Who has had their Seasonic PSU replaced by Seasonic due to RTX 3080/3090 shutdown issue

[hamilton21017]

Specifically, it is an excessive variability test that repeats "0 ⇔ 99%".
Power supply shutdown that happens while using RTX3080.
This is not just a Seasonic problem.
We are trying to solve this.

 

I experienced so many shutdowns with this Dying Light 2.
I have not experienced this with the current 2021 model.
There is a scene in the video that I want to draw your attention to.
Around the 20-50 second mark.
You can see that there is a very large variable load being applied.
Please try it if you can.
And let us know the result.

 

 

https://i.imgur.com/vFm5kaU.jpg
So why did it happen in my 2020 model?
Here's the situation that originally happened.

1) First time, when loading a game in Call of Duty: Vanguard, it suddenly shut down.
　However, this was not frequent, but once a week.
2) Purchased Dying Light 2.
　It didn't happen at the beginning of the game, but after a while, this also happened when the game was loading.
　Sudden shutdown, but the interval was severe and it happened almost every day (12 times in a row).
(3) Assassin's Creed Valhalla also shuts down when the game loads, which has not happened since the first startup.
　In DEATHLOOP, it shuts down suddenly while playing.
　This was experienced after Dying Light 2.
　What are the conditions? I predicted that it would happen when a combination of I made a prediction.

4) Replaced with June 2021 model
https://imgur.com/Yfblkza.jpg
　Symptoms (1) to (3) stopped.
　This makes it unlikely that GPU failure or game bugs are the cause

It may not happen unless these conditions are touched? It also seems to me that...
Seeking a professional answer.

 

 

[BiotechBen]

What CPU are you running?

[DreamCat04]

6 minutes ago, hamilton21017 said:

Specifically, it is an excessive variability test that repeats "0 ⇔ 99%".
Power supply shutdown that happens while using RTX3080.
This is not just a Seasonic problem.
We are trying to solve this.

 

I experienced so many shutdowns with this Dying Light 2.
I have not experienced this with the current 2021 model.
There is a scene in the video that I want to draw your attention to.
Around the 20-50 second mark.
You can see that there is a very large variable load being applied.
Please try it if you can.
And let us know the result.

 

 

https://i.imgur.com/vFm5kaU.jpg
So why did it happen in my 2020 model?
Here's the situation that originally happened.

1) First time, when loading a game in Call of Duty: Vanguard, it suddenly shut down.
　However, this was not frequent, but once a week.
2) Purchased Dying Light 2.
　It didn't happen at the beginning of the game, but after a while, this also happened when the game was loading.
　Sudden shutdown, but the interval was severe and it happened almost every day (12 times in a row).
(3) Assassin's Creed Valhalla also shuts down when the game loads, which has not happened since the first startup.
　In DEATHLOOP, it shuts down suddenly while playing.
　This was experienced after Dying Light 2.
　What are the conditions? I predicted that it would happen when a combination of I made a prediction.

4) Replaced with June 2021 model
https://imgur.com/Yfblkza.jpg
　Symptoms (1) to (3) stopped.
　This makes it unlikely that GPU failure or game bugs are the cause

It may not happen unless these conditions are touched? It also seems to me that...
Seeking a professional answer.

 

 

What wattage is your PSU? It could be the really high power spikes of 3000 series cards

[hamilton21017]

I am using an 850W power supply.
https://imgur.com/SAn94uR.jpg
Here is the info on HWiNFO.

At the time it was a 2020 manufactured model.
This has not happened with the current 2021 manufactured model.
Error history at the time.
https://imgur.com/6Ur3Jig.jpg

Did Seasonic do any work inside between 2020 and 2021?
Here is the answer from Seasonic.
https://i.imgur.com/DrCDu7V.jpg

[Slizzo]

Not sure what the OEM on my EVGA powersupply is, but I have an RTX 3080 Ti FTW3 Ultra, 10940X both under a full open loop. 850W P2 powersupply.

 

Never had any issues in Dying Light 2 besides the normal jank from that game.

[BiotechBen]

A 3080 @ "320" watts and a

10940X @ "165" W, with the transient power spikes of the 3000 series cards and the optimistic power estimates from Intel, I would probably suggest a 1k watt PSU if you don't want to deal with full load OCP tripping potentially.

[Guest]

2 hours ago, Slizzo said:

Not sure what the OEM on my EVGA powersupply is, but I have an RTX 3080 Ti FTW3 Ultra, 10940X both under a full open loop. 850W P2 powersupply.

 

Never had any issues in Dying Light 2 besides the normal jank from that game.

Your PSU is made by Super Flower.

4 hours ago, hamilton21017 said:

I am using an 850W power supply.
https://imgur.com/SAn94uR.jpg
Here is the info on HWiNFO.

At the time it was a 2020 manufactured model.
This has not happened with the current 2021 manufactured model.
Error history at the time.
https://imgur.com/6Ur3Jig.jpg

Did Seasonic do any work inside between 2020 and 2021?
Here is the answer from Seasonic.
https://i.imgur.com/DrCDu7V.jpg

Does the "2021" version have the +12V sense pin?  I've heard that to "fix" the problem, they sent out cables with the +12V sense removed.

4 hours ago, hamilton21017 said:

Specifically, it is an excessive variability test that repeats "0 ⇔ 99%".
Power supply shutdown that happens while using RTX3080.
This is not just a Seasonic problem.
We are trying to solve this.

Too many contradicting sentences for my brain to register.

 

This is or is not a Seasonic problem?

 

The problem occurred with the old PSU, but now is fine with the new PSU?

[Slizzo]

Shit, the man himself here.

 

Trust anything jonnyGURU says about all things powersupply @hamilton21017

[hamilton21017]

I am very sorry for the lack of clarity.
We have changed from the old 2020 model to the 2021 model, and the abnormal detections including errors and shutdowns have not occurred for over a month now since the changeover.

[Guest]

8 hours ago, hamilton21017 said:

I am very sorry for the lack of clarity.
We have changed from the old 2020 model to the 2021 model, and the abnormal detections including errors and shutdowns have not occurred for over a month now since the changeover.

So then answer the question:

"Does the "2021" version have the +12V sense pin?  I've heard that to "fix" the problem, they sent out cables with the +12V sense removed."

 

The wire in question is circled below:

[hamilton21017]

Understood.
I will send you the image then.
This is June 2021.
This is the current TX-850 model.

[hamilton21017]

I think it is unlikely that the "+12V sense pin" is the direct cause.
I have been using the same cable as the 2020 model with prior permission from the manufacturer.
In other words, I only replaced the unit.
The shutdown problem ended immediately after I replaced it with the 2021 model.
I am now doing daily load tests.

Here is one thing I can think of.
Could the internal components have changed?
Has anyone done a thorough comparison check of the internal components for the 2020 and 2021 models?

[Guest]

5 hours ago, hamilton21017 said:

I think it is unlikely that the "+12V sense pin" is the direct cause.
I have been using the same cable as the 2020 model with prior permission from the manufacturer.
In other words, I only replaced the unit.
The shutdown problem ended immediately after I replaced it with the 2021 model.
I am now doing daily load tests.

Here is one thing I can think of.
Could the internal components have changed?
Has anyone done a thorough comparison check of the internal components for the 2020 and 2021 models?

If it works, why do you care anymore?

 

[hamilton21017]

I would like to know the reason for this.
Currently, I have only been using it for about a month.
Maybe it will recur in 3 or 6 months.
Before that happens, we want to know why.

Why did it shut down?
The number of times it made errors was unusual.

[Guest]

10 hours ago, hamilton21017 said:

I would like to know the reason for this.
Currently, I have only been using it for about a month.
Maybe it will recur in 3 or 6 months.
Before that happens, we want to know why.

Why did it shut down?
The number of times it made errors was unusual.

It works because Seasonic fixed a well known, long standing problem they've had with their PSUs?

 

How in depth do you need to get into this?  Seems like you're just curious and there's no real reason to dig any deeper outside of this curiousity.

 

But if you want to get into the weeds, we can get into the weeds, because I know exactly what the problem was.

 

[Guest]

Here:  https://cultists.network/4904/deepcool-pq1000m-review/

 

Scroll down to the section titled "Nvidia 3xxx?"

[hamilton21017]

季節限定 850 プライム ウルトラチタン パワーサイクル (OCP?) (3090 |ページ 2 |Overclock.net

Therealjustin has asked Seasonic a question.
Seasonic responds that they have no history of sending replacement cables.
Where is the basis for your information?
Please do not write false information.

The cables are from the 2020 model without removing them from the main unit, and the only replacement is the replacement of the main unit.
I have tested it repeatedly for about a month and a half under heavy load, including games and benchmarks.
The number of times may be counted close to 100 times.

 

Yes.
There is no abnormality at all.

[hamilton21017]

I also experimented with higher load tests.
Prime95 SmallFFT and FurMark at the same time.
At this time, CPU load was 100% and power consumption of nearly 200W was maintained for a long time.
I then applied a variable load to the GPU load intermittently.
No abnormalities were observed.

[Guest]

15 hours ago, hamilton21017 said:

季節限定 850 プライム ウルトラチタン パワーサイクル (OCP?) (3090 |ページ 2 |Overclock.net

Therealjustin has asked Seasonic a question.
Seasonic responds that they have no history of sending replacement cables.
Where is the basis for your information?
Please do not write false information.

Please do not accuse me of writing false information.

therealjustin's post is from Sept. 2021.  Seasonic probably didn't know the actual root cause of the problem back then and started shipping the cables out to people after that date.

 

There are LOTS of customers that, when they contacted Seasonic about the issue trying to get an RMA, were sent replacement cables without the +12V sense pin to remedy there issue.  But all of these people I've seen get these replacement cables were within the 2022 year.

 

13 hours ago, hamilton21017 said:

I also experimented with higher load tests.
Prime95 SmallFFT and FurMark at the same time.
At this time, CPU load was 100% and power consumption of nearly 200W was maintained for a long time.
I then applied a variable load to the GPU load intermittently.
No abnormalities were observed.

You HONESTLY expect to capture these abnormalities with PC software?  Really?  What I showed in the link above (which was someone quoting me) was measured using probes attached to the +12V, power good and +12V sense.  No PC software is going to capture that.

 

Besides... the reason you're not shutting down is because Seasonic figured out the problem and fixed it.  I don't know why you're so thick to get down to the root cause when you're using sticks and rocks to determine the problem.

 

[Guest]

13 hours ago, hamilton21017 said:

I also experimented with higher load tests.
Prime95 SmallFFT and FurMark at the same time.
At this time, CPU load was 100% and power consumption of nearly 200W was maintained for a long time.
I then applied a variable load to the GPU load intermittently.
No abnormalities were observed.

Please note that there is no monitoring of the +12V sense wire in HWInfo.  So what you're showing in these screenshots is just garbage data.

 

 

 

[hamilton21017]

Perhaps there is some misunderstanding.
This image is presented only as a "load test running history" document.
Please understand that the
The model with a manufacturing date of December 2020 performed an immediate shutdown at 0 ⇔ 99%.
However, the June 2021 model did not shut down immediately after replacement for about a month and a half, even after repeated testing.
The cable came with the December 2020 model.
Perhaps that user needs to replace the cable via Seasonic.
Is that user a pre-2021 model?
That is not what I am trying to find out.
I want to know what you know.
What changes have been made to the current model? [...] 

[hamilton21017]

I have one concern and worry.
That is the "a few more months" status mentioned here.
I am worried because I bought the 2020 model new and this happened when I started up "Call of Duty Vanguard".
That was 6 months later.
Now it has only been a month.
That is why we need your help.
So I wanted to know about the differences between the internal structure of the 2020 and 2021 models.

[hamilton21017]

There has been no history of any shutdowns since March 22, when the 2021 model was replaced.
High load tests have been run daily for about 2 months now, but there are no shadows.
If we continue to do the same thing for almost 6 months, will a shutdown be more likely?
 

[Slizzo]

Look man, you're getting way too far down into it. Your PSU was replaced with one that doesn't have the issue of the previous one.

 

Continue to do what you do normally, the specific cause of your previous issue has been fixed, even according to Jon. BTW, Jon is very well respected in the community, has reviewed PSUs for a very long time and has now been working at Corsair for quite a long time (designing PSUs).

 

 

 

You're writing to the guy on the right in this video...

[Sleepycat3]

9 hours ago, hamilton21017 said:

There has been no history of any shutdowns since March 22, when the 2021 model was replaced.
High load tests have been run daily for about 2 months now, but there are no shadows.
If we continue to do the same thing for almost 6 months, will a shutdown be more likely?
 

I've run my PX-850 for 6 months now, always full load on my 5900X and tensor core load on my 3080. Zero issues since replacing the old SSR-850PD with the new March 2021-built PX-850. If it has not shutdown for you, then keep using it. If it does shut down years from now, use your warranty again.