News Technician Repairs Hundreds of RTX 4090s With Melted Power Connectors Every Month

Status
Not open for further replies.

Hresna

Commendable
Jun 18, 2022
12
14
1,515
I’d like to have heard more of his reasoning for why its a design flaw and not user error.

I’m not sure why it coudln’t be both. A connector that makes it exceedingly difficult to actually seat properly is a design flaw - but one that can be mitigated by extra caution by the user.

If he’s only getting burnt cards and not dismantling the builds, there isn’t really any way he can know for certain these connectors were “properly seated” when the damages occurred except to take users’ word for it.
 

YouFilthyHippo

Prominent
BANNED
Oct 15, 2022
216
119
760
I have been telling people since the day this problem started that it is not user error. No one ever listens. A recall is in order, but NVidia cares too much about profits. What other computer repair technician do you know had to buy a <Mod Edit> gas mask just to do his job of repairing computer components? This is insane. He's getting 25 a week? Across a full year its been, that's gotta be 1300 GPUs JUST AT HIS SHOP ALONE. Imagine the statistics across north america, or even the world. It's time for NVidia to stop deflecting blame and own their mistake.
 
Last edited by a moderator:

PEnns

Reputable
Apr 25, 2020
703
746
5,770
I’d like to have heard more of his reasoning for why its a design flaw and not user error.

I’m not sure why it coudln’t be both. A connector that makes it exceedingly difficult to actually seat properly is a design flaw - but one that can be mitigated by extra caution by the user.

If he’s only getting burnt cards and not dismantling the builds, there isn’t really any way he can know for certain these connectors were “properly seated” when the damages occurred except to take users’ word for it.
Why not call it a design flaw? I haven't heard of any other GPU that has this exact problem and of this magnitude.

Or is Nvidia so infallible in its fans' dreams??
 
Last edited:

cyrusfox

Distinguished
I have been telling people since the day this problem started that it is not user error. No one ever listens. A recall is in order, but NVidia cares too much about profits. What other computer repair technician do you know had to buy a <Mod Edit> gas mask just to do his job of repairing computer components? This is insane. He's getting 25 a week? Across a full year its been, that's gotta be 1300 GPUs JUST AT HIS SHOP ALONE. Imagine the statistics across north america, or even the world. It's time for NVidia to stop deflecting blame and own their mistake.
Show me one person who was refused an RMA... Until that happens enmass, there is no need for a recall. All the manufacturers will silently replace until they are out of warranty (3 years generally).
Northridge has a reputation, even the OEM send him cards to fix... he is a refurbisher likely unwittingly... I wouldn't extrapolate his numbers to others.

Nvidia will not need to own this as these cards sellout everywhere, there is not enough outrage. Just flame ups in the comments. Its a $2000 card(MSRP was $1600-1800 originally), maybe it will burn some users out (I for one won't ever buy another Samsung appliance for similar reasons). But overall Nvidia has zero reason to re-enter this discussion. Per MLID the 4090 is going to be phased out (very tight/nonexistent supply going forward), the coming super lines will take over. It is now understood this can happen and the mechanism whether user error or design(or a combination) has been hotly debated and its old news that has already sorted itself, new connector modification done to improve mating connector, should eliminate or lower its occurrence hopefully

Why not call it a design flaw? I haven't heard of any other GPU that has this exact problem and of this magnitude.

Or is Nvidia in so infallible in its fans' dreams??
Look at power draw of 4090, some have reported it can peak draw 780W, there were reports of it blowing PSU or causing power faults, its simply a beast, both in power draw and performance (peerless). That coupled with a new power delivery standard is prone to teething issues such as this
 
Last edited:

ezst036

Honorable
Oct 5, 2018
617
533
12,420
My opinion, which is purely my own, is that it is BOTH user error AND connector error.
there is poor to no verifiable locking mechanism on the connector
even locked, wiggling can work cable loose.

and then to user error:
According to Steve, and his testing (lot of repeats, lot of different things tried, only one way seemed to cause meltdown, there could be more but.. those are yet to be proven) seemed to be valid, only way to melt the connector is to have it BARELY connected (already almost 50% out of connector)
cutting pins (5 out of 6) out of the connector, if properly seated, did NOT cause meltdown or extreme heating up.
so high power usage is NOT the issue for melting.

and the fixes done to more recent connectors in later cards:
shorter sense pins, so if connector wiggles loose, it will lose sense connectivity BEFORE melting point
causing computer to shut down or refuse to start until connector seated properly.

so beware of the V1 cards, if they even mention which version of connector the card has.

for the record, I do not own 4090.
 

coromonadalix

Distinguished
Nov 26, 2006
129
41
18,610
well ill do a small rant about him

use an powered solder sucker, a manual one will kick on the pcb when it does a kick back, for sure he take the big approach to heat and desolder the plug at once, the ground plane is massive on many cards

you have to go fast

but theses connectors melting is a bad design at 1st, power pushed to the limit on thesess pins i would say

yeah tons of discutions already, but i stand on a bad design at start

yeah theorithical power limit to the pins blah blah you never push 80-100% current in pins the margins are to narrow
 
I’d like to have heard more of his reasoning for why its a design flaw and not user error.
design flaw of a power connector that has extremely poor retention makes user error more likely.

you dont hear of prior connector having these issues which means its bad design of newer connector as more users have managed the old connectors than the newer one w/o issue.
 

TJ Hooker

Titan
Ambassador
The RTX 4090 came out 13 months ago. AFAIK all 4090s have at least 12 months of warranty; many have more, e.g. those bought direct from Nvidia have 3 years.

So is this guy getting all these damaged cards just from people who bought 4090s in the first month it was out, and models with the shortest warranty? Or is he dealing specifically with people who have done something to void their warranty? Something about this doesn't quite add up.

Edit: On 2nd thought, he implies he has been fixing these for some time, so he must have been dealing with cards that are within their warranty period. Why would these people be sending their cards to independent repair shops while they're still under warranty?
 
Last edited:

King_V

Illustrious
Ambassador
Are all of these GPUs out of warranty?
So, he's my question. If Nvidia says this is user error, do they deny warranty claims based on it being user error and not a defect?

Now moving on to this topic/thread on a general level rather than you the specific post I replied to...

There were problems that didn't happen with older connectors. Then they also said you can't bend the cabling too much near the connector, to the point that, if I recall correctly, it was mentioned that you might want to check if your case has enough extra side clearance to accommodate this connector's needs.

Then there was a revision they actually seems to address the problems.

Yet, it seems like there are people dedicated to perpetuating the belief that it's entirely user error. Or, in what appears (to me) to be an attempt at false equivalence, mentions of why it can't be both a design flaw AND user error.

Why this dedication to blaming the user?

And even some going so far as to say that because it's a new design for a high end, expensive product, some issues should be expected. To which I'd say that might be expected, a little, for a budget product where corners were cut, but most certainly NOT for a halo product.

EDIT: fix mobile fumble-fingered typos
 
Last edited:
  • Like
Reactions: PEnns

vanadiel007

Distinguished
Oct 21, 2015
237
233
18,960
I have a 7900XTX, have unplugged and plugged the power connectors multiple times. Nothing burning, on any 3 of them.
They should have stuck with the existing power connectors on this generation, because they have proven to provide very little issues.

This had nothing to do with a better connector, but this had all to do with power supply sales.
 
The RTX 4090 came out 13 months ago. AFAIK all 4090s have at least 12 months of warranty; many have more, e.g. those bought direct from Nvidia have 3 years.

So is this guy getting all these damaged cards just from people who bought 4090s in the first month it was out, and models with the shortest warranty? Or is he dealing specifically with people who have done something to void their warranty? Something about this doesn't quite add up.

Edit: On 2nd thought, he implies he has been fixing these for some time, so he must have been dealing with cards that are within their warranty period. Why would these people be sending their cards to independent repair shops while they're still under warranty?
When Nvidia came out and said it was user error that caused the melting they basically gave AIBs a blanket to deny fixing melted cards under warranty. Physical damage is never covered under warranty unless it's proven to be a manufacturing defect. Also using third-party or modded cables voids the warranty.
 
Nov 11, 2023
3
0
10
When Nvidia came out and said it was user error that caused the melting they basically gave AIBs a blanket to deny fixing melted cards under warranty. Physical damage is never covered under warranty unless it's proven to be a manufacturing defect. Also using third-party or modded cables voids the warranty.
Exactly, which means they're most likely blocking people from using their warranty like they should and not losing money out of a scenario where people would get their GPUs replaced under warranty if they hadn't said that. Sneaky bastards.
 

valthuer

Upstanding
Oct 26, 2023
101
101
260
Can't tell if it's a user error or design flaw.

What i can do, however, is convey my personal experience, which is more than 5 months of intense gaming, at 4K 60 Ultra settings, with an Inno 3D GeForce RTX 4090 24GB GDDR6X X3 OC.

In the fear of accidentally creating the problem myself, i have refrained from using a 180 Degree adapter.

So far, i have encountered zero issues.

But, i guess, anything can happen, so i'll just have to wait and see how things play out in the coming months...
 
Status
Not open for further replies.