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

[BoggledBeagle]

That's the most painfully obvious guide I've ever read. And you said you need to be handy, have specific knowledge and skill for that?

Too bad that the majority of people who had problems with melting connectors failed in one or more steps of this guide.

 

[Maxiaid]

Too bad that the majority of people who had problems with melting connectors failed in one or more steps of this guide.

And I guess you have well documented evidence about that.

[...] we must wait for the well documented and reliable case that a correctly installed, undamaged and well made cable caused a problem.

The fact that this discussion is even a thing shows how much bullshit this connector design is. You guys can choose to be delusional. There is really no debate.

 

[ShrimpBrime]

Have you got any documentation on that?

I got the document requested.

 

[Dredi]

I got the document requested.

My mom said not to download executable code from the internet.

 

[ShrimpBrime]

My mom said not to download executable code from the internet.

I'm not that talented lol.

 

[Maxiaid]

This discussion is primarily based on the countless reports from the 4090s that have been regularly happening over the last 2 years. It has been revived because the new incidents on the 5090s have provided proof that the issue is not resolved, as we were told it was. There's, to my knowledge, at least another 5090 that melted connectors on the PSU side from a Spanish youtuber (Toro Tocho). Many other parties (JayzTwoCents, OC3D TV, der8auer) have provided further proof of abnormal temp readings and out of spec currents on the individual lanes. We're less than 1 month in, mind you. Gamers Nexus already have an investigation in the making.

 

[JustBenching]

This discussion is primarily based on the countless reports from the 4090s that have been regularly happening over the last 2 years. It has been revived because the new incidents on the 5090s have provided proof that the issue is not resolved, as we were told it was. There's, to my knowledge, at least another 5090 that melted connectors on the PSU side from a Spanish youtuber (Toro Tocho). Many other parties (JayzTwoCents, OC3D TV, der8auer) have provided further proof of abnormal temp readings and out of spec currents on the individual lanes. We're less than 1 month in, mind you. Gamers Nexus already have an investigation in the making.

You also have to be aware that "5090 is melting cables" generates a lot of traffic. Just saying.

 

[Maxiaid]

You also have to be aware that "5090 is melting cables" generates a lot of traffic. Just saying.

That's true. But I doubt big, reputable channels like those would risk losing reputation over a little bit of clout, and the 12VHPWR fiasco is well documented. See this documentary from Gamers Nexus.

 

[JustBenching]

That's true. But I doubt big, reputable channels like those would risk losing reputation over a little bit of clout, and the 12VHPWR fiasco is well documented. See this documentary from Gamers Nexus.

There is nothing reputable about 97.7% of youtubers. It's about dem clicks. There are a few exceptions (hello Daniel Owen and Optimumtech)

 

[Maxiaid]

There is nothing reputable about 97.7% of youtubers. It's about dem clicks. There are a few exceptions (hello Daniel Owen and Optimumtech)

I don't think any tech media outlet deserves my respect and credibility more than Gamers Nexus does. They've proven to be fully honest and all for the consumer. I like Daniel Owen too. Optimum, I'm neutral. Jay seems solid, der8auer too. Clear sell-outs like Linus Tech Tips I generally ignore.

 

[JustBenching]

credibility more than Gamers Nexus does.

Noooooooooooooo. He is probably the least credible of them all. Id rather watch MLID.

 

[Maxiaid]

Noooooooooooooo. He is probably the least credible of them all. Id rather watch MLID.

How come?

 

[JustBenching]

How come?

Plenty of videos but this is completely off topic. If you like him sure, whatever, my opinion is irrelevant.

 

[JAS_design]

Having used Micro-Fit connectors for years, I have no issue with the quality of the connectors, the size of the pins, the rated number of cycles, or any of that. Obviously a manufacturer other than Amphenol or Molex might produce a lower quality clone, but I'd personally only ever use the real thing from an official distributor such as Mouser, so that's irrelevant to me.

But I read the spec sheets and follow them to the letter; personally, I try not to load crimp contacts beyond about 50% of their rated maximum. My issue with the 12VHPWR / 12V2x6 is that both of the major manufacturers (Amphenol and Molex) have taken time-proven connectors and repurposed them to be used at a much higher current rating than they were ever rated for before. And to make matters worse, they're being USED at very close to that maximum. There's absolutely no margin for error, anywhere.

I don't see how, in the case of Amphenol Minitek, a contact that was previously rated for 7.5A on a 2x6 is suddenly safe at 9.5A. It may be that the contacts were previously derated to provide a built-in safety factor, and were always capable of more. But that just means the safety factor has already been reduced massively before you even get into any issues of user error, rough handling, longevity, load balancing or anything else.

The connectors are great when used judiciously, but their choice for use in the 12V2x6 specification is... questionable.

 

[Endymio]

The cable spec demands that the current in one lead should not exceed X. Nvidia has failed to produce a gpu that would guarantee that the current per lead will not exceed X. The cable CANNOT dictate the current pulled through it. The PSU CANNOT dictate the current transferred through one lead.

You're absolutely confused about the most basics of electrical engineering. The GPU has nothing to do with this lead variance. A PSU, though, can -- if it has more than one rail (a single-rail PSU can't). An AIB board, in theory, also could. However, while I don't know about the 5090, my 4090 has the connector leads all running to the same backplane, so it can't be the board either. Any current variance is a simple case of differential resistance in the connector-pin or pin-cable matings.

Now let me fix that example, and make it accurate. You design a bridge...(example elided)

Ignoring the other problems with your example, you're assuming this "current lead variance" is the problem, when it almost certainly is not. When the ends are cross-connected, the only way one lead carries more current than others if its pin mating is lower resistance -- and if that's the case, the pin is going to generate *less* heat, not more.

I'll pose this question again: last year alone, there were more than 51,000 home fires due to electrical faults, causing more than $1B in damages and more than 500 deaths. Let me repeat that: deaths. All of this from a mature wiring standard that's been around decades. Would you say there's a severe problem with that standard? If so, what one company do you pin the blame on?

You're telling me a guy whose full time job is to review hardware and has been doing it for over 15 years doesn't know how to plug in a cable

No, we're telling you those guys make MONEY by gaslighting weak-lighted individuals with click bait. And at least one of "those guys" admits his problem was with one severely-abused cable alone, and admits he couldn't repeat the problem with any other cable.

 

[eidairaman1]

My mom said not to download executable code from the internet.

He's not a new member, I'm only suspect of brand new members suddenly putting up stuff

 

[OneMoar]

you people don't get it any reason that allows this failure mode to exist is unacceptable

plugging in a cable more then a handful of times is not abuse not misuse not anything other then piss poor design

the card not current balancing is also you guessed it piss poor design

designing anything with no margin for error/safety is once again piss poor design

I don't how anybody with more then two sea shells to click together could fail to see it any other way

you can put current balance resistors and split the rails and add sense pins to try and detect if the cable is flopping around. to mitigate the melting issue but all you are doing is adding safety margin that should have been in there in the first place you are NOT addressing the underlying flaw of the connector which is its just do damn small the pins are too small the connector engagement is too shallow

and the rest of you people can stop harping on der8auer with your thinly veiled racist remarks and attacks on his ability he clearly demonstrated the faults with the design and we thank you keep where he's from out of this discussion

this isn't reddit you don't come here with a personal vendetta and allow it to color your opinion we don't do that here period

 

[Endymio]

plugging in a cable more then a handful of times is not abuse not misuse not anything other then piss poor design

You people don't get it. If a properly-handled cable fails after a few mating cycles, the issue is the quality of the components. There's nothing about the design of a metal pin fitting inside another that limits the number of insertions.

designing anything with no margin for error/safety is once again piss poor design

There's a safety margin. Not a large one -- but again, these margins only exist to protect against manufacturing defects or user error. Without those, there would be no margin required.

 

[OneMoar]

You people don't get it. If a properly-handled cable fails after a few mating cycles, the issue is the quality of the components. There's nothing about the design of a metal pin fitting inside another that limits the number of insertions.


There's a safety margin. Not a large one -- but again, these margins only exist to protect against manufacturing defects or user error. Without those, there would be no margin required.

Wrong completely and utterly wrong No good engineer is going to design something without some margin when the failure could result in property damage / somebody's house burning down

and there is certainly a lot to consider when designing metal pins that need to mate multiple times especially with gold/brass everytime you insert/ remove it you wear a tiny bit of material off

and if your connector is so poorly engineered that there is no tolerance for that you get exactly this kind of failure. worse in cases where there is high current draws or potential for current bias issues with multiple conductors.

nor should the connector be so poorly engineered that twisting or pulling or anything encountered with normal use should EVER cause any kind of failure if it does you failed as a engineer

this is down to the connector being under engineered and being asked to perform WAY outside any sane margins for safety / normal operation

and nvidia not wanting to admit they screwed up and back down

iv about had it with the blame shifting there is zero Excuse for this to ever be a problem for any user anyware period I don't care if a one fingered blind man installed the thing it should not be physically possible for this failure to occur if any decent engineer has done there jobs correctly

you can disagree with that semitment all you like but you are fking wrong and need to stop

it is entirely feasible to design a connector that will function survive any and all use cases it might reasonably encounter and have a long and robust service life why they didn't do that well it rhymes with funny

 

[Endymio]

...nor should the connector be so poorly engineered that twisting or pulling or anything encountered with normal use should EVER cause any kind of failure

I'm going to repeat this one more time. The "normal use" of twisting and turning home electrical cables and wiring causes some 51,000 home fires a year, and more than 500 deaths. Every year. You're fine with that, but sky-screaming inanely at NVidia, which hasn't caused one single fire, and hasn't had one single failure from an actual NVidia-supplied cable. Sell your nonsense elsewhere.

 

[AVATARAT]

I'm going to repeat this one more time. The "normal use" of twisting and turning home electrical cables and wiring causes some 51,000 home fires a year, and more than 500 deaths. Every year. You're fine with that, but sky-screaming inanely at NVidia, which hasn't caused one single fire, and hasn't had one single failure from an actual NVidia-supplied cable. Sell your nonsense elsewhere.

We can follow any guide to be safe, not bad, we know.
We can debate this all day - and it won't change anything.
Nvidia made crap and only Nvidia can fix it.

I'm fixing my 4080S via this "mod", but I'd rather Nvidia do it better, that's all.

 

[Maxiaid]

No, we're telling you those guys make MONEY by gaslighting weak-lighted individuals with click bait.

They make as much money off of this as off of anything else. I don't think a conflict of interest applies here. This logic could apply to any and everything these techtubers said. Clickbait with catchy thumbnails/titles is one thing and all of them do it to some extent, it's ordinary marketing, but that doesn't necessarily disqualify what they investigate or analyze in the video. I'm not trying to make a case for any of these people in particular, I personally like some and don't trust others, but the fact that this is their job doesn't mean they're full of shit.

And at least one of "those guys" admits his problem was with one severely-abused cable alone, and admits he couldn't repeat the problem with any other cable.

All this is saying is the problem is inconsistent enough to be hard to predict and replicate, but it doesn't mean it's rare or simply not an issue. You can't possibly say all cases reported over the last 2 years are due to severely abused cables. The only way to prevent this is to increase the failsafes in place and/or add proper sensors to detect abnormalities and warn the user or cut operation, just like an overheating CPU will shut down your system.

 

[Dredi]

You're absolutely confused about the most basics of electrical engineering. The GPU has nothing to do with this lead variance. A PSU, though, can -- if it has more than one rail (a single-rail PSU can't). An AIB board, in theory, also could. However, while I don't know about the 5090, my 4090 has the connector leads all running to the same backplane, so it can't be the board either. Any current variance is a simple case of differential resistance in the connector-pin or pin-cable matings.

Well, someone is confused for sure.

Name one computer PSU ever which could adjust for per lead current draw, or even per connector power draw. Bonus points if you can describe how it did that.

I can name thousands of GPU’s which did current balancing. The 5090 is not among them though, and that seems to be part of the problem. Such balancing would fix the differential resistance problem for most use cases.

This is a design issue with the GPU, mainly.

 

[Vayra86]

You people don't get it. If a properly-handled cable fails after a few mating cycles, the issue is the quality of the components. There's nothing about the design of a metal pin fitting inside another that limits the number of insertions.


There's a safety margin. Not a large one -- but again, these margins only exist to protect against manufacturing defects or user error. Without those, there would be no margin required.

... what?! Have you gotten the memo that the original redesign of the 12VHWPR connections involved pin design?

Rest in Peace 12VHPWR Connector - Welcome 12V-2×6 Connector, important modifications and PCIe Base 6 | Exclusive | Page 2 | igor´sLAB

The king is dead, long live the king. Or the other way around: The findings of the last few months are finally taken seriously and the connector is modified, work is done on a new standardization and…

www.igorslab.de

I think you need a safety margin too, your posts have absolute zero intelligence tolerance at this point. 'You people don't get it'. Get some fresh air, I think you're inhaling your own farts at this point, you're delusionally repeating points that have been discussed already ad infinitum.

32 pages for you to check back up on. I'd suggest you start at page 1. This hasn't been complicated anymore for about 28 pages I reckon, if you have a sliver of cognitive capability.

 

[AVATARAT]

Most PSUs have 1 rail nowadays, there is also a niche which has multiple rail so it can deliver X watts across each rail. He probably means that when you use 2-3 rail wiring, it will balance through the PSU - and yes, it will probably work. The point is that if it can't deliver current through any of the rails due to a bad contact, and can't deliver it through the others due to rail restriction, it will probably crash the whole machine and it will be hard to figure out why.
But here comes the question - why do we need to change our PSU's for each generation because Nvidia changes something in power usage for each generation - this is insane.