• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Check Your 4090's 12vhpwr Connectors To Make Sure They are in Pristine Condition

I still believe power cycling could be contributing to all this. Be that as it may there are plenty of people reporting certain pins melting because they are catching it early. Although I would still count those post that got deleted.
 
People that don't understand how Reddit and sticky threads work, they haven't deleted anything, theyve added them to and referenced them in the 1st post of the megathread
You need to stop spreading false information like that. They are deleting threads. Yes, the thread "should have been moved" but it was not. It was deleted.
If the thread was moved they would have used automoderator to remove them and added a comment to a link to the megathread. Or, just use a PRAW bot. All they did was just remove the post.
 
Last edited:
ATX 3.0 PSU now melting connectors too, again maybe not inserted correctly or is it an issue with the graphics cards or the silly connector standards.



https://www.reddit.com/r/nvidia/comments/yltzbt/maybe_the_first_burnt_connector_with_native_atx30/

92r54iC.jpg

rnNkDSP.jpg

tzOOlyt.jpg

ozl1S9M.jpg

H5jiMo7.jpg
Well, well, well. Look at what we have here. This guy has a ATX 3.0 PSU and using a Gen5 cable for his Gen5 connectors and still melts the cable? This throws a pretty sizeable and harden wrench into the works that this was limited to a few adapters. This is starting to look like a major design fault. And blaming people for "not connecting them correctly" will simply be ignored when the design itself comes into question. Lets be honest with ourselves people are plugging them proper.

Personally, I wouldn't want to use the GPU until the issue is identified, verified and confirmed to be fix. Spending $1600+ is a lot of money to worry about melted connectors. Specially when I know, on my end, I've done everything correctly.
:(
 
Last edited:
This guy also looks like an experienced system builder and has many graphics cards and systems and an enthusiast too.. so it is starting to make me wonder what is really going on as we have not seen even one FE with a melting issue... is it the cards for AIBS that are the problem and they didn't follow Nvidia's power advice on the reference models design that they give them... It all smells of more than one issue causing all this now to me or a combination of issues.
If only EVGA could speak freely on the situation. Because I am beginning to wonder if this has anything to do with EVGA's "last straw". Because it's eerie coincidental that they got out right when this turns up.
Just saying...
 
Honestly this scares the hell out of me as sometimes I leave my pc running projects and this would have me worried leaving the machine and walking away for long periods of time or going to be leaving it to work overnight.

They need to fix this and start making AIB cards with 8pin connectors on again as they did with 30 series.
Agreed!
I just don't understand why this isn't taken seriously. Optically, they treat this like it's within margin of error or something.
 
you can’t crimp that connector. 24 wires going into 12 wires across 4 separate pcie8 connectors. Doesn’t make any sense for crimping. You can combine 2 wire into 1 pin for crimp. But that means some of the pcie8 pin will be crimped with other pcie8 wire which is not meant to be done.

Doesn’t seem to be the solder joint being the issue so far as that. Even Igor has stated as much.
It's not that they "can't" they didn't do the proper research. Which is why the design is flawed. Think about it what you just posted for a minute. No wonder we are seeing these issues. So they cheesed out on solder.
 
Last edited:
I agree the design is flawed. It was a crap concept to try to bridge 4 pcie8 to a single 16pin. Madness. They should have gone with 2 pcie8 and limit the power to 450w and be done with that. And unlock 600w with native pcie5 support.
Solder joints are used across the industry for power connection. I mean insides of PSUs all joints are solders. It is not the issue that the connection is soldered.
I separated this because you are completely contradicting yourself in a disconnect in trying to defend soldering in this gen5 connector. You cannot say the solder is fine in and of itself and ignore the fact that the design is completely flawed. You obfuscate the solder when the entire design is flawed from top to bottom fashion. And yes, using solder with such an obvious untested new connector is certainly not an industry standard for PC gaming. The important part is "industry standard". And like what Surveyor stated you are comparing apples to oranges trying to justify it.

Using solder was a horrible choice and comes off as cheesy "work around" to me.
;)
 
Last edited:
Soldered connection is not the premise of “bad connection”. As I pointed out. Soldered wires are used everywhere in a PC and PCB and power electronics. Just because a connection is soldered doesn’t not mean it is unfit for purpose. That’s the primary point.

The design of the backward compatible 12VHPWR connector that Nvidia came up with is clearly flawed no argument there.

The issue of the adopter is that it shouldnt be done like that. Soldered joints or not. It’s a Frankenstein. Never should have existed. They should have simplified it ie power capped the card when run on adaptor and full juice on pcie gen 5 native. Or do the decent thing supply every card with an ATX3 psu.
No need to argue. I simply don't agree with you with the onset of how many "reported" connectors failed so far. Trying to salvage soldering used as "this is fine" and imply it had no baring on the failure rates we are seeing is argument in a bubble that popped a while ago.
;)
 
The 3090 Ti actually had a crimped connector. Supposedly, the reason we see solder instead of crimping is because nvidia is not using the newer single seam terminals recommended by PCI SIG. Nvidia found that when the cable is bent in a North to south orientation, the terminals come loose and pop out and reported in to PCI SIG. This was back in August. In order to meet the timeliness for the 4090 launch they couldn't wait for PCI SIG to address this, hence they developed their own solution and use solder.

The Nvidia adapter itself is flawed.


There are very few cases on the market which are wide enough for the 4090. Average width of an AIB 4090 is around 150mm. Adding 36mm on top of that gets you to 186mm. This is right on the edge even for massive tower cases.

With how wide these cases are, its baffling how nvidia didnt deduce people would do thisnor atleast bundle a 90 degree adaptor.
Yeah, the use of solder always looked liked cheesy work around to me. This whole situation is completely baffling indeed. This has turned into a well known issue amongst PC gamers. And, I seriously doubt Nv will do anything about it unless pressured to do so. Shame really. But once someone decides to go to their local news outlet, etc besides reddit we might see change.
 
Last edited:
literally 0 have caught fire... 105c to melt around 350c to ignite
That's not important. What is important is the actual damage to the connector. Regardless of how it happens a customer's confidence in the video card is eroded when they see that they can't reliably feed power to the video card.

Nvidia may not have a choice, American 4090 owners can now sign up to sue Nvidia

hmm...lets see where this goes.
 
Last edited:
Then we should have seen burnt-connector megathreads with every GPU launch.

I don't recall seeing this kind of cluster_before.

I suspect the majority of people with melted cables managed to not melt the power connectors on their previous GPU installs yet, strangley enough, must have forgot how to plug in a power cable when they purchased their 4090's.

What a coincidence.

Good point here. How can someone victim blame for not inserting the connector correctly when they were able to insert the pcie connectors just fine. For well over a decade now.

Perhaps, just perhaps the connectors are not fitting properly? Have a look:


This user suggests that these connectors are not fitting properly per jonnyguru.





Anyway, this user found out that his cables female pins are uneven. At first glance you might miss it. But if you look inside of each plastic connector you will see that the metal female pins are not even inserted.

Therefore, if you bought cables you want to check to make sure that the metal female pins are inserted evenly now.




These myriad of issues could actually be contributing to the issue. Which is why it's hard to replicate.
-connectors not fitting properly
-uneven pins
-bent cables
-using solder
-etc
Could all be factors to why they melt. Not just bent cables alone. Just an overall flawed design IMO.
 
Last edited:



Found another instance where the connections is not seating properly. This is not user error. But tolerance issues with the male portion of the connector fitting inside the female portion. It's not fitting flush.

It's pretty hard to see. I suggest 4090 owners take your phone's camera and put it on macro mode. Then take a picture of your connectors to see if the male portion of the plug is flush inside the female portion connected on the gpu.
 
Last edited:
how do you know they didn't just not properly insert it all the way in the first place then it slightly pulled out with cable management.

doesn't seem like much of a bend but probably wouldnt take much force to unseat it when it's not clicked in place
The real question is how do you prove that they did? I can only show you the results. These same people have been using pcie connectors for years w/o issue. That's is how I can prove to you that they inserted the connector properly.

It's getting quite obvious, for me, that the connectors themselves are not true to tolerance. And the cause of the imperfections to actual fit and finish is the fact that the design of the connectors is flawed. Heck, at a guess if you took a metal filer and evenly shaved off some of the plastic it would seat correctly. But that's just one of a many different issues we are seeing with these connectors. We have to see the forest before the bark of a tree.
 
Last edited:
That bend is asking for trouble and I bet that's why its getting loose at the connector end. Per Jony Gerow, bending it horizontally that much will cause the adaptor to melt for sure. Vertical bends are safe (nvidia adaptor is an exception) on third party cables as PCI SIG revised the terminals to prevent them from losing contact in a North to south bend. East to west bends continue to be very dangerous.

That case clearly cannot handle the 4090 and needs the Cablemod 180 adaptor mandatorily. People should stop trying to force it to fit.
We really need to ask ourselves why is it that pcie connectors work fine for well over a decade while these new connectors wet the bed in a matter of days... Regardless of how you see the issues unfold when compared to a try/true standard like pcie connectors they don't measure up, at all.
 
Last edited:
Back
Top Bottom