Info 12VHPWR PSA on cable bending

[Stuka87]

How come nobody seated a 3090ti wrong? Or any of the older 12 pins? You'd think with the crazy number of cards they shipped across all the miners and scalpers that someone would have burned a 30 series card too, yet the only burnt 30xx cards I've found online were standard 8pin connectors.

Anyone else remember the 3090 launch and some EVGA cards were frying power delivery caps on the back side? That problem came and went just like this power connector issue. Makes me think, it seems like it wasn't a connector problem but something that was solved in a board revision. Or it is a connector problem and the issues was a small batch of bad female receptacles, and nothing to do with the dongles?

The connector used by the 3x00 series is different than the one used by the 4x00 series.

 

[Heartbreaker]

The connector used by the 3x00 series is different than the one used by the 4x00 series.

Microfit that carries the power is the same.

If you look back at 3000 series, it was mostly FE cards that had 12 pin power, and the 3000 FE cards were all lower power than the 4090.

So if we are talking about a problem that happens .04% of the time, and you spread it among fewer cards with lower power, it just didn't manifest in enough numbers to get noticed.

 

[BFG10K]

How come nobody seated a 3090ti wrong?

Exactly right. Calling this "user error" expects us to believe 4090 owners have never installed graphics cards before, or they're incompetent compared to non-4090 users.

Anyone else remember the 3090 launch and some EVGA cards were frying power delivery caps on the back side?

Yep, don't forget 2080TI space invaders, and the 590 which routinely blew up and TPU put a hidden message "epic fail" in their review:

Hidden Message at TPU about the GTX 590 | Geeks3D

Hidden Message at TPU about the GTX 590

www.geeks3d.com

 

[Stuka87]

Microfit that carries the power is the same.

If you look back at 3000 series, it was mostly FE cards that had 12 pin power, and the 3000 FE cards were all lower power than the 4090.

So if we are talking about a problem that happens .04% of the time, and you spread it among fewer cards with lower power, it just didn't manifest in enough numbers to get noticed.

They are very similar, but as I recall, the 12 pin in the 3x00 series makes a click when you push it in. The 4x00 series 16 pin does not. Its really mushy.

 

[lopri]

you would like to think so based on the outer housing, but if you look at the pcb it is a different story. (the 3080 rog has some perspective in the focal length while the 3080 founders is nearly orthogonal, but i've tried to match up the pcie pins as close as possible). the cut and paste of the 3x 8pins is roughly the area they would have to take up. unless nv changed their basic tried/tested layout, they would have to make an even bigger pcb or abandon the through card fan hole. but they cant lose the hole and keep temps down. they made the 4090 through card fan even bigger so they lost even more pcb area.

View attachment 71449

I was thinking it's the fans, not the PCB. Compare the size of the fans from these cards to other cards. They are proportionally bigger in every dimension. For that HSF design, 3 x 8-pins are clearly getting in the way.

Also I believe some AIBs with different cooler design still use traditional 8-pin connectors.

 

[Heartbreaker]

They are very similar, but as I recall, the 12 pin in the 3x00 series makes a click when you push it in. The 4x00 series 16 pin does not. Its really mushy.

Both use standard Molex Microfit 3.0 specification for the pins and retention mechanisms.

There is the same subtle click either way.

Main factors are as I mentioned before. Lower power and fewer of them in use.

One more occurred to me. Check the orientation of the connectors for 3000 vs 4000 series:

In the 3000 series the connector is angled and sideways to the length of the card. It seems much less likely to pull the connector from the sides, which was the failure induced by GN.

The 4000 connector is straight up and rotated 90 degrees. Closing the case on a loose cable will be much more likely to put that kind of sideways force on it.

One final point I didn't make clear enough in my first post.

IMO, Many people didn't seat the connector properly on the RTX 3000 cards either.

Connector melting is an issue that so far has occurred on about 0.04% of cards, but I expect the incidence of improperly seated connectors is probably about 100X as much as that.

Not clicking the connector home isn't enough to cause this problem. You have to fail to seat the connector, then bump it enough times to work loose, and then apply sideways force to have it sit unevenly.

 

[coercitiv]

One more occurred to me. Check the orientation of the connectors for 3000 vs 4000 series:

In the 3000 series the connector is angled and sideways to the length of the card. It seems much less likely to pull the connector from the sides, which was the failure induced by GN.

One more detail to add to your list: average custom card width has increased with the 4000 series. Just one anecdotal example: Asus went from 140mm with TUF 3090Ti to 150mm with TUF 4090. Cases that could fit the last series were suddenly cramped for the new series, hence they introduced additional mechanical stress on the adapter / connector assembly.

 

[Stuka87]

Both use standard Molex Microfit 3.0 specification for the pins and retention mechanisms.

There is the same subtle click either way.

Main factors are as I mentioned before. Lower power and fewer of them in use.

The 3090Ti uses a lot more power than the 4090 (per TPU). The 3090 uses a bit less.

Having the connector at an angle did also allow the users to SEE the clip and if it was seated or not. Which I think was a boog thing. But that the fact that AIB models all used traditional 8 pins is probably why we didn't hear about it.

The 4090 being larger certainly doesn't help. So people are putting even more stress on the connector.

 

[Heartbreaker]

The 3090Ti uses a lot more power than the 4090 (per TPU). The 3090 uses a bit less.

Having the connector at an angle did also allow the users to SEE the clip and if it was seated or not. Which I think was a boog thing. But that the fact that AIB models all used traditional 8 pins is probably why we didn't hear about it.

The 4090 being larger certainly doesn't help. So people are putting even more stress on the connector.

The 3090 Ti was probably a low seller. It was near universally lambasted as massive price/power increase for minor performance bump..

 

[Aikouka]

I finally got around to testing my 4090 last night prior to putting a water block on it. Given that I didn't want to drastically alter my PC without testing the card, I just used the adapter with my Gigabyte Aorus 3080's existing 3x 8-pin setup. The biggest thing that I noticed is that the adapter seemed absolutely fine to me. I expected it to have a very minimal clicking noise and to not be sure if it was properly seated or not; however, the click was quite audible and I didn't perceive any difficulty in inserting the cable. Now, to be fair, that might be due to inserting the adapter while the card was out of the case, and plugging the PCI-E cables into the adapter while the card was in the case.

 

[sze5003]

I finally got around to testing my 4090 last night prior to putting a water block on it. Given that I didn't want to drastically alter my PC without testing the card, I just used the adapter with my Gigabyte Aorus 3080's existing 3x 8-pin setup. The biggest thing that I noticed is that the adapter seemed absolutely fine to me. I expected it to have a very minimal clicking noise and to not be sure if it was properly seated or not; however, the click was quite audible and I didn't perceive any difficulty in inserting the cable. Now, to be fair, that might be due to inserting the adapter while the card was out of the case, and plugging the PCI-E cables into the adapter while the card was in the case.

Hopefully everything is good if it clicked in and there is no issue when no you tried to pull on the wire to see if it would come out.

I dunno if it's the Nvidia drivers but I'm seeing lots of fps drops from like 114-120fps down to 40fps or even 90 for example. This often causes stutters. This was in plague tale requim, cod, doom eternal at 4k. Lots of times I'm seeing the 13900k at only 10-15% usage.

 

[CP5670]

I set up mine too. I made sure the 16-pin cable was pushed in firmly and had a straight entry, and heard the click indicating it's in place. However, that adapter looks incredibly clunky in my otherwise clean case, and I definitely want the Corsair native cable. It's currently out of stock everywhere and being gouged to 3-4x the price on ebay.

 

[In2Photos]

And now aftermarket cables are melting.

Custom 12VHPWR power adapter by CableMod has melted - VideoCardz.com

Cablemod’s 12VHPWR cable melts The first reported case of melted custom cable by CableMod. CableMod 12VHPWR cable melting, Source: Reddit After many weeks of silence, melting RTX 4090 cables are back on the agenda. This time the problem affects a custom cable designed and manufactured by...

videocardz.com

 

[Karnak]

https://www.reddit.com/r/nvidia/comments/10eyyfm/_/j4u8nwm

See the reddit thread about it. Not plugged in correctly (like all of the cables so far) and that it's "melting" is what you would expect in this case.

The connector just needs a good enough indicator that it's fully plugged in. That's not the case for common folks right now. Which is bad. Kind of a faulty design one could argue. But if it's plugged in correctly nothing will happen.

 

[Heartbreaker]

The connector just needs a good enough indicator that it's fully plugged in. That's not the case for common folks right now. Which is bad. Kind of a faulty design one could argue. But if it's plugged in correctly nothing will happen.

It does have an indicator. It clicks/locks into place. If you are paying attention, it's not a problem.

Is it foolproof? No, but that's nearly impossible to achieve.

 

[CP5670]

I heard both the adapter and the Corsair native cable make a distinct click. It's not as loud as the 8-pin connector but easily audible if you listen for it.

They should really have had the connector on the back edge of the card. That would have further reduced issues with this. Interestingly the cancelled EVGA card was the only one that had it there.

 

[BFG10K]

And now aftermarket cables are melting.

Rather ironic since Cablemod were buzzing around the reddit forums when this first started and were basically saying "wait for our cables, it won't happen there!"

So if it's not the cable, the only other explanation is that 4090 users are incompetent fools that have never installed graphics cards before. Except we only need to look at the last six months and see practically zero cases of 8pin cables melting, despite tens of thousands (or even millions) more in use.

Nah, the plug is a defective fire hazard design, plain and simple. And the worst part is that this septic firestarter is rumored to be on 4060W parts with just a 160W TDP. A solution to a problem nobody ever asked to solve.

 

[Karnak]

Nah, the plug is a defective fire hazard design, plain and simple.

Did you look at the reddit post I posted above? That cable mod adapter was not plugged in correctly. Plain and simple and that's obvious looking at it. Because you can see how far it has been plugged in. Which was not all the way in.

And yes, there is less of a feedback plugging in the 12VHPWR compared to the usual 6/8pins. That's a problem which will be fixed in the future. We know the PCI-SIG is already working on it. Best thing is to plug it in before you put the GPU in your case/mb. Then it's easier not only to plug it in but also to see if it was correctly done.

 

[BFG10K]

That cable mod adapter was not plugged in correctly.

Because the cable design is defective. That's why 8pin doesn't have the same problem, doesn't need side panel clearance "guidance", or mollycoddling when routing it.

 

[Heartbreaker]

Because the cable design is defective. That's why 8pin doesn't have the same problem, doesn't need side panel clearance "guidance", or mollycoddling when routing it.

I'd bet just as many 8 pin connectors aren't plugged in correctly either.

Nothing can force the oblivious to do things correctly.

The difference is 8 pin connectors only carry 150 watts, and not up to 600 watts.

 

[BFG10K]

I'd bet just as many 8 pin connectors aren't plugged in correctly either.

Show us 50 examples of 8pin connectors burning up in the last 3 months.

Even if you somehow manage this, there are still a thousand 8pin connectors for every 12VHPWR in use.

The difference is 8 pin connectors only carry 150 watts, and not up to 600 watts.

And that's the point. The new connector is smaller and less robust with less tolerances, yet it's expected to provide a whole lot more power. That makes it inherently unsafe by design.

 

[Heartbreaker]

Show us 50 examples of 8pin connectors burning up in the last 3 months.

Even if you somehow manage this, there are still a thousand 8pin connectors for every 12VHPWR in use.

Already answered and even acknowledged by you in the second part of your post.