Strange power (?) problem - chassis? PSU? mb?

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

matt_garman

Active Member
Feb 7, 2011
212
40
28
I have a chassis and PSU that may be cursed. :) Or at least something is askew. I thought I'd describe the problems here and see if anyone has any ideas.

The chassis is an iStarUSA S-35 and power supply is a Seasonic SS-250SU (Flex ATX). I used these parts as the foundation for my server build a few years ago. That system worked great from late 2015 until we moved to a new house in mid-2017. After the move, the motherboard (ASRock E3C224D4I) would not POST. At the time, I assumed the movers mishandled the chassis... And it gave me an excuse to upgrade to an x10sdv-tln4f, so I didn't think much about it. (I have since done further troubleshooting on the motherboard, and it's definitely mostly dead. Won't POST, and the IPMI is not usable. But the CPU and memory are fine.)

That case and power supply sat unused until recently. I re-purposed them with an ASRock h110m-itx/ac and i5-6500 build for my parents. This motherboard plus CPU (and RAM) combo had been my desktop PC for the better part of a year, and before that, ran the Zoneminder system in my new house (both without issue). I tested my parents' rebuild at my house (again no obvious issues), and sent it home with them.

When they got it home, they observed the first strange thing: plugging in the HDMI to their TV temporarily killed all HDMI inputs on their TV. After a power cycle of the TV, everything seemed to work OK. I initially didn't think too much of this, and wrote it off as a fluke.

The system worked fine for a few days, maybe a week. Then, out of nowhere, their entire LAN died. They noticed it because they were watching Netflix via Roku. The Roku actually told them their Internet connection was lost. My dad power cycled his switch, and the problem remained. So then he unplugged everything from his switch. He then plugged things back in one at a time. He plugged in everything except the system I recently built for them. Everything was fine. They went back to watching Netflix. Then my dad plugged in the network cable of that new system, and almost instantly, the network died again. He quickly unplugged it, and once again everything was fine.

Taking all these things together---my server motherboard death, the HDMI issue, and the network issues---makes me think that there is something amiss with the power supply and/or chassis, like a power rail out of spec or an intermittent short. We removed the motherboard and checked voltages on all the PSU outputs, they seem reasonable. I visually inspected all the cables coming out of the PSU, thinking maybe the insulation got stripped on one of them, but I don't see any obvious problems.

It is of course entirely possible my server motherboard issue is not related to my parents' system issues. But either way, I'm at a loss of what to test next. Anyone have any thoughts on all of this?

Thanks!
Matt

Edit: Originally had wrong PSU model as PS-250SU (1U). Actual model is SS-250SU (Flex ATX). Corrected above.
 
Last edited:

evolucian911

Member
Jun 24, 2017
36
6
8
35
I have a chassis and PSU that may be cursed. :) Or at least something is askew. I thought I'd describe the problems here and see if anyone has any ideas.

The chassis is an iStarUSA S-35 and power supply is a Seasonic PS-250SU (1U). I used these parts as the foundation for my server build a few years ago. That system worked great from late 2015 until we moved to a new house in mid-2017. After the move, the motherboard (ASRock E3C224D4I) would not POST. At the time, I assumed the movers mishandled the chassis... And it gave me an excuse to upgrade to an x10sdv-tln4f, so I didn't think much about it. (I have since done further troubleshooting on the motherboard, and it's definitely mostly dead. Won't POST, and the IPMI is not usable. But the CPU and memory are fine.)

That case and power supply sat unused until recently. I re-purposed them with an ASRock h110m-itx/ac and i5-6500 build for my parents. This motherboard plus CPU (and RAM) combo had been my desktop PC for the better part of a year, and before that, ran the Zoneminder system in my new house (both without issue). I tested my parents' rebuild at my house (again no obvious issues), and sent it home with them.

When they got it home, they observed the first strange thing: plugging in the HDMI to their TV temporarily killed all HDMI inputs on their TV. After a power cycle of the TV, everything seemed to work OK. I initially didn't think too much of this, and wrote it off as a fluke.

The system worked fine for a few days, maybe a week. Then, out of nowhere, their entire LAN died. They noticed it because they were watching Netflix via Roku. The Roku actually told them their Internet connection was lost. My dad power cycled his switch, and the problem remained. So then he unplugged everything from his switch. He then plugged things back in one at a time. He plugged in everything except the system I recently built for them. Everything was fine. They went back to watching Netflix. Then my dad plugged in the network cable of that new system, and almost instantly, the network died again. He quickly unplugged it, and once again everything was fine.

Taking all these things together---my server motherboard death, the HDMI issue, and the network issues---makes me think that there is something amiss with the power supply and/or chassis, like a power rail out of spec or an intermittent short. We removed the motherboard and checked voltages on all the PSU outputs, they seem reasonable. I visually inspected all the cables coming out of the PSU, thinking maybe the insulation got stripped on one of them, but I don't see any obvious problems.

It is of course entirely possible my server motherboard issue is not related to my parents' system issues. But either way, I'm at a loss of what to test next. Anyone have any thoughts on all of this?

Thanks!
Matt
Get and cut out antistatic material from a bag or Styrofoam like what u get with a new motherboard in the box. Place it on the chasis before mounting the motherboard so it is between the chassis and board. To rule out shorting issue from chassis. Also even if u test psu for voltages u may be misled. The best way is to look inside and test components . on another note we never purchase seasonic anymore at my work place due to their high failure rate and we had yet to purchase a seasonic psu which lasted a year. Even the 80plus 850w ones. The first incident may have been your board shorting on chassis due to impact from moving. Second could be poorly made board model but i can bet my bottom $ its that Seasonic psu.

Sent from my LG-LS997 using Tapatalk
 

matt_garman

Active Member
Feb 7, 2011
212
40
28
Get and cut out antistatic material from a bag or Styrofoam like what u get with a new motherboard in the box. Place it on the chasis before mounting the motherboard so it is between the chassis and board. To rule out shorting issue from chassis. Also even if u test psu for voltages u may be misled. The best way is to look inside and test components .
Thanks for the ideas, I'll give that a try!


on another note we never purchase seasonic anymore at my work place due to their high failure rate and we had yet to purchase a seasonic psu which lasted a year. Even the 80plus 850w ones. The first incident may have been your board shorting on chassis due to impact from moving. Second could be poorly made board model but i can bet my bottom $ its that Seasonic psu.
Somewhere along the way I picked up the idea that Seasonic PSUs were very high quality, and have standardized on them for all my personal stuff. Never had a problem with them, but if you're using them at work, you've presumably churned through substantially more. Perhaps they've gone downhill over the years? What PSU manufacturer do you use now? Even though I don't yet know for sure the problem is the PSU, I was kind of thinking of trying a new one. Not a lot to choose from in the Flex ATX form factor: FSP, Sparkle, iStarUSA (Xeal), inWin and Seasonic. Back in the day it seems FSP also had a good rep, but I haven't really researched PSUs in a while (and IIRC, Fortron/FSP/Sparkle/SPI are actually all the same).
 

evolucian911

Member
Jun 24, 2017
36
6
8
35
Thanks for the ideas, I'll give that a try!




Somewhere along the way I picked up the idea that Seasonic PSUs were very high quality, and have standardized on them for all my personal stuff. Never had a problem with them, but if you're using them at work, you've presumably churned through substantially more. Perhaps they've gone downhill over the years? What PSU manufacturer do you use now? Even though I don't yet know for sure the problem is the PSU, I was kind of thinking of trying a new one. Not a lot to choose from in the Flex ATX form factor: FSP, Sparkle, iStarUSA (Xeal), inWin and Seasonic. Back in the day it seems FSP also had a good rep, but I haven't really researched PSUs in a while (and IIRC, Fortron/FSP/Sparkle/SPI are actually all the same).
Reason why we purchase so many was because they were good. But up to a certain point we have had issue with a variety of their products up until we stopped using them late last year. For workstations or form factors which utilize atx power supplies we are basically using other brands and going through reviews from hardwaresecrets before we buy. So at this point I cant recommend you a brand. But i would do some reading and tread carefully with seasonic as of now.

Sent from my LG-LS997 using Tapatalk
 

pricklypunter

Well-Known Member
Nov 10, 2015
1,708
515
113
Canada
Everything, except the moving damage, could be attributed to incorrectly wired mains socket(s), a missing or poor ground on them, coupled with earth leakage from the power supply :)

I would bet placing a meter between the Ground on the TV HDMI socket and the Ground on the computer will show a voltage between them.
 

matt_garman

Active Member
Feb 7, 2011
212
40
28
Everything, except the moving damage, could be attributed to incorrectly wired mains socket(s), a missing or poor ground on them, coupled with earth leakage from the power supply :)

I would bet placing a meter between the Ground on the TV HDMI socket and the Ground on the computer will show a voltage between them.
Hrm, I'll forward this to my dad and have him check.

However, this system isn't entirely new for them; it's actually an upgrade to an older one. The previous system was using the same model of chassis and PSU. (To be clear: between my parents and I, we have two iStarUSA S-35 chassis, both with Seasonic SS-250SU PSUs.) The previous system has run for years without any issue. The new system I gave him was literally a "drop in" replacement for the old one. With all the issues on the new system, they've gone back to the previous one, and it's working fine (as before).

Their house does have a number of wacky electrical "features", so based on that we'll certainly get to probing. But I'd think if the mains outlet had an issue, we'd have seen evidence of it before. Definitely worth a check though!
 

matt_garman

Active Member
Feb 7, 2011
212
40
28
Everything, except the moving damage, could be attributed to incorrectly wired mains socket(s), a missing or poor ground on them, coupled with earth leakage from the power supply :)

I would bet placing a meter between the Ground on the TV HDMI socket and the Ground on the computer will show a voltage between them.
My dad measured that, and everything else he could think of in and around his mains outlet. Everything is as it should be. (At least, he took all these measurements with the original system.)

On my end, I went ahead and ordered a new PSU (FSP250-60FAG). When I received it, I wired everything up, hit the power button and... nothing. So I thought maybe I got a bum PSU. I pulled all the cables off the board, then used a paperclip to jumper the green and black pins on the 24-pin ATX header. PSU powered up as it should. So I re-connected the PSU cables to the motherboard and tried again: powered right up!

This was last night. I let the system run prime95 for a couple hours: no issues, temps were great (Noctua NH-U9S kept temps under 55C during those two hours of p95 testing!). So I killed p95 but left it on all night (idle, not doing any interesting work). When I got up this morning, I checked on it, still totally fine. At this point, the outer case cover wasn't on, and the cables were a big mess. So powered it down, tidied up all the cables and put the cover back on. My plan was to just let it run for a week or two to see if any of these wacky issues re-surfaced.

Well, after all that, I hit the power button and... nothing. :( So I repeated what I did last night: I pulled the power cables, jumpered the green and black wires, the PSU turns on. I re-connected everything once again, but it still won't power-on. So I got a spare ATX supply and connected that to the board, and it still won't turn on.

Next step I guess is to pull the board out of the case and see if I can get it to turn on in a completely open environment.

I feel like I might be a step closer to figuring this out. Perhaps there is a short somewhere between the board and the chassis? I need to put a layer of plastic or something non-conductive between the board and chassis as suggested above.

Any other thoughts?

Edit: Powers up just fine when the board is outside the case. That's even with the FSP250-60FAG PSU that is still in the case. So to me that suggests something on the chassis is shorting something on the motherboard. I can't think of what else would case this kind of behavior...
 
Last edited:

matt_garman

Active Member
Feb 7, 2011
212
40
28
At the risk of jinxing myself, I might have this fixed. The bottom of the S-35 chassis is entirely flat... almost. At the very back, near the rear IO shield, there's a slight bump that runs almost the whole width of the case. It's not very tall, maybe 1mm. Nevertheless, I covered it with a strip of electrical tape.

The other suspicious item was the protrusion for the screw for one of the case feet. I didn't measure, but it appears shorter than the PCB standoffs. But, some of the through-hole components on the motherboard have leads that aren't trimmed completely flush with the board, so it's possible that it could be creating a short---or just barely enough that subtle movements, contraction/expansion due to heat, fan vibrations, etc could cause an intermittent short. So I cut a small piece of plastic to put over that screw, and taped it down with bunch of electrical tape.

There's another one of those foot-screws sticking up near the board, but not directly underneath. I went ahead and gave it the same plastic and electrical tape treatment. The other two feet screws are at the front of the case, well away from the motherboard.

In this state, I now have the cables tidied up, and the lid back on. I've power cycled it several times, un-plugged and re-plugged it several times, so far so good.

Unless anyone has any other ideas, I'll now go back to my original plan of continuing to let this thing run for a while, and see if the mystery issues pop back up. Hopefully the mystery is solved and this can go back to being the boring system it's supposed to be. :)

Thanks all!
 

evolucian911

Member
Jun 24, 2017
36
6
8
35
At the risk of jinxing myself, I might have this fixed. The bottom of the S-35 chassis is entirely flat... almost. At the very back, near the rear IO shield, there's a slight bump that runs almost the whole width of the case. It's not very tall, maybe 1mm. Nevertheless, I covered it with a strip of electrical tape.

The other suspicious item was the protrusion for the screw for one of the case feet. I didn't measure, but it appears shorter than the PCB standoffs. But, some of the through-hole components on the motherboard have leads that aren't trimmed completely flush with the board, so it's possible that it could be creating a short---or just barely enough that subtle movements, contraction/expansion due to heat, fan vibrations, etc could cause an intermittent short. So I cut a small piece of plastic to put over that screw, and taped it down with bunch of electrical tape.

There's another one of those foot-screws sticking up near the board, but not directly underneath. I went ahead and gave it the same plastic and electrical tape treatment. The other two feet screws are at the front of the case, well away from the motherboard.

In this state, I now have the cables tidied up, and the lid back on. I've power cycled it several times, un-plugged and re-plugged it several times, so far so good.

Unless anyone has any other ideas, I'll now go back to my original plan of continuing to let this thing run for a while, and see if the mystery issues pop back up. Hopefully the mystery is solved and this can go back to being the boring system it's supposed to be. :)

Thanks all!
Like I told you. Case shorting out could be likely issue. Put the foam under the board and you can use clear nail hardener on any other areas of concern. Including the the back io area where the back plate goes. I've seen cases like these a hundred times.

Sent from my LG-LS997 using Tapatalk
 

matt_garman

Active Member
Feb 7, 2011
212
40
28
Like I told you. Case shorting out could be likely issue. Put the foam under the board and you can use clear nail hardener on any other areas of concern. Including the the back io area where the back plate goes. I've seen cases like these a hundred times.
Agreed, it was just a surprise, I've never run into this before. I'd say this is a design flaw in this case, there should be plastic covers over those case feet screws. The U-NAS NSC-800 I had a while ago, and also some mini-ITX supermicro cases come with a cut sheet of plastic that goes between the motherboard bottom and case. After this incident, I feel like that should be a standard feature on all cases (or perhaps the motherboards should come with one). Now I know why Supermicro IO shields have foam on the interior side. :)
 

PoopsPoops

New Member
Sep 16, 2017
7
0
1
42
Yeah, whenever I have weird power issues like you described with things breaking when you plug things in I usually try to rule out the chassis or physical connections by taking the mobo out and just running it on a wooden table.