Dang PC

[svetz]

My desktop hasn't been turning on reliably when I hit the start button. As the button is original equipment and the guts have been replaced a few times figured it was time for a shiny new button so I wouldn't have to keep replaying the stabby scene from Game of Thrones with the PC button.

Dug the old button out (held in with hotglue) and the old button was really sad. Reminded me of the blues-mobile at the very end of the movie. Stuck the new button in, glued it, put the lid on, and pressed the button. Nothing. Nada, zilch, niente, nichts!

WTF? Checked the ATX power switch , checked it was on the right pins... hmmm... flipped everything off and blew out all the dust.

Reassembled and without bolting everything together hit the power button.

GRRRRRR!!!!​

...and then suddenly while I'm staring at it...it boots up. Shut it down, hit the switch and it boots up. Repeat, all okay.

So I put it all back together. It doesn't boot. Then after a few minutes while I'm wondering wtf, it boots up again all on it's own. Shutdown and restart all okay three more times in a row.

So.... suspect there's something odd going on yet... possibly with the motherboard?

 

[MurphyGuy]

Unplug it from the wall for a minute and try again.

Then:
Disconnect ATX power supply connector and jumper the black and green wire with a paper clip.. if the power supply turns on, its not the power supply.
Put the connector back.

Then:
Next, disconnect all peripherals (mouse, keyboard, usb, network, etc) and try power button again. If it starts up, then one of your thingies is bad.

Then:
If it still doesn't start, disconnect hard drives and try again..

etc.
Then start pulling memory cards one at a time.

etc.
keep disconnecting stuff until all you're left with is a power supply and motherboard.. If it still won't boot, then its the mother board and you need to hop on ebay..

 

[svetz]

It booted up fine this morning as it has also in the OP.

The oddity is the delayed start-up (e.g., press button and no lights, no beep, no fan, nothing at all for a few seconds to a few minutes before it all comes alive and works normally).

Apparently, it happens to others too: https://superuser.com/questions/117...ter-pressing-power-button-of-desktop-computer:

Ultimately the machine was abandoned. I still don't know (for sure) what the cause was.

Also saw:
https://linustechtips.com/topic/124...button-and-the-computer-actually-powering-on/
https://itectec.com/superuser/windo...er-pressing-power-button-of-desktop-computer/ <- says it is the PS
http://www.izitips.com/computers/pc-power-button-not-working/ <- Says its the power supply with more detail

https://www.pcspecialist.co.uk/foru...conds-delay-after-hitting-power-button.72922/

So I installed the new PSU. Everything was fine for about a week - fans/power light came on as soon as the button was pressed. However, it has now started exhibiting the problem again.

Could be a peripheral timing out which delays the post tests I guess. I wonder if there's any history of post tests....

 

[MurphyGuy]

It booted up fine this morning as it has also in the OP.

The oddity is the delayed start-up (e.g., press button and no lights, no beep, no fan, nothing at all for a few seconds to a few minutes before it all comes alive and works normally).

Apparently, it happens to others too: https://superuser.com/questions/117...ter-pressing-power-button-of-desktop-computer:



Also saw:
https://linustechtips.com/topic/124...button-and-the-computer-actually-powering-on/
https://itectec.com/superuser/windo...er-pressing-power-button-of-desktop-computer/ <- says it is the PS
http://www.izitips.com/computers/pc-power-button-not-working/ <- Says its the power supply with more detail

https://www.pcspecialist.co.uk/foru...conds-delay-after-hitting-power-button.72922/

Could be a peripheral timing out which delays the post tests I guess. I wonder if there's any history of post tests....

My computer has always taken about 3 seconds to start doing anything after I push the button...

Push button, nothing.. wait 3 seconds and the CPU fan will do a dance where it goes forward a 1/4 turn, then backwards 1/4 turn.. kind of wiggles when it does it... stops for a second, then takes off at full speed and the computer boots as normal.

Been doing that ever since I put the new motherboard and processor combo in.

 

[wholybee]

In all likelihood it is the power supply. It could be something else as detailed above, but the symptoms are typical of a power supply failure, and in general power supply failures are quite common.

 

[svetz]

My computer has always taken about 3 seconds to start doing anything after I push the button...

Wow! Even now most of the time I push the button and stuff fires up right then and there. Does it take that long if you unplug all your peripherals? Wondering if mine is sticking during the post waiting for a peripheral to respond.

In all likelihood it is the power supply.

You could be right, if I fire up a game where the GPU starts cranking up the watts the power supply delivers.

Ran all the diagnostics I could think of, did get a caution on one of my disk drives. But in reading up on it's the type of thing that can last for years. I ordered a replacement just to be safe (if a mobo or psu dies no big deal, if I lose my data it's a disaster ... ;-)

 

[MurphyGuy]

Wow! Even now most of the time I push the button and stuff fires up right then and there. Does it take that long if you unplug all your peripherals? Wondering if mine is sticking during the post waiting for a peripheral to respond.

Yup.. It will do it with just motherboard, cpu, memory cards and power supply..

 

[svetz]

Since writing the post I haven't been able to get it to delay-start...fires up every time.
I bet's it's going to be something stupid (e.g., there's dust between cap leads that change resistance based on humidity).
I know it's going to wait for something important when I'm near the deadline for it to break.

Or perhaps it achieved intelligence and was just in its sulky teenager years?

Yup.. It will do it with just motherboard, cpu, memory cards and power supply..

I wonder if that's the new normal? My motherboard is ~4 yrs old.

 

[wholybee]

Likely not dust, but aging caps. The power supply works with a high frequency oscillator. A marginal cap won't start oscillating, or might need to heat up a bit to start. But once the oscillator is running, its fine.

I have worked with equipment that is left of for sometimes 10 or 20 years. Its quite common that it works forever until you turn it off. Once turned off, all the caps need replaced in the power supply to get it to turn on again

 

[svetz]

Bleary-eyed and stumbling about at the crack of dawn I did the morning ritual of pressing the PC start button.

Nothing... hit it again... more nothing...

Hmmm... nothing for it... walked off to the kitchen to start the coffee to crowbar my eyelids apart. Just as I'm putting the filter in I heard a beep and knew my computer had started. So, maybe a minute from multiple button hits to it actually starting. Everything working fine now. <sigh>

The big difference is it's cold today, about 73° inside, we're having a cold snap... it's 60° out there.

 

[svetz]

Added a poll to the thread to see what component you think I should replace next.

 

[wholybee]

I used to repair electronics. I am mediocre at it. Yet i have great success with the rule, it is ALWAYS the power supply. Obviously it is occasionally something else. But I can not count how many times a better tech than me spends hours or days troubleshooting something to no avail. Then I either replace the (apparently working) power supply or if not available replace every capacitor in it. And bingo, problem fixed.

Prior to that, in another life altogether, i did computer repair. Mostly only 2 things ever failed. The power supply or the hard drive. And while hard drives today are much more reliable than the half height MFM and RLL drives of the past, power supplies are not.

 

[bdserv]

Some motherboards monitor and/or control CPU fan speed and will not boot if the CPU fan is not running or not up to speed. Fan problems can result from worn or gummed up bushings, dust accumulation, or a defective fan that does not start properly [even when new].

 

[wholybee]

Some motherboards monitor and/or control CPU fan speed and will not boot if the CPU fan is not running or not up to speed. Fan problems can result from worn or gummed up bushings, dust accumulation, or a defective fan that does not start properly [even when new].

True. But the motherboards i have seen that happen with will beep an error code if a fan fails.

 

[svetz]

21 seconds this morning between pushing the button and hearing the POST beep (when the fans start spinning and the monitors first light up).
(that's a rough time going one thousand one, one thousand two... ; -)

... it is ALWAYS the power supply....

that's the winning vote so far. The thing that makes me unsure (other than post #3 where replacing the PS didn't fix it) is it seems like a PS should be a works/doesn't-work thing.

The POST test doesn't have any beep codes so it didn't find a problem. But the POST runs on the motherboard, so it might be the problem.

Saw some things saying:

...sounds like a faulty CMOS battery..

The board and batteries are almost 5 years old, but if the battery were toast the motherboard would lose it's settings and present me with a screen to set them.

Saw this:

If post state takes longer then usual, but does show motherboard splash screen then there is often a initalization issue with a certain hardware component, or usb device....If the post state takes too long while the screen stays blank then it might be something related to the memory or gpu

Interestingly, Crystal Reports doesn't like one of my disk drives and I'm planning on replacing today so that could possibly fix it.

 

[wholybee]

Power supplies are not a work/don't work thing. When capacitors fail, often they work while warm, but not cold. Turning it on and letting it warm up for a minute or two is often enough to get the power supply to start. Sometimes, it takes turning it on, waiting a few minutes, then turning it off and back on.

Also, sometimes they voltage test ok, but fail under certain load conditions. I once showed up to work to find every other technician working on a broken studio camera. All the off shift techs were called in except me, because it was my first week there. They even called in some local experts. Schematics out, drawings on the whiteboard, and pieces scattered everywhere with test equipment running. They had been there for 8 hours trying to fix it and went home tired, and told me tomorrow the were boxing it up to ship to the manufacturer.

I had it working in 20 minutes. Power supply. They tested it as good, but the -5v line dropped out under load. And since all the other voltages worked, the camera powered on with lights etc., so they didn't check the PS further.

Anyway, it is entirely possible that a capacitor or something failed on the motherboard, or on a HDD. But without testing it in a lab setting, my money is on the power supply. If i recall, you mentioned the computer had been rebuilt and upgraded, but with the same power supply. So its also older than everything else.

It has been a long time since i was a computer tech and standards have changed. The way the boot process used to work, is that the power supply first appiled a single voltage to the motherboard. The motherboard did a quick test (not the post test) then responded back to the power supply with a "good" signal. On receiving that signal, the power supply fully powered on, and the post test started. If you can google the pinnouts of the motherboard connecter, you can look for those signals.

You should be able to disconnect the power from the suspect hard drive and rule that out.

If it were my computer, even with my ability for further testing, i would buy a power supply and only begin testing if the power supply didn't cure the problem.

 

[svetz]

If i recall, you mentioned the computer had been rebuilt and upgraded, but with the same power supply. So its also older than everything else.

The case is older, from 1996 ;-)
The PS isn't original and it's marked "Gold" on the side (that scale started just after the turn of the century), so probably less than 15 years old. But probably well beyond its 3 yr warranty.

The way the boot process used to work, is that the power supply first appiled a single voltage to the motherboard.

I'm assuming that part works since it does turn on (eventually) after pressing the switch.

The motherboard did a quick test (not the post test) then responded back to the power supply with a "good" signal. On receiving that signal, the power supply fully powered on, and the post test started.

The issue is somewhere between those two I think. The sequence is: press start button, then nothing visual or auditory happened for 21s, then all at the same time the POST successful beep occurs, the fans kick on, the LED lights come one, and the monitor lights up with the BIOS manufacturer's logo. (usually, it's press the start button and near-instantly the rest happens).

I'm assuming power is flowing when the switch is first pressed otherwise nothing would ever happen.

It could be the PS isn't supplying enough juice to something and that's why the timeout in the POST occurs. It does start up right away after that first startup, but modern BIOSes don't typically run the full startup tests on a "warm start".

One of the "advanced" features of "modern" BIOSes is to "discover, initialize, and catalog all system buses and devices", that's my guess for where the issue is. Unlike @MurphyGuy I can't easily test my USB peripherals since I can't get it reliably to fail. It could also be a disk, video card, CPU, RAM, or even a chipset. If it failed reliably I could at least discount some elements by disconnecting them as you suggested.

If it were my computer, even with my ability for further testing, i would buy a power supply and only begin testing if the power supply didn't cure the problem.

I wanted to replace that hard drive anyway (it's old, running hot, and Crystal Disk is saying bad things about it), so we'll see how it goes. I might hunt around in the BIOS GUI and see if there's a way to force the full POST test.

But if it persists the power supply will be next just because I agree it's a likely culprit.

 

[wholybee]

The issue is somewhere between those two I think. The sequence is: press start button, then nothing visual or auditory happened for 21s, then all at the same time the POST successful beep occurs, the fans kick on, the LED lights come one, and the monitor lights up with the BIOS manufacturer's logo. (usually, it's press the start button and near-instantly the rest happens).

No, that beep is not the post test completing. That is the motherboard giving the "good" signal. The post test happens after the logo appears. There is usually an option in the bios to hide the logo and watch the test. You will see it test the memory, scan for and find hard drives, and sometimes report other information. That part is the post test.

Anyway, either the power supply is slow to start and give the initial test voltage, or the motherboard is slow to recognize it.

I'm assuming power is flowing when the switch is first pressed otherwise nothing would ever happen.

It could be the PS isn't supplying enough juice to something and that's why the timeout in the POST occurs. It does start up right away after that first startup, but modern BIOSes don't typically run the full startup tests on a "warm start".

One of the "advanced" features of "modern" BIOSes is to "discover, initialize, and catalog all system buses and devices", that's my guess for where the issue is. Unlike @MurphyGuy I can't easily test my USB peripherals since I can't get it reliably to fail. It could also be a disk, video card, CPU, RAM, or even a chipset. If it failed reliably I could at least discount some elements by disconnecting them as you suggested.


I wanted to replace that hard drive anyway (it's old, running hot, and Crystal Disk is saying bad things about it), so we'll see how it goes. I might hunt around in the BIOS GUI and see if there's a way to force the full POST test.

But if it persists the power supply will be next just because I agree it's a likely culprit.

Sounds like a plan.

 

[svetz]

New disk is installed, didn't fix the problem. It's running 10°C cooler (33° C) and shows no issues (but it's noisier when accessed).

No, that beep is not the post test completing. That is the motherboard giving the "good" signal. The post test happens after the logo appears.

Okay, that's good to know... so that means the problem is before the POST... so not CPU or other component-related. So that's down to the PS, the MB, or a corroded wire/terminal. It does seem to fail more reliable if it's unplugged for a bit after shutdown so PS is sounding likelier.

There is usually an option in the bios to hide the logo and watch the test. You will see it test the memory, scan for and find hard drives, and sometimes report other information. That part is the post test.

I did find the bios setting to switch the MB logo to diagnostics, but they happen after the delay (darn fast, but it looked good). Didn't see any other diagnostics I could run from the BIOS other than memory or disk (which I'd already done).

Sounds like a plan.

Wonder if there will be any boxing day deals on power supplies?

 

[svetz]

About 20 minutes for it to boot up this morning. Thought it was dead for sure, but no... it's alive! Or perhaps undead? So, gold(90%, ~$100), platinum(92%, ~$150) or Titanium(94%, $200)? About a $50 increase per 2% efficiency increase. Let's see... despite the wattage rating the PC probably pulls 80 watts on average over the 16 hrs/d it'll be on. Couple of hours per day I might run it hard at 400W. So, let's say 14x80 + 2x400 = 2000W/d.

At 100% the PC costs 2kWh/d x 365d/y x 0.14 $/kWh = $73/year. Going from gold to Platinum reduces it by 2% or ~$1.5/year in power and to titanium that's about a $3/y reduction.

In terms of heat, there's 3.41 BTU per W. So a gold is 2000W/d x 365 d/y x 3.41 or 1245 tons of cooling needed/yr. Let's assume 1 kWh per ton of AC, so

Gold = 1245 x 10% = $17.50/y​

Platinum= 1245 x 8% = $14/y​

Titanium = 1245 x 6% = $10.50/y​

So, looks like each step up is a savings of $3.50/yr? If it lasts 10 years, that's worth about 10x($3.50 + $1.5)= $50 for each 2% step.

So practically a wash for me, any other reason to go platinum or titanium over gold? I see some platinums on sale for gold prices, so that sounds like a happy medium.