Nvidia RTX 4090 power connectors melting?

Darunion said:
Power strips still catch fire all the time. We still kind of accept some level of catastrophic failure if people misuse things. We need to stop thinking in binary. Blame can exist in both the connector and the users court. Same kind of people plug things in an outlet and then shove a dresser against the plug bending it half out of the socket and slowly tearing the cord out of the strain relief. Also nvidias CM should have been made to do more torture and abuse testing. My opinions.
Click to expand...
Is everyone forgetting just how much Steve, Igor and others fucked with the connector in an attempt to get it to fail, and not a single one of them did? It wasn't until they figured out that end users were not seating it correctly that they successfully got a connector to melt. I don't think anyone could have predicted that end users were just not going to actually seat the connector all the way.
Brahmzy said:
So think about how dumb the average person is. Imagine HALF of all people are dumber than that.
Click to expand...
Regretfully, it may actually be much more than half.
 
I recently installed a Fasgear 600 watts cable. As I was unable to close the case prior. Some pics.

Seems to work fine. Let’s hope my card doesn’t shit itself.
 

Attachments

  • IMG_5968.jpeg
    IMG_5968.jpeg
    629.1 KB · Views: 0
  • IMG_5969.jpeg
    IMG_5969.jpeg
    570.5 KB · Views: 0
  • IMG_5972.jpeg
    IMG_5972.jpeg
    554 KB · Views: 0
  • IMG_5975.jpeg
    IMG_5975.jpeg
    349.1 KB · Views: 0
  • IMG_5967.jpeg
    IMG_5967.jpeg
    375.8 KB · Views: 1
Lakados said:
They’re an RMA repair centre.. they would only get them in bulk.
Nvidia probably finally had enough of them to bother sending them off and they are worth enough right now to bother paying to have them repaired.
Click to expand...
They now officially work for nvidia or one of the AIBs?
 
I couldn't imagine sending a card in for repair and having some goober fix it like that and sent back to me. Surprised he didn't microwave one.
 
Furious_Styles said:
You're joking right? Guy is an ace at tech repair.
Click to expand...
I wasn't joking. Just went on the video, haven't heard of him personally. I seriously thought it was posted as a funny video to be honest. Maybe it would have come off more serious if i watched it with sound.
 
ZeroBarrier said:
Is everyone forgetting just how much Steve, Igor and others fucked with the connector in an attempt to get it to fail, and not a single one of them did? It wasn't until they figured out that end users were not seating it correctly that they successfully got a connector to melt. I don't think anyone could have predicted that end users were just not going to actually seat the connector all the way.

Regretfully, it may actually be much more than half.
Click to expand...

Der8aur did a much better video on the topic. The problems with the connector go beyond just the seating of it (power capacity being another major issue), and if you watch it, you'll see that he didn't have a hard time unseating the connector. He also pointed out the recommended bending angle for the connector is exceeded in most standard cases.

In any case, the bottom line is this connector sucks. Period. If you have this many "user problems" with your product, I'm past the point of blaming the user. It's a design problem.

I think it's a good idea to create a new connector for higher current carrying capacity given the power requirements of modern high-end GPUs because I want to reduce my cabling where I can, but this isn't a good design.
 
Mr Evil said:
Ha! 4090's failing "in bulk" sounds even worse than merely 18 of them.
Click to expand...
Well when you phrase it like that, but more like they failed over a course of a year and sat in a warehouse for months until there was enough of them to bother with dealing with them.
 
I got one of the JoyJom adapters seeing as it appeared identical to the ModDIY one and was much cheaper at the time. I was a little nervous based on all the hullabaloo around the new connector, but I've had no issues. Just made sure I seated it properly. Personally I much prefer this new connector to the old bulkier style.
 
Lakados said:
Well when you phrase it like that, but more like they failed over a course of a year and sat in a warehouse for months until there was enough of them to bother with dealing with them.
Click to expand...

Or it was just a weeks worth, were both just totally guessing.
 
HockeyJon said:
Der8aur did a much better video on the topic. The problems with the connector go beyond just the seating of it (power capacity being another major issue), and if you watch it, you'll see that he didn't have a hard time unseating the connector. He also pointed out the recommended bending angle for the connector is exceeded in most standard cases.

In any case, the bottom line is this connector sucks. Period. If you have this many "user problems" with your product, I'm past the point of blaming the user. It's a design problem.

I think it's a good idea to create a new connector for higher current carrying capacity given the power requirements of modern high-end GPUs because I want to reduce my cabling where I can, but this isn't a good design.
Click to expand...
It's a great connector for the wrong market, put that in a server where it will never get touched and most cables are run at straight shots with clean lines in large flat areas, no problem.
A consumer case, where that shit is bent and twisted at all sorts of tight angles in an attempt to hide them from plain sight, nah, that is an accident waiting to happen. Hope the 6x2 connector fixes these issues, but I wouldn't be surprised if that itself also gets replaced come the 5000 series.
 
Lakados said:
It's a great connector for the wrong market, put that in a server where it will never get touched and most cables are run at straight shots with clean lines in large flat areas, no problem.
A consumer case, where that shit is bent and twisted at all sorts of tight angles in an attempt to hide them from plain sight, nah, that is an accident waiting to happen. Hope the 6x2 connector fixes these issues, but I wouldn't be surprised if that itself also gets replaced come the 5000 series.
Click to expand...

The problems go beyond just the mechanical characteristics of the connector, though. I think Der8auer did a great breakdown of the issues with it.

I think it’s good to move forward with a one cable solution given the trend in higher end GPUs needing more power, but this isn’t it as far as I can tell. Hopefully it will evolve to be better in the future.
 
Mr Evil said:
Yeah. And it makes sense if you think about it - 4090's specifically have this problem because 4090 buyers are bigger idiots than buyers of other cards, right?
Click to expand...
Cute, but here are some melted PCIE 8 pic connectors on Radeon cards that melted as well for your viewing pleasure.

Radeon Rx295x2_________________________________________________________Radeon RX580__________________________________________________________Radeon 7900XTX
1707393185931.jpeg
1707392953937.jpeg
1707393039170.jpeg


Pretty moronic to try and make this argument a thinly veiled fanboi attack. This issue is neither new nor has it ever been exclusive to one brand. It is a universal issue of human idiocy.
 
Last edited:
ZeroBarrier said:
Cute, but here are some melted PCIE 8 pic connectors on Radeon cards that melted as well for your viewing pleasure.

Radeon Rx295x2_________________________________________________________Radeon RX580__________________________________________________________Radeon 7900XTX
View attachment 633470View attachment 633468View attachment 633469

Pretty moronic to try and make this argument a thinly veiled fanboi attack. This issue is neither new nor has it ever been exclusive to one brand. It is a universal issue of human idiocy.
Click to expand...
I believe you are the only fanboi in this thread that can't admit that the connector that nVidia used for the RTX4090 is not adequate for the amount of power the card is pulling through the connector. And I am an nVidia fanboi but I can' admit that the make mistakes just like everyone else does.
 
ZeroBarrier said:
Cute, but here are some melted PCIE 8 pic connectors on Radeon cards that melted as well for your viewing pleasure.

Radeon Rx295x2_________________________________________________________Radeon RX580__________________________________________________________Radeon 7900XTX
View attachment 633470View attachment 633468View attachment 633469

Pretty moronic to try and make this argument a thinly veiled fanboi attack? This issue is neither new nor has it ever been exclusive to one brand. It a universal issue of human idiocy.
Click to expand...

Nah, things do fail, but the 4090 is killing this poorly designed connecter because it's pushing the connecter to it's limit. Taking the same pictures and reposting them every time does not make your statement true that it's all user error. Seems like it's just the 4090 cards that have a issue, which points not to user error but most likely pushing beyond the capabilities of the connecter and should have been caught in testing. Thus why they have abandoned this original design and went to a new connecter in the same generation of video card.

Anyone defending this designed power connecter is either a investor or a fool, let me know which you are.
 
Armenius said:
The 4090 uses less than 450W on a connector designed for 600W. That is less than 6.25A per hot 16AWG wire that can handle up to 10A at 12V each.
Click to expand...

If you don't overclock then yeah most likely, but it does have transient spikes as well. Reality is, I would never get this connecter anywhere near 600 watts, it was not designed robust enough to survive that kind of use. Also the wire gauge is not the issue, it's the small pin contact at the connecter that is driving resistance up because it's trying to flow more amperage than it can handle. We had many issues like this in the auto industry as engineers tried to cut wiring and connecter down to the bare minimum. Lets just say it didn't work out like they hoped. Blower motors were famous for melting the connecters or wiring due to running to close to the theoretical max of the wire and connecter.
 
Update on cablemod recall.

https://www.reddit.com/r/cablemod/comments/1alwbwn/voluntary_safety_recall_of_cablemod_12vhpwr/

Dear Hardware Community,

Thank you for your patience while we worked through the details of our voluntary recall of CableMod 12VHPWR Angled Adapters V1.0 and V1.1. We are voluntarily recalling all of our angled adapters to keep our customers and their hardware safe. We have found that the male connectors could become loose during cable management and system use, overheat, and melt into the GPU, posing fire and burn risks. We encourage all of our customers to participate.

This recall applies only to our angled adapters and does not affect our angled cables.

If you haven’t already, stop using your angled adapter immediately. Please consult the recall page on our website for complete instructions on how to safely remove the angled adapter from your GPU: https://cablemod.com/adapterrecall. If your GPU is damaged because of a failed adapter, we will repair it if your damage claim is submitted to us by February 22, 2024. CableMod will no longer reimburse the purchase price of a GPU that may have been damaged by a failed angled adapter.

What We’re Doing for Our Customers:

As part of this recall, CableMod is offering its angled adapter customers the choice of either a full refund or a $60 / 60€ store credit to be used towards any non-customized product in CableMod’s webstore, with free standard shipping. You will receive a full refund or store credit for each angled adapter that you purchased.

If you participated in our Early Adopter Program, CableMod is also offering a choice of either a refund or store credit in the amount of your shipping cost. You will not receive a refund or store credit for the V1.1 angled adapter that you received for free.

Please note that only one type of remedy may be selected: either all refund or all store credit. You must have a PayPal account if you elect to receive a refund.

We anticipate that the volume of claims will be very high at the beginning of this recall, which may affect processing times. Since we are a small company, please allow up to 6-8 weeks for processing. Verifying claims based on purchases made from our authorized online retailers may require additional time. We’ll do our best to process your claims as quickly as possible.

More Information:

The recall page on our website contains complete instructions on how to obtain your CableMod refund or store credit. It also provides answers to frequently asked questions about this recall. https://cablemod.com/adapterrecall/ or https://cablemod.com and click on the red banner at the top of the page. If you need additional help, please call us toll free at 888-203-3605 from 9 a.m. to 5 p.m. ET Monday through Friday or reach out to customer service at https://www.cablemod.com/support/. We’ll help you as quickly as we can.

We apologize for inconveniencing our loyal customers. We hope that this is a start to making it up to you.

Your Friends at CableMod

*This recall is being conducted voluntarily in cooperation with the U.S. Consumer Product Safety Commission under its Fast Track Recall program.
Click to expand...
 
Gideon said:
If you don't overclock then yeah most likely, but it does have transient spikes as well. Reality is, I would never get this connecter anywhere near 600 watts, it was not designed robust enough to survive that kind of use. Also the wire gauge is not the issue, it's the small pin contact at the connecter that is driving resistance up because it's trying to flow more amperage than it can handle. We had many issues like this in the auto industry as engineers tried to cut wiring and connecter down to the bare minimum. Lets just say it didn't work out like they hoped. Blower motors were famous for melting the connecters or wiring due to running to close to the theoretical max of the wire and connecter.
Click to expand...

Transient spikes don't matter as the primary consideration behind what a cable/connector can handle is how much heat it can passively dissipate. The spikes are not high enough for long enough to create a significant amount of heat, which is why only average amperage matters.

In the new power connector, the pins were lengthened. This increases the contact surface area, reducing the resistance and increasing the amount of heat that can be dissipated. The tests that I have seen shows the new connector getting nowhere near as hot, even in a full power not fully seated scenario.
 
Gideon said:
If you don't overclock then yeah most likely, but it does have transient spikes as well. Reality is, I would never get this connecter anywhere near 600 watts, it was not designed robust enough to survive that kind of use. Also the wire gauge is not the issue, it's the small pin contact at the connecter that is driving resistance up because it's trying to flow more amperage than it can handle. We had many issues like this in the auto industry as engineers tried to cut wiring and connecter down to the bare minimum. Lets just say it didn't work out like they hoped. Blower motors were famous for melting the connecters or wiring due to running to close to the theoretical max of the wire and connecter.
Click to expand...
The wiring harnesses may be ugly as sin and bulky as hell but when they work they work. I've used them for custom lecterns because they are clear and hard to mess up. Take 9 different cables and easily rewire them into one large harness that only fits one way and makes a clear *clunk noise when properly seated. Beautiful.
 
Tsumi said:
Transient spikes don't matter as the primary consideration behind what a cable/connector can handle is how much heat it can passively dissipate. The spikes are not high enough for long enough to create a significant amount of heat, which is why only average amperage matters.

In the new power connector, the pins were lengthened. This increases the contact surface area, reducing the resistance and increasing the amount of heat that can be dissipated. The tests that I have seen shows the new connector getting nowhere near as hot, even in a full power not fully seated scenario.
Click to expand...

Yep the issue is the amount of pin contact and why were seeing the failure right at the connecter. Glad to hear the new connecter is not getting hot and they increased the pin contact area.
 
ZeroBarrier said:
Cute, but here are some melted PCIE 8 pic connectors on Radeon cards that melted as well for your viewing pleasure.

Radeon Rx295x2_________________________________________________________Radeon RX580__________________________________________________________Radeon 7900XTX
View attachment 633470View attachment 633468View attachment 633469

Pretty moronic to try and make this argument a thinly veiled fanboi attack. This issue is neither new nor has it ever been exclusive to one brand. It is a universal issue of human idiocy.
Click to expand...
B-B-but that isn't possible, Ngreedia bad! /s Seriously laughing at trying to make idiocy a fanboy thing rather than just idiocy. It's also a rare defect in either brand at times.
 
Tsumi said:
This increases the contact surface area, reducing the resistance and increasing the amount of heat that can be dissipated. The tests that I have seen shows the new connector getting nowhere near as hot, even in a full power not fully seated scenario.
Click to expand...

Have any links to those tests? Curious to see the difference.
 
What is amazing to me are people citing the general idiocy of the average customer as justification for what is clearly bad cable design likely with cheap parts and high power draw. Steve at GN putting this on the connector not being fully plugged in was just silly. Toss in some of the issues some of the 3090 cards had with transient power spikes and bad power supplies? Just crazy. Seasonic of all PSU companies telling customers to use a hair dryer to warm up the cable before bending at certain angles? You want to give the general idiot public that responsibility? I don't want to deal with that either way especially not on a $1k+ product.

It is bad design and corporate profit margins driving it.
 
Faulty 12V2x6 connector found with RTX 4080 Super card — it could be a one-off problem, but caution is advisable

Igor's Lab reports that at least one 12V2x6 adapter he received with an RTX 4080 Super third-party card was unable to lock in place, which could lead to serious power delivery issues with their host graphics cards...thankfully, the issue is specifically related to this particular adapter and not the connectors in general...however, it's a potentially serious problem for gamers and professionals who need to run the adapter due to compatibility restraints...

https://www.tomshardware.com/pc-com...be-a-one-off-problem-but-caution-is-advisable
 
polonyc2 said:
Faulty 12V2x6 connector found with RTX 4080 Super card — it could be a one-off problem, but caution is advisable

Igor's Lab reports that at least one 12V2x6 adapter he received with an RTX 4080 Super third-party card was unable to lock in place, which could lead to serious power delivery issues with their host graphics cards...thankfully, the issue is specifically related to this particular adapter and not the connectors in general...however, it's a potentially serious problem for gamers and professionals who need to run the adapter due to compatibility restraints...

https://www.tomshardware.com/pc-com...be-a-one-off-problem-but-caution-is-advisable
Click to expand...
Literally just posted here:
Lakados said:
Just gonna throw this one in here... Because new drama is good drama.
https://www.igorslab.de/en/nvidia-a...ibution-a-chance-find-or-a-one-off-slip-up/3/
Click to expand...
polonyc2 said:
do the new Super cards come with the new 12V-2x6 connector?...or does this overheating issue only apply to 4090 cards?
Click to expand...
Any 12VHPWR connector can overheat with a high enough power draw and improper seating. By now most new cards that use the 12x6+4 connector should be shipping with the 12V-2x6 connector rather than the 12VHPWR, especially nVidia ones. It is possible some AIBs would try to use up their existing stock of 12VHPWR connectors before switching to the 12V-2x6 connector.
 
You must log in or register to reply here.
Back
Top