It's happening again, melting 12v high pwr connectors

[ShrimpBrime]

Not sure I'd trust ChatGPT or whatever on this, but will admit a fuse might be present (haven't seen one in any PSU review yet, but have not seen them all either).

Mind you if you blow a fuse, your PSU is basically dead. But I guess the wire would be ok.

Yeah, not sure. Where's that Jonny Guru guy when you need him??

 

[Frick]

So I dont know, I asked Google AI, which I will copy paste here. Not trying to duke it out, but genuinely thought OCP was on both sides.

Here's that copy paste.

Sure a fuse would work, but not if your cable is rated for less than what the fuse is (or it's defective).

Not sure I'd trust ChatGPT or whatever on this, but will admit a fuse might be present (haven't seen one in any PSU review yet, but have not seen them all either).

Mind you if you blow a fuse, your PSU is basically dead. But I guess the wire would be ok.

I've replaced fuses in PSUs. They're commonly shrinkwrapped these days.

Spoiler

 

[mkppo]

Yeah, not sure. Where's that Jonny Guru guy when you need him??

Ares is the PSU guy now with his cybernetics stuff. Some say he's a bit biased but the man certainly knows a PSU inside out.

HWBusters

 

[Endymio]

Lol what - so you are saying it was not nvidias choice to go from 8 pin pice to 12 pin ? Nvidia fanboy trying to twist the narrative..

Point out which word you failed to understand and I'll substitute a simpler one. NVidia chose the standard. OEMs chose the specific cabling and connectors. Those OEMs who chose quality parts aren't seeing issues:

"we have yet to see an undamaged Corsair 12V-2x6 cable fail, and we sell millions of power supplies, no exaggeration....."

If plastic melts at 250c, it's a fire hazard. That's above the flash point of the paper the receipt for your 5090 GPU was printed on. It's a fire hazard.

Do you often insert your paper GPU receipts into your power connectors? The melting I've seen appears to be micrometer-scale arcing within the molex connector itself, at which point current stops flowing. If it isn't reaching the ignition point of what these connectors are composed of, which is I believe Nylon 6 (~430C), it can't start a fire. Which is why you've never, ever seen one. Stop the gaslighting.

he just proved the problem is the card, it should not allow that to happen even if there is something out of spec in the cable.

Where does insanity like this come from? Out of spec is out of spec. Your comment is like claiming the card shouldn't allow anything catastrophic to happen even if fed 3-phase 250V.

 

[ShrimpBrime]

Sure a fuse would work, but not if your cable is rated for less than what the fuse is (or it's defective).



I've replaced fuses in PSUs. They're commonly shrinkwrapped these days.

Spoiler

View attachment 384928

The fellas at Warp9 had come to the idea/conclusion the cable was kinked or a sharp bend might had caused the melt down.

For sure, double checked the cable and it was properly rated. The PSU was Antec CP series 1000w. I also had an 850w CP series that had an unstable 12v rail drooping below 12v. Which some people say is OK, but for most of its life I never say that and the 1kw version didn't have the issue. Nor does my HX1050 watter. So I dunked the 850w because of it.

What is 600w DC converted pull on the AC side. My fuses (old school 1965 screw in type) ar rated 10 amps for most of the house. Not even sure I could run a 5090 lol.....

Do you often insert your paper GPU receipts into your power connectors? The melting I've seen appears to be micrometer-scale arcing within the molex connector itself, at which point current stops flowing. If it isn't reaching the ignition point of what these connectors are composed of, which is I believe Nylon 6 (~430C), it can't start a fire. Which is why you've never, ever seen one. Stop the gaslighting.

That's not the point of my comment.

Don't troll me dude.

 

[Panther_Seraphin]

Yeah, not sure. Where's that Jonny Guru guy when you need him??

Working for Corsair in the PSU division (No Joke)

 

[Endymio]

That's not the point of my comment.

Don't troll me dude.

And yet you troll the entire board? If you were speaking English, the point of your comment was to falsely claim that this micrometer-scale arcing inside a molex connector was some sort of "fire hazard". Until you show up with videos of computers on fire, you are well advised to moderate your remarks back to reality.

 

[ShrimpBrime]

And yet you troll the entire board? If you were speaking English, the point of your comment was to falsely claim that this micrometer-scale arcing inside a molex connector was some sort of "fire hazard". Until you show up with videos of computers on fire, you are well advised to moderate your remarks back to reality.

It's not a false claim.
Total fire hazard.

Moderate this screen shot of pictures with PCs that have caught fire.

 

[freeagent]

I would fucking cry.

With rage.

Damn.

 

[mkppo]

NVidia chose the standard. OEMs chose the specific cabling and connectors. Those OEMs who chose quality parts aren't seeing issues:

Nvidia co-designed and chose the standard. And it's sub-standard.

Don't even for a second think Nvidia has no blame to be appointed here.

 

[Dragam1337]

Nvidia co-designed and chose the standard. And it's sub-standard.

Don't even for a second think Nvidia has no blame to be appointed here.

Indeed. But the tool you're replying to is just a keyboard warrior who is constantly moving the goalpost - don't waste any energy on him.

 

[Bomby569]

Where does insanity like this come from? Out of spec is out of spec. Your comment is like claiming the card shouldn't allow anything catastrophic to happen even if fed 3-phase 250V.

yes the card should have measures in place to not let this happen, what is even the doubt here?

 

[freeagent]

Yeah, not sure. Where's that Jonny Guru guy when you need him??

He does post here now and then. Haven't seen him around for some time now.

 

[evernessince]

In relation to both JohnnyGuru and Aris coming out saying you can't pass 20A though a single wire of a 12VHPWR / 12V2X6 cable, it looks like Aris has changed his stance to you can after Derbauer's latest video:

What I don't get is, why didn't they at least test this before making statements. People were lead to doubt Derbauer and understandably after two experts came out saying 20A+ isn't possible on a single cable.

Now that we know that Derbauer's results are valid, we need someone else to confirm them. I'd also like to see testing done for other modes of failure.

 

[eidairaman1]

He does post here now and then. Haven't seen him around for some time now.

Probably busy with corsair

 

[ShrimpBrime]

In relation to both JohnnyGuru and Aris coming out saying you can't pass 20A though a single wire of a 12VHPWR / 12V2X6 cable, it looks like Aris has changed his stance to you can after Derbauer's latest video:

What I don't get is, why didn't they at least test this before making statements. People were lead to doubt Derbauer and understandably after two experts came out saying 20A+ isn't possible on a single cable.

Now that we know that Derbauer's results are valid, we need someone else to confirm them. I'd also like to see testing done for other modes of failure.

Just a few variables I ponder about.

You can pass 20a at 12v.
But can you pass 20a at 12.8v?

Or maybe on the lower end, 11.8v does the current increase for compensation.

How does it go?
If resistance increases, (at the connector) the current should lower or be lower at the card side of the connector.

They could have done 12p connector, but utilize the same male female pins as the 8 connectors.

Space should not be a problem with a card that carries a fkn 4 slot cooler. (Im0)

 

[mama]

Ares is the PSU guy now with his cybernetics stuff. Some say he's a bit biased but the man certainly knows a PSU inside out.

HWBusters

Yes, but his latest rant was not edifying and defensive.

 

[Panther_Seraphin]

How does it go?
If resistance increases, (at the connector) the current should lower or be lower at the card side of the connector.

As long as the following is true
Card is still demanding the same amount of power at that point in time then the following applies

In simple terms voltage drops, current goes up, which then in turn increases wire/conductor temperature which increases it resistance slightly again. Which is where thermal considerations on wires/conductors get taken into acount for "Derating" due to heat etc.

 

[ShrimpBrime]

As long as the following is true
Card is still demanding the same amount of power at that point in time then the following applies

In simple terms voltage drops, current goes up, which then in turn increases wire/conductor temperature which increases it resistance slightly again. Which is where thermal considerations on wires/conductors get taken into acount for "Derating" due to heat etc.

Yes, it seems to be a cascading effect.

Maybe some Gold plating on the design they utilize now would be of some benefit?

 

[Endymio]

It's not a false claim.
Total fire hazard.

So you post of bunch of fire photos that have nothing to do with NVidia cards and pretend it's somehow relevant? What's with the disinformation campaign?

 

[ShrimpBrime]

So you post of bunch of fire photos that have nothing to do with NVidia cards and pretend it's somehow relevant? What's with the disinformation campaign?

Provided what you asked for.

Until you show up with videos of computers on fire,

Which for whatever reason cant fathom the issue could catch a PC on fire, or a fire hazard, you're just plain wrong.

Anything that moves a lot of current is a fire hazard period.

Anything that causes an arc, is a fire hazard.

And PCs have caught fire, can catch fire and will catch fire, albeit it somewhat rare, but most people don't pull 600w or more just on their gpu.

Maybe no PCs will catch fire...

But that doesn't mean it isn't a hazard.

 

[SmookinJoe]

Mine ran fine for 1.5 years then melted...

Because you posted I went and looked at all my 4090's using a laser thermometer...some connectors were showing 39C-45C some of the cables leading up to the OEM adapters were showimg 50C.
I had to be very careful not to disturb the wiring (because of EVERYTHING in this thread) and the Coolermaster HAF Evo lid comes in contact with the adapter wiring when I slide it to open so I had to lift it above that wiring as I opened it. I couldn't lift it too far because it has an exhaust fan attached to the lid.
In My honest opinion....I sense the cabling was less pliable than stock. In other words it seemed to hold it's shape due to being HEATED...
So currently things appear fine but I am not unplugging anything,,,and If I do I will probably replace that section of cabling.
Nothing appeared to be in trouble..no aroma etc.
Totally visual, inspection, with light contact with the cable far away from the adapter and point of insertion of 12 pins.
Hopefully I will not see any issues..I hope!

Sorry not the best pictures
But you get to see a bit of my farm..

 

[Endymio]

Provided what you asked for.

You provided FUD. These cards have started no fires, and the evidence so far points to out-of-spec cables, not an NVidia design flaw.

Anything that moves a lot of current is a fire hazard period.

All high-end graphics cards and cpus, power supplies, home air conditioners, microwaves, refrigerators, dryers, large-screen TVs, and floodlights move a lot of current and are thus fire hazards. So why single out NVidia for your venomous spite? Going out on a limb here, but if I search your posts, I imagine I'll find diatribes from you against their pricing, marketing, and pretty much everything else they do too.

 

[ShrimpBrime]

You provided FUD. These cards have started no fires, and the evidence so far points to out-of-spec cables, not an NVidia design flaw.


All high-end graphics cards and cpus, power supplies, home air conditioners, microwaves, refrigerators, dryers, large-screen TVs, and floodlights move a lot of current and are thus fire hazards. So why single out NVidia for your venomous spite? Going out on a limb here, but if I search your posts, I imagine I'll find diatribes from you against their pricing, marketing, and pretty much everything else they do too.

I dont know man. A few modern connectors look just as burnt as they do melted, cause to melt fuckin plastics, you need a heat great enough to create a fire.

Or maybe you have a different understanding of what fire or something that is fire capable is.

I see no entire PC has been set a blaze. Is that the response you're looking for here? I never said it did.

I said it's a fire hazard and that's exactly what it is. IDGAF what you have to say beyond that.

Capiche?

I own a 4060 and a 4070 super. Had a 4060 ti 3060 3060 ti 3050 GTx 980 780 680.

Don't tell me shit about what I think about NV pricing. They've had plenty of my money through the years fella.

 

[TheDeeGee]

Just stick to the 8-Pin connector, each of the 4x 12V wires is rated for 7 Amps (336 Watts Total), so even a 5090 would only need 3 of them and would be well within overspec range. Just make sure GPU's have load-balancing for each connector (2x2 should be fine), switch to 4-spring connectors and bob is your uncle.