• 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

z10m said:
My corsair cable also leaves tiny gap when bent but it doesn’t heat up at all as I tested it with thermal camera.


06-A102-A7-16-A4-4-D47-B15-D-A7-D530218-CEF.jpg
Click to expand...

It's great that Nvidia are bundling Thermal Cameras with every GPU to check this though
 
pastymuncher said:
@z10m Could you tell me what thermal camera are you using please?

Why on earth have Nvidia stuck the power connector half way down the card? It's right in the way there.
Click to expand...

It is because they have gone with an ultra wide and very short PCB. The power connectors are actually on the end of the PCB and everything else past that point is heatsink.

I agree it is a hideous design and even if it worked perfectly well I would still hate it aesteticly. I would most likely have a 4090 in my PC at the moment if it was not for these design decisions by NV. I went with a 7900xtx and the deciding issue for me was the PCB shape.

Edit - I did not pay too much attention to the video but iirc when Jayz2cents looked at the cancelled EVGA 4090 they had routed the power connection to the far end of the card so it would have been easier to hide.
 
Last edited:
Jimbo said:
My cablemod cable still causes black screen 100% fan spin so I took their advice and set maximum power in the Nvidia control panel. However it's no good because it constantly keeps fans at 30% instead of off at idle. So I'm going to arrange rma of cablemod cable and try the corsair one or the default Nvidia one that came with it. Disappointed as this cable cost what it did.
Click to expand...
I had this issue a couple of times with the cablemod cable on my Inno3d Ichill X3 4090.

I thought it was the drivers instead once I started using the Seasonic cable that came with my Vertex GX PSU 0 issues since, And if you look closely or anyone who has a new Vertex PSU they'll see the cable and sense cables are built far better than the cablemod sense wires.

The Seasonic look not only thicker gauge they are tight in the connector, The cablemod sense cables are thinner and feel cheap and flimsy to me like your thin tin wire used to fix circuit board repairs.

Not what I expected from a cablemod cable at the cost it was after now seeing how Seasonic have done it.
 
Jimbo said:
I've got a replacement cablemod cable on the way, so let's see if they have improved any. I hope so because as you say it's not a cheap cable.
Click to expand...
Unfortunately it's just the same, I've had a replacement a few days ago and they haven't changed the quality of the sense cables, I'm sticking with the PSU cable for now, Hopefully yours works with the replacement without the black screens.

Quite frightening seeing it for the first time, Screen went total black and GPU fans locked at 100% had to pull the power from the PSU wouldn't reset, I was only browsing the web when it happened, Thought the GPU was toast when I first saw it.
 
Well I've got the new cable in and so far it's been perfect. No crash and 100% fan spin and I've deliberately left it idling where it would happen with the old cable. Saying that I can't remember if the old one just developed the problem after a period of time. But so far so good. :)
 
Jimbo said:
Well I've got the new cable in and so far it's been perfect. No crash and 100% fan spin and I've deliberately left it idling where it would happen with the old cable. Saying that I can't remember if the old one just developed the problem after a period of time. But so far so good. :)
Click to expand...
Glad it's working well, Mine did work initially and seemed to develop that fault after a little while, I've stuck with my PSU cable and never seen the issue since.

Cablemod seem to be getting a bit of flack lately for their products including the new 90 degree adapters build quality with wobbles and mis aligned connectors etc.

Haven't used them much for products other than risers but the first cable I got which was faulty didn't bode well with my confidence of their products.

Hopefully yours is now sorted.
 
This thread seems to have been buried despite the issue still being quite relevant and resurfacing sometimes only to be buried again in the 4000 Series thread.

Below are a few videos from Northridge Fix that adds another valid point of view.




Not every 4090 (and 4080) that has been sold has evidence of this melting/burning connector (not sure of the %) so perhaps it's not the design of the card as suggested by Northridge Fix. As GN has shown if you mess around and don't fully seat the connector you can make it happen, but that shouldn't be taken as to why it's happening.

There is so much data missing here to be able to see the whole picture. For example, what power supply was being used by each user of the damaged GPUs - what was the manufacturer, model, power rating and efficiency rating of each unit? What type of 12VHPWR cable was being used - was it the the 3 dimple, 4 spring or some other design of internal crimped connector? Was the damaged cable supplied by Nvidia, the PSU manufacturer or third party? What was the gauge of the wire being used? What pins were actually damaged and does it differ in each reported case?
 
Mysterae_ said:
There is so much data missing here to be able to see the whole picture. For example, what power supply was being used by each user of the damaged GPUs - what was the manufacturer, model, power rating and efficiency rating of each unit? What type of 12VHPWR cable was being used - was it the the 3 dimple, 4 spring or some other design of internal crimped connector? Was the damaged cable supplied by Nvidia, the PSU manufacturer or third party? What was the gauge of the wire being used? What pins were actually damaged and does it differ in each reported case?
Click to expand...

Not so easily nailed down but also what kind of stuff the user was doing - it might be certain games or applications are causing higher power spikes than normal, etc.
 
Rroff said:
Not so easily nailed down but also what kind of stuff the user was doing - it might be certain games or applications are causing higher power spikes than normal, etc.
Click to expand...

For sure.

I'm interested in the power supply side of things because not all of the six 12V pins are burnt in the examples I've seen. Why's that?

Excluding the signal contacts, each contact is rated at 9.5A maximum, with 6 of them giving a rated maximum power of 684W draw at 12V (the other six contacts are ground). If the card was pulling 600W and the power supply (or connector) drops even one contact then that's 10A per contact for 5 contacts, 12.5A per contact for 4 contacts etc, maxing out the rated current for the contact. If the voltage fluctuates that has an impact on the current draw too. These ratings are for the Amphenol manufactured connectors, other suppliers may be rated lower or higher.
 
Mysterae_ said:
For sure.

I'm interested in the power supply side of things because not all of the six 12V pins are burnt in the examples I've seen. Why's that?

Excluding the signal contacts, each contact is rated at 9.5A maximum, with 6 of them giving a rated maximum power of 684W draw at 12V (the other six contacts are ground). If the card was pulling 600W and the power supply (or connector) drops even one contact then that's 10A per contact for 5 contacts, 12.5A per contact for 4 contacts etc, maxing out the rated current for the contact. If the voltage fluctuates that has an impact on the current draw too. These ratings are for the Amphenol manufactured connectors, other suppliers may be rated lower or higher.
Click to expand...

10A per contact shouldn't be a problem, 12.5A probably pushing it. Outside of computer use I've used cables and connectors like this for projects and you can usually push at least 15% over the stated maximum sustained before you have issues but the maximum duration for a significant spike can be very brief.

I've noticed with Ampere GPUs you can get at times, especially with certain applications, very brief but extreme spikes - unfortunately outside of the 2Hz resolution of my PC wattage monitor to register accurately.
 
Last edited:
Mysterae_ said:
This thread seems to have been buried despite the issue still being quite relevant and resurfacing sometimes only to be buried again in the 4000 Series thread.

Below are a few videos from Northridge Fix that adds another valid point of view.
Click to expand...

All those cards were sent to Northridge Fix by Cablemod, accumulated over months. They weren't sent in by random customers experiencing the burning issue
 
Last edited:
I've had my Inno 4090 since launch and not had any issues with the connector since day 1, Have used now totalling 3 different cables and have plugged and unplugged numerous times already.

I do not believe this is a manufacturing defect as it would have happened to thousands of cards.

Something is at play whether it's mostly user error not connecting properly, Faulty cables that don't have their pins seated properly and are half and half making contact or people pushing the power target and overclocking as much as possible.

Maybe it's the cards that go over 450w are the ones that are burning and users are not admitting they have pushed 600w through them.

I know the connector is rated at 600w but I wouldn't like to run anymore than 450w through mine even if I could.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom