Hi all, have been spending a couple months now trying to troubleshoot the weirdest grounding issue I’ve seen in my 30 yerrs of building my hardware …
I have been going a little on the tangent with my latest build (documented here: Epyc on Asrock ROMED6U-2L2T - Proxmox Build - #6 by MadMatt ) that includes an Epyc motherboard, in a stacked Thermaltake Core V20 configuration, running proxmox with vfio passthrough to two GPUs and a bunch of other VMs …
The case is adjacent to my desk, on a small stool to allow for the robot cleaner to be able to swipe below, there are multiple cables going back and forth between my desk and the case such as:
2x displayport cables
4x usb cables going to 2x USB hubs and 2x audio interfaces
1x stream deck
The pc is powered through an UPS, one power cable to a power strip that also powers the 2x USB hubs, no other shared devices
House grounding has been tested and working (Italy, schuko plugs everywhere)
The ‘issue’ I have been having is that I can fairly regularly trigger a reset of the whole thing by either:
building up some static electricity and brushing any metal part of the case - this one is always true
(this is the weird one) pushing away my chair with my hands on the side of the desk, slightly flexing the desk - this does not happen always, I may go a couple of weeks without reboots, while sometimes it happens multiple times per day, whenever I shove off the desk
When the reset happens, it is only the motherboard, the BMC controller does not reboot, and the USB hubs attached to the same power strip stay on. In the six months I have been using the setup, I have never had a reset that could not be reconduced to either brushing/touching the case or pushing away from my desk
Things I have tried:
replaced the power supply
checked there are no extraneous things grounding any electronic inside the case
added an additional ground by building a one wire ‘thing’ hooked to the metal side of the case on one side and plugged to my house ground on the other
removing various USB parts of my setup to identify possible cable shorts/bad cables
reducing the number of ‘things’ connected to my setup
removing everything and adding things one at a time
using different USB hubs
swearing, lots of it, to no avail apparently
The only think I did not change at this point is the case, that to be honest has become my main suspect as instead of having the usual brass standoffs for the mobo it uses raised supports cut from the base metal sheet:
I have a new case on order with a more ‘traditional’ motherboard support structure
If anyone has some feedback that would be welcome, my fear is that when I move everything into the new case if this keeps happening I am really out of options …
When you mention you moving and your PC resets it makes me think a wonky connection myself. I would look at front panel connections myself.
Best advice for you with the new case is to take everything back to nothing connected and carefully redo it. Also you could try and see if the case is not on your desk does it reset?
Best of luck tho, the weird ones are the hard ones.
That is the weird part … the case is not on the desk, it’s on the side, and other than the usb cables there’s no direct contact between case and desk, and I have repeatedly tried to switch/detach/move/fiddle with the usb cables with no apparent result
Initially my thoughts were a usb cable producing a short by being somewhat pinched, but there’s no such thing as far as I can tell, all cables are either over the desk or under it in a 2by4 cable rail, where there is a lot of free room/they can move freely…
While doing another round to triple check I didn’t do something silly, I may have found one possible reason for a short though, I am using two of these slimsas to pci extenders:
so yesterday I changed the standoffs/screws to use ones that do not extend over the copper round circle around the hole … will see if it makes any difference …
… and scratach that as well, two reboots this morning, both right after taking my hands off the desk/rolling my chair backwards …
btw, I am using a cordless keyboard so definitilely no grounding/continuity between keyboard frame and anything in the pc …
Have you tried disconnecting the front panel switches from the motherboard? The board/switches for the front panel might be shorted to the case/ground in a weird way. I’ve seen all kinds of strange restart behaviour boil down to a poor quality front panel.
Is your motherboard using that stand off that you highlighted in the picture since I have the exact same case and never had any issues at all but that stand off is not used on my board. I am probably missing something but oh well.
Not yet, went overboard and ordered a whole new case … excellent suggestion though, I might as well disconnect them on principle, as the motherboard I am using has a BMC anyway, and the few times I start/stop it I do it from the web interface anyway …
New case arrived today, transplanted the hardware, positioned the slimsas extender in a better/safer config, will report back on whether it keeps rebooting…
Who ever said your wall socket can supply what Epyc requires? I don’t see in your initial post if you tried your computer on another electrical ring (my bad if you didn’t already try this I’m not reading the whole thread). I’ve had something similar with a multi gpu mining rig (three 290x) just crapping out on a bedroom electrical ring but running just fine on a living room electrical ring.
Old house? Apartment? Look closely around where you live and see if you believe someone was really really paying attention to the details and doing it right?
Thanks for the suggestion, I ‘think’ (you never know) I’m covered on power delivery, I have a 3000VA UPS feeding power to almost all my pc components through a single 3x2.5mm2 run, and a smaller 1500VA one feeding power to the networking stuff (3x switches, starlink, dsl modem, 2x Microserver gen10 NAS boxes) in the attic …
Update: after almost three weeks of switching cases, and I have not had a single occurrence of the spontaneous reboot … I’m about to call it a problem with the old case/the old case wires/switches …
Another update, for closure … I have found what causes the issue and … drum roll … it was not the case …
After weeks of trial and error (and sometimes getting reboots even with the new case) I now can 100% reproduce the spontaneous reboot and it turns out the faulty component is … the fabric of my chair
I can now cause a reboot whenever I feel like it by:
Wiggling my arse on my Secret Lab Titan chair (softweave Fabric)
Pushing off the desk while sitting, it works for distances of as much as 1 meter/4feet from the desk
standing up from my chair, making sure my calf /back of the knee contacts the chair’s cushion
getting an electrostatic discharge
having my 3K+ dollar setup reboot like it was a Spectrum ZX80 with a wiggly power connector
Floor is hard tile, no contact between me and the desk, no contact between the desk and the PC other than for two audio interfaces, connected through USB cables.
I have testedt it with all USB cables disconnected from the case, stilll triggers a reboot
My guess is that the static discharge travels through the metal in the tiles’s mortar and then finds its way to the best grounded thing in my room that happens to be my PC … and triggers a reset
The solution … after verifying my case is properly grounded, after validating I have no weird contacts between motherboard and case, all cables, changing all the usb cables, replacing the power supply, having an electricina chech the house ground in the sockets in my room was …
Installing a cotton towel between my arse and the chair’s cushion … no more electrostatic discharges, no more reboots
