Enclosure Crashing when riser is turned. Sandwich case requires severe bend.

[2335416fvc]

I switched from and EATX case to a Sliger SM580 recently and have had huge stability problems. I narrowed it down to the riser that came with the case. Like any sandwich style case the riser comes from the top of the case, makes a 270 degree turn, goes down the back of the motherboard, and makes a sharp 90 degree turn before plugging in at a 90 degree connector. I had a spare 16x PCIE riser on hand and, if fairly straight, had no stability issues. With the included riser, it would crash within seconds of any GPU load. After Sliger kindly sent me a replacement riser I've been having the same issue. But... if I let the riser cable be mostly straight it does not crash. Do I just have two bad risers or is there something inherently wrong with this case design? I ran all this hardware before the case came in without a riser on my big case without issues. Motherboard is a Asus Strix Z270-i. I've tried setting to PCIE 2, no change. Crashes experienced with a Vega 64 Nitro+ and my old GTX 770. I can induce a crash by simply articulating the cable 60 degrees or so and running 3D Mark. Might my motherboard be incompatible with such a sandwich case due to riser turns? I've reinstalled Windows twice and removed any unnecessary peripherals.

Solved edit: bad PCIE power plug on the EVGA PSU.

 

[BaK]

I can induce a crash by simply articulating the cable 60 degrees or so and running 3D Mark.

If by doing this everything stays connected, this seems to point out a problem with the riser.

Are you able to check how your spare riser works if you bend it like the ones you got from Sliger?

 

[Nanook]

Sorry to hear. That sounds like a bad riser (or two of them in a row).

 

[2335416fvc]

If by doing this everything stays connected, this seems to point out a problem with the riser.

Are you able to check how your spare riser works if you bend it like the ones you got from Sliger?

I was thinking it was something other than the riser when I used my spare straight one and it might have problems if it were smushed in odd angles but when I gave it a gentle curve and laid the GPU on top of the case it worked fine. But yeah, odd crimps caused problems with all three it seems.

About inducing a crash... I'll do a 20 minute Firestrike stability test with the cable straight and it'll be fine, then I'll carefully maneuver the card in its normal place with as gentle a curve on the riser as possible and do the test again. Fail-crash-black screen in seconds. On the old GTX 770 it would freeze the screen while on the Vega the screen goes black.

 

[Nanook]

Sorry if dumb question... Did you shut down the pc before adjusting the cable?

 

[2335416fvc]

Sorry if dumb question... Did you shut down the pc before adjusting the cable?

Not a dumb question. I did not shut down between adjusting the position of the GPU. The cable stayed plugged in throughout. I was able to use the system for a minute to start the stress test between moving the GPU and doing the test that resulted in a crash. I then tried the test again but with no fans on the GPU and case fans at max. GPU super thermal throttled, but no crash. I put the fans on the GPU and run all fans at max with the riser in the "correct" crimped position. Very cold, crashes almost immediately. I relax the GPU and lay it with all fans on. Runs well, no problems. I put the GPU back standing up but pulled out from the system and suspended with zip ties. Riser is not stressed. No crash. EGPU success! I've also rerouted the internal 120v power cable away from the riser for fear of EMI. No change with that idea.

 

[Nanook]

I've only seen crashes like that when I unplugged my GPU (by mistake) while testing new components. Based on that, it sounds like there is either a short, or a broken trace in your riser(s). I've used 6-7 risers before (Thermaltake, Phanteks, DanCase, Sliger), and have only came across one bad riser (in the Thermaltake G3).
Also can you double check your PCIE power cable(s) as well? I assume you had to switch from ATX to SFX PSU.

 

[BaK]

But yeah, odd crimps caused problems with all three it seems.

Would be very bad luck...
I am not following the Sliger threads, is such an issue happening to others too?

With the included riser, it would crash within seconds of any GPU load.

[... ]then I'll carefully maneuver the card in its normal place with as gentle a curve on the riser as possible and do the test again. Fail-crash-black screen in seconds.

Does it mean you have no issue with curved risers when idling?
If so, checking the power part of your build wouldn't indeed be useless.

 

[2335416fvc]

I've only seen crashes like that when I unplugged my GPU (by mistake) while testing new components. Based on that, it sounds like there is either a short, or a broken trace in your riser(s). I've used 6-7 risers before (Thermaltake, Phanteks, DanCase, Sliger), and have only came across one bad riser (in the Thermaltake G3).
Also can you double check your PCIE power cable(s) as well? I assume you had to switch from ATX to SFX PSU.

Astute observation there. Yes, I went from an ancient Antec non-standard PSU (P180? P190? case) to an EVGA 650GM. Last night I tried a spare EVGA 1000 G3 and have been unable to crash on gaming. Chrome no longer works with hardware acceleration enabled (fixable with a startup flag on the app) so thats weird. I might try a different wiring setup with the SFX PSU again. The SFX PSU, with bent cable, would crash. For cable cleanliness I was using the double headed 8-pin but it can run two independent wires.

 

[2335416fvc]

Would be very bad luck...
I am not following the Sliger threads, is such an issue happening to others too?

Does it mean you have no issue with curved risers when idling?
If so, checking the power part of your build wouldn't indeed be useless.

I was asking on reddit and other Sliger users are not having problems. I'm personally worried about defaming Sliger because I really like their designs and customer service. They sent a replacement riser with virtually no friction. They've been tossing ideas at me on what to test as well as other users.

Idle and CPU centric tasks are okay. Switching to a big-ass PSU for testing gave no errors except a new Chrome error I hadn't experienced before. As noted above, I'm refining my thoughts on what the power issue may be with additional tests.

Besides all that, I pulled the offending GPU/riser combo and tested on a known good R5 1600 system and could not crash no matter how severely I molested the riser cable. [testing intensifies]

 

[Nanook]

Astute observation there. Yes, I went from an ancient Antec non-standard PSU (P180? P190? case) to an EVGA 650GM. Last night I tried a spare EVGA 1000 G3 and have been unable to crash on gaming. Chrome no longer works with hardware acceleration enabled (fixable with a startup flag on the app) so thats weird. I might try a different wiring setup with the SFX PSU again. The SFX PSU, with bent cable, would crash. For cable cleanliness I was using the double headed 8-pin but it can run two independent wires.

You may want to confirm bad pair power cable by testing the gpu direct mounted to the motherboard.
If it’s the SFX pcie power cables, it’s an easy fix. If it’s the SFX PSU itself, you should replace the SFX PSU. Either way, you will have to contact EVGA

 

[2335416fvc]

You may want to confirm bad pair power cable by testing the gpu direct mounted to the motherboard.
If it’s the SFX pcie power cables, it’s an easy fix. If it’s the SFX PSU itself, you should replace the SFX PSU. Either way, you will have to contact EVGA

I've narrowed it down to VGA1 on the PSU, I think. VGA1 and VGA2 in use together powering the GPU, crash. VGA1 in use, double cable, crash. VGA2 double cable, good. So yeah, bad port on the PSU.

 

[Nanook]

Nice troubleshooting. Good luck with returns and with replacement PSU!