Info 12VHPWR PSA on cable bending

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Saylick

Diamond Member
Sep 10, 2012
3,646
8,223
136
So, apparently PCI-SIG had some concerns regarding the 12-pin power cable, but I'm guessing some higher ups at Nvidia said, "Nahh, ship it!"

At first I thought it was the adapter that was the issue, yet there's clearly a picture of the 12-pin cable being plugged into what I presume is a PSU that natively supports the 12-pin cable. The solution to all this seems to be to limit the power rating of the cable itself, but that would more or less eliminate the whole point of using just one cable.

 
Reactions: Tlh97 and Leeea

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
The example of the molten pins in the vids were from a vertically mounted GPU with very slight bending far from connector.

I recall the vertical one going to the Gigabyte 4090 having a decent amount of cable droop likely caused by the weight of the adapter. That's why it seems like there needs to be some sort of support. It's also why I was surprised that the CableMod cable's flange wasn't an actual support, or at least that's how it appears based upon those images. I'd be tempted to grab a Velcro tie or a zip tie and use the flange as one.

EDIT:

Cablemod is working on a right angle adapter that can be used in conjunction with the adapter cables Nvidia cards come with in the box.


Even though I'll vertically mount when on water, I'll probably grab one so long as it isn't prohibitively expensive. (The Global store appears to ship from China, or at least that's where my cables came from.) I'd rather have that available if I need it rather than have to wait!
 
Reactions: Tlh97 and Leeea

sze5003

Lifer
Aug 18, 2012
14,267
655
126
I recall the vertical one going to the Gigabyte 4090 having a decent amount of cable droop likely caused by the weight of the adapter. That's why it seems like there needs to be some sort of support. It's also why I was surprised that the CableMod cable's flange wasn't an actual support, or at least that's how it appears based upon those images. I'd be tempted to grab a Velcro tie or a zip tie and use the flange as one.

EDIT:



Even though I'll vertically mount when on water, I'll probably grab one so long as it isn't prohibitively expensive. (The Global store appears to ship from China, or at least that's where my cables came from.) I'd rather have that available if I need it rather than have to wait!
I can't find my p600s case accessory box that has the vertical mount adapter but I would imagine even that would cause horizontal flex in the Nvidia adapter cable?

If I end up finding a 4090 this week at microcenter this means it's sitting unopened until I either get an atx 3.0 PSU or cablemod releases the 90 degree adapter.

I don't like sitting on parts not tested but I will buy the microcenter warranty with that card if I do get one.
 

MrTeal

Diamond Member
Dec 7, 2003
3,650
1,899
136
How are the pins on the new connectors configured? Do all the ground and 12V pins go to common planes, or are they separated on the GPU so that it can try to force equal current through each pin?
 
Reactions: Leeea

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
Even 8-pin PCIe connectors have had failures like this. Not sure this isn't anything other than user error or simply a bad plug. Stuff happens. When we see droves of them coming in then we might be able to assume this is an issue.

IOW, put down the pitchforks for now.

The difference is that four 8pins has the same power being dispersed across a larger number of pins. So there is less current draw per pin with four 8 pins compared to 12 pins. A high resistance pin (due to loose connection or something) on an 8pin is less likely to hit critical temps due to having less current draw.

The 8pin connectors were also more robust as they were physically larger.
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
How are the pins on the new connectors configured? Do all the ground and 12V pins go to common planes, or are they separated on the GPU so that it can try to force equal current through each pin?

In reference to my earlier post, and based off my experience with power delivery testing/engineering, the load is likely not being evenly distributed across the 6 power pins. But I do not have any proof, as I have not tested any of these. Its still possible that the power draw is even, but one of them is higher resistance, and because we now have fewer pins handling the same power, a high resistance pin is going to generate a lot more heat than it would with four 8pin connectors.
 
Reactions: Leeea

In2Photos

Platinum Member
Mar 21, 2007
2,159
2,234
136
I bet the issue is the strength on the pins themselves on the wire side. There isn't much room for material in the female portion of the pin so with ANY stress on the connector itself it opens up the female pin, making the connection bad and causing the resistance issue along with potential arcing.
 
Reactions: Leeea

mindless1

Diamond Member
Aug 11, 2001
8,412
1,586
126
The new connector was designed by ignorant engineers and only fit for a controlled environment, not crawling under your desk at 2AM trying to figure out why your PC won't post, lol.

That stated, sometimes I'd do a hackish repair on misc. things and often find that the solution for this kind of dodgy connector problem, is you plug the connector in (this is important so the contacts are correctly aligned), then you coat the back side (where wires come out) with epoxy, getting a "tiny" bit into the connector shell but not enough to foul the contacts. In the case of the connector shown in the video that was wrapped in cloth (tape?), that would need removed first and any adhesive residue cleaned off.


Don't overdo it, use a tiny total amount of epoxy so you aren't risking it flowing down into the contact area of the connector. In some situations like an automotive environment, the connector and wires where entering it, may need washed off first for good epoxy adhesion.

In some situations hot glue will also work instead of epoxy, but I would not use it on a high current connection as the heat may soften if not melt it.
 
Reactions: Leeea

MrTeal

Diamond Member
Dec 7, 2003
3,650
1,899
136
In reference to my earlier post, and based off my experience with power delivery testing/engineering, the load is likely not being evenly distributed across the 6 power pins. But I do not have any proof, as I have not tested any of these. Its still possible that the power draw is even, but one of them is higher resistance, and because we now have fewer pins handling the same power, a high resistance pin is going to generate a lot more heat than it would with four 8pin connectors.
Looking at a couple board shots, it seems like the 6 pins go into a single plane. That's kind of what you'd expect, though it's different than it used to be where generally each 8 pin power connector was a separate 12V input and fed specific phases of the VRM.

But yeah, it should be pretty obvious to anyone that you will have less margin with a dense 12 pin Microfit connector running up to 50A at 600W when the pins itself are rated at 9.2A. Even with perfect balancing you'd have less than 1A margin. Compared to a physically larger 8 pin connector at 12.5A (150W) over 3 pin pairs, with some Minifit-Jr headers rated up to 13A per pin.
 

mindless1

Diamond Member
Aug 11, 2001
8,412
1,586
126
How are the pins on the new connectors configured? Do all the ground and 12V pins go to common planes, or are they separated on the GPU so that it can try to force equal current through each pin?
That would be up to the card designer, but there isn't a good reason not to route all 12V pins from the connector, to the same power plane, wouldn't be a PCB routing issue when they are all adjacent on the same connector.

In some different types of circuit design like analog audio, there could be some merit to isolate the noise of one subcircuit from another but in this case they are all just feeding VRMs with their own feedback loop regulation for GPU and memory, so these VRMs isolate memory power from the huge GPU fluctuations. That's besides the fans which aren't noise sensitive.

Otherwise, there is no reasonably way to try to force equal current through each pin, besides the mechanical aspect of everything similar, same pin type, same wire gauge/length to it, ample PCB copper connection to power plane.
 
Reactions: Tlh97 and Leeea

amenx

Diamond Member
Dec 17, 2004
4,132
2,412
136

Aris puts a 12VHPWR cable through testing. He said old 12 pin hit 100c and was fine, 12+4 Straight 55c, Bent 57c. 505W sustained and he's not concerned, suggests poor connection increasing resistance and therefore more Amperes flowing through a poor connection. Could it be just a bad batch of cables, user error plugging the card in? He is going to do more tests when Asus send him a 4090 and using proper methodology reach a conclusion that isn't a knee jerk "Ngreedia bad" but a reasoned data led reason.
So the design of the 12VHPWR is perfectly fine under perfect conditions. Problem is manufacturing quality inconsistency allows too much room for problems to occur. Dont they take such things into account when they design things like this? FFS!
 
Reactions: Leeea

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
So the design of the 12VHPWR is perfectly fine under perfect conditions. Problem is manufacturing quality inconsistency allows too much room for problems to occur. Dont they take such things into account when they design things like this? FFS!

Don't be too quick to blame the engineers that were working on the design. Its not unusual for them to be given a set of requirements that they have to meet. Even if they say "this wont work", the brass will say "make it work". Been in that position far too many times.
 

Furious_Styles

Senior member
Jan 17, 2019
492
228
116
So the design of the 12VHPWR is perfectly fine under perfect conditions. Problem is manufacturing quality inconsistency allows too much room for problems to occur. Dont they take such things into account when they design things like this? FFS!

Welcome to the trials and tribulations of the early adopter. If you're not prepared for issues like this, don't join the club.
 

Saylick

Diamond Member
Sep 10, 2012
3,646
8,223
136
So the design of the 12VHPWR is perfectly fine under perfect conditions. Problem is manufacturing quality inconsistency allows too much room for problems to occur. Dont they take such things into account when they design things like this? FFS!
I don't see how Nvidia couldn't have encountered this issue in their own labs, because I highly doubt they tested the GPUs in perfect conditions either. Nvidia engineers are still normal people who just plugged in the 12-pin cable without much thought regarding minimum bend radiuses, etc, and they probably tested a reference card inside an actual case, not an open air rig, to determine if the reference cooler was satisfactory. Any kind of long term testing would have hopefully exposed the power connector issue.
 
Reactions: Leeea

moonbogg

Lifer
Jan 8, 2011
10,643
3,100
136
I would never use that connector out of principle. If all it takes is a little uneven loading to make it melt, then screw that thing forever. My OCD would eat me alive with that in my case.
 

amenx

Diamond Member
Dec 17, 2004
4,132
2,412
136
I would never use that connector out of principle. If all it takes is a little uneven loading to make it melt, then screw that thing forever. My OCD would eat me alive with that in my case.
I think that connector is OK if it were rated at less power. For cards in the 300-350 watt range and no more than 3 x8-pin cables connecting to it. For a 4090, I think they should have 2 of them with no more than 2 x8-pin cables connecting to each to more evenly balance load.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |