be quiet! Dark Power PSU Owner Reports Melting of 12VHPWR Connector

[kawice]

It dont matter if they sold 100.000 PSU's and only one report, or 10 reports. One is too many knowing it can set your house on fire.

These things are extremely sensitive.

It does matter every device has a failure rate and return rate.

Even a car can randomly be set on fire or even explode (the electric ones) it doesn't mean we should jump to conclusions and stop using all of them.

If there was real or high fire hazard I'm sure proper safety certification organization wound ban selling those PSUs. An bequit or any other company would have to recall them.

I have a 4070ti that I put super glue on the power connector. Sure its not as hot as 600w 4090 but the sudden connector power melting drama got me scared that time.

If this is not a joke you actually increased fire hazard since super glue is flammable.

I really don't get why we do not get lower power rated cables with those PSUs or GPUs. I mean only 4090 actually needs 600W cable.

Other cards could use 450W or 300W cables and therefore power would be limited by PSU itself and GPU driver.

Lower power, lower electric current, lower heat on connector, lower fire hazard.

WTH happened with this idea?

 

[Bomby569]

It does matter every device has a failure rate and return rate.

Even a car can randomly be set on fire or even explode (the electric ones) it doesn't mean we should jump to conclusions and stop using all of them.

If there was real or high fire hazard I'm sure proper safety certification organization wound ban selling those PSUs. An bequit or any other company would have to recall them.

No, you know why electric cars immolate themselves, there is always a risk and the causes are investigated, failures happen on any part.
But if a single cable was the cause of several fires i can guarantee you that cable you'd be discontinued, recalled and the matter would be investigated. And if the company knew and continue to sell them them were in for a bad time, including jail time.

 

[remunramu]

GG, now you can't pull it out when it does melt

I hope not. I think I still can be pull it out with a slight force. I'm not that stupid enough to pour a lot of glue on it. Just a tip in corners to prevent them being push out by itself.

If this is not a joke you actually increased fire hazard since super glue is flammable.

No joke I was really scared since its the most expensive GPU I ever bought. I think its fine I didn't pour all over it. Just a tip in corners.

 

[STSMiner]

WTH happened with this idea?

View attachment 301336

That idea is used, but also abused, only need to ground the sense pins GPU side to trick (allow) the full wattage (600 watts), the GPU will pull what ever it needs at that point up to it's max limit from any PSU, as in the case for the RTX 4090 FE - thats 600 watts as defined in the BIOS, as for all other cards that's dependent on the limits set within the BIOS for the GPU as well, but I've seen GPU's pull over that max limit in some cases as well.

Myth Busted: This is how NVIDIA’s 4-way adapter for the 12VHPWR port of the GeForce RTX 4090 really works! | igor´sLAB

Who doesn’t know it, the funny adapter from NVIDIA, with which you can merge four PCIe 6+2 pin cables to be able to run graphics cards with up to 600 watts? If you only connect three plugs…

www.igorslab.de

As for the pure 12VHPWER cable designed for / come with ATX 3.0 PSU's that actually have the 12VHPWER connector, that will also depend on how those sense pins are wired to both connectors (GPU side to PSU side), most will be wired to allow full wattage (600 watts) as these PSU's are supposed to have the logic circuit built into them.

 

[DarkChib]

Yeah I get it that its likely consumer error but if the damn design is soooooo crappp that this can happen...

Well its a shockingly bad design that these events can happen from just being 1 or 2 millimetres not homed into the socket

Cant say I have ever had any issues with other prior lead designs like they are having with this one in the near 29 years of building and maintain PCs personally and professionally.

 

[pavle]

Too little connector, too thin metal parts for 600W; they heat up, how much - depends on design or being plugged in properly. Not worthy of being part of any standard really.

 

[R0H1T]

You don't have so many issues with heaters(radiators?) consuming 2kw+ in homes! Nvidia should provide an additional "kit" for fireproofing with everyone of these

 

[PapaTaipei]

This problem highlighted something I couldn't belive. No youtubers and no websites are telling the truth while the reason is known to even a 10yo with basic electricity knowledge:
THERE IS MORE POWER GOING THROUGH SMALLER/LESS CABLES=MORE HEAT=NEEDS MUCH MORE PRESSURE TO HAVE A TIGHT CONNECTION, AND MORE MEETING SURFACE.

 

[the_finn]

I have a Be Quiet atx3.0 PSU and in the manual it says "Do not bend horizontally" and "do NOT bend less than 40mm" when using the PCIe 5.0 cable... Just saying....

 

[caroline!]

You don't have so many issues with heaters(radiators?) consuming 2kw+ in homes! Nvidia should provide an additional "kit" for fireproofing with everyone of these

free extinguisher with the purchase of any 4000-series graphics card!!!

 

[RogueSix]

Not surprised at all that this is happening with a beQuiet PSU. That company produces nothing but garbage and is one of the most overrated gigs in hardware land.

 

[evernessince]

I am furthermore unconvinced that this is widespread or even a cause for concern, statistically speaking.

Widespread no. Cause for concern? Yes, undoubtedly everyone who buys a card with this connector should be ensuring that they seat it fully as it is known to be hard to plug in all the way. By definition it is something of concern people need to take in mind to ensure special care. The current statistics already include the fact the people who are purchasing these cards are already taking precautions to ensure, more so than other connectors, that it's fully plugged in. Any statistics already have have a level of concern for this connector's unique issue baked into them.

In addition, we still don't know how many of these instances we'll see when the market fully saturates with this connector and how many of them fail as their wear increases. As of right now the current failure numbers are a best case scenario for this connector. People are taking pre-cautions to ensure it's fully connected because it's mostly enthusiasts that purchase the card and these connectors are brand new with no wear. Obviously the total number of burned up connectors will increase as saturation increases but it'll be interesting to see if the rate increase as wear and less savvy users get their hands on the cards.

I have a Be Quiet atx3.0 PSU and in the manual it says "Do not bend horizontally" and "do NOT bend less than 40mm" when using the PCIe 5.0 cable... Just saying....

It's a known issue with the connector due to the way it's designed. Unfortunately a lot of people are not going to care about that or don't have a choice due to case restrictions. Not sure if bending alone can cause the connector to burn up but then again the PSU should be robust as it typically lasts multiple platforms.

Based on the tests Igorslab and GamersNexus did, yes, I believe so with the cables being AWG16.

Yep, the cables aren't the issue. It's litteraly just the pin design and locking mechanism that are the problem here. Intel has suggested a version with an updated pin design but they should also improve the locking mechanism as some users have reported the cable backing out over time even when they fully clicked it in.

 

[caroline!]

#61

Bad connector is bad, such wow. Sometimes size does matter, oh and pins should be round and solid instead of square and hollow.

 

[AGlezB]

Do Intel Arc GPUs even use the 12VHPWR?
A case of Intel saying to nVidia "here, take this and run" and nVidia dying on the 12VHPWR hill instead of Intel?
This might do for a funy SNL comedy sketch.

 

[STSMiner]

Do Intel Arc GPUs even use the 12VHPWR?

Yes, but not on their consumer class cards right now, but that could change with their next gen card, just the data center side for now (see article below for info on that).

Intel Data Center Max GPU "Ponte Vecchio" Implements 16-pin 12VHPWR Connector

The swanky new Intel Data center Max GPU "Ponte Vecchio" is the company's first product to implement the 12+4 pin ATX 12VHPWR power connector, which the company helped design as part of the ATX 3.0 spec. The PCI-Express add-in card (AIC) form-factor variant of the GPU comes with a single 12VHPWR...

www.techpowerup.com

 

[LabRat 891]

Can't wait for 'adapters' to hit the market that plug into the 12VHPWR, provide 'battery terminals' and include 0000 AWG copper wire leads.

 

[ARF]

I think the most normal thing is that nvidia gets sued. And the court investigates the issue, and sets a multi-billion fine, and bans the said connector. Which is anything but connector. It's a crap..

 

[Sora]

THERE IS MORE POWER GOING THROUGH SMALLER/LESS CABLES=MORE HEAT=NEEDS MUCH MORE PRESSURE TO HAVE A TIGHT CONNECTION, AND MORE MEETING SURFACE.

No.

I think the most normal thing is that nvidia gets sued. And the court investigates the issue, and sets a multi-billion fine, and bans the said connector. Which is anything but connector. It's a crap..

This only happens in lala land where you live, since lalaland is devoid of fact or reality.

 

[ARF]

This only happens in lala land where you live, since lalaland is devoid of fact or reality.

What is your "fact" ?

If you, by any sad chance, have got a "diploma" in Electrical engineering, you must throw it in the trash. NOW!

 

[Godrilla]

Widespread no. Cause for concern? Yes, undoubtedly everyone who buys a card with this connector should be ensuring that they seat it fully as it is known to be hard to plug in all the way. By definition it is something of concern people need to take in mind to ensure special care. The current statistics already include the fact the people who are purchasing these cards are already taking precautions to ensure, more so than other connectors, that it's fully plugged in. Any statistics already have have a level of concern for this connector's unique issue baked into them.

In addition, we still don't know how many of these instances we'll see when the market fully saturates with this connector and how many of them fail as their wear increases. As of right now the current failure numbers are a best case scenario for this connector. People are taking pre-cautions to ensure it's fully connected because it's mostly enthusiasts that purchase the card and these connectors are brand new with no wear. Obviously the total number of burned up connectors will increase as saturation increases but it'll be interesting to see if the rate increase as wear and less savvy users get their hands on the cards.



It's a known issue with the connector due to the way it's designed. Unfortunately a lot of people are not going to care about that or don't have a choice due to case restrictions. Not sure if bending alone can cause the connector to burn up but then again the PSU should be robust as it typically lasts multiple platforms.



Yep, the cables aren't the issue. It's litteraly just the pin design and locking mechanism that are the problem here. Intel has suggested a version with an updated pin design but they should also improve the locking mechanism as some users have reported the cable backing out over time even when they fully clicked it in.

If the cable is the issue then how did Cablemod drop the ball and have 20 reported cables burned on record?
The only fix I see is to have a cable with thermal sensors on both sides with color coded and or digital user feedback etc etc. Thoughts?

 

[Voltaj .45 ACP]

no suprise here. never accept this 5.0 standart anyway. 2019 seasonic gold FTW and phanteks version of course. using 2 focus an a amp! 750w for years. 600w with single cable? yeah sure brilliant!

 

[STSMiner]

There are plenty of reports on the internet with this melting issue with the 12VHPWR connector, and if you delve in deep enough you will see that it's mostly related around the version of female pin with dimples that is used, CableMod, a 3rd party that makes cables and special connectors are not the only ones that have made the "mistake" with using the wrong female pins in the connectors for the 12VHPWER products, below is an image of their right angle adapter, it's using the female pins with dimples as does their other cables with the 12VHPWER connector (not sure if they changed over to the correct pins here).

If the cable is the issue then how did Cablemod drop the ball and have 20 reported cables burned on record?
The only fix I see is to have a cable with thermal sensors on both sides with color coded and or digital user feedback etc etc. Thoughts?

Because they also screwed up and used the wrong pins for their cables and adapters, you can see the pins are the ones that have dimples in the image below and elsewhere on the internet if you look for them.

These pins with the dimples (ASTRON) are not up to the job and should not be used in the connectors, they don't provide enough surface area for power to flow with those dimples for 600 watts.

The image below shows the female pin from ASTRON, they are not ideal for the job, they are weak due to the two slots in it for a start vs the NTK female pin.

The NTK pins allow for more surface contact on the male pin when they are connected together and allow better power flow and reduce the heat, the pin does not have the two open slots to weaken it either (splay apart), I have not seen any GPU with melted connector using these pins (RTX30x0 FE, RTX3090 Ti, RTX 4090 FE, RTX 4080 FE).

I think the most normal thing is that nvidia gets sued. And the court investigates the issue, and sets a multi-billion fine, and bans the said connector. Which is anything but connector. It's a crap..

This is not going to happen, it's not an Nvidia issue, they did not design this power connector standard.

 

[evernessince]

If the cable is the issue then how did Cablemod drop the ball and have 20 reported cables burned on record?
The only fix I see is to have a cable with thermal sensors on both sides with color coded and or digital user feedback etc etc. Thoughts?

Did you read my comment? I specifically said the cable WAS NOT the issue.

Yep, the cables aren't the issue. It's literally just the pin design and locking mechanism that are the problem here.

There are multiple instances of me pointing out that the connector is the source of the issues throughout my comment. I even specifically point out the pin design.

 

[ARF]

This is not going to happen, it's not an Nvidia issue, they did not design this power connector standard.

They must take the responsibility for using it. It is very unsafe, and could lead to your PC components on fire.

AMD is wiser. AMD doesn't use it. Do you know why?

 

[STSMiner]

They must take the responsibility for using it. It is very unsafe, and could lead to your PC components on fire.

AMD is wiser. AMD doesn't use it. Do you know why?

It's not an Nvidia issue, the 3rd parties are not following the specs / guidelines for this connector, and Nvidia did not design it, Intel designed it.