News Nvidia Provides Update On RTX 4090 16-Pin Adapter Fiasco

[ohio_buckeye]

Well for one I feel like the USA should be building these factories here to insulate from world events. Just an opinion, but that could partially resolve the tariff.

 

[TJ Hooker]

I mean, way back, didn't Nvidia even explicitly say that the connector could only be plugged/disconnected a relatively few number of times?

I'm kind of thinking that was a hint that there might be issues. Why can't this connector last as long, for as many changes, as a 6-pin or 8-pin PCIe connector?

12VHPWR connector is rated for 30 mating cycles, which is the same as the old 6 and 8 pin Mini Fit Jr connectors are rated for.

 

[King_V]

12VHPWR connector is rated for 30 mating cycles, which the same as the old 6 and 8 pin Mini Fit Jr connectors are rated for.

Wait, really?

Dang . . .

 

[InvalidError]

Wait, really?

Dang . . .

And just about every other PC connector before those too. In general, internal connectors aren't intended for repeat plug-unplug cycles after the assembly line. Some PCB headers even have a bad record for the plastic frame leaving its pins behind when attempting to disconnect a cable because it often takes less force for pins to rip through the board-side connector shell than to separate the two plastic shells.

 

[OffMyRocker]

Its the AMPs that kill you

 

[King_V]

And just about every other PC connector before those too. In general, internal connectors aren't intended for repeat plug-unplug cycles after the assembly line. Some PCB headers even have a bad record for the plastic frame leaving its pins behind when attempting to disconnect a cable because it often takes less force for pins to rip through the board-side connector shell than to separate the two plastic shells.

I'll admit, I was reminded of this issue after I'd made that post by recently fighting a standard 24-pin PSU board connector.

Wrestling with them and saying "are they SUPPOSED to be this stubborn? Are they SUPPOSED to threaten the damage the motherboard if I have the audacity to want to unplug it?"

 

[spongiemaster]

Not entirely Nvidia or AMD's fault when TSMC jacks up their prices by 25% on the single most expensive part of a GPU. Then in the US there is still a 25% Tariff Tax implemented by Trump and kept by Biden that isn't helping things any either

The tariffs related to GPU's were temporarily removed almost a year ago. That exemption is scheduled to end on the last day of the year, but that's irrelevant to the current market prices.

 

[spongiemaster]

And just about every other PC connector before those too. In general, internal connectors aren't intended for repeat plug-unplug cycles after the assembly line. Some PCB headers even have a bad record for the plastic frame leaving its pins behind when attempting to disconnect a cable because it often takes less force for pins to rip through the board-side connector shell than to separate the two plastic shells.

The ones I break all the time are the terribly designed USB3 front panel headers on motherboards. The cable design sucks and is impossible to route in tight builds.

 

[Phaaze88]

View: https://www.youtube.com/watch?v=ig2px7ofKhQ


Looks like no one is safe; they're all at risk.

 

[InvalidError]

Looks like no one is safe; they're all at risk.

30min story short: "connector manufacturer doesn't matter if you don't plug it in properly."

One of the most impressive bits in that video IMO is how snipping four out of six 12V pins (two pins left) still wasn't enough to cause anomalous temperatures. So the connector itself has a ~3X safety margin when seated correctly.

 

[Phaaze88]

30min story short: "connector manufacturer doesn't matter if you don't plug it in properly."

One of the most impressive bits in that video IMO is how snipping four out of six 12V pins (two pins left) still wasn't enough to cause anomalous temperatures. So the connector itself has a ~3X safety margin when seated correctly.

So, by your logic(assuming), you would say the main cause of these failures is user error. Some of them have stray bits of carbon from manufacturing burning up too.
'Cause, what I gathered from the info is that these plugs are a B-! to plug in correctly. If it doesn't click, you'd best do it again.

 

[spongiemaster]

So, by your logic(assuming), you would say the main cause of these failures is user error.

Yes.

Post from Steve Burke:

Debris on its own did not result in failure. It took a combination of things to get the connector to melt. Debris did not contribute to a failure unless the cable was mounted improperly, which is a user error, and the cable was bent in a particular direction.

 

[InvalidError]

So, by your logic(assuming), you would say the main cause of these failures is user error. Some of them have stray bits of carbon from manufacturing burning up too.

Why would all traces of carbon have to come from manufacturing? People don't leave in clean rooms, it could have come from hair, dust, lint, Doritos, pizza crumb, etc. and even if one whole bump was lost to foreign object, the connector still has 35 more. A 3% loss of contact surface wouldn't cause a catastrophic failure on its own.

 

[Phaaze88]

Why would all traces of carbon have to come from manufacturing?

I didn't say they all did.

 

[InvalidError]

I didn't say they all did.

Just wanted to make sure people didn't get the funny idea of blaming all traces of carbon in connectors on manufacturing from "Some of them have stray bits of carbon from manufacturing".