Troubleshooting freeze-ups on Acer Veriton X4610G How can I pinpoint the random freezes?

billsey

Typically there are two kinds of failures that affect power supplies. They either don't work at all for some voltages or they provide really noisy power. If they don't work at all it's easy to test with a VOM. Disconnect from the motherboard, jumper the PS_On signal to turn the supply on and measure at the main connector. They should all be reasonable to match the ATX specs for that connector. Noisy power is much tougher to diagnose, it's best measured with a scope instead of a VOM. Any time the noise level exceeds the threshold voltage the system starts getting confused as to whether something is off or on, which gives weird symptoms. By far the easiest test for that issue is to try a different supply. Since your system has the smaller form factor for it's supply it's unusual for someone to have an alternate just sitting around from an old computer. But, you can take an ATX supply and mount it externally with your case open to run the tests...

Larryodie

Speccy is a hand tool to monitor real time but to record the events, when it crashes is very unlikely for any program to accomplish that as the bus is STALLED.

Here is my real time SPECCY

SPECCY.jpg

Larryodie

Look for unsteady voltages (fluctuations) as that is what bad filter condensers will do and go from low to higher as they warm up. (usually)

Larryodie

I reread your opening statement and you say that the display is still visible when it freezes. This should be a log of when and what was running when it happened. ?

I'm still studying Speccy as to when I can save a snapshot

YouTube has some instructions videos that I'm studying.

Here is one.

https://www.youtube.com/watch?v=l83QxXtKPL4

Marty11

https://community.acer.com/en/discussion/comment/1175909#Comment_1175909

Wow, that's a wealth of experience. It's great having you guys helping me, thank you very much (that also goes for @billsey of course).

Now I know what to look for in voltage fluctuations related to failing capacitors.

I tried Speccy, but just like HWiNFO it gives me weird readings. I made three screenshots related to voltage measurements one of Speccy, HWiNFO and BIOS:

When Speccy is in view whilst freezing, I should be able to see the last voltages on screen. But first I need to get it to log correct voltages.

On HWiNFO forums I read that they need detailed information from the manufacterer in order to interpret sensor readings, as they are motherboard (vendor) specific. And they don't have that for my system.

@billsey I don't have a scope, nor would I know how to use one. I do know my way around with a multimeter (VOM). The BIOS voltage readings look okay and the system starts, so the PSU is obviously not dead.

I currently still have 2 suspects.

Suspect 1: Noisy power. I'd like to tackle that with some sort of voltage monitoring and logging. HWiNFO would be able to, if it only would know how to read and interpret the right sensor information...

Suspect 2: Onboard graphics or graphics memory failure. I ran a video stress test, but nothing froze.

The tapping test would be next. But beforehand I need to get all of my data safe. As it is my main PC, it's kind of crucial.

I still have a Windows dump file from one of the freezes. Would that be easy enough to analyze?

For the moment the system seems to be working reasonably steady, although the graph is a bit misleading because the system was off for two day (and nothing fails of course when it's off, so stability artificially increased). See the stability monitor graph below. Darn I missed a hickup yesterday.

Marty11

Finally i tried the trial version of AIDA64 Extreme, but still no valid Voltage sensor readings:

Larryodie

Billsey is right about using an oscilloscope (I had 3) for noise but as learned the SMPS operation, noise (ripple) was most always caused by bad filter caps. Most of the time a visual on the big caps and the little ones for venting at the top or swelling was enough to replace them with quality 105 degree replacements from known on-counterfeit vendors like Mousers or Digi-USA. I'd but 100 at a time on several values as om televisions we had to repair not replace complete units.

But I learned that a meter would reveal bad filters too if not visible. Bring up the unit cold and monitor the stand-by volts or operation volts. Usually a bad 5V supply would start at 3.9V or so then rise to like 4.8V and steady out (with load).

The ATX supplies are 12V only .

I accept Speccy as it is and look for any drastic changes. It has a lots of good system info too that is hard to find on windows.

If it's working the don't mess with it.

billsey

Those types of utilities can't show power supply issues, since they are only able to view motherboard voltages, not power supply voltages. Most motherboard voltages are generated onboard from the PSU ones. Your system uses a standard ATX power supply, not the 12VO ones that newer systems do, so you have lots of different voltages on that 24 pin connector.

Marty11

https://community.acer.com/en/discussion/comment/1176183#Comment_1176183

Thanks for your reply. I read up on the Google Whitepaper about 12V PSUs immediately.

You wrote: "Most motherboard voltages are generated onboard from the PSU ones.". Might you know which voltage line from the PSU would be used mostly to generate the others? That will be my first target to monitor (need to get me some clamps to leach on to those lines with my multimeter). I suspect +12 or would it be -12?

billsey

Well, there are three different voltages that are most predominant on that connector. 3.3V, 5V and 12V. If I were to make a guess I'd say the 3.3V is the one used to generate CPU and memory voltages, 5V is ideal for TTL logic so is likely used for many of the external chips on the motherboard and 12V is used primarily to drive things like fans and the PCIe bus.

microchipmatt

We are having the same issue with a series of Veriton VX4690G_W. It happens randomly, and we cannot reproduce it, but we have seen it on several users' computers in our enterprise. The same Image works on many other models from other manufacturers, and I can reproduce this behavior in Linux (I cannot reproduce it on the fly, but if I leave the computer, eventually it happens), so it seems more of a hardware issue than an OS issue. It doesn't seem like anyone has gotten any closer to the cause. Are others seeing this in their enterprise? Any closer to a solution? It goes into a deep sleep-like mode; the user cannot use their keyboard and mouse, and we must reset the machine hard. I set ALL of the energy settings never to sleep nor spin down, and as I said, I do the same in Linux, and if I leave it long enough, the same thing happens.

billsey

You might try configuring to disable sleep modes entirely, to see if that bypasses the error completely. Then turn on monitor sleeps without sleeping the computer to see if it's a GPU issue, etc..

Marty11

Is there a beeping pattern when you turn on the computer (before boot)?

My Veriton X4610G makes these beeping sounds: http://sndup.net/4qnj2 .

Does anybody know what these beeps mean?

The computer starts without a problem after the beeps have sounded.

The instability issue seem to come and go with winter (I've re-seated all memory dimms).