I've never seen this - Part 2 - Not a Motherboard issue.

BoiseTech

Limp Gawd
Joined
Mar 15, 2018
Messages
442
I recently built a Threadripper 3960 system with a 1200 watt power supply. It works great 80% of the time, but occasionally when you push the power button the power supply "clicks" and you get nothing, you have to flip the hard on/off switch on the PSU and try again to get it to boot. Sometimes it takes 4/5 tries.

I've switched the PSU, videocards, re-seated everything with no luck. Granted the PSU switch was with another, same model 1200 watt PSU.

I thought it was a video card grounding out on the chipset heatsink, but I just reproduced it with a smaller video card in that doesnt touch the heatsink.

Any ideas?

Specs:

Threadripper 3960x
MSI TRX40 Creator
4 x Corsair Vengence RGB 32gb
2 x 2tb Corsair PCIe 4x SSD
2 x EVGA 2080 Super FTW3
Dark Rock Pro 4 TR4
Seasonic 1200 watt PSU
 
I recently built a Threadripper 3960 system with a 1200 watt power supply. It works great 80% of the time, but occasionally when you push the power button the power supply "clicks" and you get nothing, you have to flip the hard on/off switch on the PSU and try again to get it to boot. Sometimes it takes 4/5 tries.

I've switched the PSU, videocards, re-seated everything with no luck. Granted the PSU switch was with another, same model 1200 watt PSU.

I thought it was a video card grounding out on the chipset heatsink, but I just reproduced it with a smaller video card in that doesnt touch the heatsink.

Any ideas?

Specs:

Threadripper 3960x
MSI TRX40 Creator
4 x Corsair Vengence RGB 32gb
2 x 2tb Corsair PCIe 4x SSD
2 x EVGA 2080 Super FTW3
Dark Rock Pro 4 TR4
Seasonic 1200 watt PSU

Do you have a larger/different model PSU to test with? Even if you OC I would imagine 1200W is enough, but according to this (mostly irrelevant) link, your setup would use around 1100W without any OC. But to be fair that site also said my old system needed 1600W minimum and I ran it off of a 1275W for years.

Also, try without any fans/extra things plugged in (to the mobo and PSU if possible), ive seen USB headers and case fans cause all sorts of weird issues.
 
Thanks for the input!

I've dropped it down to a 1660 (non ti, non super) And replicated the issue. The PSU I'm using is a platinum rated, and when it does boot, I've hit it with prime95, cinebench, AND furmark with no issues / no shut downs. I haven't considered bigger, but I have a 1200 watt EVGA on its way to test, this seasonic is a "version 1" standard ATX size, so maybe there are issues with it?
 
Do you have a larger/different model PSU to test with? Even if you OC I would imagine 1200W is enough, but according to this (mostly irrelevant) link, your setup would use around 1100W without any OC. But to be fair that site also said my old system needed 1600W minimum and I ran it off of a 1275W for years.

Also, try without any fans/extra things plugged in (to the mobo and PSU if possible), ive seen USB headers and case fans cause all sorts of weird issues.


Agree with pulling the fans off the mobo.

I generally run the CPU fan(s) off the mobo, but all other case fans I power directly from the PSU.
You can get converter cables and splitter cables to give you enough fan power connections.

It's better not to run all of that fan power through the mobo when it's not necessary.

May not fix the OP's problem, but still a best practice thing.


.
 
Agree with pulling the fans off the mobo.

I generally run the CPU fan(s) off the mobo, but all other case fans I power directly from the PSU.
You can get converter cables and splitter cables to give you enough fan power connections.

It's better not to run all of that fan power through the mobo when it's not necessary.

May not fix the OP's problem, but still a best practice thing.


.

I missed that on his post. I've done that! All fans are run on a sata powered header. Still having the issue :(

I have a new motherboard and different power supply on the way.
 
Sounds like the PSU isn't liking the large surge on boot and is trying to protect itself. If you're local to 29856 I have a 1000w (EVGA) and a 1200w (XFX) you could test (but not many people are local to me though, lol). If the PSU detects a short on startup it'll shut down to protect itself. The larger M/B and VRMs could confusing it at boot thinking it's overloaded (even though once the initial surge is over it would go back to normal). Maybe the PSU is to sensitive (not long enough boot up delay). Seems once it gets passed this initial draw that it is perfectly fine. Maybe call or ask the PSU manufacturer if they have tested it with Threadrippers before?
 
Sounds like the PSU isn't liking the large surge on boot and is trying to protect itself. If you're local to 29856 I have a 1000w (EVGA) and a 1200w (XFX) you could test (but not many people are local to me though, lol). If the PSU detects a short on startup it'll shut down to protect itself. The larger M/B and VRMs could confusing it at boot thinking it's overloaded (even though once the initial surge is over it would go back to normal). Maybe the PSU is to sensitive (not long enough boot up delay). Seems once it gets passed this initial draw that it is perfectly fine. Maybe call or ask the PSU manufacturer if they have tested it with Threadrippers before?

Yeah, I have an EVGA 1200 on its way, I would have grabbed an EVGA to start, but they were out of stock :(

Thanks for the offer, you're right, I'm in 83701!

We'll see how the EVGA goes, I may do a 1600 watt one, if this also doesnt work.
 
Yeah, I have an EVGA 1200 on its way, I would have grabbed an EVGA to start, but they were out of stock :(

Thanks for the offer, you're right, I'm in 83701!

We'll see how the EVGA goes, I may do a 1600 watt one, if this also doesnt work.
Ok, good luck, please let us know if it fixed it so if someone has similar issues we can have a better understanding.
 
Wouldn't be surprised if this is a mobo issue. I have high doubts that the two 1200Ws you have are both defective or that you need more than 1200W on boot. What a PITA though, gl!
 
Yah, I had an issue one time where a fan header was touching a metal bracket on my GPU and causing similar issues. Worth looking around.
 
I've had a similar issue, but not as intense. In my situation, my power supply didn't like trying to boot up all of my hard drives at once and that "surge" to get them spinning caused it to stop. If I tried to turn it back on in the next few seconds, after they (and all the case fans) had already started spinning, things booted up like normal and were absolutely fine until it ended up at a cold shutdown again. I just got in the habit of doing a weird double-boot if I ever turned it on. Hit power, wait for a click, hit power again... (I think I had 6 hard drives and two DVD burners back then...)
 
Last edited:
I've had a similar issue, but not as intense. In my situation, my power supply didn't like trying to boot up all of my hard drives at once and that "surge" to get them spinning caused it to stop. If I tried to turn it back on in the next few seconds, after they (and all the case fans) had already started spinning, things booted up like normal and were absolutely fine until it ended up at a cold shutdown again. I just got in the habit of doing a weird double-boot if I ever turned it on. Hit power, wait for a click, hit power again... (I think I had 6 hard drives and two DVD burners back then...)

That used to be a common problem many years ago even with a single hard drive connected.
It had to do with the PSU's large caps charging time and the inrush current of the total load.

The PSU would not assert the "Power Good" signal (pin 8 on the 24-pin connector) in the required time.
The mobo would not see power_good and would hold the Reset line to keep the system from booting.

So the first power up/cold boot would always fail. Turn the power off and back on again and it would boot.
The PSU caps would still be charged and the PSU could assert power_good and the system booted.

It was a chronic problem that affected many combinations of PSUs and mobos.

The problem has been mostly fixed these days through better PSU design and I believe they have changed
the timing on when the mobo does the pass/fail test on power_good.

I imagine it's still possible to have that same issue these days due to things like your large number of drives
and the PSU not being able to provide power_good when it's expected by the mobo.

You could try looking at PSUs with better transient/surge ratings to see if that fixes it.

.
 
Yeah, that was with my old system. I don't have any issues with my modern one.
 
Back in the day you occasionally had to worry about not having enough load on the 3.3/5v rails too. I had old cheapo PSU's that wouldnt boot if I only had the 4 pin CPU and 20 pin mobo connector plugged in, had to throw a couple of drives and some fans on it or else it wouldnt boot.
 
Back in the day you occasionally had to worry about not having enough load on the 3.3/5v rails too. I had old cheapo PSU's that wouldnt boot if I only had the 4 pin CPU and 20 pin mobo connector plugged in, had to throw a couple of drives and some fans on it or else it wouldnt boot.

Yep, there has been some sketchy stuff out there.
Still is I'm sure but the major brands know better for the most part.

Some of the bad PSUs out there just had power_good tied to 5v so it was always "good" even if it wasn't. lol
That's just crap engineering there.

.
 
Replaced the motherboard AND the silverstone 1200w with an EVGA 1200w, no issues so far.

FYI I had re-sat everything, uninstalled and re-installed the previous motherboard in the past with no luck.
 
Last edited:
Replaced the motherboard AND the silverstone 1200w with an EVGA 1200w, no issues so far.

FYI I had re-sat everything, uninstalled and re-installed the previous motherboard in the past with no luck.

Do you still have the old PSU? I think its too early to tell, but if you continue to not have any issues with the new setup, try running the new board with the old PSU to see if the issue comes back. It might help narrow down if the issue was with your old board or PSU.
 
Do you still have the old PSU? I think its too early to tell, but if you continue to not have any issues with the new setup, try running the new board with the old PSU to see if the issue comes back. It might help narrow down if the issue was with your old board or PSU.

Was still in its return window so I sent it back.
 
I had this issue with a power supply in a house I lived in when I lived in England. It was due to the old power lines and there generally not being enough current, so the PSU was basically protecting itself and not booting. Only way it would boot most of the time was doing the same thing you have to do - Flipping the power switch and hoping it boots.
 
This guy is the gift that keeps on giving.

So we never got around to setting up this PC for the owner. We have a new issue.

LogMeIn won't stay connected. Every other PC in our network (with the same settings) is just fine. We chalked it up to crappy software on LogMeIn's part, and decided we'd use RDP on this machine.

Well... We just purchased a Digi AnywhereUSB for some security dongles we use, and it experiences constant disconnects. Every other system on our network (with the same settings) is fine. I'm now leaning towards an issue with this machine.

I just dropped in an Asus Zenith II Extreme Alpha motherboard (replacing the MSI that was in it before) but no luck. I've tried the 3 different onboard network adapters, and a few external ones. No luck. Everything about this PC works great, but the AnywhereUSB problem is a deal breaker. It gets great cinebench scores, and passes furmark and other stress tests without overheating.

I used Ryzen Master to drop the cores down to a single CCX and 6 cores - didnt help. I broke the mirror and installed on a single SSD, no luck. I tried a fresh install of Windows off a CD, no luck. I tried a fresh install of windows off a new USB created today - No luck. I updated windows to latest version and updated drivers to their latest version - No luck.



Anybody have any other ideas?

Rolled back software on LogMeIn and Digi and they both work fine now... What a pain.
 
Last edited:
Network drops sounds like a network connectivity problem.

Check the ethernet cable connection at the computer, wall, switch, etc.

If on wifi, check the wifi signal where the computer is, fix low signal, update driver, etc.
If it's a wifi problem, try a wired connection if at all possible.

ETA: Try to narrow down the cause of the problem early on in the troubleshooting.
You don't want to be replacing the motherboard, messing with cores, and re-installing Windows
for a simple network problem.

.
 
Last edited:
Call a priest?
Last time I had one that survived everything being replaced, we literally took the chassis outside and lit it on fire. Replacement worked perfectly.

As an official Sr support engineer for a fortune 100 company, I wrote (on company letterhead) that we recommended sacrificing a chicken. Microsoft came back concurring (officially) and recommended escalating to a goat. The hardware vendor let us burn the box.
 
Try disconnect all but power switch from board header ,reset sw. 'HDled etc. and see if that helps.
 
Whenever I encountered really weird stuff like this I completely disassembled the system and built it in a cardboard box to see if there was maybe a connector sticking out the bottom of the board that was making contact with something in the case...extreme measure...but in several instances, there were things like mfgr defects and a piece of the case stuck up too high and contacted the mobo causing a ground...or a screw fell in the case and lodged itself in such a way that it ramdomly caused issues.

Hope you find the cause!
 
So, semi-long term update. Logmein is still broken, but Digi is working. I rolled back and its fine. The PC is working great other than logmein and their support team transitioned to a new company so we have to start at square one.

We still haven't fully transitioned over to it, but so far so good?
 
Guys.... why the suggestions of loose screws and other hardware troubleshooting when the only issue is a logmein problem?

You don't replace the water pump to try to fix a problem programming radio stations on the car radio. lol

.
 
Did you say you have other machines at the same location that work fine with logmein?

Does it connect and then drop or no connection at all?

.
 
Yeah, over 50 machines at the same location with no issues. It connects, and drops at random between 5 seconds to 20 minutes.
 
Is it on wifi?

Try doing a ping test to your ISP DNS server and let it run for a while.
Look for any dropped packets.

example below:

ping 4.2.2.1 -t

Does it have "DNE Lightweight Filter" listed in the items on the network adapter properties?

Turn off IPv6.


.
 
Last edited:
Board has 2 onboard wired nice and wifi, tried all 3, plus a usb wifi and pcie asus 10gb card. Same result on all 3, reformatted 3 times and tried swapping the motherboard with like model AND a different brand. All no luck.
 
Try a different case? Or a goat?
If the goat fails, may have to go to a steer, or possibly ox sacrifice. If that doesn't work, human sacrifice would be the last step, but I'd probably just throw it out instead of going to that extreme.
 
Ya I'm at the point where I'm convinced that Logmein has compatibility issues with something in the system. Everything else works fine.
 
Did you try building it on a cardboard box to eliminate case shorts/weird grounding/capacitance issues?
 
I hadn't, but I mean, there are no other issues? Ping tests are fine, all my other network based software works?
 
Back
Top