processor you need at least 16A of current. Wow. That calls for serious copper lines in the power supply cable. On the DC side. So a power supply with higher rating is recommended just for wire thickness. If the processor has a bit more moderate power requirements and on-board video is used, any power supply of 200 Watts should suffice. And yes, we are talking here about the ampacity of the DC power cable.

jlinkels

Hya

Thanks for replies.

Yes, overheating should have been ruled out. But symptom and room temperature do not correlate. Cooling fan is working.
I guess the title of this thread should have been "Can power supply cause random hardware down after healthy period?"

Penguin 1 was happy for a while and went south. So, brain (motherboard, memory and disk) was transplanted, then revived as Penguin 2.
Penguin 2 was happy for a while and went south.
The brain of Penguin 1 found another skin (power unit, case and keyboard) and is happy.

If power supply is kaput, Penguin 2 should not survive, I guess. Is there anything like honeymoon period?

I will update.

cheers