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

[Bomby569]

Well I do not want burn my 260€ psu and 1000+€ gpu, but I need a new gpu. AMD seems to still have issues with my favorite game and the 9070XT will be a reball king. We pretty much lost the battle when GN called it user error. We need LTT (since he loves nvidia so much) to make a big news about this.

sure on influencer screwed us, and another influencer (a braindead, unreliable one at that) will save us

 

[Vayra86]

Well I do not want burn my 260€ psu and 1000+€ gpu, but I need a new gpu. AMD seems to still have issues with my favorite game and the 9070XT will be a reball king. We pretty much lost the battle when GN called it user error. We need LTT (since he loves nvidia so much) to make a big news about this.

What some influencers say on YT is of zero relevance. What matters is sales, cold hard $$$.
Everything else is bullshit

It seems increasingly difficult it seems for people to understand the value of measurements and facts. It really is that simple. User error doesn't even come into play here.

The tolerances are too low.
The design is too small, inviting far too much risk on manufacturing variance
The QC is clearly lacking and may need to be given some sort of Seal of quality garantuees carried by the industry to acknowledge its a safe cable.
Given the fact its a GPU oriented cable, it should also be specced for sufficient un-and replug activity - at least enough to last two lifetimes of a graphics card, as these cables get reused.

Until those four points are fixed I won't be touching this at all. I don't even care what GPUs are left to buy at some point. It just ain't happening.

 

[Bomby569]

How Nvidia made the 12VHPWR connector even worse.​

 

[dragontamer5788]

How Nvidia made the 12VHPWR connector even worse.​

I came here to post this video.

The 4090 and 5090 have a singular shunt resistor on their PCB for the entirety of the 575W load. The cards cannot balance or measure power across the different pins at all.

 

[ty_ger]

What engineering? Every good engineer knows you can "flow" maximum 10A per pin as thick as in this "12V high power" connector.
That is maximum, where is the safe value? Divide that by at least 2 because you have to take into account the contact surface. And you are already at ~300W (safe maximum), not 600 anymore.

That was my point.
The engineering controls they have chosen to adopt -- because they are basicly free -- have not been effective. Moving the sense pins back a bit, talking about the quality of the pins and importance of QC, and educating the consumer about the dangers, has clearly not been effective. The connector is not robust and it should not be the consumer's job to foll9w a magic recipe for success. More engineering controls are needed; or a complete redesign.

The product launch rush with little testing certainly helps no one.

 

[stahlhart]

Here is a now two-year-old CableMod PRO ModMesh extension (one 12VHPWR to four 8-pin PCIe) that has been installed between a Corsair HX-1200 and an MSI RTX 4090 Suprim X Liquid, and has been running pretty much 24/7 since it was purchased. The 12VHPWR end gets pulled from the GPU every couple of months for inspection and a light droplet of DeOxIt D100L on all of the contacts.

I'm not even going to try and speculate what might be the root cause of the problem some users are experiencing, but parroting, painting with a broad brush and inciting internet angry mobs are not helping.

 

[BoggledBeagle]

Looking at the 12 tiny pins, why do not we have just 6 massive tabs instead? The overall size of the connector would be the same.

There could be some cutouts on the plug connectors sleeves, and in the end of the plug travel, some protrusions in the socket would through these cutouts push on the metal reeds of these connectors, so that they would contact the tabs securely, even if they lost some of their original springiness.

View attachment 384352

Some locking mechanism, which would be usable only when the plug is fully inserted and which would actuate some contact enhancing mechanism surely would not be that hard to develop.

 

[TheDeeGee]

View attachment 384356

Here is a now two-year-old CableMod PRO ModMesh extension (one 12VHPWR to four 8-pin PCIe) that has been installed between a Corsair HX-1200 and an MSI RTX 4090 Suprim X Liquid, and has been running pretty much 24/7 since it was purchased. The 12VHPWR end gets pulled from the GPU every couple of months for inspection and a light droplet of DeOxIt D100L on all of the contacts.

I'm not even going to try and speculate what might be the root cause of the problem some users are experiencing, but parroting, painting with a broad brush and inciting internet angry mobs are not helping.

You inspecting it every couple of months and spraying it with deoxid says enough about the connector doesn't it?

Should be plug in and forget.

 

[N/A]

I'm not even going to try and speculate what might be the root cause of the problem some users are experiencing, but parroting, painting with a broad brush and inciting internet angry mobs are not helping.

Owning an Nvidia card has become a major issue. It's impossible to use it safely without 6-amp clamps and real-time monitoring or thermocouples.

 

[Vayra86]

View attachment 384356

Here is a now two-year-old CableMod PRO ModMesh extension (one 12VHPWR to four 8-pin PCIe) that has been installed between a Corsair HX-1200 and an MSI RTX 4090 Suprim X Liquid, and has been running pretty much 24/7 since it was purchased. The 12VHPWR end gets pulled from the GPU every couple of months for inspection and a light droplet of DeOxIt D100L on all of the contacts.

I'm not even going to try and speculate what might be the root cause of the problem some users are experiencing, but parroting, painting with a broad brush and inciting internet angry mobs are not helping.

I don't think anyone is denying there are also cables and GPUs working properly. The issue is the luck of the draw and the amount of things you get burdened with to pull luck in your favor.

Such as your inspection.

 

[Splinterdog]

I just upgraded a client's PC with a Palit 4070 Ti which came with an adaptor cable, so I'm hoping he doesn't have similar problems. Nothing to report so far, thank goodness.

 

[Dragam1337]

This is Music for AMD marketing department, they can now claim in adverts that they cards free of issues of melting 12P connector

Honestly, it's THE strongest argument for getting an amd gpu - and it's a quite strong one.

 

[dragontamer5788]

I don't think anyone is denying there are also cables and GPUs working properly. The issue is the luck of the draw and the amount of things you get burdened with to pull luck in your favor.

Such as your inspection.

Only one shunt resistor on the 4090 and 5090.

That's a big mistake. Cables like this WILL have unbalanced loads and the card should be compensating for that. If you fail to balance the current across cables, you get melted cables and fire hazards.

This doesn't look like a connector problem to me. No one can possibly expect a little plastic thing to evenly apply pressure to all 12 pins in all scenarios. The EE should have build some degree of sensing and compensation.

 

[adilazimdegilx]

If a new thing replacing the older design is also making it more susceptible to user error than it's not really a user error in my book.

Imagine if you invent a new revolutionary brake system for cars that works extremely well at 30psi but completely fails if you press harder yet you release it like that and just tell people to 'not press too hard'. You cant claim a user error for that. It's your responsibility to make sure that user cannot apply more force than 30psi without willingly and intricately removing whatever safety measure you came up with. Then, it becomes a user error.

That being said, failures always happen and will happen. You cant make it 100% safe. However, what you can do is making sure it is at least as safe as what it's replacing. That's where this cable fails. In worst case, one of those tiny cables will have to pass up to 50A by itself which is well over their limits. So they have to either make that each cable can withstand that power or make sure that it failsafe when it happens. First one would require something like a thicker cable but second one is not really a rocket science, easy and pretty cheap (bullzoid explains quite well how).

But they didnt put that in their 2000$ (if you believe their MSRP which doesnt exist in real life anymore) GPU.

 

[remekra]

<quietly turns the power limit on 5080 back to default from 450W>

 

[Bomby569]

incredibly Nvidia gets away with this. If it was Intel or AMD all hell would brake lose, all the influencers would bring the pitchforks years ago. You have to wonder how Nvidia compensates all these people so well

 

[dragontamer5788]

That being said, failures always happen and will happen. You cant make it 100% safe. However, what you can do is making sure it is at least as safe as what it's replacing. That's where this cable fails. In worst case, one of those tiny cables will have to pass up to 50A by itself which is well over their limits. So they have to either make that each cable can withstand that power or make sure that it failsafe when it happens. First one would require something like a thicker cable but second one is not really a rocket science, easy and pretty cheap (bullzoid explains quite well how).

This new video doesn't imply that it's a cable issue.

Indeed, any electrical engineer worth their salt knows and understands how hard it is to balance power multiple copper lines. The 3090 did balance two wires at a time.

But it seems like the 4090 and 5090 have only one shunt resistor, meaning that the EEs who designed the board see the 6 pins of power as just one cable. !!!!!

So no balancing, no sensing. And that's after widely reported 4090 issues.

incredibly Nvidia gets away with this. If it was Intel or AMD all hell would brake lose, all the influencers would bring the pitchforks years ago. You have to wonder how Nvidia compensates all these people so well

Stock bro. They're likely shareholders who knows the name of the game. Insult your company and the company price drops.

 

[AVATARAT]

How Nvidia made the 12VHPWR connector even worse.​

But RT, DLSS and FG just work, who cares about cable, just replace it when it melts down.

 

[R0H1T]

I'm not even going to try and speculate what might be the root cause of the problem some users are experiencing, but parroting, painting with a broad brush and inciting internet angry mobs are not helping.

Yeah running 24x7 at idle ain't it

Stock bro. They're likely shareholders who knows the name of the game. I sult your company and the company price drops.

They're selling $1000 GPU with additional $5-10k markups to "AI" enterprises, you think they care about plebs

 

[TheDeeGee]

Anyways, Der8auers video now has proven that the FE has a design flaw, because it lacks per pin sensing, which the ASUS Astral 5090 does have and thus can balance the load over all six 12V cables.

 

[Gigaherz]

I swear to god if I have to read one more thread about people defending this dogshit connector I am going to snap.

What is wrong with you? Seriously? Buyers remorse for owning one ?

I have been connecting 8 pins left and right for the past 10 years. Literally hundreds if not thousands of plug in and out on the same three test PSUs and hand full of Test GPUs. They look like theyve gone through a war and yet they still work. BECAUSE THEY ARE DESIGNED TO LAST. Opposed to what we have now. Everyone on the higher Hwbot ranks has a similar setup and story to tell. 3rd party, pre used my ass. This should still work!

The 12VHPWR and updates are simply inacceptable. There is no sane counterargument.

 

[Dragam1337]

I just replaced cable for my 4090, cause voltage from 12pin had started dropping to 11.6 during gaming, thus was immidiate danger of overheating - using official cable from corsair, and with new one it's back up to 11.8v during gaming.

 

[N/A]

Anyways, Der8auers video now has proven that the FE has a design flaw, because it lacks per pin sensing, which the ASUS Astral 5090 does have and thus can balance the load over all six 12V cables.

No, it doesn't balance, it just pops up a message or something.

Why does one pin closer to the pcb get disproportionately higher amps, is it the shorter path of less resistance. very counterintuitive so far at least two FEes are like that. I have no idea.

 

[Dragam1337]

Anyways, Der8auers video now has proven that the FE has a design flaw, because it lacks per pin sensing, which the ASUS Astral 5090 does have and thus can balance the load over all six 12V cables.

I would really like to know if my gigabyte 4090 gaming has per pin sensing...

 

[_roman_]

Here is a now two-year-old CableMod PRO ModMesh extension

I'm glad that you are in the good scenario. Cool you have a working product.

I do not want to be in the bad scenario in any given time period.

I'm kinda lucky with getting bad products. Fans at full speed = pwm not working with a new ASROCK 6600XT challenger 8GB D graphic card. That graphic card was bought some days after the official release. I doubt anyone else used that card before myself. Dying Corsair power supply unit after 1.5 years, dying presonus eris 3.5" speakers, bad enermax psu cables out of the box. I think the quality control is the end consumer sometimes. Replace a bad unit when the consumer complains in the one year warranty. Before that say anything so there is no need to replace it. RMA only allowed for the first buyer, not for second hand graphic cards which are in the one year warranty window, ...

Basically when the brakes do not work for a car it is okay. Just when 10 percent of cars are crashing it's acceptable, right? Or do we care to reduce the ~200 death people in year 2024 in austria? It's always about the small numbers of bad products which are going public.

--

I also have a current clamp TRUE RMS multimeter.

Around minute 15

Is it not the power supply unit fault when one wire gets according to roman something around 10 Amps and the other wire only 2 amps? AFAIK what i remember, roman was talking about the 12V DC wires from the power supply unit to the graphic card

Those tech websites and youtubers should look into power supplies also. How much current is allowed per wire? How is the schematics in the inside for the ASUS SFX loki power supply unit (or wahtever they call that "quality" product)? Opening a psu is not that hard. Than checking from the output connectors the schematics is not that hard.

How is the circuit designed on the power supply unit, that the wires get evenly the amps? (watt = volt * amps)

How is the circuit designed on the graphic card side to evenly distribute the amps for the wires?

My older ATX psu from 486 and such have all those wires soldered together on the printed circuit board. There was no circuitry to even distribute the amps. Note: that is a very old power supply which i use for certain purposes.