• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Check Your 4090's 12vhpwr Connectors To Make Sure They are in Pristine Condition

you can’t crimp that connector. 24 wires going into 12 wires across 4 separate pcie8 connectors. Doesn’t make any sense for crimping. You can combine 2 wire into 1 pin for crimp. But that means some of the pcie8 pin will be crimped with other pcie8 wire which is not meant to be done.

Doesn’t seem to be the solder joint being the issue so far as that. Even Igor has stated as much.
It's not that they "can't" they didn't do the proper research. Which is why the design is flawed. Think about it what you just posted for a minute. No wonder we are seeing these issues. So they cheesed out on solder.
 
Last edited:
It's not that they "can't" they didn't do the proper research. Which is why the design is flawed. Think about it what you just posted for a minute. No wonder we are seeing these issues. So they cheesed out on solder.
I agree the design is flawed. It was a crap concept to try to bridge 4 pcie8 to a single 16pin. Madness. They should have gone with 2 pcie8 and limit the power to 450w and be done with that. And unlock 600w with native pcie5 support.

Personally I don’t see anything of an issue with the soldering on most of the connectors seen. The Igor’s on looked bad cos he took such liberty at butchering the connector. Others who have taken more meticulous care in opening the connector up show clearly the soldering is done well.

Solder joints are used across the industry for power connection. I mean insides of PSUs all joints are solders. It is not the issue that the connection is soldered.
 
I agree the design is flawed. It was a crap concept to try to bridge 4 pcie8 to a single 16pin. Madness. They should have gone with 2 pcie8 and limit the power to 450w and be done with that. And unlock 600w with native pcie5 support.

Personally I don’t see anything of an issue with the soldering on most of the connectors seen. The Igor’s on looked bad cos he took such liberty at butchering the connector. Others who have taken more meticulous care in opening the connector up show clearly the soldering is done well.

Solder joints are used across the industry for power connection. I mean insides of PSUs all joints are solders. It is not the issue that the connection is soldered.
The insides of PSUs aren't subject to the same strains as a power connector.
 
I agree the design is flawed. It was a crap concept to try to bridge 4 pcie8 to a single 16pin. Madness. They should have gone with 2 pcie8 and limit the power to 450w and be done with that. And unlock 600w with native pcie5 support.
Solder joints are used across the industry for power connection. I mean insides of PSUs all joints are solders. It is not the issue that the connection is soldered.
I separated this because you are completely contradicting yourself in a disconnect in trying to defend soldering in this gen5 connector. You cannot say the solder is fine in and of itself and ignore the fact that the design is completely flawed. You obfuscate the solder when the entire design is flawed from top to bottom fashion. And yes, using solder with such an obvious untested new connector is certainly not an industry standard for PC gaming. The important part is "industry standard". And like what Surveyor stated you are comparing apples to oranges trying to justify it.

Using solder was a horrible choice and comes off as cheesy "work around" to me.
;)
 
Last edited:
I separated this because you are completely contradicting yourself in a disconnect in trying to defend soldering in this gen5 connector. You cannot say the solder is fine in and of itself and ignore the fact that the design is completely flawed. You obfuscate the solder when the entire design is flawed from top to bottom fashion. And yes, using solder with such an obvious untested new connector is certainly not an industry standard for PC gaming. The important part is "industry standard".

Using solder was a horrible choice and comes off as cheesy "work around" to me.
;)
Soldered connection is not the premise of “bad connection”. As I pointed out. Soldered wires are used everywhere in a PC and PCB and power electronics. Just because a connection is soldered doesn’t not mean it is unfit for purpose. That’s the primary point.

The design of the backward compatible 12VHPWR connector that Nvidia came up with is clearly flawed no argument there.

The issue of the adopter is that it shouldnt be done like that. Soldered joints or not. It’s a Frankenstein. Never should have existed. They should have simplified it ie power capped the card when run on adaptor and full juice on pcie gen 5 native. Or do the decent thing supply every card with an ATX3 psu.

I think someone linked a video of some dude talking about the Seasonic adopter which is a 450w adaptor with 2 pcie8 going into the 12VHPWR.
 
Last edited:
Soldered connection is not the premise of “bad connection”. As I pointed out. Soldered wires are used everywhere in a PC and PCB and power electronics. Just because a connection is soldered doesn’t not mean it is unfit for purpose. That’s the primary point.

The design of the backward compatible 12VHPWR connector that Nvidia came up with is clearly flawed no argument there.

The issue of the adopter is that it shouldnt be done like that. Soldered joints or not. It’s a Frankenstein. Never should have existed. They should have simplified it ie power capped the card when run on adaptor and full juice on pcie gen 5 native. Or do the decent thing supply every card with an ATX3 psu.
No need to argue. I simply don't agree with you with the onset of how many "reported" connectors failed so far. Trying to salvage soldering used as "this is fine" and imply it had no baring on the failure rates we are seeing is argument in a bubble that popped a while ago.
;)
 
I am not convinced solder is the problem. Igor initiated this train of thoughts and he s back peddled.

There is something else at play here. We won’t be able to find out until more people volunteer some information and be transparent about it.

The fact the burnt out is not easily replicable or even replicable kind points the finger at end user.

But as GN Steve said - the design can be at fault if the design is so poor that user error becomes inevitable or more common as the result of poor design.
 
Last edited:
Soldered connection is not the premise of “bad connection”. As I pointed out. Soldered wires are used everywhere in a PC and PCB and power electronics. Just because a connection is soldered doesn’t not mean it is unfit for purpose. That’s the primary point.

The design of the backward compatible 12VHPWR connector that Nvidia came up with is clearly flawed no argument there.

The issue of the adopter is that it shouldnt be done like that. Soldered joints or not. It’s a Frankenstein. Never should have existed. They should have simplified it ie power capped the card when run on adaptor and full juice on pcie gen 5 native. Or do the decent thing supply every card with an ATX3 psu.

Imagine a few failures have been because they've jammed them in cases that are really too small causing excess strain at connection - particularly if connector not fully seated to start with.
 
I separated this because you are completely contradicting yourself in a disconnect in trying to defend soldering in this gen5 connector. You cannot say the solder is fine in and of itself and ignore the fact that the design is completely flawed. You obfuscate the solder when the entire design is flawed from top to bottom fashion. And yes, using solder with such an obvious untested new connector is certainly not an industry standard for PC gaming. The important part is "industry standard". And like what Surveyor stated you are comparing apples to oranges trying to justify it.

Using solder was a horrible choice and comes off as cheesy "work around" to me.
;)
The 3090 Ti actually had a crimped connector. Supposedly, the reason we see solder instead of crimping is because nvidia is not using the newer single seam terminals recommended by PCI SIG. Nvidia found that when the cable is bent in a North to south orientation, the terminals come loose and pop out and reported in to PCI SIG. This was back in August. In order to meet the timeliness for the 4090 launch they couldn't wait for PCI SIG to address this, hence they developed their own solution and use solder.

The Nvidia adapter itself is flawed.
Imagine a few failures have been because they've jammed them in cases that are really too small causing excess strain at connection - particularly if connector not fully seated to start with.
There are very few cases on the market which are wide enough for the 4090. Average width of an AIB 4090 is around 150mm. Adding 36mm on top of that gets you to 186mm. This is right on the edge even for massive tower cases.

With how wide these cases are, its baffling how nvidia didnt deduce people would do thisnor atleast bundle a 90 degree adaptor.
 
Last edited:
I think the whole PCI-E setup needs revisiting and redesigning.

It would be nice if GPUs pulled all their power from the PCI-E slot with extra power connectors plugging into the bottom of the motherboard.

This would make cable management much neater with much shorter cables going from the PSU to the edge of the motherboard.

I don't see why this could not be done as I have run CPUs that pulled more than 600W when overclocked.
 
I think the whole PCI-E setup needs revisiting and redesigning.

It would be nice if GPUs pulled all their power from the PCI-E slot with extra power connectors plugging into the bottom of the motherboard.

This would make cable management much neater with much shorter cables going from the PSU to the edge of the motherboard.

I don't see why this could not be done as I have run CPUs that pulled more than 600W when overclocked.
The mac pro 2019 use's no power connections for the GPU's its all pulled direct from mother board like the W6800XT duo 2 x 6800xt GPU's on one board, and you can run 2 of them cards side by side with infinity link so 4 6800 gpu's all pulling power direct from mother board no wire connections at all.

https://www.apple.com/uk/shop/product/MJ093ZM/A/radeon-pro-w6800x-duo-mpx-module. and you can run 2 of these linked with infinity fabric.
 
Last edited:
I think the whole PCI-E setup needs revisiting and redesigning.

It would be nice if GPUs pulled all their power from the PCI-E slot with extra power connectors plugging into the bottom of the motherboard.

This would make cable management much neater with much shorter cables going from the PSU to the edge of the motherboard.

I don't see why this could not be done as I have run CPUs that pulled more than 600W when overclocked.

The whole inside of PC cases need a redesign starting with the motherboard
 
I think the whole PCI-E setup needs revisiting and redesigning.

It would be nice if GPUs pulled all their power from the PCI-E slot with extra power connectors plugging into the bottom of the motherboard.

This would make cable management much neater with much shorter cables going from the PSU to the edge of the motherboard.

I don't see why this could not be done as I have run CPUs that pulled more than 600W when overclocked.
The mac pro 2019 use's no power connections for the GPU's its all pulled direct from mother board like the W6800XT duo 2 x 6800xt GPU's on one board, and you can run 2 of them cards side by side with infinity link so 4 6800 gpu's all pulling power direct from mother board no wire connections at all.
The whole inside of PC cases need a redesign starting with the motherboard


It would be nice to see a separate thread about this with a poll.

It would be great to see people's views about what they think needs changing and why.
 
i always thought of moving mobo cmos battery to case, just for easier replacement.. so maybe youd have a cmos battery hub near the power button/topside and then it powers the thing though wires connected to the mobo
 
Cablemod is also thinking off pulling their native ATX 3.0 cables. Apparently the only safe cable on the market is their 3/4x8pin to 12VHPWR native cable for ATX 2.0 PSUs as per their rep.
 
Back
Top Bottom