• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Check Your 4090's 12vhpwr Connectors To Make Sure They are in Pristine Condition

Soldato
Joined
13 Mar 2004
Posts
4,712
Location
Norwich
Have there been any reports of melting connectors with anything other than the "official" nvidia adapter?

Was just having a look at the one that came with my FE and the quality of it isn't great.
The sleeving and fabric wrap is a bit loose and the crimp terminals inside the 12VHPWR end don't look very solid, can see two seams in them which could flex apart under pressure and cause the arcing.

The Corsair 12VHPWR cable I am using seems much more solid as far as the terminals and wire gauge.

Avoid the adapter and use a proper cable!
 
Caporegime
Joined
22 Nov 2005
Posts
45,418
Their idea is that bending either pulls the pins out completely or partially and partial loss of contact results in less area for current to pass through and even without these issues the pins are operating very close to their maximum design.
thats not true someone intel? or whomever it was said the adapter can handle 700 odd watts the cards only need 450 at 100%

More will come - the connection should be idiot proof as its consumer grade, if you can only bend a certain amount or you lose connection integrity then the cable should be designed so people cannot do that...
it's pretty stiff at that point, it's clear you aren't supposed to bend them, seems with a certain case people are just bending them anyway.


shoulda got a p600s and then they don't meed to force anything :D
 
Last edited:
Soldato
Joined
8 Jan 2003
Posts
3,711
Location
Scotland
thats not true someone intel? or whomever it was said the adapter can handle 700 odd watts the cards only need 450 at 100%


it's pretty stiff at that point, it's clear you aren't supposed to bend them, seems with a certain case people are just bending them anyway.


shoulda got a p600s and then they don't meed to force anything :D
It's everyone's fault bar nVidia?

When there's a fire hazard, there should be no compromises and quick action should be taken.
 
Last edited:
Caporegime
Joined
22 Nov 2005
Posts
45,418
It's everyone's fault bar nVidia?
no one even knows if it is the bending that caused ir, someone did a video where bending it to an extreme angle made a 2c difference in temp to the connector.


Maybe these 3 people never fully inserted the cable in the first place, how many GPU did Nvidia sell? must be like 0.0001% of GPU sold so far
 
Associate
Joined
16 Aug 2017
Posts
1,049
It's everyone's fault bar nVidia?

When there's a fire hazard, there should be no compromises and quick action should be taken.
Well, technically NVIDIA did not design that thing - it was intel and PCI-Sig. But NVIDIA seems to be the only vendor actually using it and pushing for it, at the moment. However, that cable included in the box, they ordered that and delivered to all AIBs too, so that is on them.
 
Associate
Joined
19 Jul 2015
Posts
489
no one even knows if it is the bending that caused ir...
Have you still not read the PCI-SIG report that's been posted multiple times in this thread? Maybe you cant read it because it's an image, so here's a complete transcription (which will also be of benefit to search engines and translators):
Test conditions
* 55A continuous per CEM spec
* Cable routed with a reasonable bend radius (such as in a desktop case)
* Test cases below

[diagrams showing 30mm bend radius]

* All six header pins joined together in respective PCB planes per spec
* Benchtop/lab environment
* Hand plugging - non energized (cold plug)
* Room temperature - 26C

Observations
* images below show overheating of the power connector at mating point. Multiple suppliers and designs have failed
* Cables with low cycles and without bend condition have not failed
* Failures observed on both rows of pins depending on load direction
* Hot spots observed @~2.5hrs, melting 10-30hrs
* Note - Also observed after high mating cycles ~40, straight plug w/o side load

[photos showing melted connector]
[thermal photos showing hot connector]

Possible Root Causes
* Resistance variation between pins leads to unbalanced current
* As bending causes high resistance in other pins, more current transfers to the lowest resistance
* Reduction in contact area caused by side load condition and loss of contact in mating interface. Possible deformation of receptacle

[diagram of plug sitting straight in socket]
[diagram of plug sitting ****-eyed in socket, reducing contact area]
EDIT: I see I have fallen victim to the Scunthorpe problem, with the perfectly innocuous "****-eyed" censored.
 
Last edited:
Soldato
OP
Joined
8 Jun 2018
Posts
2,827
What do you think their priority was while cable managing. Connection integrity or aesthetics?

Yeah, people doing "cable management" bend the living crap out of their cables trying to get them to "look" tidy, I'd say they would be much more likely to have a problem over people who just fit the cables how they fall and leave them like that

my 3080 cables go around the end of the card and then through the motherboard plate and it looks quite tidy, but it only works like that because the FE card has the connector already canted 45 degrees from the PCB instead of 90 degrees perpendicular to the PCB the away AIB's do it.



Perhaps they were reassured by Nv Senior Technical Marketing Manager who is also a PCIe-Sig member to Jays2cent. That they had nothing to worry about. They tested those cables and had absolutely no issues (I will even assume they did so inside a closed case). No need to worry it just works, man!!! How ironic if you ask me.

No were in that letter does it state to make sure that you "don't bend those cables". I find culpability in this letter.
 
Last edited:
Caporegime
Joined
22 Nov 2005
Posts
45,418
Alright, so it seems that you're trolling rather than failing to read, but for the benefit of others reading the thread, I'll pick out the relevant part of what I posted above:
yea I'ma bend a cable that warns you shouldn't bend it before 35mm, a cable that takes a lot of force to bend...

you wouldnt bend a heatsink on your cpu to force it to fit your too small case
 
Last edited:
Caporegime
Joined
17 Mar 2012
Posts
48,002
Location
ARC-L1, Stanton System


Perhaps they were reassured by Nv Senior Technical Marketing Manager who is also a PCIe-Sig member to Jays2cent. That they had nothing to worry about. They tested those cables and had absolutely no issues (I will even assume they did so inside a closed case). No need to worry it just works, man!!! How ironic if you ask me.

No were in that letter does it state to make sure that you "don't bend those cables". I find culpability in this letter.

Nvidia have been so successful with pulling the wool over our eyes it has now because the default MO.
 
Soldato
OP
Joined
8 Jun 2018
Posts
2,827
Nvidia have been so successful with pulling the wool over our eyes it has now because the default MO.

Loose connection testing by GALAX with a 12VHPWR connector. Loose connection = 100C+ Good connecttion = 60-70C That's while the connector is pulling a sustained 1530W load. There's more here and it is possible that bending is not the main issue here, force contact is.
...
That's just 1 test done by GALAX and it was a sustained load for 10-15 mins. In PCI-SIG's testing, the melting occurred around 10-30 hours of prolonged runs. GALAX will have a more detailed video on this coming out soon. Hopefully, we can get to the bottom of this.
...
Highly likely that's the case and based on GALAX's tests, the bending issue is being blown out by the press.
...
We should not have to worry about the specific orientation of a cable to this degree. It should just work without issues.
...



It is my opinion that the best way to use a 4090 moving forward is to use a PSU that is ATX 3.0 spec'd with a Gen5 connector. And avoid using any adapters.
I am not sure that the gen5 cabling comes with the psu yet. Even still that has to be scrutinized after this fiasco.

But the question remains if consumers actually find this creditable or not. I am sure there are some who simply fine this as all waffling about.









PSU Example

Doing it this way comes at an added cost. A new PSU and cables specific for 12vhprw isn't cheap when you factor in cost for a 4090 as well.
 
Last edited:
Caporegime
Joined
22 Nov 2005
Posts
45,418
It is my opinion that the best way to use a 4090 moving forward is to use a PSU that is ATX 3.0 spec'd with a Gen5 connector. And avoid using any adapters.
or just don't bend the connector near the pins... and the people saying oh look the pins come out, yea after they strip away all the hard material protecting the wires.....

What connector looks like this? with bare wires and no sheath...
6p4x3sD.jpg


it's like stripping away your house walls and wondering why the rain is coming inside
 
Last edited:
Soldato
OP
Joined
8 Jun 2018
Posts
2,827
or just don't bend the connector near the pins... and the people saying oh look the pins come out, yea after they strip away all the hard material protecting the wires.....
Or...just don't buy adapters when you should be buying a ATX 3.0 PSU with Gen 5 connector support.

What connector looks like this? with bare wires and no sheath...
Cheaply made connectors.


Dear PCI-SIG Member,
Please be advised that PCI-SIG has become aware that some implementations of the 12VHPWR connectors and assemblies have demonstrated thermal variance, which could result in safety issues under certain conditions. Although PCI-SIG specifications provide necessary information for interoperability, they do not attempt to encompass all aspects of proper design, relying on numerous industry best-known methods and standard design practices. As the PCI-SIG workgroups include many knowledgeable experts in the field of connector and system design, they will be looking at the information available about this industry issue and assisting in any resolution to whatever extent is appropriate.
As more details emerge, PCI-SIG may provide further updates. In the meantime, we recommend members work closely with their connector vendors and exercise due diligence in using high-power connections, particularly where safety concerns may exist.
Thank You,


Perhaps one shouldn't hold such confidence in the implementation and assembly of a adapter that clearly show's itself as inferior/defective when they are bent/curved.

:cry:
 
Last edited:
Back
Top Bottom