NVIDIA Responds to Melting Cables, Warranty Concerns, & 12VHPWR Adapter Failures
video description
Date: 2022-11-18
Related videos
Comments and reviews: 15
Hurk
on the subject of FOD, i used to work for a machining company that did electrical connectors for both private industry and government aerospace contracts. the cost differences for making pins that are completely deburred and smooth vs what you see in the video here are extreme. mind you this is pricing from 20 years ago, but simple folded/cut tin plate likely used in these, then heat treating and electrolysis re-plating would generally have a cost of under 1 cent to 3 cents per unit. however, polished, deburred, plated, heat treated ends that would meet extremely strict guidelines to prevent issues like this were closer to the 15-20 cents per unit range. thats just the pin manufacture btw. we did not do plastic/wire work, so we would sell those pins to the contract that actually would make the connector.
on the high extreme level, where we would serialize lots, do complete tension testing, radioscopy, xray, etc, like gamers nexus had done to verify the parts, the cost could easily be over 1 per pin. the most expensive pins i ever saw made (and are now declassified) were for the SR-71 program, which each pin was made as part of a set of 50 for a specific connector and the cost per pin came out to 10 each.
reply
on the subject of FOD, i used to work for a machining company that did electrical connectors for both private industry and government aerospace contracts. the cost differences for making pins that are completely deburred and smooth vs what you see in the video here are extreme. mind you this is pricing from 20 years ago, but simple folded/cut tin plate likely used in these, then heat treating and electrolysis re-plating would generally have a cost of under 1 cent to 3 cents per unit. however, polished, deburred, plated, heat treated ends that would meet extremely strict guidelines to prevent issues like this were closer to the 15-20 cents per unit range. thats just the pin manufacture btw. we did not do plastic/wire work, so we would sell those pins to the contract that actually would make the connector.
on the high extreme level, where we would serialize lots, do complete tension testing, radioscopy, xray, etc, like gamers nexus had done to verify the parts, the cost could easily be over 1 per pin. the most expensive pins i ever saw made (and are now declassified) were for the SR-71 program, which each pin was made as part of a set of 50 for a specific connector and the cost per pin came out to 10 each.
reply
Joori
it's disappointing people need to be instructed on how to plug in a power connector... It's even more disappointing that an extra brain cell wasn't utilized to go maybe i'll use a different power connector cause you know, the adaptor that comes with the cards is ridiculously short and thick cables that are short will be hard to manage/mold. I mean christ, my rog strix 4090 cost me near 4k AUD, but i already had ordered a cable mod substitute that was longer before my card arrived because there was no way i was going to use the junk adaptor provided. Not after spending so much money on a freaking gpu. BTW my rog thor platinum II psu came with a very nice sleeved 12vhpwr 2 x pcie connector so i used that instead of the 4 x pcie cable i got from cablemod. The adaptor that came with the card went in the garbage. I should mention it's much easier to see if the plug is pushed in fully or not on the rog strix compared to the FE and the click sound is pretty audible as well.
reply
it's disappointing people need to be instructed on how to plug in a power connector... It's even more disappointing that an extra brain cell wasn't utilized to go maybe i'll use a different power connector cause you know, the adaptor that comes with the cards is ridiculously short and thick cables that are short will be hard to manage/mold. I mean christ, my rog strix 4090 cost me near 4k AUD, but i already had ordered a cable mod substitute that was longer before my card arrived because there was no way i was going to use the junk adaptor provided. Not after spending so much money on a freaking gpu. BTW my rog thor platinum II psu came with a very nice sleeved 12vhpwr 2 x pcie connector so i used that instead of the 4 x pcie cable i got from cablemod. The adaptor that came with the card went in the garbage. I should mention it's much easier to see if the plug is pushed in fully or not on the rog strix compared to the FE and the click sound is pretty audible as well.
reply
Zixxth
I don't have a 4090 just to be clear, but this seems more and more like a design problem than a user error honestly speaking. If the latching mechanism is this bad that it comes off with cable management then blaming the user for the problem is purely bad taste and a bad take. I know you are not doing that but you kinda implied it in the previous video. A better latching mechanism like the 6-pin or 8-pin normal cables would be better or something that detects if it is fully seated as you said but then if it does slide out over time, it's just going to lose power on random occasions over time, which seems like a different problem. So the whole fiasco would be fixed with just a better latching mechanism added to the design but that's up for PCIE SIG I guess.
reply
I don't have a 4090 just to be clear, but this seems more and more like a design problem than a user error honestly speaking. If the latching mechanism is this bad that it comes off with cable management then blaming the user for the problem is purely bad taste and a bad take. I know you are not doing that but you kinda implied it in the previous video. A better latching mechanism like the 6-pin or 8-pin normal cables would be better or something that detects if it is fully seated as you said but then if it does slide out over time, it's just going to lose power on random occasions over time, which seems like a different problem. So the whole fiasco would be fixed with just a better latching mechanism added to the design but that's up for PCIE SIG I guess.
reply
Wombel
Thats the BS statement that everyone was waiting for. Just some numbers for context. How many cables where you send? How many cables got Jay? Even with two of the biggest tech YT got X amount of calbes. I would say 10.
And then Nvidia also got send a lot for RMA of course. And then they claim its only 50 globally? For how dumb they take us?
And you still claim mostly user error and then show this dumb design and how its almost impossible to plug it in correctly . At this point you lost all creditability and just chill for nvidia.
Really disappointed in you guys. Your findings might point into the right direction but the way you lean to Nvidia is crazy. And to add to that you then defend Nvidia even more with their warranty.
reply
Thats the BS statement that everyone was waiting for. Just some numbers for context. How many cables where you send? How many cables got Jay? Even with two of the biggest tech YT got X amount of calbes. I would say 10.
And then Nvidia also got send a lot for RMA of course. And then they claim its only 50 globally? For how dumb they take us?
And you still claim mostly user error and then show this dumb design and how its almost impossible to plug it in correctly . At this point you lost all creditability and just chill for nvidia.
Really disappointed in you guys. Your findings might point into the right direction but the way you lean to Nvidia is crazy. And to add to that you then defend Nvidia even more with their warranty.
reply
Quackfoo
So Nvidia recommends plugging in the adapter the exact same way I did it. It took me 30 seconds to figure out what it took Nvidia a couple of weeks to figure out... Personally I'm tired of arguing with people about how blaming Nvidia for them not plugging the cable in correctly is Nvidia's fault. Anyone who does their own work on pc's should know that making sure the connector latches are engaged when dealing with power. I wouldn't plug my motherboard's power in and not make sure the latch was engaged so why would you not do the same with your very expensive video card. This nanny state where people expect their lives to be bubble wrapped and fool proof is annoying.
reply
So Nvidia recommends plugging in the adapter the exact same way I did it. It took me 30 seconds to figure out what it took Nvidia a couple of weeks to figure out... Personally I'm tired of arguing with people about how blaming Nvidia for them not plugging the cable in correctly is Nvidia's fault. Anyone who does their own work on pc's should know that making sure the connector latches are engaged when dealing with power. I wouldn't plug my motherboard's power in and not make sure the latch was engaged so why would you not do the same with your very expensive video card. This nanny state where people expect their lives to be bubble wrapped and fool proof is annoying.
reply
Ciel
Nvidia could also just keep the current 12vhpwr design by applying a white line on the cable pin housing indicating just how far 'in' the connector should be inserted (till it is no longer visible). That is, if the failure cause truly is due to user error. That would be the cheapest solution Nvidia would likely take. If Nvidia does a full-on redesign of the connector itself, it would indicate that an inherent design flaw with Nvidia's specific design spec for the 3rd party cable fab was the issue all along and all this was just an effort to save face for having their 1600 flagship product launch with faulty power cable design.
reply
Nvidia could also just keep the current 12vhpwr design by applying a white line on the cable pin housing indicating just how far 'in' the connector should be inserted (till it is no longer visible). That is, if the failure cause truly is due to user error. That would be the cheapest solution Nvidia would likely take. If Nvidia does a full-on redesign of the connector itself, it would indicate that an inherent design flaw with Nvidia's specific design spec for the 3rd party cable fab was the issue all along and all this was just an effort to save face for having their 1600 flagship product launch with faulty power cable design.
reply
Masterofworlds
wow. i got a 4080, and i know its not as likely to burn, i had issues getting it to boot at first because my 12+4 wasn't plugged in all the way. i had it plugged in maybe like half way, but that felt normal because it was really tough to push it any further, so i assumed that it was normal. i had even unplugged and replugged it in during my troubleshooting, and still couldn't get it to work. eventually i did realize that it had to be pushed really hard all the way in and it started working immediately. i think even having a picture of how it should look on the packaging would make this way less of an issue.
reply
wow. i got a 4080, and i know its not as likely to burn, i had issues getting it to boot at first because my 12+4 wasn't plugged in all the way. i had it plugged in maybe like half way, but that felt normal because it was really tough to push it any further, so i assumed that it was normal. i had even unplugged and replugged it in during my troubleshooting, and still couldn't get it to work. eventually i did realize that it had to be pushed really hard all the way in and it started working immediately. i think even having a picture of how it should look on the packaging would make this way less of an issue.
reply
sixspeeddeath
If a high current connector can be installed improperly and appear visibly to be fully inserted, it's a defect in design.
I'm sure the connector was improperly installed due to carelessness in some cases, but it's an overall failure of a design.
No redundancy in current path or pin contact to accommodate for debris, and a connector latch that requires extra effort to insert is not ready for consumers.
You shouldn't have to shove the housing in, or perform voltage drop tests on a connector to ensure it's plugged in properly.
reply
If a high current connector can be installed improperly and appear visibly to be fully inserted, it's a defect in design.
I'm sure the connector was improperly installed due to carelessness in some cases, but it's an overall failure of a design.
No redundancy in current path or pin contact to accommodate for debris, and a connector latch that requires extra effort to insert is not ready for consumers.
You shouldn't have to shove the housing in, or perform voltage drop tests on a connector to ensure it's plugged in properly.
reply
KK
I think nvidia came up with the answer at least a week before based on their own and the AIBs investigation. This could be seen from MSI release of their 12vhpwr connection guide released days ago, I am sure MSI did consult nvidia on it. They probably just waiting for the right time to release a properly worded response. They probably also waited for a 3rd party(GN) to release the findings and then acknowledged it so it doesn t make them insensitive by calling their customers user error directly.
reply
I think nvidia came up with the answer at least a week before based on their own and the AIBs investigation. This could be seen from MSI release of their 12vhpwr connection guide released days ago, I am sure MSI did consult nvidia on it. They probably just waiting for the right time to release a properly worded response. They probably also waited for a 3rd party(GN) to release the findings and then acknowledged it so it doesn t make them insensitive by calling their customers user error directly.
reply
Doityourself
LOL - I can assure you AMD will not have this same issue since they used more power pins. Admit it Steve.! The RTX-4090 is a poor power design and you know they are going to fail later on. I am willing to bet that no RTX-4090 connector will last more than 3 years of use before it fails. ( 3000 hours of heat cycling before it fails ) If your telling me all 3 dimples on the pins have to (must) make contact or the pins will burn up - that in its self is a bad design.
reply
LOL - I can assure you AMD will not have this same issue since they used more power pins. Admit it Steve.! The RTX-4090 is a poor power design and you know they are going to fail later on. I am willing to bet that no RTX-4090 connector will last more than 3 years of use before it fails. ( 3000 hours of heat cycling before it fails ) If your telling me all 3 dimples on the pins have to (must) make contact or the pins will burn up - that in its self is a bad design.
reply
Green
7:19 .. that's not fair given the included Nvidia adpater has a thirty use limit.
To be honest I think it must be illegal to sell a power product with usage limited to thirty uses. It immediately implies fragility of the adapter, and risk with use.
I would not use an Nvidia card with one of those adpaters. I would only a cable from my PSU manufacturer, or one like it. One cable with a solid plug on the GPU end.
8:34 .. OK Nvidia addressed my point.
reply
7:19 .. that's not fair given the included Nvidia adpater has a thirty use limit.
To be honest I think it must be illegal to sell a power product with usage limited to thirty uses. It immediately implies fragility of the adapter, and risk with use.
I would not use an Nvidia card with one of those adpaters. I would only a cable from my PSU manufacturer, or one like it. One cable with a solid plug on the GPU end.
8:34 .. OK Nvidia addressed my point.
reply
Mikey
Consumer product design in modern times NEEDS to facilitate prevention of user error! If it takes some significant amount of care for a user of a consumer product to plug in a simple power connector and then to verify that they did it correctly in order to prevent a potential fire hazard or avoid product damage, then it s a BAD consumer product design. Like, USB cables are keyed for a reason, and the contacts are different length for a reason
reply
Consumer product design in modern times NEEDS to facilitate prevention of user error! If it takes some significant amount of care for a user of a consumer product to plug in a simple power connector and then to verify that they did it correctly in order to prevent a potential fire hazard or avoid product damage, then it s a BAD consumer product design. Like, USB cables are keyed for a reason, and the contacts are different length for a reason
reply
kathrynck
well that addresses the cable itself.
Still, if i got a card which could spike to 600w usage... in a computer which is pulling 300 watts itself, in a room with a phone, a vape, a wall clock, a lamp, etc. also plugged in... it would be incredibly easy to peak over 1300 watts and trip the breaker (which isn't necessarily ideal for the house, or the computer).
PS: There's a reason I'm subbed to GN, and would never consider unsubbing.
reply
well that addresses the cable itself.
Still, if i got a card which could spike to 600w usage... in a computer which is pulling 300 watts itself, in a room with a phone, a vape, a wall clock, a lamp, etc. also plugged in... it would be incredibly easy to peak over 1300 watts and trip the breaker (which isn't necessarily ideal for the house, or the computer).
PS: There's a reason I'm subbed to GN, and would never consider unsubbing.
reply
reviewforthetube
There first response well should say half asked response was it isn't a worry and that some people aka Jay were making to big of a deal about it and worrying about something that isn't happening but I guess everyone forgot about that response? Also not to bad of a failure rate the 3000 series could of been covered to around the same failure rate but for other reasons. Yet never have we seen things covered this way before?
reply
There first response well should say half asked response was it isn't a worry and that some people aka Jay were making to big of a deal about it and worrying about something that isn't happening but I guess everyone forgot about that response? Also not to bad of a failure rate the 3000 series could of been covered to around the same failure rate but for other reasons. Yet never have we seen things covered this way before?
reply
Catt
I will blame nvidia though for a: lack of proper click or any feedback that the adapter provides and that there was no included right angle adapter. It is a failure on nvidias part and regardless of the here and now the design does not show a good long term outlook. Imagine the cable pins weaken from stress?
Christ blaming users is like eat shit pay me for my garbage product.
This is a failure, do not blame users.
reply
I will blame nvidia though for a: lack of proper click or any feedback that the adapter provides and that there was no included right angle adapter. It is a failure on nvidias part and regardless of the here and now the design does not show a good long term outlook. Imagine the cable pins weaken from stress?
Christ blaming users is like eat shit pay me for my garbage product.
This is a failure, do not blame users.
reply
Add a review, comment
Other channel videos
