Intermittent powerdown

Topic

New Reply

Do I have a good one! The computer in question is a desktop Systemax using XP3 fully patched. Earlier this year I had massive no power problems with the unit and to solve the problem, I obtained another motherboard, power supply, and CPU. I had replaced the mobo and supply (same type supply,)used the original cpu and it had worked fine for some time. Lately, at odd times and over a period of 2-3 weeks, the unit just shuts off. By turning off the power switch on the rear of the unit to let the supply reset (10-15 seconds) then powering it back up, it works. An app showing the various temperatures within the unit shows proper temps and looking at the event log gives no clues. The only apps really running is the Weather Channel and a browser streaming a local radio station outside of the usual background stuff: System Suite 14, various system items.
Anyone have an idea about what may cause this or a program that may point this out? Thanks.

Reply | Quote

Replies

Give your system’s memory a good workout with Memtest86 for no less than 6-8 hrs.
If you’ve got an add-in GPU card you could also load test it and observe.
Prime95 can be used to load test your processor but I doubt that it is the problem.

It isn’t beyond the realm of possibility that your new PSU could be at fault too, but that’s a bit tougher to test.

I wouldn’t completely rule out a software driver issue too.

Reply | Quote

1. If you have added extra devices (attached USB devices, PCI, PCIe cards, etc.) Disconnect one at a time.
2. Brute force method: Try a higher Wattage Power Supply. Since you have already changed to new PS and had no effect, maybe higher Wattage PS is needed. No harm to try it. I would try 40% larger than your existing PS Watt rating, or larger even.
3. I assume you did it already: Maybe this should be first to try: Virus scan.

Reply | Quote

That’s what I did for the first powerdown problem but at least it was constant so was easy to check. My problem now can sometimes take a couple weeks to manifest and typically would occur within 1 to 1 1/2 hr of turnon (actually I would only use the machine mornings for about the same length of time.) Once it happened during bootup. The last time around I let it run into the afternoon with no problem then turned it off. The next day it was dead. Go figure.
As for drivers, the only change recently has been drivers for the new graphics board; all else remained the same. I do have a somewhat heavier ATX supply (400 W) but for some reason it won’t operate in the system. Wierd.

Reply | Quote

Do I have a good one! The computer in question is a desktop Systemax using XP3 fully patched. Earlier this year I had massive no power problems with the unit and to solve the problem, I obtained another motherboard, power supply, and CPU. I had replaced the mobo and supply (same type supply,)used the original cpu and it had worked fine for some time. Lately, at odd times and over a period of 2-3 weeks, the unit just shuts off. By turning off the power switch on the rear of the unit to let the supply reset (10-15 seconds) then powering it back up, it works. An app showing the various temperatures within the unit shows proper temps and looking at the event log gives no clues. The only apps really running is the Weather Channel and a browser streaming a local radio station outside of the usual background stuff: System Suite 14, various system items.
Anyone have an idea about what may cause this or a program that may point this out? Thanks.

What power supply are you using? What is your system setup? More information is needed.
Did you use the same RAM from the old motherboard?

Judging by your statement this is not happening under heavy load. So, it might not be the psu itself but rather another component from your old build.

Environmental issues may be an issue as well, again more information is needed.

Edit: You mention having a “heavier” psu of 400 watts. That itself is kinda low. Other specs to take into consideration are amperage on the 12v rails(especially if using a discrete GPU).

Fill this out if you have time:

Cpu:
Mobo:
psu:
gpu(if discrete):
hard disk/s and optical drive/s:
cooling(if applicable):

Reply | Quote

Lately, at odd times and over a period of 2-3 weeks, the unit just shuts off. By turning off the power switch on the rear of the unit to let the supply reset (10-15 seconds) then powering it back up, it works.

What are settings for Power Configuration (can also be observed using the PowerCfg program)?

Reply | Quote

Just fishing around …
1. Take out and re-seat the video card on its slot. There could be a nearly short at the connections. Heat may cause the short while cold is just enough to unshort the short.
2. If your motherboard has built-in video (such as svga). Set video to svga (so to use Windows default video driver). Go to Control Panel-System-Device and delete the plug-in video card. Hard power off. Remove the video card from the slot. Reboot and try it over night. Sometimes advanced or sophisticated video cards may present too much transient load (not a constant load).
3.Quote: “… for some reason [400W PS] won’t operate in the system.” 400W and won’t work. Hmm… If the 400W PS is good, it could be a hint. Try 1-3, then try a 600W-800W or higher PS.

Reply | Quote

BAck again. I had to get another supply, a 600-W. Corsair to try out with the unit. It is really getting snippy now in that it shuts down quickly. I first tried out the original supply which turns on and goes down in a couple seconds. The 400W supply does not turn on at all and the Corsair shut down before a complete bootup I did have the extra memory from the other mobo I bought (1.5 GB) installed with the original 2 GB but pulled them with no change. Gonna pull the new video card to check that out . . .oops, I need it to properly log in as I use a password and have another user to select from at log-in. Rats. I’d hate to have to get another card just to check out the computer.

Oy, vey! I just tried to turn on the system and now it is locked up at the “loading personal preferences” stage. Will power down and try again. While I’m thinking on it, there is a lead on the 24-pin plug that changes from low to high if the PSU is shut down by the computer. Does anyone know from where that kind of error signal is generated? CPU? Mobo?

Also, here is the list of what I have in the Systemax: MOBO K8N NEO 4, CPU- AMC Athlon 64 II 4400+, GPU EVGA GT520 card, 3-SATA HDD, 1-DVD burner IDE type, 1- CD burner, IDE type, 1-floppy drive, 1-multicard reader. Cooling for the CPU is by heatsink/fan.

Reply | Quote

While I’m thinking on it, there is a lead on the 24-pin plug that changes from low to high if the PSU is shut down by the computer. Does anyone know from where that kind of error signal is generated? CPU? Mobo?

Easiest and a definitive answer (no speculation) occurs by using a meter to measure wires as listed below. You are asking about the power controller – a circuit that determines when a PSU powers on or off and also when the CPU is permitted to execute. No faster and more definitive way exists to answer your question without below numbers using a 3.5 digit meter.

With its black probe connected to the chassis. Set the meter to 20 VDC. Touch its red probe to the purple wire where it enters a nylon connector on the motherboard. That should read about 5 volts (when computer is not trying to power on). That number must be reported to three digits.

Do same for the green and gray wires both before and when the power switch is pressed.

And finally monitor any one red, orange, and yellow wires as the power switch is pressed. Report what those voltages attempt to do.

Your question is about the green wire. Voltages on all above wires are relevant to your question AND define the operation of many components that define a supply system. The system is more than just a PSU. Those numbers mean the next post answers your question, defines or exonerates suspects, and provides answers you do not yet know to ask.

Reply | Quote

I think you have a short somewhere, or something you did before that created a short. PS shuts down right away because of shorts or overload. Not easy to overload a 600W PS. If the short is on the 12V rail, 600W/12V=50Amps, you can do spot welding!
Westom shows the correct way. As an everyday Joe ‘garage guy’ we simply look for shortcuts and easier methods. I would pull the newly touched card/memory/hard drive/what have you (whatever you did before this happens). Then pull out add-in devices one at a time. At the end, if your PC has only motherboard and still shuts down a 600W supply …
Either you have a failed motherboard or something like metal piece fell on the motherboard creating a short.

Reply | Quote

With the Corsair supply in place, I got the following voltages: with the system in the dead/shutdown mode, Purple-5.04V, green- 3.83V, all others 0V. After a power supply reset and power up, purple-5.02V, green, 0V, gray-5.08V, red-5.09V, orange-3.36V, yellow-12.16V. All voltages were shown by a Fluke 105B scopemeter. Using the RECORD function, all voltages remained steady while running until shutdown with the green wire rising to 3.83V.

Curiously, a couple of times the system hung at the processor initialization phase and stayed there for as long as I let it. I’m beginning to think I’ll need to call Father Damien.

I repair home theater equipment for a living and in those items, the CPU sends out the shutdown signal. Would the computer CPU do the same to the PS? Even so, what type(s) of protection lies in it’s innards or does the processor do all the sensing? What’s next?

Reply | Quote

With the Corsair supply in place, I got the following voltages: with the system in the dead/shutdown mode, Purple-5.04V, green- 3.83V, all others 0V. After a power supply reset and power up, purple-5.02V, green, 0V, gray-5.08V, red-5.09V, orange-3.36V, yellow-12.16V.

OK. This is what you saw. The power controller needs 5 volts for operation. That 5.04 volts should be when the computer was powered off – only connected to AC mains. (And why the power cord must be removed from the wall receptacle before making any hardware changes.) Green wire is the power controller telling the PSU to power on. 3.83 is slightly low compared to most systems. But more than enough (above 2.0 volts) to tell a PSU to stay off.

When the power button is pressed, then the power controller dropped the Green wire to less than 0.7 volts. The PSU then powers on. A voltage monitor in the PSU then monitors other DC voltages and should increase to more than 2.4 volts in less than 2 seconds.

This part is troubling. That power good signal should not be same as the 5 volts. It is typically less. But many third party supplies ‘forget’ to include electronics for the Power Good (gray) wire. Instead connect that wire directly to the 5VDC (red) wire. Then unexpected power loss can cause data corruption. Your numbers for a gray wire should be slightly lower if the required power good semiconductors exist.

So the gray wire remain above 2.4 volts and green wire remains below 0.7 when the CPU hung? If yes, then the power system is apparently good; reason for that hang resides elsewhere.

Your red, orange, and yellow wires (when fully loaded by all computer components) are well above what would be problematic. For example, ATX specs say red wires should remain about 4.75 volts. That means any voltage below 4.83 indicates a failure. Obviously your voltages are quite robust, well above that problematic minimum, and do not even suggest a possible power system short. Move on to other suspects and don’t even look back.

CPU sends messages to the power controller for various shutdowns such as hibernate or restart. The power button also sends a message to that controller. But all power cycling is controlled by the power controller that is telling the PSU (via green wire) to stay on.

Reply | Quote

I agree with the others that you very likely have a short somewhere.

But here’s something you can check, just to make sure — download and run Speedfan, to see if some component in your system is overheating.

In order to boot into Windows (so that you can run Speedfan), you probably will have to go into Safe Mode. See if you can successfully get into Safe Mode.

Also, see if you can boot into VGA Mode.

(Safe Mode and VGA Mode are accessed by hitting the F8 key several times when you first turn on the computer.)

If you can successfully boot into VGA Mode, then your problem is video-related.

If you can’t successfully boot into VGA Mode, but you can successfully boot into Safe Mode, then at least you are making progress.

Group "L" (Linux Mint)
with Windows 10 running in a remote session on my file server

Reply | Quote

Check that the screws to mount the mother board are in the correct holes..

I had a similar experience once replacing a mother board…

Reply | Quote

Well, whatever the reason, the problem has been slowly creeping more and more to the fore, as in something slowly breaking down more and more often. This does not sound like misplaced mounting screws. I had been monitoring the unit via Speed Fan for quite some time before with no unexpected readings shown. Quite the mystery.

Reply | Quote

I had been monitoring the unit via Speed Fan for quite some time before with no unexpected readings shown. Quite the mystery.

1) So what do system (event) logs report? Windows discovers problems, reports it in the logs, and then works around the problem so that you might not see it. Windows is a poor tool for finding problems – especially intermittents. But the logs are a source of hard facts – especially those that appear to make no sense.

2) So the system just hangs and the monitor still displays as is? Or does the monitor blank out? Does the computer in the keyboard still work when the main CPU hangs? Do you really know the entire system crashed or just the video interface has frozen?

3) Heat is often an excellent tool for finding intermittents. Ideal temperatures for all computer hardware is below 40 degrees F and above 100 degrees F. A hairdryer on high is a perfect tool for isolating an intermittent to one location. For example, heat the DRAM board for about two minutes. Does the system hang? Then move on to other suspects; until you find a semiconductor that is 100% defective because it becomes unstable at 100 degrees F.

First hardware becomes intermittent at 70 degrees and will crash at 100 degrees. Eventually a semiconductor defect (created by a manufacturing defect) will start failing often even at 70 degrees. Heat is a diagnostic tool.

Reply | Quote

I also once had a cracked motherboard that caused similar problems…

Reply | Quote

Another shot in the dark, applicable only for motherboard 4 years old, or over 5 years; 7 years definitely a candidate:
Modern motherboard has on-board switching voltage regulators right near the cpu. It is to supply power to cpu at heavy transient current draw. Close-by decreases long wire/cable resistance (0.05 Ohmx20A=1V drop!). Transient loading from cpu could be quite high. Frequently these are multi-phase switching regulators to support transient loading.
You can identify the on-board regulator by the bank of ‘towering’ electrolytic capacitors (frequently 10-12 of them) near the cpu socket.
Because of high ripple current, these capacitors could heat up internally (not detectable from outside), and slowly losing capacitive efficiency. This will lower the regulator efficiency over time. At a critical point, the efficiency is simply not good enough to support transient overload. Overloading, even momentarily, will trigger a shutdown.
Because regulator efficiency goes down slowly over time, the point of malfunction is highly irregular, and depends on external trigger (a heavy current spike). (And it has nothing to do with external PS, no matter how muscular it is).
Fortunately, a way to verify it is to parallel-connect all the on-board capacitors with new capacitors. If it works, this is the cause. Of course, this is for quality control and failure analysis people.
My 2 cents.

Reply | Quote

For Westom: The event logs seem to give no proper information about the problem. Anything around the crash time (before the problem started coming on so quickly) just showed the information sign. As to the processor hang, this was at bootup with the monitor showing nothing.

For Scaisson: both of the mobos I have for/in the unit have been around a while so I would not be very surprised about bad ESR. Some of them do look slightly bulgy on top so I’ll get some caps to put in the original board then install it and see what happens. The quick shutdowns now are during the loading phase where demands on the CPU are greater. Weeks ago, it would shut down occasionally after over a hour run with a browser doing radio streaming.

Reply | Quote

Some of them do look slightly bulgy on top so I’ll get some caps to put in the original board then install it and see what happens.

Not just any caps. Critical for those caps are other parameters that affect ESR including its temperature rating. Design of that adjacent power supply is found in Intel requirements for the original Pentium. A CPU must consume from well less than 1 amp to well over 10 amp – in microseconds. Another reason why just any electrolytics are insufficient for that location.

Above discussed temperature test would have identified that problem. Heat is a powerful diagnostic tool. Any part that fails at 100 degrees is completely defective and will probably get worse with time.

An intermittent failure does not necessary coincide with a timestamp on an error message. Reason for a crash may be due to a failure reported minutes or even an hour earlier. The error message may not identify the defect. But can be an important symptom that points to (with additional facts) to what is actually defective. Any error messages (flagged in red) should be included in a list of symptoms.

Reply | Quote

Granted, heat can be a problem, but with the quick powerdown AND the side of the box open, heat now is no problem. ESR sure seems to be the main culprit and I will try to get things back in order as time permits.

Reply | Quote

Granted, heat can be a problem, but with the quick powerdown AND the side of the box open, heat now is no problem.

Nothing even suggested heat is a problem. “Heat is a powerful diagnostic tool.” Use heat to find what is otherwise an intermittent defect. No facts currently exist to suspect ESR. That was only one suspect based only upon speculation.

Reply | Quote

Back on the air again. I had to obtain motherboard caps and put them in (oy, such a task!) this, on the original mobo, get back to my regular job of fixing audio equipment then putting in the original mobo. Interesting results, I found. (1) I still have the problem. (2) For some odd reason, if I run it in safe mode works, it does. No sign of overheating. I ran Malware Bytes full scan and removed 2 items. I am using a 650W Corsair P/S now. (3) I removed the driver for the Nvidia video card, rebooted into regular XP and within 5 minutes, bang! (4) A peculiarity occurred in that a mystery found hardware screen popped up and, of course, would not install. I did find “unknown device” in device manager and removed it. (5) Bios date was way off. That got resolved but the desktop, which has a wallpaper changer that has a calendar showing did not change at all (Groundhog Day!). Only on the last reboot did it change properly. I’m beginning to think of upgrading the OS to Windows 7 32-bit.

Reply | Quote

(2) For some odd reason, if I run it in safe mode works, it does.
I’m beginning to think of upgrading the OS to Windows 7 32-bit.

Do more testing in safemode if your system is not shutting down with it.
If your thinking of upgrading to Windows 7, why not just try reinstalling XP clean?

Reply | Quote

Can you do a physical check of the motherboard’s capacitors to check for swelling tops ?

Reply | Quote

I haven’t done any more to the Systemax due to snow and the need to do my primary job. It seems there must be a basic software glitch that is available in standard boot that does not load in safe mode. This implies the need to do a clean install of either my XP or jump to Windows 7. I did the upgrade advisor and can do it fairly easily. At least I have a backup (which I might try first) if things go to pot (Macrium Reflect-yay!) or do the install. I also have other partitions that carry a number of my programs and data so I won’t lose them. I just have to find the time.

Reply | Quote

UPDATE: I could not use Macrium Reflect to install the backup image; I got shut down in the middle of the restore, rendering the install inoperative-not bootable. I then tried to do the non-destructive OS install; the system did not give me the proper option to do so. I bit the bullet and did a partition format and new install. Since the new install is SP2, I put in my SP3 disc to install the update and the *%#) system shut down! I then tried a live Ubuntu disc to get into the system and fiddle with the partitioning system on the 3 HDDs I have reasoning the the HDD upon which I have C: partition (along with a data and a separate programs partition) might have sector problems. No luck: the Ubuntu was getting ready to get onto the desktop and bang! Twice! The peculiar thing is that I can sit in Safe Mode all day long. Any ideas, please?

Reply | Quote

This may be way off in left field, but I wonder if there is some defective hardware device which draws as much power as it can, and it is the cause of your problems. In other words, the more powerful your power supply, the more power the device draws. Hence, the problem seems to occur quicker when you have a more powerful power supply. Also, perhaps it drawing as much power as it can is part of the problem — it won’t stop at the normal amount.

A lot of functionality disappears when you go into Safe Mode, so maybe said device isn’t operating when you are in Safe Mode.

What about if you boot into VGA Mode? (I’m not sure if it is called VGA Mode in XP; that’s what it was called in Windows 3.1.) Basically, it is low-grade video mode. If there is a video issue which is causing your troubles, then booting into VGA Mode might allow things to work. VGA Mode strips out most of the video functionality, so that if video is the culprit, you can still use Windows.

You access VGA Mode from the same screen that you access Safe mode.

Group "L" (Linux Mint)
with Windows 10 running in a remote session on my file server

Reply | Quote

Actually, I had removed the video card driver earlier to check for funky drivers and have not been able to reinstall it, so I should imagine that the default mode for the card with no driver would be VGA.

Reply | Quote

I would still try rebooting into VGA mode. That will be the least resource-intensive video mode, with the least possibility of conflicts.

Group "L" (Linux Mint)
with Windows 10 running in a remote session on my file server

Reply | Quote

I’ll try that next chance I get along with double checking the memory chips. Thanks.

Reply | Quote

Another update: I ran a memory check using Memtest 4 with no problem, neither a chip fail nor a power shutdown. Trying a new video card did not work; I still got a shutdown. Trying a pair of Ubuntu live discs gave the shutdown. Now, since I have changed the mobo, the power supply, the video card and even unplugged the HDDs in the system (I thought possibly there was bad sectors on the boot drive-unprovable because I would get shutdown before a chkdsk run would complete), the only thing left would be the CPU: the only common factor through the various iterations in the test regimen. I do have a new one around somewhere but my servicing area got moved within the shop so things are scrambled. At least if I have to get another one, they are cheap.

Reply | Quote