Jump to content
Corsair Community

Recommended Posts

Posted

Hi Corsair!

I bought an rm850x and I thought it worked great. However, long story short, it (secretly) managed to destroy one mother board and cripple another before I realized it had to be the PSU and replaced it with the old one.

I wanted to feedback this rather bad fault condition. Tried to do it in a ticket, where the support offered to replace it but I believe the feedback part got lost. I'll try it here instead.

Longer story. It appeared to work wonderful. Stress tested with CPU and GPU no problem, played some games, quiet while performance was great. I put the computer to sleep over night, next morning it doesn't wake up from suspend. Cold booting works however. But, ethernet (on MB) doesn't work. Fast forward. Another night of suspend and next morning it does not wake up. Also does not cold boot any more.

Unfortunately I guessed it was the old MB that had an issue or didn't survive the maintenance.

Managed to find another similar MB. Also worked wonderfully at first. Stress tests no problem. Everything performing while being quiet and cool. Next day, however, almost same story. Didn't wake, cold boot worked but ethernet (on MB) was gone, as was all the rear USB ports. At this point I removed the PSU and replaced it with the old one.

I also took some measurements powering the supply alone outside of the computer. To sum up. It appears to work in system. Measured voltages (no load) when power on manually outside system also look ok. Only irregularity I have been able to measure is a spike on the 5V standby power line when turning it off (removing pull down on PS_ON pin). There is a really odd quick jump from 5V to 5.35V before it returns to 5V. After a few repetitions also observed spikes as high as 5.5V. This would be outside of a 5% tolerance.

This could actually explain why issues only were discovered after suspend. PSU is powered down on suspend and the spike on the standby voltage rail zaps some chips. Standby voltage also checks out with damaged functionality. Ethernet would be connected because of Wake on LAN functionality as would be USB for being able to wake up by keyboard and such.

Seems like a great product... otherwise. Was really happy with the quiet operation until it started to destroy things 🫠 😬. Please help make sure that this fault condition is tested for the future and let me know if I can bring any other information. I'll attach the screenshots of the "spike" on the 5V standby rail for reference.

Best regards,

Christoffer

5V_SB_load70mA.png

5V_SB_load5mA.png

5V_SB_load25mA.png

Posted

The specification for the overshoot at turn on / turn off is 10% of the rail nomilal voltage, so the PSU is fine.

Posted

Hmm I am not familiar with the concept of overshoot when there's nothing to "shoot" for. To be clear, it's the standby voltage rail going from 5V to 5V, that is, shouldn't change. Everything else is turning off.

The PSU is most certainly not fine hehe. Granted, it does not have to be my observed behaviour that killed the motherboards, but it did kill the motherboards.

Let's say 10% would be fine. The observed behaviour still look really weird to me for a quality power supply, leaning on the very edge of that spec. And, I did some more measurements after the initial post, with a higher bandwidth probe for a higher frequency response. I'll attach the result.

In the attached screenshot there's high frequency oscillations appearing during the "shoot" with tops towards 6V, that'd be 20% from nominal.

5V_SB_x10.png

Posted

well the thing to shoot for is 5v. overshoot is the temporary spike when turning on or off a supply.

if i read the scale correctly, the voltage spikes to 5.4v, the rest is just noise that is filtered by the motherboard. your measurement was with the motherboard connected or floating cable?

Posted

The constant inherent noise is baked into the thicker yellow line already. We're at 200mV per division which makes the signal look rather clear. I didn't dismiss it as environmental noise because it appears exactly for the duration of the of the spike.

I'll sort the rest of the captures later.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...