Wednesday, January 22nd 2025

NVIDIA Claims 16-Pin Power Connector Issues are Over, No More Melting

During a recent press event in South Korea, NVIDIA addressed concerns about power connector safety for their upcoming RTX 5090 graphics card. The new GPU will consume 575 watts of power, marking a massive 225-watt increase from its predecessor, the RTX 4090. The previous generation RTX 4090 faced significant issues with melting 12VHPWR power connectors, especially with third-party adapters, where incomplete connections led to overheating and connector damage. When questioned about potential risks with the RTX 5090's higher power draw, NVIDIA representatives stated they've implemented an updated 12V-2×6 power connector across the RTX 50 series. Unlike the 12VHPWR 16-pin connector, the new 12V-2x6 has sense pins having recessed further back to ensure proper contact before the GPU can request higher power outputs.

"It is expected that such issues will not occur with the RTX 50 series," a company representative explained during the Q&A session. "After about two years, we believe these problems have been resolved." While the company maintains that user error was the leading cause of failures, the extended timeframe required for developing and shipping revised connectors raised questions about the initial design's reliability. Despite NVIDIA's assurances, the RTX 5090's exceptional power requirements could potentially amplify any unexpected technical issues. The company's previous experience showed that problems became more pronounced in cards with higher power demands, as demonstrated by the RTX 4090 having more incidents than the lower-powered RTX 4080 series. So, more power means more trouble, but the company has worked on it to ensure no future problems arise.
Sources: QuasarZone, via VideoCardz
Add your own comment

65 Comments on NVIDIA Claims 16-Pin Power Connector Issues are Over, No More Melting

#26
Dammeron
john_Nvidia used Gamers Nexus' credibility to silence everyone that was saying that there was a problem with the connector and unfortunately Steve was insisting on the user error story for way too long after that initial video. Now Nvidia says that the user error problem wasn't a user error problem but something that was fixed with a new revision of the connector.
Hecate91Gamers Nexus went with the user error claim since it worked in a test bench environment, but he made a video later on saying the 12v connector is a dumpster fire and wasn't fool proof enough
^THIS

It wasn't that the connector was faulty and caused fires - the failure rate is still few % at most - there are plenty of cards and cables running properly without any problems. The problem is that the connector can become unsafe with a moderate ease by the user.

What nV claims now is not that they fixed the defect, but that they made it more fool proof. Whether it's true or not... we shall see.


As for GN - his jokes may not be funny and videos are sometimes on the verge of being boring, but he is pretty throughrough with his testing and 'investigations'.
Posted on Reply
#27
Readlight
You need solder on larger wires and connector.
Posted on Reply
#28
Random_User
freeagentIts fine. And its better than looking at 3x 8 pins hanging out of the side of your card.

Imagine if there was 4? Man that would look even goofier.
Cover the side panel/glass, and problem solved. People supposed to look at the screen, not onto the connectors.
Besides, if they work without issues, there's no problem. At least the 8-pin ones do not have the required bending angle issue.
Posted on Reply
#29
freeagent
Random_UserCover the side panel/glass, and problem solved.
Who spends thousands to do that?
Posted on Reply
#30
Tomorrow
freeagentWho spends thousands to do that?
Never seen industrial builds?

I myself run thousands worth of hardware with a solid panel. The case is not even on my desk where i could look at it.
Oh and there zero RGB in it or in any of my peripherals. Just the way i like it.
Posted on Reply
#31
Prima.Vera
Battler624450W to 575W is 225W yes.
125... Yeah, I know..
Posted on Reply
#32
john_
TomorrowOh and there zero RGB in it or in any of my peripherals. Just the way i like it.
Big mistake

Posted on Reply
#33
Crackong
DammeronWhen GN tried to recreate the connector melting they even cut most of the cables so that the load goes only through 1-2 of them and nothing happened. 20-30% more power going through it won't change anything - it's all about removing the risk of contact points barely touching each other, creating high resistance.
While GN also told us the connector design is a dumpstar fire, without audible clicks it is wide open for installation error, and without a physical lock( it is just a friction lock ) the connector could walk itself off just by the force from doing cable management.
Hecate91its interesting Nvidia blames it on user error even though they quickly revised the connector.
true.

If it ain't broke , don't fix it.
If they fixed it, it was broken.

Every revision costs them millions of dollars, they won't fix it for no reason.
Posted on Reply
#34
x4it3n
Let's hope so because when I bought my 4090, I was definitely checking the cable pretty often... Even now I still check from time to time just to make sure!
I bought a Seasonic Prime TX-1600 Noctua Edition (ATX 3.1 / PCIe 5.1) and am waiting for the 5090 for that and DisplayPort 2.1 (UHBR20) so I can use DSR/DLDSR on my 4K@240Hz QD-OLED monitor.
Posted on Reply
#35
trsttte
Here I though there were never any power connector issues :rolleyes:

As broken as the class actions system is, the statements from this press event should be exhibit A on one against nvidia for both selling this and pushing partners to do the same.
Posted on Reply
#36
Rightness_1
There are many repair shops that do not share nv's opinion on this not happening for 2 years now, or shall we call it for what it is... an outright lie.
Posted on Reply
#37
user556
The user error argument is bullshit and a cover up for the shitty 12VHPWR plugs on the cables.
Posted on Reply
#38
DudeBeFishing
Macro DeviceSimply switching good ol' 8-pin to mandatory AWG14 would've fixed this "issue." (by "issue" I mean having more than one power cable)
12AWG wire and a single XT60 connector should be enough, although voltage drop might be an issue. But then you'll have stupid users plugging in lithium batteries to power their GPUs.
Posted on Reply
#39
Macro Device
DudeBeFishingstupid
Hey, it's 2025, watch your tongue! They ain't stupid, they are creative!
Posted on Reply
#40
Hecate91
Dammeron^THIS

It wasn't that the connector was faulty and caused fires - the failure rate is still few % at most - there are plenty of cards and cables running properly without any problems. The problem is that the connector can become unsafe with a moderate ease by the user.

What nV claims now is not that they fixed the defect, but that they made it more fool proof. Whether it's true or not... we shall see.


As for GN - his jokes may not be funny and videos are sometimes on the verge of being boring, but he is pretty throughrough with his testing and 'investigations'.
Exactly , there isn't enough of a safety margin, even if the failure rate is 1-2% thats still too high for what is supposed to be a better connector. There are several tech repair youtubers I know of still repairing melted 4090 connectors, NorthridgeFix claims to get in 4090's with melted connectors every day. And I doubt people with the money to throw around buying $2k graphics cards are too dumb to plug them in correctly.

I'm not saying I don't like GN at all though, GN is one of my favorite tech youtubers, he doesn't hold back on companies and is willing to take a loss in the interest of helping the tech community.
Posted on Reply
#41
DudeBeFishing
Macro DeviceHey, it's 2025, watch your tongue! They ain't stupid, they are creative!
Green is not a creative color.
Posted on Reply
#42
trsttte
DudeBeFishing12AWG wire and a single XT60 connector should be enough
I think you're joking but why not, be it XT60 or any of the dozens of established high power connectors why did they need to go with this shitty one!? :wtf:

This will never make sense to me
Posted on Reply
#43
DudeBeFishing
trsttteI think you're joking but why not, be it XT60 or any of the dozens of established high power connectors why did they need to go with this shitty one!? :wtf:

This will never make sense to me
I'm semi joking. It's smaller than the 8 pin and doesn't need precise soldering. They will need 12 AWG fine stranded wire, otherwise you'd run into wire bending issues, and maybe stress on the PCB. There's plenty of other high current 2 pin connectors, even ones that lock with sense pins. I think we can all agree that moving to a connector with less surface area while increasing GPU power draw, was a terrible idea, and that there were way better options available to condense the power connectors.
Posted on Reply
#44
chrcoluk
freeagentYes, because it is really hard to plug something in. Difficulty. /s
Some cables are to be fair. They can be made too tight, and with poor click mechanics.

The 4000 series connector and cable design was just idiotic, they literally could not have made it worse, lets put it all together.

New connector and cable that requires cable to not be bent for first 3cm.
Combined with, taller GPU that sits much nearer to edge of case meaning the majority of cases dont have 3cm clearance.
Combined with, right angled connector sticking straight out, even more of a shocker, is it was angled the previous gen, someone went out their way to make it right angled for 4000 series.
Combined with a stiff cable.

Its as if they only tested it on a bench or something.

Of course we then had the original design as well that used sense pins which were allowed to make contact with partially inserted cables.

The 5000 series after the bad PR from the issues, clearly had a thought process behind the design, but there is no excuse for how bad they got it wrong on the 4000 series.
Posted on Reply
#45
_roman_
Warning here comes a troll post.

NVIDIA Claims 16-Pin Power Connector Issues are Over, No More Melting

There are no issues with it. Not worth discussing. There are only user errors. It's the buyers fault. It's the users fault. - End of troll post - I am sorry.
Posted on Reply
#46
Caring1
AirisomGN concluded it was user error, quickly washed their hands of it, and NVIDIA ran with it so they didn't have to honor repairs under warranty. Pretty sus move from GN imo. NorthRidgeFix deemed it a design issue with board power delivery and the connector itself. I just find it hard to believe that all of these people aren't insertting the connector properly after all of the buzz surrounding it and the high cost of these 4090s. But we'll see with the 5090 if they fixed it.
Why don't you just say you don't understand how the user error occurred.
freeagentYes, because it is really hard to plug something in. Difficulty. /s
Apparently it is from the number of connectors I've seen not seated correctly, or the cables pulled that tight, they place undue strain on those connectors.
Posted on Reply
#47
trsttte
DudeBeFishingI think we can all agree that moving to a connector with less surface area while increasing GPU power draw, was a terrible idea, and that there were way better options available to condense the power connectors.
Not only better options, but also very easy to find, they could have just used a 12 pin minifit instead of microfit. Even if they were already thinking of placing the connector vertically like it is on the 5090, a 12pin minifit would still be small enough to fit like that (it's around ~26.4mm, it's almost small enough to fit a 20.32mm single slot card!)
Posted on Reply
#48
Airisom
Caring1Why don't you just say you don't understand how the user error occurred.
If the connectors are melting from an engineering failure, then it isn't the user's fault. If the engineers don't allow a sufficient margin of error or don't have sufficient protections in place to ensure reliability, then it is an engineering failure, not user error.

I think we can all agree that the 12VHPWR connector is a failed design. The 12V-2x6 has been used for a while now, and I haven't heard of any reports of those melting, so that indicates that they fixed the shortcomings of the connector. All that's really left is confirming that this connector can handle a 5090 in a variety of builds.
Posted on Reply
#49
wNotyarD
AirisomAll that's really left is confirming that this connector can handle a 5090 in a variety of builds.
Especially as only the FE has the angled connector on the board so far, everyone else is putting the cable perpendicular against the case side.
Posted on Reply
#50
freeagent
12V2X6 seems good, and is what I am referring to when 12VHPWR is used.. my fault there..

But this is still a thing..


Posted on Reply
Add your own comment
Feb 22nd, 2025 12:57 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts