• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Design analysis discussion of Nvidia-supplied 12VHPWR connectors

This. Im not sure why people are quoting 300V or 150V. Its 12V coming from the rail....

The important electrical property to measure here is Amps.

Just goes to show how much people think they know.
Since we have experts on the matter here what is the solution seems everything recommend is still at fault and burning except running a power limit unless someone had a burned cable with a power limit. We are 32 days post launch of the 4090 and still no real solution. Seems Nvidia is more focused on 4080 launch than resolving anything. Your turn!
 

Since we have experts on the matter here what is the solution seems everything recommend is still at fault and burning except running a power limit unless someone had a burned cable with a power limit. We are 32 days post launch of the 4090 and still no real solution. Seems Nvidia is more focused on 4080 launch than resolving anything. Your turn!
What I would do right now if I had a 4090 card: buy a cheap fan controller with temperature sensors and glue a sensor to the 12VHPWR connector. No fans required, they would be of no use. It only needs to be able to sound an alarm when it's too hot. No, it's not a joke.
 
What I would do right now if I had a 4090 card: buy a cheap fan controller with temperature sensors and glue a sensor to the 12VHPWR connector. No fans required, they would be of no use. It only needs to be able to sound an alarm when it's too hot. No, it's not a joke.
Glue where, the metal part is hidden under plastic, make a small incision, drill a hole or stick on the back side of the PCB. The contact point is more accurate.
 
Last edited:
What I would do right now if I had a 4090 card: buy a cheap fan controller with temperature sensors and glue a sensor to the 12VHPWR connector. No fans required, they would be of no use. It only needs to be able to sound an alarm when it's too hot. No, it's not a joke.
Too hot as in greater than.... lets say 65 degrees Celsius?
 
What I would do right now if I had a 4090 card: buy a cheap fan controller with temperature sensors and glue a sensor to the 12VHPWR connector. No fans required, they would be of no use. It only needs to be able to sound an alarm when it's too hot. No, it's not a joke.

or literally just buy a third party adapter for like $20.
 
or literally just buy a third party adapter for like $20.

True but what about the voiding warranty part? If the cable is not factory shipped and burns then wouldn't you be risking it? Also are 3rd party cables from cablemods guaranteed not to faulter?

Update Cablemod says they have 5 weeks backlog before your cable is shipped at an additional $15 for shipping FYI.
 
Last edited:

Since we have experts on the matter here what is the solution seems everything recommend is still at fault and burning except running a power limit unless someone had a burned cable with a power limit. We are 32 days post launch of the 4090 and still no real solution. Seems Nvidia is more focused on 4080 launch than resolving anything. Your turn!
Company attempts to avoid blame and push it to someone else. News at 11.
 
Too hot as in greater than.... lets say 65 degrees Celsius?
Seems right to me, if the sensor is attached to the plastic shell, as close to the contact points as possible. Now about the type of glue ... frankly I don't know. Hot glue or anything rubber based could ungluewhen hot, right when it shouldn't, and cyanoacrylate would be impossible to clean completely if you ever have to RMA the card.
 
Company attempts to avoid blame and push it to someone else. News at 11.
It's complicated. One manufacturer seems to be making a better plug than the other, so it's not entirely wrong to blame the one making a bad adaptor. On the other hand, if tolerances weren't so tight, this likely would never be an issue in the first place.
 
It's complicated. One manufacturer seems to be making a better plug than the other, so it's not entirely wrong to blame the one making a bad adaptor. On the other hand, if tolerances weren't so tight, this likely would never be an issue in the first place.
Seen a comment on Videocardz claiming the supposed better brand also faulted what a cluster fudge of a situation although I didn't question the validity of that claim.
 
Screenshot_20221116_144052.png


 
Check out the Gamer's Nexus video that just dropped they got multiple cables to burn including the " good one" FYI.
Link
 
Check out the Gamer's Nexus video that just dropped they got multiple cables to burn including the " good one" FYI.
Link

I've just seen the video. They talk about the connector as being made of copper/tin/ coated with nickel.

But here's my problem which I already posted in this thread which no-one bothered to check. Not once in that video has any makers of the adaptors mention "Gold plating" of the pins. We are most used to Gold Plating of our cables, so the question is why the change.

The second thing that bothers me is the latching of the cable so that it is locked in-place. Let me this clear, if it's not latching locking in-place that's a manufacturing fault full stop. You can clearly see in the video the cable has the ability to lock in-place, so if it's not doing this that's a fault.

I could go on about plastic housing of the connector, but I think I have said enough.
 
I've just seen the video. They talk about the connector as being made of copper/tin/ coated with nickel.

But here's my problem which I already posted in this thread which no-one bothered to check. Not once in that video has any makers of the adaptors mention "Gold plating" of the pins. We are most used to Gold Plating of our cables, so the question is why the change.

The second thing that bothers me is the latching of the cable so that it is locked in-place. Let me this clear, if it's not latching locking in-place that's a manufacturing fault full stop. You can clearly see in the video the cable has the ability to lock in-place, so if it's not doing this that's a fault.

I could go on about plastic housing of the connector, but I think I have said enough.

Gold plated contacts is not common. Ni/Sn plating is much more prevalent to the point of being nearly universal. Gold offers little benefit outside of environments with high corrosion risk.
 
Made a quick edit.

Is this a possible solution for the sideways wiggle?

5HwT7Xu.png
 
Made a quick edit.

Is this a possible solution for the sideways wiggle?

View attachment 270264

Possibly. What might be a better solution for dual retention is latches on the ends, like some automotive connectors. That way you're only worried about squeezing between two points instead of four.

Alternatively, now that I think about it, there could be two pins like you have illustrated on the female connector, but with a wide tab for the latch on the male side so the user only has to squeeze in one place. The trick, though, is that we're only solving one of the issues with this. Apparently, a small number of properly-inserted and -stressed connectors are still failing. Uneven current loading seems the obvious culprit; what's less apparent (still) is why.
 
Did anyone gather statistics to know which pins overheated most often? This might be an important clue.
 
Did anyone gather statistics to know which pins overheated most often? This might be an important clue.

Nvidia might be the only entity with enough samples for meaningful statistics, and they're not sharing AFAIK.
 
Possibly. What might be a better solution for dual retention is latches on the ends, like some automotive connectors. That way you're only worried about squeezing between two points instead of four.

Alternatively, now that I think about it, there could be two pins like you have illustrated on the female connector, but with a wide tab for the latch on the male side so the user only has to squeeze in one place. The trick, though, is that we're only solving one of the issues with this. Apparently, a small number of properly-inserted and -stressed connectors are still failing. Uneven current loading seems the obvious culprit; what's less apparent (still) is why.
Another idea is have a latch on both "top" and bottom" of the connect to account for the orientation.
With the FE having the latch on the bottom is weaker at resisting the torque of the cable and side panel twisting the plug downward when the card is mounted in most cases.
This combine with a wider latch you suggested should account for the torque on both axis.
The 12V pin being on the side without the latch is also concerning, if the connector starts to cam out a bit, those pins will be the first one to have bad contact.
 
Last edited:
Back
Top