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

[Vayra86]

Heavy heavy


Typical US outlet circuit, 110-125VAC, 15AMPs= 1875Watts.

PIE. 125*15

Seems like in two generations of Nvidia GPUs you need to emigrate to Europe to get the required wattage from your wall sockets to run a gaming rig.

We can pull 3500W through a wall socket.

 

[ARF]

So much for user and third-party cable blaming.
First-party MSI cable included with 5090 melted.
Source:

https://www.reddit.com/r/pcmasterrace/comments/1kuf6wu/rtx_5090_power_connector_melted

Alternative source: https://videocardz.com/newz/another-msi-yellow-tipped-power-cable-burns-on-geforce-rtx-5090

Great! It seems things are getting worse, and no one thinks how to solve the problem.

 

[N/A]

Why on earth are they still using this 4x8-pin to 12-pin converter? And we don't get the whole picture. The CPU cooler is close, 35mm of flex not respected.

So much for user and third-party cable blaming.
First-party MSI cable included with 5090 melted.

 

[JustBenching]

Seems like in two generations of Nvidia GPUs you need to emigrate to Europe to get the required wattage from your wall sockets to run a gaming rig.

We can pull 3500W through a wall socket.

Uhm, no. Nvidias gpus that have similar performance to their competition draw less power. If you want even more performance than that then "you have to move to europe" regardless of vendor.

 

[Vayra86]

Uhm, no. Nvidias gpus that have similar performance to their competition draw less power. If you want even more performance than that then "you have to move to europe" regardless of vendor.

I don't see the competition release >350W GPUs yet tho.

Also, it was just a joke, you can stow your camp bullshit right now.

 

[JustBenching]

I don't see the competition release >350W GPUs yet tho.

Because they don't have products at that performance tier. If they did, their gpus would draw even more than nvidias.

EG1. Someone quoting you doesn't neccessarily mean they are talking to you specifically, you already know that nvidia is more efficient (hence you were joking), I was trying to make it obvious to the rest of the readers cause they might get the wrong impression and buy the wrong product for their usecase.

 

[Chomiq]

Interesting solution from Seasonic:

Basically if it detects anything wrong with the current on any of the 6 pins it will shut down the PSU. Plan for Version 2 is to support sending signal to motherboard.

 

[Solaris17]

Interesting solution from Seasonic:

Basically if it detects anything wrong with the current on any of the 6 pins it will shut down the PSU. Plan for Version 2 is to support sending signal to motherboard.

View attachment 401316

Technologically thats pretty sick; kind of wild its needed though.

 

[freeagent]

kind of wild its needed though.

Absolutely.

Pretty rough crash course for everyone involved, from engineering right down to the consumer.

Nothing that a little bit of testing couldn't solve before shooting them out the door though.

Classic case of an over inflated ego calling the shots.

 

[TheinsanegamerN]

Because -- either today or before it -- there has been on "en masse" melting. However it is, like the SWPE described above, yet another example of mass hysteria, with the proximal cause in both cases being a media campaign aimed at the credulous.

No, they were not melting en masse. Reports of 8 pin connectors melting are hard to find and pictures typically showed the connector not being seated correctly.

The 12v2 even if seated correctly will melt because any increase in resistance is resulting in more power being shoved onto the other pins.

If a pin is rated for 12A, it doesn't mean it fails if it carries 12.1A. Any reputable manufacturer designs in (and tests for) a degree of safety factor. What we're seeing now is which cable makers are reputable, and which are not.

Oh you mean like the 8 pin connector had?

It's not just cheap cables melting. Nvidia's own adapters melt too. So do the upgraded MSI connectors. The design just flat out sucks. The safety margin of the 12v2 is WAY too low compared to pretty much any other high power connector used in any other industry. You cant just manufacture the cable to have above spec safety margins.

Why on earth are they still using this 4x8-pin to 12-pin converter? And we don't get the whole picture. The CPU cooler is close, 35mm of flex not respected.

Because there are hundreds of thousands, if not millions, of high wattage PSUs that work perfectly fine but do not have 12v2 native connectors? Believe it or not most people dont dump expensive PSUs every time they upgrade a GPU.

My rosewill platinum 1000w still works perfectly fine 12+ years later.

Technologically thats pretty sick; kind of wild its needed though.

Is it wild though? USB C has had this since it's origin because if you're going to shove 100+w through a small cable, you need to make sure the load is balanced. What's wild to me is WHY ON EARTH the new standard had no hardware mandated to make sure power was actually being split. Once again those multi trillion dollar corpos and all their engineers failed at such a basic task.....

 

[Zyll Goliat]

Source:LINK

 

[LittleBro]

Interesting solution from Seasonic:

Basically if it detects anything wrong with the current on any of the 6 pins it will shut down the PSU. Plan for Version 2 is to support sending signal to motherboard.

Amazing feature. It will be lot of angry gamers in upcoming years, I guess. It's like starting the campaign over and over again, unable to reach saving point, because, well, PSU shuts down. Instead, mighty Seasonic could have come with a real solution that could manage current distribution among wires.

 

[ty_ger]

Amazing feature. It will be lot of angry gamers in upcoming years, I guess. It's like starting the campaign over and over again, unable to reach saving point, because, well, PSU shuts down.

When the connector burns 5 minutes later, you'll miss a save point too.

Instead, mighty Seasonic could have come with a real solution that could manage current distribution among wires.

Not really. Check out der8auer's video I posted a couple days ago. He explains why that doesn't make sense. This is the reason: if one wire has lower current, it's because the pin is making poor contact and contact resistance is higher. If you try to balance by pushing more current down that wire, you'll just burn up that pin faster. The best solution is to provide an alarm when that happens, and shut down the PSU if the current continues to balance worse.

 

[londiste]

All I got to say is - if this was a design issue - we should be seeing a LOT of incidents. Like, hundreds - possibly thousands of them. I can't fathom how it's a design issue (which means it should plague ALL cables) with failure rates at 0.05% or something.

This is probably very simple - the problem really is not a common one. However, if you do run into the problem, you get spectacular results

Is it wild though? USB C has had this since it's origin because if you're going to shove 100+w through a small cable, you need to make sure the load is balanced. What's wild to me is WHY ON EARTH the new standard had no hardware mandated to make sure power was actually being split. Once again those multi trillion dollar corpos and all their engineers failed at such a basic task.....

Can you think of another widely used mainstream standard with 6 or more live pins? Nobody thought of the worst case scenarios where all that power intended for 6 wires ends up going through... like maybe 2

 

[chrcoluk]

I don't see the competition release >350W GPUs yet tho.

Also, it was just a joke, you can stow your camp bullshit right now.

To be fair if I was a GPU maker, I would refuse to release such a product, if a card needs that much juice it shouldnt be a product in my opinion.

 

[Endymio]

To be fair if I was a GPU maker, I would refuse to release such a product, if a card needs that much juice it shouldnt be a product in my opinion.

You remind me of a PC mag editorial in the 1980s. The author was angrily offended by the introduction of PCs with more than 640K of RAM -- not gigabytes or megabytes, but kilobytes. His rationale was, that since any PC program could easily fit within that figure, providing more would simply lead to "sloppy programming".

More than a half-century before that, I read there was a similar sentiment expressed when consumer automobiles began appearing with 50+ horsepower. Engines consuming that much power were wasteful of fuel and indeed dangerous. Some could actually catch fire and explode!

 

[JustBenching]

You remind me of a PC mag editorial in the 1980s. The author was angrily offended by the introduction of PCs with more than 640K of RAM -- not gigabytes or megabytes, but kilobytes. His rationale was, that since any PC program could easily fit within that figure, providing more would simply lead to "sloppy programming".

More than a half-century before that, I read there was a similar sentiment expressed when consumer automobiles began appearing with 50+ horsepower. Engines consuming that much power were wasteful of fuel and indeed dangerous. Some could actually catch fire and explode!

People just confuse power draw with efficiency. The 5090 isnt winning any efficiency awards, but its not bad at it either. Its the 4th most efficient gpu according to the tpu review. I dont know why its such an issue.

 

[freeagent]

down. Instead, mighty Seasonic could have come with a real solution that could manage current distribution among wires.

Looks like you have solved everything! Make it happen captain, get it to production before someone steals your idea.

 

[lilhasselhoffer]

Wait, you actually think that asus implemented (only on it high end super expensive model) a way to measure the cable amperage because they are actually worried about the cables melting or just because they wanted to take advantage of the mass hysteria?


The same principle applies to every facet of the human life. But it always comes down to cost and practicality. The safer something is, the more impractical and expensive it usually is.

The asus rog model is the perfect example. All the naysayers worrying about cables melting, why, there is a model that solves your issue. What, "it's too expensive" I hear you say? Oh well.. That's the point.

I haven't the slightest clue where your absolute crap statements are coming from.

You quoted me saying stuff about milspec not being the same as a viable spec, and talking about how a complex thing made with many parts that each "almost never fail" can rapidly become "there are failures." You then pull from betwixt your nethers crap about ASUS and their attempt to address the problem with unbalanced loading, something that was demonstrable and repeatable, as a thing that is mass hysteria. WTF?


Are you incapable of processing language, or intentionally attempting to have a conversation that I didn't have? I mean, there's changing goal posts, and then there's changing games like you're attempting to pull here. I'd call it dishonest, but that doesn't seem adequate. What I'm seeing here is someone who literally cannot win the argument with an honest statement, so they decide to have a parallel argument with someone that...is whatever strawman they feel like.

For the record, I believe line level monitoring is a solution. It's the solution that has you saving pennies on a connector and spending dollars on a solution, but it is by definition a solution. I would also suggest that the two better solutions are going back to 6+2 connectors or using a more robust fastening system...but you can definitely solve your problem by ignoring the root cause and dealing with all the symptoms. I'm sure you wanted less nuance there...but my opinion stands on facts. The best solution is a less crap specification, one that doesn't require it to be milspec, doesn't require a special piece of electronic monitoring, and most definitely doesn't start the conversation with ASSUMING it's 100% user error as the opposing side stated from the word go. Of course, you're welcome to defeat as many strawmen as you want, but the simple truth is that Nvidia precipitated this to save money. When a car company does something like this there's a recall, but people want to pretend Nvidia should be immune. Cool. I know what your sacred cow is.

 

[Endymio]

...mighty Seasonic could have come with a real solution that could manage current distribution among wires.

Looks like you have solved everything! Make it happen captain, get it to production before someone steals your idea.

His idea is even more revolutionary than you think. Stranded copper wire is nothing more than a bunch of independent current-carrying strands. Why not monitor the load on each individual strand? That would be even safer and more robust still!

 

[danc]

View attachment 401328
Source:LINK

The bending is all over the place. I think this case is more user error.

 

[LittleBro]

When the connector burns 5 minutes later, you'll miss a save point too.

Not really. Check out der8auer's video I posted a couple days ago. He explains why that doesn't make sense. This is the reason: if one wire has lower current, it's because the pin is making poor contact and contact resistance is higher. If you try to balance by pushing more current down that wire, you'll just burn up that pin faster. The best solution is to provide an alarm when that happens, and shut down the PSU if the current continues to balance worse.

Makes sense. Then best and only solution would be ditching this connector for good and bring new one, a properly made one. Solution to this problem is not monitoring, PSU shut downs or any kind of warning system. Solution is to solve root cause of the problem, which is insufficient robustness of connector. Solution is something that eliminates the problem, not mitigates or asks user for monitoring stuff.

 

[_roman_]

Soldering everything in the power supply unit is not the best. The power supply manufacturer has to think before he sells something. This was not the problem before because of the 150Watt limitation afaik.

120€ for a power supply unit but stupid design which do not use a proper circuit design to technically limit the output power. 600 Watts are 600 Watts. A bad power supply unit design can not be blamed on the products connected after the power supply unit. Regardless if the graphic card board design is bad or not.

Also the cables and cable connectors usually are shipped with any power supply unit. It's the fault for those cheap garbage power supply units. What is so hard to make for each 12 V DC output single cable a separate circuit in the power supply unit which monitor and restricts the current to the allowed one? Nothing. Simple easy. Including temperature sensors in the cable is also not that hard.

Designing a power supply unit seems to be something new. It was never done before as it seems. All those devices work with the white or black smoke in the first place.

 

[LittleBro]

His idea is even more revolutionary than you think. Stranded copper wire is nothing more than a bunch of independent current-carrying strands. Why not monitor the load on each individual strand? That would be even safer and more robust still!

Strands need to be isolated if you want to measure different values between them. Had you known this, maybe you'd have formulated your joke/insult differently after moderator's encouraging response.

 

[micropage7]

at some points something cross my mind why don't they use cable like this one

the current connector whether from 3rd party or stock hardly handle the heat and the current