Users report melting 16-pin RTX 4090 power adapters, again

Daniel Sims

Daniel Sims

Posts: 919   +35
Staff
Here we go again: Half a year after Nvidia's last official statement regarding damaged RTX 4090 cable adapters, new cases indicate the problem hasn't faded. The blame has been passed between Nvidia, user error, and the specification publisher. Switching to ATX 3.0 power supplies is still the safest bet.

At least three new reports of melting 16-pin adapters for Nvidia's flagship GeForce RTX 4090 graphics card appeared on Reddit this week. The complaints could signal something is still wrong with the connectors six months after the controversy emerged.

The problems started in October when multiple 4090 owners said their power cables had burned or melted. In November, Nvidia acknowledged 50 cases worldwide, saying the users had improperly connected the adapters. However, Igor's Lab questioned the soldering quality of the supplied adapters. One user opened a class-action lawsuit against Team Green, and AMD took the opportunity to take a jab at its competitor in a holiday tweet last year (below).

The RTX 4090 uses 16-pin 12VHPWR connectors and natively supports ATX 3.0 power supplies. Users who don't want to spend extra to upgrade from their ATX 2.0 PSUs can use the bundled adapters on their 8-pin connectors.

One of the new incidents occurred after around six months of use. The user claims to have set the connectors correctly and securely. Similarly, a second user claims his properly-connected adapter went up in smoke about six or seven months after installation. The user primarily used the PC to train AI models rather than play games. A peculiar aspect of the second report is that the affected PC contains a second RTX 4090 that hasn't experienced any problems.

A third report comes from someone using a third-party Corsair SF750 platinum adapter. Assuming it was correctly connected, this instance could suggest that the adapter isn't the problem. Even stranger is that the user limited the GPU to 70 percent power and paired it with a modest Ryzen 7600 CPU, so the PSU certainly wasn't overtaxed.

Reports of 4090 connectors melting or burning in ATX 3.0 supplies haven't appeared. So the best safeguard against the issue is probably to avoid using any adapter and simply upgrade to an ATX 3.0 PSU, which probably wouldn't cost much extra compared to the $1,600 graphics card. Those who can't avoid using the adapters can check the status of their 16-pin connection with a new feature in GPU-Z.

Permalink to story.

https://www.techspot.com/news/98819-users-report-melting-16-pin-rtx-4090-power.html

 
VitalyT

VitalyT

Posts: 6,557   +7,522
Thanks to nVidia, we now can use a PC case to dry clothes. A few more years, their cards' meltdown may come with radiation fallout.
 
  • Like
Reactions: wizardB
godrilla

godrilla

Posts: 771   +425
" Those who can't avoid using the adapters can check the status of their 16-pin connection with a new feature in GPU-Z."
What would be considered abnormal or prerequisite behavior or markers?
I set my parameters to highest 16 pin power and highest 16 pin voltage. After Firing up Vermitide 2 power peaked at 405 watts and voltage at 12.3 volts. Are these normal?
I have a similar setup since launch 10/22 4090 Suprim liquid at stock ( quiet bios ) set power to 95% seen in latest Precision X1 by default. PSU I am using is the 750 watt sfx platinum by Corsair. While not relevant I was using the 9900ks all core oc 5.1 ghz and upgraded to the 7700x at launch with all core oc to 5.65 ghz. I also have only 3 of the 4 the 8 pins cables connected to the PSU side and use an open H210 itx case to minimize cable angle strain. I am using the factory 16 pin adapter.
If anything happens I will definitely post it here.
 
  • Like
Reactions: b1k3rdude
Axeia

Axeia

Posts: 89   +98
Didn't Gamers Nexus already figure this out?
From my memory:
1) The connector is a bit crap making it hard to tell if it's properly seated or not. Not having it seated properly might cause lessened contact surface between the well.. contacts. Less contact surface = more heat
2) Besides it being easy for the connector to not properly be plugged in bending the cable close to the connector (which will happen in basically any case because these cards are ENORMOUS) risks some of the pins not making proper contact as the strain on the cable pulls them away from where they should be. Again, less contact surface = more heat
3) For some reason the cards will still pull full power even if one or two pins aren't connected. More power over fewer pins = more heat
4) Low quality metal with debris in it can cause poor contact = more heat

Basically it's a badly designed connector but it's poor contact one way or another that will make things melt. Plugging it in all the way and avoiding bending the cable close to the connector are steps you can do yourself to prevent things from going wrong. If I owned one I would probably get some 90 or 180 degree angled vhpwr adapters. That way you can greatly reduce the likelihood of cable strain being an issue. If you mess around with the side panel it won't mess with the cable and it being there can serve as a reminder to pay attention to what you're doing whenever you mess around with stuff yourself. Cablemod does both 90 and 180 degree adapters - they're pricy for what they are but hey compared to what they'll be plugged into it's a small price to pay.
 
hahahanoobs

hahahanoobs

Posts: 4,823   +2,769
I'm certainly not dismissing the issue, but if you paid $1600 for a graphics card and are using an adapter is crazy. It's like using graphics drivers on the discs that came with them. It's bare minimum support to make the hardware functional. I don't even have a 40 series card but I still grabbed Corsair's 12VHPWR cable for my RMx PSU for $30CAD so I'm ready in case I need it in the future.
 
DSirius

DSirius

Posts: 624   +1,289
TechSpot Elite
Axeia said:
Didn't Gamers Nexus already figure this out?
From my memory:
1) The connector is a bit crap making it hard to tell if it's properly seated or not. Not having it seated properly might cause lessened contact surface between the well.. contacts. Less contact surface = more heat
2) Besides it being easy for the connector to not properly be plugged in bending the cable close to the connector (which will happen in basically any case because these cards are ENORMOUS) risks some of the pins not making proper contact as the strain on the cable pulls them away from where they should be. Again, less contact surface = more heat
3) For some reason the cards will still pull full power even if one or two pins aren't connected. More power over fewer pins = more heat
4) Low quality metal with debris in it can cause poor contact = more heat

Basically it's a badly designed connector but it's poor contact one way or another that will make things melt. Plugging it in all the way and avoiding bending the cable close to the connector are steps you can do yourself to prevent things from going wrong. If I owned one I would probably get some 90 or 180 degree angled vhpwr adapters. That way you can greatly reduce the likelihood of cable strain being an issue. If you mess around with the side panel it won't mess with the cable and it being there can serve as a reminder to pay attention to what you're doing whenever you mess around with stuff yourself. Cablemod does both 90 and 180 degree adapters - they're pricy for what they are but hey compared to what they'll be plugged into it's a small price to pay.
Click to expand...
Northridgefix proved that, even with a cablemod correct inserted, the power connector on Nvidia card is melting and catching fire, and only on Nvidia card side connector.
 
Last edited:
  • Like
Reactions: daffy duck, Axeia and godrilla
Prosercunus

Prosercunus

Posts: 347   +272
Early on when I read about this issue, I knew when I bought my RTX 4090, I would get a compatible ATX 3.0 PSU. No issues whatsoever, despite the cable being reasonable smooshed up against the tempered glass in my Corsair 5000D.
 
  • Like
Reactions: Burty117
NeoMorpheus

NeoMorpheus

Posts: 1,539   +3,353
Since I dont have neither cate for nvidia gpus, didnt pay much attention to this.

What I do find “funny” is how quiet all the Tubers and media outlets have become on this issue after it was discovered.

Worse, the same places instead have doubled down in their daily efforts in shoving the 4090 down our throats, together with RT/PT and of course, the magical fake frames generator called dlss3.

But boy did they all jumped hard on AMD for the bad batch of coolers in the 7900 xtx and didnt let that one go.
 
  • Like
Reactions: wizardB, fb020997, Vayven and 2 others
DSirius

DSirius

Posts: 624   +1,289
TechSpot Elite
NeoMorpheus said:
Since I dont have neither cate for nvidia gpus, didnt pay much attention to this.

What I do find “funny” is how quiet all the Tubers and media outlets have become on this issue after it was discovered.

Worse, the same places instead have doubled down in their daily efforts in shoving the 4090 down our throats, together with RT/PT and of course, the magical fake frames generator called dlss3.

But boy did they all jumped hard on AMD for the bad batch of coolers in the 7900 xtx and didnt let that one go.
Click to expand...
Gone are days with double standards as a norm. Now we witness the rise of "Triple Standards" from corporations like Nvidia and how they deal with public in general, their customers, and reviewers (Youtubers).
 
  • Like
Reactions: wizardB, fb020997, Kam7r and 1 other person
takaozo

takaozo

Posts: 660   +1,056
No matter how hard you try to beat the laws of physics they will get back to you. Tech Jesus only made it worse.
Small contact surface and pressure = more resistance and that = more heat. More heat = metal expansion and that = less contact pressure. And since the connector on PCB has enough copper to dissipate the heat only the external connector will heat up big time.

You don't have to be a engineer in electronics to figure this out.
 
Last edited:
J

Jack77

Posts: 23   +23
Personally I think that over time more and more users will experience issues. As there is no locking mechanism, the cable can get loose when working on the computer.
 
  • Like
Reactions: wizardB
K

KofeViR

Posts: 295   +135
Jack77 said:
Personally I think that over time more and more users will experience issues. As there is no locking mechanism, the cable can get loose when working on the computer.
Click to expand...
What are you talking about. There is a locking mechanism on 12VHPWR. My cable clearly clicks and then don't move at all.
 
  • Like
Reactions: godrilla
Burty117

Burty117

Posts: 4,772   +3,207
I used the adaptor that came with my 4090 for 4 months, no issues at all. Then I waterblocked the GPU and put it into a new PC, this time with an ATX 3.0 PSU and still not had a single issue.

I know two more people with 4090's, one of them has also waterblocked the card, neither of them have had any issues or would call the connector a problem.

I'm not saying this isn't an issue but everyone I know with a 4090 don't have any issues and haven't even mentioned it to me, until I specifically asked if they had issues with it.
 
K

KofeViR

Posts: 295   +135
Burty117 said:
I used the adaptor that came with my 4090 for 4 months, no issues at all. Then I waterblocked the GPU and put it into a new PC, this time with an ATX 3.0 PSU and still not had a single issue.

I know two more people with 4090's, one of them has also waterblocked the card, neither of them have had any issues or would call the connector a problem.

I'm not saying this isn't an issue but everyone I know with a 4090 don't have any issues and haven't even mentioned it to me, until I specifically asked if they had issues with it.
Click to expand...
It's 100% user fault - it only happends if cable is not fully and properly inserted.
People might as well get used to this cable, because it's the new standard for GPUs.

AMD better change soon, because many new ATX 3.0 PSUs don't even have regular 8 pins for GPU anymore. Only 12VHPWR.
 
SiJiL

SiJiL

Posts: 13   +9
If the issue is with the cable connector that inserts into the card, how would using an ATX 3.0 PSU solve the problem? You still have the same connector going into the graphics card - the only difference is the end that goes into the PSU. Surely the same amount of power is going through the cable? Or does using an ATX 3.0 PSU come with some additional features that prevent the overheating.
 
K

KofeViR

Posts: 295   +135
SiJiL said:
If the issue is with the cable connector that inserts into the card, how would using an ATX 3.0 PSU solve the problem? You still have the same connector going into the graphics card - the only difference is the end that goes into the PSU. Surely the same amount of power is going through the cable? Or does using an ATX 3.0 PSU come with some additional features that prevent the overheating.
Click to expand...
It might, if user don't make sure it's in. However, this is a non-issue for people that actually knows anything about hardware. Easier than before. Both my 12VHPWR adapter and native cable makes a loud and obvious click. Way less cable management with the new connector and the sense function makes sure the GPU don't pull more power than PSU can deliver.

Anyone that think the old solution is better, has zero experience with the new one. Because it's better for sure; 600 watt sustained with 1800 watt spikes allowed + Sensing so GPU don't pull more than PSU can actually deliver + Less cables inside the case + Cleaner look.

AMD will be forced to use 12VHPWR on Radeon 8000 series. They should have done it on 7000 series as well IMO. Instead they made it an actual selling point, which is laughable really - because the new solution is better and can replace 3-4 x 8 pin. Many 7900XTX use 3 x 8pin.
 
Last edited:
neeyik

neeyik

Posts: 2,740   +3,331
Staff member
godrilla said:
What would be considered abnormal or prerequisite behavior or markers?
I set my parameters to highest 16 pin power and highest 16 pin voltage. After Firing up Vermitide 2 power peaked at 405 watts and voltage at 12.3 volts. Are these normal?
Click to expand...
Those figures are normal. What you want to look for is the voltage consistently dropping below 12 volts. If the sensor is constantly reporting, say, 11.8 volts at full load, then there's something amiss.
 
  • Like
Reactions: godrilla and takaozo
Geralt

Geralt

Posts: 1,356   +2,183
To pay so much money to have this serious problem, no way. That happens when you go crazy about power consumption. It's full Nvidia's fault despite its fanboys try to hide it. No new high-end GPU for me this generation.
 
  • Like
Reactions: wizardB and yRaz
Axeia

Axeia

Posts: 89   +98
DSirius said:
Northridgefix proved that, even with a cablemod correct inserted, the power connector on Nvidia card is melting and catching fire, and only on Nvidia card side connector.
Click to expand...
As luck would have it YouTube suggested exactly this video to me today.
Yeah even those adapters aren't a sure fix but they should reduce the risk at least, the connector just seems flawed to begin with though and NVIDIA should stop using it and whoever is in charge of coming up with the standard needs to go back to the drawing board.
I really don't want this connector in my PC, hell - I see little use for it anyhow as I don't want a card that needs so much power that it needs one of these connectors in the first place.
 
  • Like
Reactions: wizardB
b1k3rdude

b1k3rdude

Posts: 12   +14
Axeia said:
Didn't Gamers Nexus already figure this out?
From my memory:
1) The connector is a bit crap making it hard to tell if it's properly seated or not. Not having it seated properly might cause lessened contact surface between the well.. contacts. Less contact surface = more heat
2) Besides it being easy for the connector to not properly be plugged in bending the cable close to the connector (which will happen in basically any case because these cards are ENORMOUS) risks some of the pins not making proper contact as the strain on the cable pulls them away from where they should be. Again, less contact surface = more heat
3) For some reason the cards will still pull full power even if one or two pins aren't connected. More power over fewer pins = more heat
4) Low quality metal with debris in it can cause poor contact = more heat

Basically it's a badly designed connector but it's poor contact one way or another that will make things melt. Plugging it in all the way and avoiding bending the cable close to the connector are steps you can do yourself to prevent things from going wrong. If I owned one I would probably get some 90 or 180 degree angled vhpwr adapters. That way you can greatly reduce the likelihood of cable strain being an issue. If you mess around with the side panel it won't mess with the cable and it being there can serve as a reminder to pay attention to what you're doing whenever you mess around with stuff yourself. Cablemod does both 90 and 180 degree adapters - they're pricy for what they are but hey compared to what they'll be plugged into it's a small price to pay.
Click to expand...
Intel upgraded the spec of the female connector pins to prevent the issue, but most if not all of the adapters on the market atm are using the old/original spec.
 
  • Like
Reactions: DSirius
B

Belgeek

KofeViR said:
It's 100% user fault - it only happends if cable is not fully and properly inserted.
People might as well get used to this cable, because it's the new standard for GPUs.

AMD better change soon, because many new ATX 3.0 PSUs don't even have regular 8 pins for GPU anymore. Only 12VHPWR.
Click to expand...
All serious brands provide both. My dark power 13 got them.
 
  • Like
Reactions: wizardB
You must log in or register to reply here.

Similar threads

midian182
Nvidia is giving away RTX 4060, 4060 Ti, 4080, and 4090 cards this summer
Replies
9
Views
137
pcnthuziast
pcnthuziast
N
Enthusiast proves AMD's RX 7900 XTX can reach RTX 4090-like performance levels
2
Replies
37
Views
384
hwertz
hwertz
Shawn Knight
Asus announces a pair of GeForce RTX 4090 cards featuring compact designs
Replies
14
Views
209
loki1944
L

Latest posts