Nvidia RTX 4090 power connectors melting?

He did say he had to unclick it to pull it off so it was clicked in all the way. peopel are saying even if its clicked in you gotta push more to make sure there is no gap.
Somoene else showed a zoomed-in bit of the pic claiming marks on it showed the connector wasn't fully engaged, FWIW.
 
Somoene else showed a zoomed-in bit of the pic claiming marks on it showed the connector wasn't fully engaged, FWIW.
Like I said it shouldn't be a rocket science with this. I still don't like this plug, go veritical or go home kinda situation with these cards. The user said it was clicked in then on top you gotta make sure even if its clicked in its pushed in lmao. Its all weird to me that much attention has to go in to a connector and cables. Never had this issue with 8 pin, its clicked its done. Don't have to push harder, that is what clicking it in should do. I think the cable can become lose if its being pulled on and get lose even if clicked in, thats the real issue to me.
 
Like I said it shouldn't be a rocket science with this.
Absolutely. This is a shit design.

Maybe nVidia should've brought back screw terminals.

term8.jpg
 
I was using a cablemod cable on my 4090 for several weeks cause the Seasonic complimentary cable was delayed. Last weekend I switched it out for the seasonic one and did not see any damage to the cablemod cable. The seasonic cable is much nicer than the cablemods one and way more flexible. I never had any problems with the stock Nvidia adapter either, but cable management was a pain.
 
Purpose-built 12VHPW melted. Don't tell me the user that bought and installed that was not aware of needing to seat it properly.
Wouldn't say that. Just saying the design on these 12VHPWR is crap so just clicking in often isn't enough. So one really does have to make sure.

But I agree the fundamental issue is a design one.
 
No matter how idiot-proof you make anything, there will always be an idiot to out-idiot the idiot-proof. Never underestimate the stupidity of humanity. :coffee:

From the looks of it, Nvidia didn't try very hard to make it user friendly, quality testing was obviously not a priority for them, otherwise this issue would have shown up in test labs.
 
No matter how idiot-proof you make anything, there will always be an idiot to out-idiot the idiot-proof. Never underestimate the stupidity of humanity. :coffee:
Not like this. I mean my 8 pin clicks I am done. To say you have to push it hard then clicks you are obviously looking for trouble lmao. That just means click isn't enough and you can still have a little pull on cable somehow and over time it can shit the bed. Not something user should have to deal with.
 
I'm using a CableMods cable and a FE card. That little plastic bend indicator thing is pressing on the top of the cards heatsink. It made me a bit nervous and I almost removed it before finally leaving it on. I did manage a very good click out of it and tugged on the cable to ensue there was no way for it to wiggle loose. So far so good but yeah a little scary to see.
 
Yeah universal design is when a locking mechanism for a cable engages it is supposed to be inserted the correct amount.

Now whether or not that is what truly happened who knows, it could have been not engaged then the act of him going in to grab the cable caused it to go that extra bit to engage hence why he needed to unclip it.

But if he did in fact clip it in then I'm sorry no extra force should be necessary unless the design is fundamentally broken and it is asinine to call it user error.
 
Look closely at the other picture:

1674087923825.png


Especially on the right side, looks like it was several mm out of the socket...

I've had multiple Cablemod cables with 0 issues. Cablemod has clear directions, gotta be able to read.

"But it was clicked in!?!!"

What do you expect he is going to say? "I fucked up..."?? Nope.

That lower line is sus.... it wouldn't surprise me if a couple of these have been caused to melt on purpose... The hate is strong. Someone else on that reddit thread showed where it is apparent it was also plugged in crooked. So if it truly clicked in as the user said, it obviously got loose somehow. More likely is that it was never fully inserted, and became unseated a few mm, then unseated even more.

[H] Forum posters: LET'S GO BACK TO MOLEX! Those never burned!

1674088924432.png

Oh wait...

Something to note, the connectors have meted, but there has never been a fire as far as I have heard.
 
Last edited:
I've have yet seen a good answer to why the 3090/3090Ti did not seem to have this sort of problem with a similar power envelope. With way more of them in service including mining farms etc. What is different besides sense pins? Users less intelligent?
 
Bigger cards and a cable with 4 dang pcie power connectors on a short adapter make the cable management trickier. The connector is tough to push in all of the way. I suspect if the 4 wiresets to the 4 pcie connectors had been about twice as long, this would have been a much rarer (if at all) issue. Trying to pack away 4 big pcie connectors all bunched together and only 4 ish inches from the 12hpwr connector? That wasn't very smart. Longer cables and maybe even 2 different lengths so that the connectors would not all 4 be competing for the same space would have made all of the difference.

I had a 3090 with only 2 cables that ran to the PSU, that thing pulled up to 417W for many extended hours of gaming and had 0 issues. This is over a 2 year lifespan.

My 4090 at 100% power pulls 443W. 6% more power. I'm using Cablemod cable tho and not that shortdicked adapter the card came with.
 
Bigger cards and a cable with 4 dang pcie power connectors on a short adapter make the cable management trickier. The connector is tough to push in all of the way. I suspect if the 4 wiresets to the 4 pcie connectors had been about twice as long, this would have been a much rarer (if at all) issue. Trying to pack away 4 big pcie connectors all bunched together and only 4 ish inches from the 12hpwr connector? That wasn't very smart. Longer cables and maybe even 2 different lengths so that the connectors would not all 4 be competing for the same space would have made all of the difference.

I had a 3090 with only 2 cables that ran to the PSU, that thing pulled up to 417W for many extended hours of gaming and had 0 issues. This is over a 2 year lifespan.

My 4090 at 100% power pulls 443W. 6% more power. I'm using Cablemod cable tho and not that shortdicked adapter the card came with.
I was thinking similar with the bigger ass cards + more cables adding weight making cable management tighter/harder. I am wondering if heat and weight/bends of the cables, heat cycles as well is actually causing some of these cables to overcome the lock? Anyways something is different enough to cause a bigger problem with basically the same type of connection.
 
Look closely at the other picture:

View attachment 542718

Especially on the right side, looks like it was several mm out of the socket...

I've had multiple Cablemod cables with 0 issues. Cablemod has clear directions, gotta be able to read.

"But it was clicked in!?!!"

What do you expect he is going to say? "I fucked up..."?? Nope.

That lower line is sus.... it wouldn't surprise me if a couple of these have been caused to melt on purpose... The hate is strong. Someone else on that reddit thread showed where it is apparent it was also plugged in crooked. So if it truly clicked in as the user said, it obviously got loose somehow. More likely is that it was never fully inserted, and became unseated a few mm, then unseated even more.

[H] Forum posters: LET'S GO BACK TO MOLEX! Those never burned!

View attachment 542721
Oh wait...

Something to note, the connectors have meted, but there has never been a fire as far as I have heard.
yea some user is going to risk burning things on purpose to make a point thats already been made? I get it you don't wanna believe it but to say people are just being shady is pretty ignorant. The user said it was latched pretty good when he had to pull it off becuase he couldn't really do it easily.

The fact that it has never been an issue but all of sudden you have to baby sit an adapter tells alot. May be they should make that design more user friendly or may be its the damn weight who knows pulling on cable in vertical positions. Like I said these cared are best in vertical mount otherwise if you got any pull on the cable it can may be mess up the connection even if its clicked in?
 
I was thinking similar with the bigger ass cards + more cables adding weight making cable management tighter/harder. I am wondering if heat and weight/bends of the cables, heat cycles as well is actually causing some of these cables to overcome the lock? Anyways something is different enough to cause a bigger problem with basically the same type of connection.

yea most likely. I have never had to worry about GPU connector lmao. I honestly woudn't put this card in horizontal position just for the sake of things. Vertical is the way to go with these so there is least amount of pull on cable.
 
  • Like
Reactions: noko
like this
yea some user is going to risk burning things on purpose to make a point thats already been made? I get it you don't wanna believe it but to say people are just being shady is pretty ignorant.

Anything is possible. There is probably 1 rich dickhead out there who would do it. Hell, youtubers take shit apart often destroying the item in the process. You think they give a shit about $1600 (or $0 if they were supplied a review card) to make a high click thru video?
I agree in this case it is a low chance, but I do not think the chance is 0 either.

The user said it was latched pretty good when he had to pull it off becuase he couldn't really do it easily. The fact that it has never been an issue but all of sudden you have to baby sit an adapter tells alot.
Look at the pictures, do you really believe that? At the minimum, too much tension was placed on the cabling. I have not tried to 'pull' the cable out for hours on end on mine... If there is some weakness in the clip I can see one coming loose if it is pulled taught, then it starts heating up, the plastic softens a little, then the clip lets go. That is a probable scenario. It seems obvious to me that there should not be tension on the cables plugged into pcie cards.. avoiding that isn't "babysitting" the connector. I used a 12vhpwr powered card for 2 years prior to these melting connectors without issue.
May be they should make that design more user friendly or may be its the damn weight who knows pulling on cable in vertical positions.
Completely agree that not enough thought and some dumb choices were made in that adapter design. A couple of tweaks to the connector on the adapter side would solve the problem. Position sense pins to be the first to disconnect as the plug comes out a few mm, the card senses that, powers off, user checks the cable. Using a metal clip, or just a thicker plastic would help them from unplugging. Thin those plastic fingers around the adapter contacts a few thousandths so that it inserts more easily. Some redesign of the official connector came out in mid October, a few weeks into the issue. Not sure what changes were made. But it is easily correctable, and probably already has been.
Like I said these cared are best in vertical mount otherwise if you got any pull on the cable it can may be mess up the connection even if its clicked in?
Just use some common sense, don't put tension on the cable, make sure you plugged it in, and it will be fine.
 
One wonders, if PC's had PC speakers in them still-----would these 4090's have been screaming?

With a PC speaker, and no PCIe power connected to a GPU, you normally get some crazy beeping.
 
The design is faulty. Period.

Once somebody dies in a fire, it will be taken seriously.
Will it? You hear of iPhones bursting into flames all the time, we blame it on the users and move on. I googled to find how many house fires are blamed on PCs and not much. I feel like it would melt, smoke, but not much fire.

But yeah hope no one would die or any real fires from a gaming PC, it would be a sad day.
 
One wonders, if PC's had PC speakers in them still-----would these 4090's have been screaming?

With a PC speaker, and no PCIe power connected to a GPU, you normally get some crazy beeping.

I did that once, the GPU had its own beeper, cause the PC didn't have a speaker. Anyway, poor connection isn't going to make the GPU scream, until it melts enough. From experience with melty connections in pinball, it usually takes a lot more melting than that picture for the connection to actually break; that one looks like it's got a ways to go before the connectors desolder themselves or the crimps come out.
 
Just use some common sense, don't put tension on the cable, make sure you plugged it in, and it will be fine.
This. I feel it's common sense (which I know isn't that common in this world sometimes), that if you spend that much on one single piece of hardware you should make sure it's fully plugged in no matter what. I agree it's not the best design, but if you take your time everything will be fine.
 
Like I said it shouldn't be a rocket science with this.
Spend some time looking at the builds people show off on reddit. Half of these guys couldn't put together a NUC if it arrived already built.

Once somebody dies in a fire, it will be taken seriously.
As long as you're using components that actually comply with the law and aren't made of literal garbage, there is zero risk of actual fire. Smoke and [brief] light? Yes. A conflagration? lolno.
 
Spend some time looking at the builds people show off on reddit. Half of these guys couldn't put together a NUC if it arrived already built.


As long as you're using components that actually comply with the law and aren't made of literal garbage, there is zero risk of actual fire. Smoke and [brief] light? Yes. A conflagration? lolno.

I get it man. But all of sudden plugging in a GPU adapter is such a huge mess right? I had a 4090 and my 12vh cable was coming loose on one side like molex would lmao. Its poor craftsman ship all around. I have never had that happen on 8 pin even with severe bend.

All I am saying make the connector more error proof on such an expensive card. Apparently this was not an issue on 3090 ti etc. The size of these cards and the bends that cable get and may be the pull on them I think is getting the connection lose at times. Why I am saying vertical position is best for this monster cards.
 
I mean it’s pretty clear to anyone with an inkling of common sense that stuffing 12 (it’s 12 right?) 16AWG wires into a hilariously tiny plastic connector housing is going to lead to cable/connector stress. What a shit design - probably in the name of make it smol except it turned into smoldering. Hah.

IIUC that Jon Gerow post correctly then the issue is the pins inside the connector shift and cause extra resistance? If so - making the connector housing deeper and maybe have extra guide tabs on the terminator pins would solve the problem.
 
I have Corsair's cable and it's rigid as hell and sticks out more than you'd think. It's a tight fit in my 5000D and it pushes against the glass ever so slightly. Probably not enough to matter, but I'm going to keep my eye on it. The instant other cable options start showing up (especially a right-angle one), I'll be hopping on board. For what it's worth, the 4-cable adapter my card came with was far less rigid and pokes out less than Corsair's 1-cable solution. It's ugly and kind of a pain in the ass, but it doesn't come in contact with my glass panel at all.
 
I have Corsair's cable and it's rigid as hell and sticks out more than you'd think. It's a tight fit in my 5000D and it pushes against the glass ever so slightly. Probably not enough to matter, but I'm going to keep my eye on it. The instant other cable options start showing up (especially a right-angle one), I'll be hopping on board. For what it's worth, the 4-cable adapter my card came with was far less rigid and pokes out less than Corsair's 1-cable solution. It's ugly and kind of a pain in the ass, but it doesn't come in contact with my glass panel at all.

I have the corsair cable, any significant bend the cables pop out the back like a molex connector. Far from rigid unless they started soldering them in or something. That absolutely should not happen on a plug designed to push 600w.
 
This. I feel it's common sense (which I know isn't that common in this world sometimes), that if you spend that much on one single piece of hardware you should make sure it's fully plugged in no matter what. I agree it's not the best design, but if you take your time everything will be fine.
Some of us feel it's common sense (which we know isn't that common) for a company selling GPUs at this price to make sure the design of the power connector doesn't need special treatment and monitoring. You know, basically how it's mostly always been.
 
Why not just solder a pigtail straight to the card from the factory with a connector.
 
Because some people have ATX 3.0 power supplies.

I think the idea was to have a female 12VHPWR connector on a pigtail, which your ATX 3.0 power supply lead could plug in to.

Speaking of...

Why not just solder a pigtail straight to the card from the factory with a connector.

Ever try to plug a 4-pin Molex splitter or extension in? Now picture that, but with 16 pins with beefy conductors fighting you the whole time.

Probably the most cancerous connection I could imagine. I'd rather lose a less-favorite body part.
 
I was thinking a 10ft extension cord and plug it onto a 110 wall outlet, just bypass the PSU, I am sure their next card may need 220 or 221 what ever it takes.
 
Back
Top