Can DDR5 cause PSUs to go into protection mode?

Hardware Hub PSU
1

I have a very weird issue. I have encountered faulty/failing RAM before (and I have been so unlucky that both my last builds had failing RAM).

I built myself a new gaming/editing rig with a 7950X3D, G.Skill 2x32GB 6000CL30 Trident Z5 Neo RAM on an ASUS X670E-A Gaming motherboard with a Seasonic Prime 1300 TX PSU. First my CPU got damaged by dear ASUS and their ridiculous voltages when EXPO settings were active. Received a new CPU.

Then I started to get issues with the WLAN module, program crashes, reboots. I tested everything, looked at memory dumps, used my RAM in default mode (4800Mt) and even then the PC kept crashing. Determined that it must be the motherboard. Replaced the motherboard and the crashes became a lot fewer.

But now the crashes were a sudden shutdown of the PC due to the PSUs protection coming on. I was dumbfounded and tested everything again. RAM was fine, no errors when testing. There were never any memory dumps and nothing else indicating anything else than a PSU related issue. So I replaced the PSU and cables. Still crashing. So I replaced the GPU as it seemed to be the only other component able to cause a PSU protection shutdown. Still the same crashes.

Switched off motherboard voltage monitoring after consulting Seasonic. No change. Remembered we bought a small freezer and thought spikes/phase shifts from the compressor turning on/off might cause it, so I got a power rail with overcurrent, overvoltage protection and a phase filter. Still crashes.

After weeks of trying I finally ran my RAM in default mode again and now the crashes are gone even though I ran an 18h and a 28h memory test.

I have never thought that RAM issues could lock up a PSU in protection mode, especially with no program crashes or errors showing when running a RAM Test. Just BAM! PC off, protection mode.

My current guess is that the change from the RAM voltage regulation moving from the motherboard to the RAM itself can cause this behaviour. Has anybody else come across a similar error?

I feel really dumb chasing this issue and replacing nearly all components in my PC. Half a year and a couple hundred dollars wasted.

2

Perhaps. As far as I understand, on DDR5 the modules receive 5V and do their own regulation. If the motherboard feeds 5V directly from the PSUs 12 pin ATX connector into the modules, a current spike or so of the module could trip the PSU. This is me just hypothesising, I don‘t know for sure.

3

Thanks for your reply! Guess it’s a bit of an unknown considering DDR5 is still very much new.

4

I would maybe raise the question “How much trouble ASUS’ Expo actually caused”.

I had a case, where I was testing the cpu for overclocking limit, and at some point got a situation, where my PCI ethernet card stopped seeing network (I can relate this only to overclocking, because besides that there was nothing I had done in hours). The card was visible, all looked ok - BUT I suddenly could not connect to the router, which was sitting right next to the PC (by ethernet only, and other ethernet devices were ok). And the onboard ethernet chip didn’t have this problem.

After 2 hours of debugging it suddenly started to work again. Suddenly. Without any explanation.

6

Apologies for reviving an old thread, wanted to share my own experience with this issue.

I have a 7950X3D, a seasonic prime 1300 titanium, and corsair dominator at 6000mhz, on a crosshair hero.
I rebuilt my pc after years of having a 4th gen intel, and switched to AM5 just as the exploding expo fiasco came out. Due to that,t i always ran expo off ever since Gamer Nexus posted the first video.
Thing is my computer, more often than not, at idle, would just turn off like somebody turned off the switch at the back of my PSU. This happened very frequently, 3-4 times a day.

Temperatures were ok, (cooling with 420 AIO), SOC voltage was always in the safe range, and irrespective of what i was doing, often my pc would just turn off, and to restart, i had to flick the PSU switch on and off.

I even bought a UPS to isolate the pc cause the click the PSU was making when turning off, made me think a spike in my voltage line would cause an over sensitive OCP to kick in. Not even that helped.

The thing that “fixed” it was staying, on top of bios updates.
After a certain update (cant really tell, which, because i flash new bioses as soon as they come out), the issue just simply went away.

My guess? ASUS’s crap OCP and the seasonic prime being overly sensitive to spikes caused the problem. Now im running EXPO I without any issue. I run a custom curve optimizer that nets me 39K in Cinebench R23. Now, I don’t know if my CPU was damaged by either EXPO or the frequent hard power offs, but it performs like a dream.

I’m more than happy to help you troubleshoot.

1 Like
7

Try disabling your onboard WiFi. The AMD WiFi module uses a donkey-poo grade MediaTek chip. I would advise ripping it out and replacing with an Intel AX210 even if it’s not the culprit.

8

I read a lot about this. And this is the only reason I yet keep expo off inmy kingston 6400 (its on Asus Qv) on my new threadripper system.
Yet I don’t know what Asus or AMDs policy is regarding str5 Sockets.
Sad to read how much problems you guys had.
I Postet my built on another thread. Since now it’s pretty solid…

What video do you refer to on GN?

9

To throw it out there, any part in a system can cause OCP/SCP under the right conditions, because power goes in from the power supply in 12v/5v/3.3v rails, and comes back out to the power supply over ground wires. Can be really hard to diagnose, and the tuning of boost behavior, or even memory timings could cause weird power fluctuations.

I’ve heard similar stories to this with HEDT Intel systems back in the day, where the motherboard had overly sensitive undercurrent/undervoltage protection, and the power supply having transient dips caused the motherboard to trigger power protection. Allegedly a bios update fixed it for most but not all people having issues, but switching to entirely different PSU platforms often solved the problem as well.

2 Likes
10

Who knows what was causing it on my end, maybe my 3090ti? Glad its gone away. it was quite annoying to deal with.

1 Like