Nvidia RTX 4090 power connectors melting?

If all of the 12v pins are connected together anyway, then what is the point of using 12 pins (not counting the 4 sense pins)? Why not just two large robust connectors, one positive and one negative? That's basically how every other 12v device in the world works. Something like this:

12awg_banana.jpg

Brings to mind the old saying: "Keep it simple, stupid"
 
Or on the side facing the front of the case. Literally anywhere is better than what the standard has been.

Cards are already too long for many cases. Putting the connector facing case front would extend the overall length even further
 


His whole "test" was a joke. It was all done in open air with plenty of airflow, of course he won't come across problems.

There's a reason that power testing of wire gauges have two standards, one in open air and one in a loom/cable assembly. You have drastically less heat rejection capability in a wire surrounded by other hot wires. You further have even less in a hot box case that has its ability to remove heat compromised by a huge brick effectively cutting the case in half, unless you have a monster case like a Corsair 1000D.

If all of the 12v pins are connected together anyway, then what is the point of using 12 pins (not counting the 4 sense pins)? Why not just two large robust connectors, one positive and one negative? That's basically how every other 12v device in the world works. Something like this:

View attachment 522001

Brings to mind the old saying: "Keep it simple, stupid"

Large gauge wires with single plugs are just as susceptible to failure as cable assemblies with more connectors. The connector itself is the failure point, any little bit of oxidation, dust or metal imperfection that starts the heat generation process can cause the whole thing to go up in flames. Seen it happen on all sorts of high power equipment, AC and DC alike. The crimp on the wire can also degrade over time, especially in more humid environments. Oxygen gets in between the wire strands and oxidizes the crimp from the inside out.
 
I really hope some AIBs planned on using regular 8 PINs for the lower end cards. If Nvidia allowed them. I see why they wanted to move to a single connector from 3-4, but for lower end cards it wasn't an issue as power draw would be lower and I am assuming "3080" 12GB and lower could get away with two 8 PIN.
 
Large gauge wires with single plugs are just as susceptible to failure as cable assemblies with more connectors. The connector itself is the failure point, any little bit of oxidation, dust or metal imperfection that starts the heat generation process can cause the whole thing to go up in flames. Seen it happen on all sorts of high power equipment, AC and DC alike. The crimp on the wire can also degrade over time, especially in more humid environments. Oxygen gets in between the wire strands and oxidizes the crimp from the inside out.

I disagree. With a single large connector you aren't having small individual parts breaking off, such as what is shown in the video above. A single large connector is not going to have issues if you bend the cable a bit either. And I couldn't help but chuckle a bit when you talked about oxidation, dust, humidity, etc. The connectors that I posted a picture of are actually the connectors that I use on my Traxxas RC Trucks. I drive these trucks through dirt, dust, mud puddles, sand, salt-water (beach), rain, snow, pretty much every harsh environment imaginable, and obviously worse than what the interior of any PC would ever encounter. In 15 years I've never had a connector fail. These trucks use large LiPo batteries that can discharge thousands of watts.
 
All of this makes the ending of this video from 2021 even funnier, and quite foretelling...



The top comment from Captains Workspace:
2022 Update: Jokes on me... Freaking GPU sticks as standard accessory, couldn't have made that up! :O What a time to be alive.
What a time to be alive, indeed. :whistle:
 
So far it looks like the adapter is just bad news. ATX 3.0 PSU is the way to go. I won't be gaming on my 4090 until I get an ATX 3.0 PSU. Simple as that.
 
The more I read about these connectors the more they sound like they're a bad design all around but those adapters are ridiculously bad and anyone involved in choosing or okaying those particular adapters shouldn't have a job making those sorts of decisions. I'd really like to see them either go back to the drawing board and create a more robust connector or just go back to multiple PEG connectors.

This is the third straight Nvidia series with hardware issues emerging shortly after launch on their high end cards, WTF is going on over there.
 
So far it looks like the adapter is just bad news. ATX 3.0 PSU is the way to go. I won't be gaming on my 4090 until I get an ATX 3.0 PSU. Simple as that.
Problem is finding one now, and who knows when more will hit the shelves.
 
Ya looks like the issue is really just Nvidias shoddy adapter and its build quality.
I wonder though with a card this size... would it have been so crazy to have just added ALL the connectors. I get it adding 4 connectors is ugly... but really add a removable flap on the top that reveals the old 8 pins. If the card detects power from a rear mounted 12VHPWR don't accept power from the others. I mean they would have had to spend a few more pennies on connectors. But it didn't have to be ugly if they hid them.... and this would have solved adapter issues. Power adapters are always bad news... but when you are talking about 450+ watts just add both options to the cards.
 
I disagree. With a single large connector you aren't having small individual parts breaking off, such as what is shown in the video above. A single large connector is not going to have issues if you bend the cable a bit either. And I couldn't help but chuckle a bit when you talked about oxidation, dust, humidity, etc. The connectors that I posted a picture of are actually the connectors that I use on my Traxxas RC Trucks. I drive these trucks through dirt, dust, mud puddles, sand, salt-water (beach), rain, snow, pretty much every harsh environment imaginable, and obviously worse than what the interior of any PC would ever encounter. In 15 years I've never had a connector fail. These trucks use large LiPo batteries that can discharge thousands of watts.

Unless the PCBs inside that thing are encased in resin or some other water impermeable barrier, and the connectors don't see any liquid, I call complete BS. There's absolutely no way you're pushing thousands of watts though a motor or PCB that's had liquid ingress, especially salty liquid. That thing will corrode in a matter of seconds.

I've worked in the AV world, and film projection world. We have cables carrying low voltage high amperage DC. Xenon arc lamps are DC devices that operate in the thousands of watts envelope, and I've seen those have spectacular failures with very beefy connectors because of the issues I cited earlier.
 
Long story short, don't use Nvidia's shiite. Get a cable from your PSU manufacturer.
 
The gb doesn't meet atx 3.0 specs and yeah, they build/sell junk.

The MSI is a really solid unit but isn't available until the end of November. There's a review at TPU.

The TT might be decent. I've heard their toughpower units are the only ones worth buying. I haven't seen any reviews yet.
So a PSU can have a 12VHPWR cable and not be ATX 3.0 compliant?
 
or new 24v standard which would decrease the current by 1/2 and thus heat. It is not the power the cabling is handling that heats up the plug but how many amps is going through it. The wire, pins, sockets all have resistance which restricts current flow and converts electrical energy into heat. The more current flow, the more heat generated.
Power is directly proportional to amps, P = I x E, so yes, it is a power issue.
 
After the original link/story was posted I was counting down until we got the LTT/JZ OUTRAGE video.

Alas, a rift may be forming between the tech-misinformation-generators like JayZ and MLID, and the review operations that still care at least a little about their word and their reviews meaning something, the ones still seeming interested in getting to the truth rather than just sensationalizing.

HardwareUnboxed taking Jayz to task

1666929048482.png


1666929076628.png


1666929132551.png
 
Last edited:
Power is directly proportional to amps, P = I x E, so yes, it is a power issue.
Yes but because heat is current square, usually the more tension with the less current you have to transport power to less heat issue.

That why there some transmission line when they use DC try to go for super high voltage for little has possible actual current moving around

When moving large volumes of electricity, it’s better to increase voltage instead of current intensity (amperage), in order to reduce energy losses and limit the total cost of transmission (building additional power lines, for example). A large portion of the power generated by Hydro-Québec is transmitted using 735-kV lines.

Those 735,000 volt line can send 2,000 millions of watts with only 2700 amps.
 
Yes but because heat is current square, usually the more tension with the less current you have to transport power to less heat issue.

That why there some transmission line when they use DC try to go for super high voltage for little has possible actual current moving around

When moving large volumes of electricity, it’s better to increase voltage instead of current intensity (amperage), in order to reduce energy losses and limit the total cost of transmission (building additional power lines, for example). A large portion of the power generated by Hydro-Québec is transmitted using 735-kV lines.

Those 735,000 volt line can send 2,000 millions of watts with only 2700 amps.
Correct, P = I x E so if you want the same power and reducing heat by lowering current you can increase the voltage. There is another way to reduce heat. P = I^2 x R, so by reducing resistance by making the pins larger you can reduce heat without changing voltage or current.
 
  • Like
Reactions: noko
like this
Buildzoid weighs in on this frilly fracas - I trust him more than probably any techtuber since he doesn't clickbait or sensationalize, or is even interested in his subscriber count.

 
Unless the PCBs inside that thing are encased in resin or some other water impermeable barrier, and the connectors don't see any liquid, I call complete BS. There's absolutely no way you're pushing thousands of watts though a motor or PCB that's had liquid ingress, especially salty liquid. That thing will corrode in a matter of seconds.

Yes the electronics are waterproof; the PCB in the electronic speed control (ESC) is encased in resin, the receiver sits within a small waterproof box, and the brushless motor is fully enclosed. They are designed to be used in wet environments. But the connectors between the battery and the speed control have no special water-proofing other than the basic protection that comes from the male connector being fully inserted into the female connector. The connectors are still subjected to considerably more environmental exposure than they ever would be inside a computer case.

Another cool thing that the speed control can do is log data... Is 235 Amps a lot?

castlelog.jpg
 
This would the Corsair 1200watt power supply making love songs called Total Bull Shit on Tour to the power connection of RTX 4090 FE = A lab a ma means to be Man made and huge mistake!

ttps://youtu.be/J2Q0gHA9vXY



 
The "say anything to generate traffic" ecosystem begins to self-pleasure violently the moment anything with controversy potential hits their nose, and when no controversy exists they will find it necessary to amplify a non-troversy, or just outright invent it. Guaranteed that if MLID or JayzTwoBraincells ever found themselves in a courtroom for their FUD impacting a corporation financially, they'd take the Alex Jones "hey man it was just entertainment" defense.

But it does seem like a rift is forming between the tech-misinformation brigade, and the operations that still care at least a little about their word and their reviews meaning something, the ones still seeming interested in getting to the truth rather than just sensationalizing.

HardwareUnboxed taking Jayz to task

View attachment 522094

View attachment 522095

View attachment 522096
I used to enjoy watching Jay a long time ago, but he has gotten progressively worse as time goes on. Glad to see that someone else has seen through his ridiculousness.
 
All of this makes the ending of this video from 2021 even funnier, and quite foretelling...



The top comment from Captains Workspace:

What a time to be alive, indeed. :whistle:

so glad this video made it back into my life lolol thank you
 
The "say anything to generate traffic" ecosystem begins to self-pleasure violently the moment anything with controversy potential hits their nose, and when no controversy exists they will find it necessary to amplify a non-troversy, or just outright invent it. Guaranteed that if MLID or JayzTwoBraincells ever found themselves in a courtroom for their FUD impacting a corporation financially, they'd take the Alex Jones "hey man it was just entertainment" defense.

But it does seem like a rift is forming between the tech-misinformation brigade, and the operations that still care at least a little about their word and their reviews meaning something, the ones still seeming interested in getting to the truth rather than just sensationalizing.

HardwareUnboxed taking Jayz to task

View attachment 522094

View attachment 522095

View attachment 522096
Meh they all run the same game. Although I do agree that Jay should do more research sometimes before speaking.
 
I used to enjoy watching Jay a long time ago, but he has gotten progressively worse as time goes on. Glad to see that someone else has seen through his ridiculousness.

the problem is they get big, as in big subscriber counts. they then need to keep the people watching in order to make a living. So they start going off the rails until they right themselves or crash spectacularly. its how youtube fame workss.
 
The "say anything to generate traffic" ecosystem begins to self-pleasure violently the moment anything with controversy potential hits their nose, and when no controversy exists they will find it necessary to amplify a non-troversy, or just outright invent it. Guaranteed that if MLID or JayzTwoBraincells ever found themselves in a courtroom for their FUD impacting a corporation financially, they'd take the Alex Jones "hey man it was just entertainment" defense.

But it does seem like a rift is forming between the tech-misinformation brigade, and the operations that still care at least a little about their word and their reviews meaning something, the ones still seeming interested in getting to the truth rather than just sensationalizing.

HardwareUnboxed taking Jayz to task

View attachment 522094

View attachment 522095

View attachment 522096
Thanks for posting this, that was hilarious to read.
 
Unless the PCBs inside that thing are encased in resin or some other water impermeable barrier, and the connectors don't see any liquid, I call complete BS. There's absolutely no way you're pushing thousands of watts though a motor or PCB that's had liquid ingress, especially salty liquid. That thing will corrode in a matter of seconds.

I've worked in the AV world, and film projection world. We have cables carrying low voltage high amperage DC. Xenon arc lamps are DC devices that operate in the thousands of watts envelope, and I've seen those have spectacular failures with very beefy connectors because of the issues I cited earlier.
I'm sure the loads you work with are more constant than what happens in the RC world. They also don't move.
I have a 1/10 3s buggy myself and can attest to the hundreds of amps and harsh environments these connectors can handle and also haven't had one fail or deteriorate to warrant replacing.
Also what was missed is that also comes with amazing airflow over everything, considering these RC cars are moving in excess of 55 MPH so not super comparable to stationary things other than these connectors can take insane abuse.
And then there are RC boats...
 
If all of the 12v pins are connected together anyway, then what is the point of using 12 pins (not counting the 4 sense pins)? Why not just two large robust connectors, one positive and one negative? That's basically how every other 12v device in the world works. Something like this:

View attachment 522001

Brings to mind the old saying: "Keep it simple, stupid"
They have to work with regular ATX power supplies, so at some point, you'd still have to adapt the 8 pin power cables to the single terminal style. Seems to me that the answer is to get your hands one of the proper cables from your power supply manufacturer, or to upgrade to an ATX 3.0 power supply, which I gather is supposed to just come with one.

Not everything uses the two fat wire solution, though. It's pretty common for aerial surveying equipment (camera systems, lidar, etc) to have three positive (+28VDC) and three negative wires that add up to a nominal 50A capacity. I don't think it's a standard, but there is one particular Amphenol bulkhead connector that gets used pretty commonly, and I've seen it in more than half of the survey aircraft I've installed in.
 
I'm sure the loads you work with are more constant than what happens in the RC world. They also don't move.
I have a 1/10 3s buggy myself and can attest to the hundreds of amps and harsh environments these connectors can handle and also haven't had one fail or deteriorate to warrant replacing.
Also what was missed is that also comes with amazing airflow over everything, considering these RC cars are moving in excess of 55 MPH so not super comparable to stationary things other than these connectors can take insane abuse.
And then there are RC boats...
Got a 6S kraton and the motor can peak ~3KW. Soldering connectors for that isnt fun haha. Think they should just hang a XT90 connector off the new rtx cards.
 
Got a 6S kraton and the motor can peak ~3KW. Soldering connectors for that isnt fun haha. Think they should just hang a XT90 connector off the new rtx cards.
Up until someone plugs a lipo into their GPU
 
Back
Top