Nvidia 40-Series power connectors, melting, and burning.

[Chiyawa]

1 hour ago, Shzzit said:

These things just didint burn up for no reason . I’m willing to bet 100% user error .  Probably wrenched the shit out of it to fit in a case.

Meh, you never seen how many slots burn out yet. I even seen a SATA power connector got melted so bad, and it was from HDD back plane of a server. The server was sitting there working perfectly fine for close to 10 years, then one day it decided to went with magic smoke (HDD might be the culprit here).

 

Yeah, human error might cause this issue, but it is not always the case.

 

And oh, if you say these things doesn't burn out for no reason, then may I remind you of Samsung Galaxy phone that combust into flame even though they are still seal in the box? Is it the user error then? If only 1 or 2, then maybe it's user error or the product just got 'unlucky', but if it is a wide spread stuff, then it is manufacturing error or designing error, which for this 12VHPWR case it is probably design error.

[Uttamattamakin]

21 minutes ago, Chiyawa said:

Meh, you never seen how many slots burn out yet. I even seen a SATA power connector got melted so bad, and it was from HDD back plane of a server. The server was sitting there working perfectly fine for close to 10 years, then one day it decided to went with magic smoke (HDD might be the culprit here).

 

Yeah, human error might cause this issue, but it is not always the case.

 

And oh, if you say these things doesn't burn out for no reason, then may I remind you of Samsung Galaxy phone that combust into flame even though they are still seal in the box? Is it the user error then? If only 1 or 2, then maybe it's user error or the product just got 'unlucky', but if it is a wide spread stuff, then it is manufacturing error or designing error, which for this 12VHPWR case it is probably design error.

What you describe would be ... normal.  Given so many years and so many connectors this would happen randomly from time to time.  Nothing is perfect. 

 

This connector has been out for a matter of weeks, only on a relatively tiny number of cards, and now it's having this issue.  This is a issue which would get a lot of products recalled. 

[Lightwreather]

8 hours ago, Gamer Schnitzel said:

Well deserve Nvidia for treating gamers like crap. Cannot wait to buy an AMD card.

Considering this (12VHPWR) is a standard developed by PCI-SIG and is part of the ATX 3.0 spec, it seems likely AMD would be using it on their forthcoming cards…

[AlwaysFSX]

3 hours ago, Shzzit said:

These things just didint burn up for no reason . I’m willing to bet 100% user error .  Probably wrenched the shit out of it to fit in a case.

If the connector isn't able to maintain contact with the pins because you need to close your side panel it is 100% the failure of PCI-SIG for designing a bad connector.

That's it, that's all there is to this. It's not a user error issue.

[vetali]

7 minutes ago, AlwaysFSX said:

If the connector isn't able to maintain contact with the pins because you need to close your side panel it is 100% the failure of PCI-SIG for designing a bad connector.

That's it, that's all there is to this. It's not a user error issue.

"You're holding it wrong!"

 

Maybe don't equip your connectors with crappy split terminals and you won't need a power cable support bracket to make sure it doesn't sag along with your card.

[AlwaysFSX]

36 minutes ago, vetali said:

"You're holding it wrong!"

 

Maybe don't equip your connectors with crappy split terminals and you won't need a power cable support bracket to make sure it doesn't sag along with your card.

The mental gymnastics about a terrible connector is mind boggling.

[Bananasplit_00]

16 hours ago, Zodiark1593 said:

Could quality control of the connector (either end) be a potential issue here?

We don't know how common it is if it's even a real issue but that could play a part.

17 hours ago, CHICKSLAYA said:

It's a user error issue. If everyone mounted their gpu vertically and didn't bend the crap out of their cable it would be fine

Don't agree as we've been treating PCIe power connectors like shit since they appeared and not had widespread issues. But again, we really don't know if this is a common issue yet so it's hard to say anything for sure.

[WereCat]

18 hours ago, starsmine said:

Are a lot of us forgetting that 75W still comes from the PCIe slot itself for our analysis?

 

RTX 4090 is using only around 10W from PCI-e

 

source - Igors Lab (mentioned and demonstrated by der8auer here):

 

 

[ZetZet]

6 hours ago, Lightwreather JfromN said:

Considering this (12VHPWR) is a standard developed by PCI-SIG and is part of the ATX 3.0 spec, it seems likely AMD would be using it on their forthcoming cards…

ATX 3.0 did not make ATX 2.xx obsolete. AMD can use the older standard for as long as they want. 

[Mihle]

6 hours ago, Lightwreather JfromN said:

Considering this (12VHPWR) is a standard developed by PCI-SIG and is part of the ATX 3.0 spec, it seems likely AMD would be using it on their forthcoming cards…

Scott Herkelman works in AMD i think.

[Lightwreather]

1 hour ago, Mihle said:

-snip-

Huh, that's quite interesting.  

[AndreiArgeanu]

On 10/24/2022 at 10:31 PM, CarlBar said:

 

According to JC they'll be using the 12 pin as well. Though depending on lead time they may change that.

they changed it obviously. AMD cards won't burn up anymore or maybe they never inteded to use it in the first place. Who knows.

For reference Scott Herkelman is Senior Vice President and General Manager at AMD Radeon.

[MageTank]

Worry not, the top engineers are already on it:

 

You might be seeing these designs sooner rather than later.

[Rex Hite]

Maybe people serious about reliable connectivity will cut off the plugs and directly solder the connections. Then they'll post pictures of their soldering  efforts and get comments on what they did wrong.

Or maybe, if the power requirements keep going up with new cards, the power cables will come permanently attached.

Maybe another solution is that just a pigtail is permanently attached  so the cables from the power supply don't place any bending force on the board.

Maybe the video card companies will just say that people shouldn't be so anal-retentive about cable management and to leave the cables alone.

[Kinda Bottlenecked]

https://new.reddit.com/r/nvidia/comments/yeb9e7/i_think_my_asus_4090_tuf_oc_adapter_is_melting_too/

 

 

[Demonking]

Yeah, I'm happy to be on this 3090 tuf, looking at this mess.

[Senzelian]

Igor's lab has released a video just a few hours ago, whichs shows a few images from a cut open Nvidia adapter cable:

 

 

 

4x 14AWG wires going to 6 pads. Two of those wires connected to 4 pads, the other two wires only to 1 pad each.

Now put 1 and 1 together and suddenly it becomes quite clear why these connectors seem to always burn at the edges.

 

He also mentions that each pad is only 0.2mm in thickness

 

 

 

On top of all that it seems that the wires are coming off the pads when being bent, or the pads break off entirely.

 

 

 

Video: 

Article:

https://www.igorslab.de/en/adapter-of-the-gray-analyzed-nvidias-brand-hot-12vhpwr-adapter-with-built-in-breakpoint/

 

 

I've gotta agree with him. Looks like low quality junk to me.

In the video he also mentiones to have tested this with someone from be Quiet and that he has also tested their cable, which doesn't seem to have the same issue.

[Spotty]

8 minutes ago, Senzelian said:

Igor's lab has released a video just a few hours ago, whichs shows a few imaged from a cut open cable:

 

4x 14AWG wires going to 6 pads. Two of those wires connected to 4 pads, the other two wires only to 1 pad each.

Now put 1 and 1 together and suddenly it becomes quite clear why these connectors seem to always burn at the edges.

English version of the written article available here: https://www.igorslab.de/en/adapter-of-the-gray-analyzed-nvidias-brand-hot-12vhpwr-adapter-with-built-in-breakpoint/

Quote

• The problem is not the 12VHPWR connection as such, nor the repeated plugging or unplugging.
• Standard compliant power supply cables from brand manufacturers are NOT affected by this so far.
• The current trigger is NVIDIA’s own adapter to 4x 8-pin in the accessories, whose inferior quality can lead to failures and has already caused damage in single cases.
• ... [ read the article for the rest ]

 

That explains why in all of the photos I've seen of burnt cables it's the two end pins (1 & 6) that go first.

Seems like it's an issue with Nvidia's adapter using cables soldered on to the terminals that is the issue. 12VHPWR cables using individual 16AWG wires with crimped terminals should be fine. If you can see individual wires going in to each terminal (like below) then it should be safe(r).

[Kisai]

1 hour ago, Spotty said:

 

 

That explains why in all of the photos I've seen of burnt cables it's the two end pins (1 & 6) that go first.

 

Seems like it's an issue with Nvidia's adapter using cables soldered on to the terminals that is the issue. 12VHPWR cables using individual 16AWG wires with crimped terminals should be fine. If you can see individual wires going in to each terminal (like below) then it should be safe(r).

 

Maybe not even that. It seems like the the manufacturer was not informed about how to attach the pins, and the "foil" that bridges them isn't equal to the 14AWG wire that's going into it.

 

My guess is that the manufacturer either tried to cut costs (which is a known issue when dealing with Chinese suppliers that aren't explicitly told every last detail, down to metals, plastics and thread-count of fabrics used) or NVIDIA left open something to interpretation that passed a mechanical QC but wasn't adequately tested with real hardware.

 

At any rate, if this is the case, then aftermarket cables that are not produced with the same connector supplier should be OK. If the wires are intended to be bridged, they should be bridged with sufficient wire instead of foil inside the connector.

[starsmine]

Nvidia's way of making the cables may be whack, but PCI-SIG has been able to recreate the issue in their own lab so I'm hesitant to think its the only issue with the connector. 

[Spicedaddy]

2000$ GPU bundled with 1$ power adapters, what could go wrong?

 

Soldering multiple cables together in that small connector is stupid. The PSU cables by Corsair, Seasonic, etc. are much better than the nVidia adapter. The 12VHPWR connector is not great, but using good cables and not maxing it out to 600W for 2fps more would also help.

[Uttamattamakin]

11 hours ago, Senzelian said:

Igor's lab has released a video just a few hours ago, whichs shows a few images from a cut open Nvidia adapter cable:

 

 

 

Article:

https://www.igorslab.de/en/adapter-of-the-gray-analyzed-nvidias-brand-hot-12vhpwr-adapter-with-built-in-breakpoint/

 

 

I've gotta agree with him. Looks like low quality junk to me.

In the video he also mentiones to have tested this with someone from be Quiet and that he has also tested their cable, which doesn't seem to have the same issue.

 This is great work . 

10 hours ago, Spotty said:

English version of the written article available here: https://www.igorslab.de/en/adapter-of-the-gray-analyzed-nvidias-brand-hot-12vhpwr-adapter-with-built-in-breakpoint/

 

That explains why in all of the photos I've seen of burnt cables it's the two end pins (1 & 6) that go first.

Seems like it's an issue with Nvidia's adapter using cables soldered on to the terminals that is the issue. 12VHPWR cables using individual 16AWG wires with crimped terminals should be fine. If you can see individual wires going in to each terminal (like below) then it should be safe(r).

Another part of the solution for this is as I said on the first page of the thread. 

"The fundamental physics of the situation means there is an inherent hazard that has to be managed.  i.e. it may be the case that next generation graphics cards, at least at the high end, should only be used with ATX 3.0 power supplies.   "

 

If one is going to buy a 1600 dollar GPU they should protect that investment by going for a good quality ATX 3.0 powersupply that has this connector on it.  Then using a good quality connector from the power supply to the card not an adapter.  When dealing with cramming high amounts of energy and power into small spaces one has to respect that when things go wrong they can go VERY wrong.  There is a hazard that one must manage and mitigate.  

It appears that the firm that made those adapters did not respect that they were dealing with a non trivial amount of power that was going to be sent down those wires.  That if anything at all goes wrong it goes catastrophically wrong. 

Don't smoke while pumping gas.  Don't mix water with electricity.  Don't run a 1600 dollar GPU on an old power supply using a cheap adapter.  (i.e. everyone knows not to plug their flagship smart phone into a cheap gas station charger for long periods of time for the same basic reasons.)  

[Uttamattamakin]

5 hours ago, Spicedaddy said:

2000$ GPU bundled with 1$ power adapters, what could go wrong?

 

Soldering multiple cables together in that small connector is stupid. The PSU cables by Corsair, Seasonic, etc. are much better than the nVidia adapter. The 12VHPWR connector is not great, but using good cables and not maxing it out to 600W for 2fps more would also help.

Just like plugging a 2500 dollar smart phone into a $2 charger then wondering why the battery that has so much energy packed into it might explode.  We all have to respect the hazard that some of this technology can present if things go wrong. 

[Senzelian]

Cablemod is now selling a right angle adapter to avoide bends. Whether this actually helps and won't just increase resistance even further is questionable imo.

 

 

But they also have another, in my opinion better, solution. An entire adapter cable, that plugs directly into a PSU. If this is of better quality than the Nvidia crap, then it should solve the issue entirely and also provide a better look.

 

 

https://store.cablemod.com/product/cablemod-rt-series-pro-modflex-sleeved-12vhpwr-pci-e-cable-for-asus-and-seasonic/

https://store.cablemod.com/product/cablemod-e-series-pro-modmesh-sleeved-12vhpwr-pci-e-cable-for-evga-g7-g6-g5-g3-g2-p2-t2/

https://store.cablemod.com/cablemod-12vhpwr-right-angle-adapter/

 

 

[HungryHamster]

13 minutes ago, Senzelian said:

Cablemod is now selling a right angle adapter to avoide bends. Whether this actually helps and won't just increase resistance even further is questionable imo.

 

 

But they also have another, in my opinion better, solution. An entire adapter cable, that plugs directly into a PSU. If this is of better quality than the Nvidia crap, then it should solve the issue entirely and also provide a better look.

 

 

https://store.cablemod.com/product/cablemod-rt-series-pro-modflex-sleeved-12vhpwr-pci-e-cable-for-asus-and-seasonic/

https://store.cablemod.com/product/cablemod-e-series-pro-modmesh-sleeved-12vhpwr-pci-e-cable-for-evga-g7-g6-g5-g3-g2-p2-t2/

https://store.cablemod.com/cablemod-12vhpwr-right-angle-adapter/

 

 

The right angle adapter isn't being sold yet, but it will be available for "pre-sale" on October 31st.

 

Also, I was under the impression that while their adapter cable is certainly better than the Nvidia one, this isn't a substitute for the right angle adapter because it is still prone to the same issue with bending.