[conspire] Computer Gremlins -- bad AC power?

[Dana Goyette]

Ever since I moved down to San LUIS OBISPO, I've had way more weird crashes and hardware failures than I've ever had before.  Can anyone suggest what might be the cause?  Is our power circuitry messed up in some way?

* LED bulbs in one room dim and brighten when starting and/or stopping (not sure whether it's 'and' or 'or') the washing machine, especially if they're on a dimmer at <100%.

* LED bulbs dim randomly even when the washing machine isn't running, if dimmer is set low.

* Random crashes every few days on my desktop when playing a certain game, both on Windows and Linux (Wine+DXVK)
  -- Note that that game (Final Fantasy 14) is the only one I regularly play, and it's generally only in the evening, which is when the family runs laundry.
  -- I've run other graphically intense games with no crashes.

* ECC RAM, and MemTest86+ says it passes.

* GPU (RX 5700) seemed to go bad (started getting crashes even without games), sent for RMA... replacement still crashes in games.

* Got a second, similar GPU (Radeon Pro W5700)... still crashes.

* Replaced my power supply... still crashes.

* Secondhand Supermicro server board suddenly failed (spark sound and a smell of magic smoke) one time when I nudged the PSU cables.
  -- When plugged into desktop PSU, a red LED blinks instead of the green IPMI heartbeat LED; from what I can tell, that means power fail.
  -- Tried it with an old PicoPSU instead of the desktop power supply, and the board started smoking at the ATX power connector.
  -- I guess a real PSU has short detection, but the PicoPSU does not!  Could be exciting to plug it in outside...

* New CPU and motherboard, new RAM: still crashes.

* New CPU and motherboard, old RAM: still crashes.
  -- The crash's error record mentions uncorrectable errors in the L1 cache.

* I remembered that this new CPU was banged around in shipping, so I RMA'd it. Replacement hasn't had enough soak time yet to see if it still crashes.

* Laptop (NVIDIA GPU) crashed with a black screen one time the very instant the washing machine stopped to change cycles.   Same laptop has been stable at my parents' place.

At this point, I bought a UPS (SMC1500C), on the suspicion that maybe the wall power itself is bad.

* Desktop LCD randomly died (software backlight control works, but no image, not even of OSD).
  -- I realized the "green mode" was enabled on the UPS (bypasses AVR components when the input voltage is okay), so I disabled that option.

So, how likely is it that the real problem is bad wall power?  My dad is the landlord, and he briefly tried to check out what was up with the dimming, but couldn't track it down.

The small UPS running our cable modem and router shows the input voltage randomly drifting between 121 and 125 volts.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://linuxmafia.com/pipermail/conspire/attachments/20210930/948d5b00/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ups-input-voltage.png
Type: image/png
Size: 5369 bytes
Desc: not available
URL: <http://linuxmafia.com/pipermail/conspire/attachments/20210930/948d5b00/attachment.png>