News RTX 5090 cable overheats to 150 degrees Celsius — Uneven current distribution likely the culprit

[InvalidError]

Funny you mentioned XT-60's ... I've used those before in various applications. They are certainly better than current garbage connectors used in PCs, BUT they are solder only (no crimping) ... which is fine, but it does require quality soldering.

Soldering them is easy: tin wires, drop a piece of solder wire (I need 10mm when I do EC5s) in the cup, heat the cup, push the wire in, wait for the wire to drop in all the way when the solder wire melts, keep the heat on for 3-5 more seconds for the tinned wire's solder to melt, remove heat and wait for the pin to cool. A production line should have little trouble handling this process, though it would be much slower than crimping.

A crimpable version of XT-60 would be simple enough to make: extend the solder cup walls enough to accommodate space for a crimp die and a transition zone from the crip side to the connector side. The crimp wall will also need to be thicker to hold crimping force. That would make the connector at least 1/2" longer and possibly slightly thicker/wider.

 

[InvalidError]

Now the issue we have on the melting gate 2.0 is that, it is NOT visual inspection-able

Temperatures are very easy to check. At the very low end of the tech ladder, you can check with fingers. You can use contact thermometers, thermochromic indicators, IR thermometers or even an IR camera.

No shortage of inspection methods ranging from $free to $expensive.

 

[YSCCC]

Temperatures are very easy to check. At the very low end of the tech ladder, you can check with fingers. You can use contact thermometers, thermochromic indicators, IR thermometers or even an IR camera.

No shortage of inspection methods ranging from $free to $expensive.

O, yeah, you're right, it is literally the user's responsibility to use a cable rated to be reused a few dozen times to go to do thermal checking for their brand new graphics card costing a car, very reasonable approach and you can't even visually see if it's going bad, you need to torture test it and see if it heats up, and what's the dangerous temp? 50C? 60C? higher on the cable? It makes perfect sense.

 

[TJ Hooker]

We got another one..

https://www.laptopmag.com/laptops/gaming-laptops-pcs/another-melted-cable-nvidia-rtx-50-series

Looks like Toms was a little quicker than laptopmag, there's already an article here on that issue from a few days ago. https://www.tomshardware.com/pc-com...su-redditor-reports-another-50-series-failure

 

[danny009]

Oh no! My overpriced overhyped RTX 9999 cable melts!

Anyway.

 

[V8VENOM]

A lot of scalpers are worried ... hence a lot of posts in defense of nVidia across this thing we call WWW ... and those that sell "Buy BOTs" ... it does make one chuckle.

There is no way 23+ Amps should be going thru one 18-16 AWG wire ... for nVidia to defend this as "normal" and it's ok if it only melts the wire and doesn't cause a fire is beyond pathetic. We should be holding companies like nVidia to a MUCH higher standard of engineering, ethics, and responsibility.

It's a connector spec by PCI-SIG, yeah ok, but nVidia provided the details of what they need to PCI-SIG (because NO ONE else is using this spec'd connector for their GPUs other than nVidia) and I'm sure nothing was mention of 23A thru a single 18-16 AWG wire.

 

[V8VENOM]

The crimp wall will also need to be thicker to hold crimping force. That would make the connector at least 1/2" longer and possibly slightly thicker/wider.

Yep, and that would a problem for tight fitting cases. There needs to be 90 degree, 180 degree, and even 45 degree connectors (possible 60 degree) to avoid wire strain. The specification NEEDS a serious update from what we've been living with from 40+ years. But NO ONE wants to do it because of compatibility and cost ... which is ridiculous when you're talking safety. The company I work for iRating is everything, our insurance costs skyrocket if our iRating (safety rating) goes up due to incidents. nVidia should be held to the same standards.

But some person that lives pay-check to pay-check will be complaining to no end if the price of motherboards, GPUs, PSUs goes up $15-$20 ... doom themselves to failure (probably the same people trying to scalp these GPUs rather than actually work for a living).

 

[InvalidError]

O, yeah, you're right, it is literally the user's responsibility to use a cable rated to be reused a few dozen times to go to do thermal checking for their brand new graphics card costing a car, very reasonable approach and you can't even visually see if it's going bad, you need to torture test it and see if it heats up, and what's the dangerous temp? 50C? 60C? higher on the cable? It makes perfect sense.

We are supposed to regularly inspect power cords to catch wear, damage and other issues before it becomes a problem too. Almost nobody does that.

As for what the safe temperature is, MiniFitJr connectors are spec'd for three years at 65C. So if you want your cables to last for at least three years, you have to stay under that.

 

[InvalidError]

Yep, and that would a problem for tight fitting cases. There needs to be 90 degree, 180 degree, and even 45 degree connectors (possible 60 degree) to avoid wire strain.

A better option is standardized connector placement and orientation so there doesn't need to be 50 connector variants in the first place.

 

[YSCCC]

We are supposed to regularly inspect power cords to catch wear, damage and other issues before it becomes a problem too. Almost nobody does that.

As for what the safe temperature is, MiniFitJr connectors are spec'd for three years at 65C. So if you want your cables to last for at least three years, you have to stay under that.

And we're supposed to clean dust accumulations for HS or other stuffs regularly, which, should involve temporary remove the power cord and then re-plug it in, now you're suggesting we should do that and replace the whole bloody cable every time.

FYI a sensible piece of electronics will stop draw power and have sensing circuitry to warn you for replacement, power cords have fuses for exactly that purpose to protect the whole thing and not bam, your gear metled. Stop suggesting unrealistic routines/workaround for defending a flawed design as "Fine"

 

[InvalidError]

FYI a sensible piece of electronics will stop draw power and have sensing circuitry to warn you for replacement, power cords have fuses for exactly that purpose to protect the whole thing and not bam, your gear metled.

Fuses do not protect against loose or broken connections, frayed cables and other "soft" faults that can produce sparks, heat and eventually cause fire. Take a lamp cord, strip the end, splay out the strands, plug it in and gently brush the strands across each other. Lots of noise and sparks but no trip on regular breakers and no blown fuses either. That is why arc-fault circuit breakers were invented.

Little to no hardware actually monitors incoming AC for power cord anomalies. They will pull whatever current they can until the connection gets so bad that they max out. Only then will loads like PCs lose power due to the PSU going into under-voltage lock-out or boot-loop from the PSU no longer being able to pull enough current to run. The PSU isn't doing any sort of cord quality check. You don't want your PC randomly shutting down because the PSU tried to be clever and misinterpreted a power glitch or transfer on/off of a UPS as a cable failure.

 

[V8VENOM]

A better option is standardized connector placement and orientation so there doesn't need to be 50 connector variants in the first place.

That would be nice but typically avoided due to interference of higher current vs. lower current ... especially given the fine line between too much and too little voltage/current for CPU, RAM, etc.

 

[YSCCC]

Fuses do not protect against loose or broken connections, frayed cables and other "soft" faults that can produce sparks, heat and eventually cause fire. Take a lamp cord, strip the end, splay out the strands, plug it in and gently brush the strands across each other. Lots of noise and sparks but no trip on regular breakers and no blown fuses either. That is why arc-fault circuit breakers were invented.

Little to no hardware actually monitors incoming AC for power cord anomalies. They will pull whatever current they can until the connection gets so bad that they max out. Only then will loads like PCs lose power due to the PSU going into under-voltage lock-out or boot-loop from the PSU no longer being able to pull enough current to run. The PSU isn't doing any sort of cord quality check. You don't want your PC randomly shutting down because the PSU tried to be clever and misinterpreted a power glitch or transfer on/off of a UPS as a cable failure.

Again, mixing concepts of extreme cases of arching electric fires with normal sensible use case. The case you mentioned are, VERY visually checkable for any of the connectors, while the 12VHPWR isn't. you are literally suggesting ppl should thermal image the card once a while with hand or imaging camera, so we should do that to every power cord in home appliances and gas pipes if that makes sense, or else your home bursting into flames are our own faults.

There are reasonable things to expect users to do it properly and STATED in the manual (like literally "DO NOT USE THE PLUG IF WIRES ARE STRIPPED in basically every piece of electronics in home appliances), and there are EXCUSES of not properly designing something to have enough headroom in plug cycles or ways to visually inspect if anything going wrong, if it can't be done visually in case shown in De8auer's video, then you have to get protection circuitry to protect the expensive piece of product.

Did Nvidia do any of those? I bet they will dare say you should just throw away a 5 times used cable and buy a new when the connector is rated to be used for 30 cycles, nor they will dare put in the manual that you should run furmark and thermal image/touch the cable to check for temperature.

 

[Peksha]

Fuses do not protect against loose or broken connections, frayed cables and other "soft" faults that can produce sparks, heat and eventually cause fire. Take a lamp cord, strip the end, splay out the strands, plug it in and gently brush the strands across each other. Lots of noise and sparks but no trip on regular breakers and no blown fuses either. That is why arc-fault circuit breakers were invented.

Hahaha, more unhealthy suggestions?

We are still waiting for you to stop spreading FUD in this thread and show how you fearlessly left your 5090, sent to you as a gift from NV, to render all night under full load without supervision.

We can even enhance the effect - I will also do this experiment by disconnecting two of the three 8pin cables from the GPU. You will cut off (with all necessary safety) 4 of the 6 +12V power cables on your connector😉

 

[V8VENOM]

The reality is any 5090 owner is ROLLING the dice and they know it … the ONLY reason a few are defending this disaster of a 5090 GPU is because they want to scalp it to some poor sap that doesn’t know any better. The 5090 issues just keep piling up and nVidia’s only response is the same response as before, we’re investigating … in other words their exhaustive (joke) investigation will reveal … drum roll … user error.

 

[palladin9479]

Cable and PSU manufacturers have repeatedly demonstrated that their "perfect" HPWR cables can pass 1000+W fine. 600W isn't supposed to be remotely problematic unless something got damaged or has a manufacturing defect.

As already described, 12V HPWR can take 600W sustained tops, and that's assuming near perfect conditions. The wires and pins are simply too small for the amount of power that is trying to be pushed through them, toping out at 8.5 to 9.5 amps per pin. The eight wire PCIe connector has similar per-wire ratings but we use a ton more wires.

https://www.reddit.com/r/pcmasterrace/comments/1io4a67/an_electrical_engineers_take_on_12vhpwr_and/

In a 600W 12VHPWR connector, each pin is being asked for 8.33A already. If you have 8.5A pins, there is functionally no headroom here, and if you have 9.5A pins, yeah that's not great either.

The 12V HPWR standard is good for up to mid range cards, anything above really should be using two of them. With something like a 5090, there is functionally zero safety margin and the moment a connection has issues, bad things will start to happen. This has been demonstrated already, both in real world and testing scenarios.

 

[V8VENOM]

nVidia WILL see their day in court, don’t mess around with safety … doesn’t matter how large nVidia’s legal team is, they will end up paying for this issue, gone on too long and getting worse.