While I, and hopefully the entire userbase that follows Jay Steve or Alex, certainly appreciate the work each one of them devotes to investigative journalism backed by proven and professional techniques as well as decades of experience, certainly do not appreciate or advocate for any of them to work against each other and basically execute a tech pissing contest about Nvidia's melting 4090 power connectors. Grow up boys. Work together. Stop throwing rocks at each other. You're making us, the followers, the subscribers, the bread and butter of a chunk of your income, cannon fodder for this silly bickering. Work it out. Post something collaborative. We're not appreciative of tabloid tech news and updates.
I don't think anybody is against anybody. I made the video from my point of view and the way I see it. Opinions can differ. If all people shared similar opinions, the world would stop and no further innovations or advancements would be made. Looking at this from 3 points of view is not enough. That's where the viewer comes in. I mentioned at the end of the video, |let me know what you think, do you agree with GN, Jay or Me?" The topic is left neutral and open to further investigation. The way i see it, the issue remains an Nvidia and not a user problem. What I said should be taken with an open mind and not as someone attacking anyone.
It seems like you have almost the same opinion just you place all the blame on the 4090 having a design flaw connector, and Steve is saying it's both user error and a design flaw. Either way the conclusion is the same that the 4090 12vhpwr connector is designed wrong and needs to be fixed
After reading your comment, it is clear that you are too ignorant to realize how ignorant your comment was showing you'd be a hard one to teach, and I cannot be your teacher, but I will hope that you will read your one-sided very simple comment once more, despite how hard it may be, and realize it has only collapsed on itself due to you crying heavily over the page you are writing on, smearing the letters as you try to make an argument about how men shouldn't argue when it comes to different opinions like the one you just gave. Also while speaking on everyone else's opinions as if you know all. Broaden your horizon. Speak for yourself. Not everyone else 👌
Endorphin, how you came to that conclusion after watching this video, I have no idea. Quite honestly it doesn't even sound like you watched the video. None of the names you mentioned have been fighting or arguing against each other with regards to the 4090 connector. They simply came to their OWN conclusions. Obviously with opposing viewpoints someone will be correct and someone won't. To the community however the fact that all these techtubers are willing to go to great lengths to showcase the issue, only benefits us the consumer.
which cable, the 3rd party cable ?, how many of these melted while using the standard cable, if I was looking for a failure point I would be testing the 3rd party cable to see if the materials used are up to the job i.e. can carry the current without heating up, when conductors heat up they increase in resistance to the current which them causes more heat and eventually total failure
If you know a car is going to burn down when you start it, would you still buy it? If you buy one, knowing there is a problem, then the only fault is yours
@Geoff H They had this problem from the start, when only 1st party cables were available. It isn't the cable, it is the connection at the pins. This has all been well documented at this point.
gtx series: nothing bad happened with cable connectors rtx 2000: nothing bad happened with cable connectors rtx 3000: nothing bad happened with cable connectors rtx 4000: cable connectors melting users: - oh no, I forgot how to connect cables
The thing is also... There's wayyyy more people now building their own pc's than before! And let's not forget the crypto miners niche also... So, more people can equal to more mistakes/accidents. It's important to not to jump to conclusions and investigate FIRST
Whatever ends up being the case, these cards cost WAY too much money for them to be damaged permanently by slight disconnect. Nvidia should fix this issue and take responsibility.
It's not just about way too much....as current market stands, the flagship GPUs now cost thousands of dollars ....on top of that, this users bought this flagships just its to melts because supposedly user error....we gave away too much money to this corporate to treat the consumer like shit!
@@triadwarfare Probably has more to do with the video card gouging that only recently has cooled down left a very bad taste in a lot of peoples mouths. People are now just plain refusing to upgrade until they absolutely have to.
You are absolutely right, people didn't suddenly forget how to plug in a power connector, and ONLY on 4000 series gpu's. Frankly, it's a disgusting practice when any company refuses to accept responsibility for their products and instead tries to gaslight their customers.
I’m on a MSI Ventus 3x 4090 with the 90degree cable mod adapter and Cable mod PSU cable. I’m also an independent IT consultant and PC builder who has built thousands of PC’s in my time. I have been keeping an eye on my system since the 4090 was installed. Non of my cables are bent or under tension I’m using an AIO and 140mm fans so little vibration is in the system. The GPU is power restricted to 80%. About once a week I check the cable is seated home and about half the time it has worked it’s way out a small amount. Thinking it could be vibrations from the fans I replaced them all with Noctua ones. Still happens. From what I can tell it’s due to thermal expansion. With the connector repeatedly heating and cooling with use it’s causing the connector to work it’s way out then once it’s far enough out to increase the resistance it starts to melt or arc. This is totally a design flaw. All the cards should be recalled and the connectors replaced with larger more substantial ones before a fire is caused and someone dies.
There's a reason why 12V was always supplied over mutiple cables, to spread the power evenly. Combining all that power into one socket on a power hungry card is a disaster waiting to happen.
Well with hwinfo you'll see atleast on my 3080. It's far from evenly. But I agree with you I like the fact that the power is running through 3 connectors instead of one.
Another vote for THIS.. And that's where actual proof could come out that it's a bad design. If under load the individual pins are measured for actual current draw, then there just might be a smoking gun (pardon the pun) in that some of them will show a very large imbalance in current from one to another.
Intel made the spec, and they chose not to use it. Should have been nVidia's first clue, but they've proven unable or unwilling to "read the room" as of late. On just about everything to do with consumers anyway, they probably don't care because their AI/enterprise side is cashing checks.
You are totally right Alex. We never had these problems when we had the older style of connectors because they were oversized for the job. The connectors now are theoretically up to handle the power of the cards..
If you look at it from another angle: I bought 4090 with the money i collected for 2 years. And now i'm learning soldering and electronic repair while waiting for the delivery of my wannabe mini laboratory to arrive, soldering station, heatgun, thermal camera, microscope, consumables, the whole shabang. I'm always wanted to learn e. repair, but was very lazy about it. Now i have the most powerful gaming GPU in the world (that works, hope to not jinx it), and soon skills and equipment to fix it if needed. Thanks for a little push 4090!
I have the feeling the connector itself is undersized for the 500+Watt. On my previous card there were 3 connectors on the GPU to run with 300-400W and now we have 1, which is way smaller, which runs 500+.
I think if you're suddenly seeing a massive uptick in melting GPU cables when neither the connector or the average skill and ability of the users changed then there really is no other alternative than it's the GPU deaign.
People has built computers for like over 20 years now with GPU:s that need external power hooked up to them. And this seems to be first time that the connector melts. So in my eyes the reason is VERY obvious, the connector has to be too undersized for the application. It can't be that suddenly hundreds or even thousands of people don't know how to connect that connector right.
6 and 8 pin atx connections have not failed because of their design, ever, and ATX standards have been around about +/- 50 years now. Defects and mistakes happen, but this connector is failing year one, repeatedly. Some of it was user error, sure, but it's all nVidia's fault that it happened in the first place.
Well maybe there are thousands of people who don't know how to connect cable so Nvidia should change design, provide sufficient training on card installation or put requirement that only qualified electrician may install 4090 in your computer.
Other connectors have melted before, this is just blowing up more because they're new cards and the connector was causing controversy before the 40 series even released.
@@stanimir4197 I'm not young first of all. Second, I'm not talking about form factors. ATX power supply connections can be found in things as old as IBM Selectrics. I know when you think you know everything, you tend to neckbeard it, but probably save the fedora for the real dance, pal.
@@redragongaming "The shill gambit, a type of ad-hominem and poisoning-the-well rhetorical move, occurs when one party dismisses another party's arguments by proclaiming them to be on the payroll of some agency."
@@Ryan_DeWitt Your comment doesn't disprove what i just said, it only enforces the facts and those are with me. You can't prove Steve isn't involved with Nvidia because he is. Of course he's gonna side with the corporation, doing the opposite could mean penalties to him. Once you're involved with the rules you gotta play by the rules, otherwise no one's going to like the hard cold truth. I'm happy to say Alex and Jay are on the right track on this one, corporations are still made up of people and they gotta respect the customer, not shift the blame from one another.
@@redragongaming Do you have proof Nvidia is paying him off to have his opinion? The burden of proof is on you since you are the one making the claim. It is one thing to say you disagree with his conclusion, a whole other to make a "he is a paid shill" accusation. This may come as a shock, but people may disagree with you and that doesn't mean they are getting paid somehow to do so.
@@redragongaming Go back to a "how to argue 101" class then return to this comment section just to start deleting your below-the-grass-level-brow comments. The case of Hitchens's razor in full public view here. One can hope something was at least learned.
I like Steve voicing out many manufacturing design errors and fighting for it. But in this one am siding with Alex and Jayz, it's Nvidia's fault to use a fragile connector like that.
@@realtuner3560 Great example of you're damned if you do, damned if you don't. Don't try to do advanced testing, "you know nothing, get some good equipment and people who know what they are doing." Get the good equipment and consult with people who know what they are doing "You are just pretending to be smart."
@@realtuner3560 Yeah man i figured out Steve doesn't actually knows so much as he pretends to. I even found a webpage with him talking about a problem with the PCI Express Downstream Switch ports that weren't working properly on a motherboard, after that he posted steps onto fixing the issue but surprise surprise these steps were not relevant, the so called steps seemed to be nothing but click bait, it's how i know he's most likely a sellout.
@@Ryan_DeWitt Let it to a certified lab and run a test procedures for connectors like the ones used in automotive industry. Period. You dont need to put a show. Get your test done under standards in a certified lab.
@@38911bytefree This isn't the auto industry. It is the gaming industry. That is what youtube is. Info with entertainment. Channels that post cold info with zero entertainment do not last long. As Alex likes to say, you get a lot of scientist in the youtube comment sections. Plus they do consult and test beyond their capabilities or equipment that they do want to run are sent off to these places.
It's amazing that the new connector made it to users. Even from a laymans point of view, the change from 2/3/4 8pin plugs, to a single plug would make me expect failures. Perhaps it could be executed properly, but it just seems undersized to achieve any sort of thermal isolation. There should be larger gauge wires and larger pins, at a minimum, but a more sophisticated latching mechanism would help as well. Software or sensor checks against the connection would be great.
Was super paranoid about this and made sure my connecter was fully seated numerous times. Literally not even 20 minutes ago while playing Diablo 4 I smelled a plastic burning smell from my not even a month old PC and knew where it was coming from. Opened up my case and the connecter is hot to the touch. There’s definitely some melting and I’m scared to pull it off.
We have never seen as many melted connectors as we have on this nvidia 4080 - 90 series. It is very rare to see any melted 8-pin connector of the millions of video cards that exist in the world. If we only take this percentage of failure that we saw in the last 4 months in this generation 40... I believe that would be enough to carry out a recall and review this type of connector. For me Alex your point of view is the best!
I agree with Alex. Throughout my many years of tinkering with electronics, I've never had a connector melt because it was plugged in partially. Most of the time it just won't make an electrical connection, and a re-seat fixes it. The only time a near connection poses a problem is with high voltage/amperage where electricity may arc across a small gap. Repeated arcs will melt even metal as seen in arc welding. The only other time they melt is when I run too much amperage over a connection but that was me being lazy and knowingly pushing the envelope on underrated wire/connectors. If the connector fails through regular usage, it is just plain bad design. If you need a degree in electrical engineering to properly plug in a connector, it's also bad design in any commercial equipment. Imagine if plugs of any home appliance would burn your house down if they were not absolutely seated 100%.
I also think that NVIDIA's fault is more precisely the connector on the cards and the high consumption of 400+ W, a simple solution is to return the old connectors that we have used for the last 10+ years and which AMD cards still use. If I said something wrong or wrote I apologize in advance. Thanks for the great videos NorthridgeFix, I follow you regularly, have a great day. 👍❤💯🔔
AMD was going to use the same connector but then saw all the problems and was like "oh hell no" and immediately redesigned their boards. Intel *designed* the connector (and didn't fully test it or engineer it properly), not the standards board, and even they don't use the connector for their motherboards or GPUs. Nvidia just doesn't want to admit fault because recalls are expensive, and they have a company policy of never admitting fault even when they screw up, and attacking anyone that criticizes them.
@@radoslavsk8591 I don't know but the cable on my MSI A1000G PCIe5 power supply says Cleary 600W on it.(meaning it can support 600Ws) I think is a combination of some 4090 models, but also power supply models, maybe people are trying to overclock cards that cannot go pass 400Ws or near 600Ws.
Great video Alex! The 3 of you should get together and talk, and maybe test some of those cards. That would make an incredible video. Steve is more protective, I think, because they have done extensive testing on that problem already. Jay is more on track with what you're thinking; you could also show Jay how to resolder pins on CPUs, the right way. Wonderful work Alex, as always!
A recall needs to be issued for the 4090 cards. NVidia has enough resources to do their own analysis. They should hire Alex to test before releasing their product. 4090 is a fire risk waiting to happen.
I'm with you, I trust someone who dedicates every day to fixing the poor work of video card manufacturers more than someone who receives gifts from video card manufacturers...
I think a part that Steve missed is also the broken power connectors that seem to be a problem with those cards. There seems to be more going on than just not pushing in the connector all the way. A 250 gpu backlog is pretty insane, good luck!
These cards are huge. In my case, the CPU AIO hoses weren't long enough to get around the card, and I couldn't mount the fans. It's possible people are trying to install these into cases that are too narrow and these issues happen when trying to install the side panel.
It might be problem with the card itself but I think the point about cablemod sending them was that cablemod was like a buffer for the cards and collected them from longer period of time and then sent them all to you in short period of time which made more cards coming in shorter period of time than if users had sent them as they failed themselves.
I'd say the video itself squashes this one as both the video here and from Steve at GN state eight cards from cable mod. Whereas this video showed a whole rack of cards from users waiting to be fixed. Those eight cards aren't making or breaking this numbers wise.
@@Stubbies2003 true, but that is a fact that only came out in this video so hard to blame GN or J2C for that when it was completely unknown info to them
@@Henrik_Holst Yeah Jay has already leaned towards something wrong with the card at this point. Steve didn't shut the door in the video either. He left it open for something that wasn't known at the time of making the vid.
I agree with you totally. I wonk on automotive electronics. GM and their micro connectors for years have been too light to take the amp load within the circuit. The small terminals in the connector will start to heat up due to the fact they are too small. A fuse is installed in a circuit to be the weak link in case of failure on that circuit. If you have a weak terminal at the connector that is too small, you have made the terminals into a fuse. Another weak link. They may not burn as quickly as a fuse but they will sit and simmer causing melting problems in the connector.
If it is always the same pin that melts, perhaps it is something related to the power that is used on this specific line. it means that probably one power phase consumes more than the others when feeding the gpu. it would be interesting to test the consumption of this line to find out if this is the problem
The fact the 3090 had the same connector and we didn't see large amounts of burnt and destroyed connectors means that there is something wrong about the 4090's.
4090's can be pushed to higher power maybe? it's a waste on such an efficient node and architecture, but it's possible nonetheless. There's always that one person that cranks every setting to the absolute limit and then runs some excruciatingly GPU intensive benchmark that maxes out every square mm of the GPU chip.
People pushing 1,000 watts of power through a small connector is why they melt then go boom!! No plastic separation of metal pins, then they touch each other… this is whY high wattage applications use big gauge wiring and big pins and a connector that is huge that separates pins so they never touch no matter how hot connector gets…
Alex, I follow Jay and GN also, but if possible, maybe you can continue this topic with them, as I do believe that you 3 will deliver a proper RCA. Again, thank you for all the work you put in!
I completely agree with your explanation Alex. In my understanding, it's a real shame how engineers are not able to design this card to manage a electron flow. That is the very first lesson to learn for a proper project. Engineering sucks.
25+ years of building PC's. Same deal, 4090 connector melted and dealing with getting it fixed... Simply put, how is it that countless PC's over the last 25+ years have never had "Oh the user needs to plug it in fully" issue. Yet, as if by magic once Nvidia put out a new first gen connector, all of a sudden it's the users problem? Sounds like a great concept! New Business Idea: I'm going to design a parachute where as it works 100% so long as you pull the rip cord at "just the right speed". Any faster or slower, you die. Remember... that's on YOU for not timing your pull correctly! Get it right user! *sigh* Not only that these connectors that are melting ARE connected tightly. It doesn't matter. If it's fully clipped in any amount of wiggle, bump or otherwise should not be an issue. Doesn't happen with this commonality with SATA connectors. Doesn't happen in mass with molex connectors. Sure, some of them do at times and can of course but no defined pattern like we're seeing here. Nvidia improving their connector in a few ways should say it all. Why improve something that doesn't have an issue?
NVIDIA is weighing the possible loss of reputation against the cost of recall to see if they really need to do anything at all. They are still counting on the shiesty sales agents to confuse customers and the fanboys to continue to show motherly love towards NVIDIA and defend them to the death. They are waiting to see if those two factors will be enough to drown out the customer complaints. This is how large scale business with a huge fanbase make business decisions. It's a natural phenomena. ✌🏽
@@hgbugalou It isn't a beast though. The 4000 series is hilariously awful. The 4060 Ti is debuting to be in a power scale of the 3060 Ti/3070? Bitch what? Why is a NEWER card in literally the same power tier as the old generation? The 4090 doesn't do anything the 3090/Ti didn't either other than arbitrary "we locked off DLSS 9001 for ONLY 4000 series, because raisons" nonsense. It's an unsafe and largely unneeded mess the 4000 (de)generation.
let me get this straight. lets say a car manufacturer has this special lug nuts for securing the tires down, user can install the tires correctly, torque them down correctly but there is still a chance of the tires to wiggle a bit if you turn them too much and some tires end up falling off. would that be user error? NO. i dont know what Steve is smoking. the evidence of that 90 degree adapter melting while fully seated is all the evidence you need. this is on NVIDIA.
The problem is everyone is right, its a poorly designed system which invites error from the user (partial connection, pressure on cables) and design fault (material ingress, mechanical stress fault).
No one in their right mind would even think to blame the melting cable saga on “user error”, unless of course, you’re trying to protect your sponsorship deals. I mean come on, it’s a cable for gods sake, we all know how to plug one in and it’s insulting to be lectured otherwise.
Steve really wanted to convince us all it was user error. To an absurd level even. And guess what the cable and the connector are all being replaced, so i guess he was absolutely wrong, or was he wrong? maybe he was just doing someone a favor, no recalls. Thanks Steve.
I think you're absolutely right, it's a problem Nvidia has to solve, it cannot be solved by anyone else. I see the problem in an overstress of contact area which is considerable smaller than in the older solutions with up to 3x 8-pin connctors. Even if you would use some clever current balancing circuit to even out the currents every one of the 6 power pins has to carry they are just f*cking small for the job. If you take Tyco MQS contacts as a comparison they are rated for 7,5 Amps per pin @ 80 degs C. Multiply this by 6 and you get a maximum wattage of around 540 W under optimal conditions and ONLY if every contact takes an even amount of the load. I can not get how mad an engineer might be to think that stunt may go OK taking 3 to 4 8-pin power outlets and put them on 12 pins in a smaller form factor connector. This is absolutely bonkers, no need for doing so if your education in physics basic course at the university has left some marks. Even if you put the female connector straight into the card and give it a firm press, there will be a problem coming, maybe from micro vibrations of the contacts, reducing the effective area for the current, perhaps pulling small sparks, getting warmer, thermal changes may lead to changes in the dimensions of the metal pins and contact tongues, perhaps even some micro wearout happens after some time. The video card is not a static thing, there are fans moving, they speed up and down, other moving parts in the computer, even the user loading a CD, pressing a button, plugging in USB devices and so on. If you push the contacts far to its physical limits without any need to do so beside marketing reasons ("it is MY CONNECTOR" - Gollum image suggested while reading this) you will get into trouble. The trouble is there. And the trouble is exactly this connector or better the way it is used: too much amperage for the contacts, way off any reasonable specification limits I as an electronics engineer and a long year developer of automotive electronics would consider to be safe in any way. This connector is not right for up to 600 watts of power. Not if you wanna be on the safe side. I thought a lot lately of replacing my 3090 by a 4090. But hell no, this connector seemed wrong to me the moment I first saw it months ago. And I'm feeling confirmed. I will not buy any graphics card forcing me to use this for sure. There is a need to revert to the old connectors in my opinion. But I do not think Nvidia is ticking this way. Which is a shame.
Same. I was about to get myself a new 4090 but as I kept looking for the best brand I came to learn about this SIGNIFICANTLY DANGEROUS issue. People: this could set your god damn house on fire. It's no joke. I am not willing to risk it.
Jay agreed with you and so do I, if 4090 cards are coming in bulk melted on the pcb nvidia should recall them. Even if it was user error for 2000 the card should have built in protection. I have limited my card to 300w until this is resolved
Alex I am 100 percent in agreement with you that it is related to Nvidia! Too many issues that it can’t be user related! I have been in IT field since 1991 and built many pc and never have had a melting issue on connectors until 4090! So I am def going to say it is manufactured issue!
Mine just went up in smoke today. I was one of the people warning others to be careful with the connector late last year and was super cautious myself. Only ever plugged it in once and made sure it wasn’t bent excessively, etc.
Glad I watched this video. I under the full impression that this was determined to be user error. As long as you attach the cable correctly, you're good. Seems like that's not the case based on this video, or at least not the case all for all cards. I was going to wait for a price reduction and get a 4090 then, but I've decided the melting risk isn't worth it. I'll wait for either a redesigned 40-series card with a safer connector or just pass on the 40 series completely and see what they do with the 50 series. I'll also give AMD a serious look finally, as this may be the time. Once Nvidia and their partners release a card with a redesigned power connector, that will be the time to buy - until then these cards are just disasters waiting to happen.
At 18:19 you can see that the clip is ON, but there is still a lot of wiggle room there that could allow the connector to disengage. I could see this inadvertently happening in the power cable routing back to the power supply. Maybe that is putting stress on the connector even though it is fully plugged in. Since Alex has seen melted connectors on the card with the Cablemod installed, that's where the plot thickens.
It can only disengage if its physically touched after its installed or its under extreme pressure which the user should obviously know if they installed it.. Through my experience I've installed hundreds of gpus and no matter how tight the wires were once things go on they normally don't move. This is only happening with 4000 series and that should be enough evidence by itself.
So all the other GPU's that user's buy they only do not plug the cable properly on the 4090 but they do on all the others, how strange. I still say it is a problem on the card, it has to be, who not all other crads coming in with burnt connecteors?
In the past I have joked with Alex saying he is a PC surgeon, the thing is I meant it too, he knows his stuff and for him to make a stance like this I'm with him all the way. To him electronics is second nature, he know a fault when he see it.
Oh, I don't know.... Could possibly have something to do with it uses much, more power than any other previous card out there. More power=More heat. Any card could have an issue but you wouldn't know it if the power draw wasn't enough to create enough heat to exceed the melting point of the plastic connector.
I’m with you Alex. Looking at the “burn line” as Steve calls it I would imagine that the end of the plug in the cable and where it makes contact with the pin in the socket on the GPU. I wonder if there is an issue with vibration burning these up, I had it at work once where we had to apply a connector lubricant to a molex type plug as the excessive current draw and vibration burnt out the pins
Thanks for this, Northridge. I was never happy with Steve's soft declaration that it's a user error -- when does it go from being "user error" to "poor manufacturer design" if you have THAT MANY "user errors"? But I agree with Jay that Nvidia will try to ride this out, until the 5090 comes out. We shouldn't let them, but there's too many "user error" parrots are out there.
I totally agree with you. This connection on a card that also likes to consume 500 W with overclocking is an absolute botch. 500 W at 12V is a good 40 amps. If the plug is not 100% seated, the contacts will inevitably get very hot. The manufacturers have saved there in the wrong place and instead of the usual 6 and 8 pin connections they have developed a new one that simply cannot handle this performance. Greetings from Dresden / Germany
The plug is poorly designed, and yes it is a big flaw for it to melt regardless of if its plugged in all the way or not. Basically if there's any chance of melting due to improperly plugged then that should be redesigned in a way to prevent it ever happening as you said.
dude, saying bullshit like "regardless of if its plugged or not" is idiotic. Because if thats the case who cares if its plugged, or power supply is weak, or motherboard has some serious manufacturing flaw. I mean, people are idiots and even if they do something wrong they will blame it on others. If cable is the issue, i agree that there should be some type of protection, like card not drawing power at all and is always in idle. Some older cards if i remember correctly could work without power cable but at like 50% power or something. Good way for protection.
@@jaycqc8136 that does not matter, cable goes into connector, connector goes into the card. Not the point i and he wrote. Maybe end of cables into the connector produce too much heat, maybe users OC the card too much. The point is it should have some kind of "protection" if CONNECTOR is not CONNECTED properly. Next time when you read try to understand the point properly.
The problem is the 12V power supplied from the PC power supply. Its the current that heats the connector, not the voltage, so by halving the current by doubling the voltage you could solve this issue easily. When PC's first came out they only needed 12 and 5 volt power (with lower current -12 and -5 volt rails and over time +-3V). PC's nowadays need more power (i.e. amps) to work, but at 12 volts and 500 watts (rendering graphics with Blender for example), you will be drawing more than 40 amps!!! The chips on the card don't use 12 volts. They are stepping down the voltage to say 3.3V/1.2V etc. A lot of the graphics card PCB space is just the on-board power supply to make the lower voltages. If the card could take 24V you would need only need approximately half the current (i.e. amps) to run the card as the incoming voltage would still be stepped down to voltages that the chips need anyway. Of course, this would mean another standard of power supplies with specific outputs for GPU's, but it would make the issue of over heating connectors go away.
I think Steve is trying to save face...I think he's in error...the 4090 was probably drawing too much current at certain operating conditions which caused the melting.
I totally agree with you because if a product has such high chances to fail in hand of end-user then it's the designer's fault. Its been years since we are using PCI-E power connectors for graphics card but never had so much issues with their connectors, cable mods, cables, etc. I'm not a super computer geek but I am gonna put some points I have in my mind- 1. Low melting point plastic used in connector 2. Female pins are not thick enough to handle that much watts 3. Play between in male and female pins inside connectors even when the plastic body of the connectors feels tight and firm I would like to mention again I'm no computer geek. Just putting what I had in mind.
agree with you, so many cards are having the fault, when designing a product for use, you have to design it so user problems don't happen at all or very unlikely to happen
Alex, you should run this card under thermal camera and measure the max temperature under load. This max t is the problem, the plastic used to make the connector cannot take this max t. It simply melts around the pins delivering so much power.
@@ariewijaya1679 soldering reflow in factory is done by machines in seconds. When heat is generated during hours of playing, it causes the plastic to eventually melt and liquify around the scorching pins.
@@ariewijaya1679 nop. those things are laser soldered, search for videos in YT you'll find it fast... and even if they weren't its easy for a electrical joint to heat way more than that after all thats how we melt steel together...
Get a 3 way Collab with Jay and Steve! All of you bring failed cards, and work together. This cuts out any “maybe this, maybe that” from the analysis process, and each of you can share your personal strengths. Plus, GN has excellent connections to industry diagnostic tools.
@NorthridgeFix Hi Alex ! Thank you for your work and your explainations. Isn't it possible to remove the connector and plug wires with sensors (temperature, amp, voltage) between the card and the connector to measure and understand what's happenning here ? I don't understand why nobody tried to measure and find the real problem.
I'm deeply invested in this! Love your videos, Alex! I'm interested to know whether anyone has watched and tested this card with a thermal camera to see where and when the heat starts building up in the connector. I believe that there is a huge current draw on those specific pins when the card is under high processing.
Didn't Gamers Nexus do this in their 12vhpwr deep-dive? I guess the heat camera is not close enough to figure out specific pins, or that they modified their cable too much.
Just want to say first off that I appreciate all the work you all do. Northridge, Jay, Steve, etc. My opinion. NVIDIA needs to man up and accept that they f**ked up and issue a recall and fix all 4090's released to date and figure out why this is happening. imo, Id 86 that connector. Go back to the 8 pins. Id rather have 3-5 connectors that are PROVEN to be sturdy and reliable no matter the application/situation than use 1 connector that is more a fire hazard than anything else in my house. imo, they are trying to crank too much power into these cards. They rushed the 40 series by overpowering them and it is now showing. Now wonder EVGA left the game. They saw what was coming.
I'm pretty sure it's way more productive for players like NFix, GN and JayZ just to check resistance of each connector's pin rather than overthinking who to blame :D Or even better - desolder a healthy pin from a new card and test it against max amps allowed AND designed by the manufacturer
On my 4090 I had to push much harder than I was used to, to properly seat the cable. I can totally see someone not pushing it all the way in. In fact there are documented examples of this happening. Doesn't mean I think all the failures are related to that, just that it's certainly a concern.
Long time viewer, I appreciate your videos and enjoy learning from you. I am very familiar with the whole 12VHPWR saga affecting users of 4090s. What I would like to point out is that in the case of the 4090 you received Cablemods connector still attached, you are stating for a fact that the cable had been properly seated and that it still melted. However, you don't actually know if the connector suffered prior damage/melting, before the user noticing some instability lets say, and then pushing it in fully. But by then, the damage to the connector was critical and then continued melting. Unless you have emperical evidence otherwise, I think you cannot jump to a conclusion that you know for a fact the connector was properly connected at all times, and that it's purely a connector issue. While it is possible that your assumption is correct, people should remember that correlation does not imply causation.
yap, cuz that makes sense! A guy that is aware of this problem and even buys an adapter to make sure that he isn't putting too much stress on the connector by bending the cables, most likely did that or maybe burned the connector on purpose and then, while still hot, inserted the adapter just to feed this circlejerk about melting connectors. Im curious about what is realy causing it to fail (most likely a number of factors and one if for sure the heat generated by the current) but the common thing is 4090's and their connectors so it's pretty obvious who should be responsible
Incompatible pins, obviously. If you run that much current through them, you gotta certify the matching plug's contact resistance to be below some safety limit when it's "plugged all the way in". Neither Nvidia nor Cablemod with their Chinese suppliers bothered to come up with a contact resistance certification, so they both should pay for the RMA.
As a product designer there is no debate here. A product needs to work for the intended user base. If there are excessive fails of one type, either the design or the user instruction is insufficient. It happens sometimes in products, especially high-performance products. I am no PC expert but it looks to me this board uses a lot of power. This will always increase the risk of power related design challenges (such as avoiding overheating). The cause of the fault is not the ultimate question. By definition any unexpected, non third-party related. unacceptably high fail rate demonstrates a product failure, especially on a high-end product. It happens, it is not a crime, but in my opinion the manufacturer needs to address it.
The 4090s need a three cable adapter that DOESN'T plug into the card's native plastic connector, but instead is soldered to the card in place of the original connector. Problem solved.
Any connector, plug or anything that connects into something else needs to be designed with the rigors of regular plugging, unplugging, cable bending and ideally withstand at least some lateral forced placed on it. Even though you're not meant to be repeatedly unplugging and replugging a GPU's power cables, they can sometimes require some force to connect properly. Even if it was caused exclusively by cables not being seated correctly, then it's still mostly on nVidia. Throughout the years as I've built my PCs I learned that you often need to apply some force to connect GPU power connectors and ensure they are seated properly. There is more often than not a subtle felt feedback given by the locking lever latching onto the connect once it is seated. There's nothing special or unique about that, and that's the hallmark of good design. nVidia's 4090 cables and connectors came with warnings not to bend the cable within a certain distance from the connector. I mean, what the fuck nVidia?! They're basically admitting that they've made a structurally weak cable, most likely exacerbated by their ridiculously dense connector.
Obviously the connector. New connection, new issues. I have little faith in gamers nexus, and shared your views on his first tests. He pretty much helped the sale of that card while nvidia were sinking. Now almost everyone has the same issue. No time for influencers.
I agree with a little of everything. I believe Jay and Alex are correct in that the problem bottom line lies with Nvidia. I believe Steve is correct t in his assumption that, because of bad design, it enables errors the user may never be aware of. The cable gets seated, and it feels tight, but it can walk itself out just enough. Like Jay mentioned, there is still give even when fully seated. Also have to remember Steve is going off of the small sample size he tested and, he did say he was willing to do more testing but, still believes that in the end, it does lie with the user to insure everything is correct. Yes and No about that, if the connection feels tight and secure, especially with a cable mod connector and it is still melting because even that has wiggle, it's because of bad design on Nvidia. I'd love to watch the three of them sit down and discuss all of this together and do testing.
People put way too much importance on what Gamer Nexus says, as if those clowns are experts in the field lol. For the record, Steve has never held Nvidia accountable for the burning 4090 cables and AMD for their recent fried CPUs---it's always "user error" and "Asus' fault."
It becomes clear that Nvidia simply chose the wrong connector for this level of DC current flow. The connector is simply inadequate to carry the current required by the card.
I agree north ridge on this it's not user fault it's a defect and to be honest I agree there is far too much power going through that 12v connector there should be 3 other connectors to share that 600 watt load as seems to me 12v connector cannot cope with the power consumption
It would be interesting to see what happens if you hard wired matching connectors to the cards. Having the larger gauge wires and higher amp rating connectors , may be the key, to success.
@@--_DJ_-- The wires are connected to a very small connector.I think I heard that the board draws 300 watts of power.Those connectors might ,on a good day handle 100 . Also, look at tha 12 volt cables, if wires and connectors of that size were on the board. there would be no issues.
thanks Alex for taking a stand on this. i have been working on computers for forty years now and have never seen damage of this type. The other technicians need to at least take into consideration their view isn't the only view. Only a fool thinks they know everything, a wise man listens to other people.
I agree with the guy that is fixing the cards that the problem is or appears to a design issue with the OEM connector on the card. I have experienced issues with connectors of various types in the past and some of those connectors I have fixed by using a zip tie or multiple zip ties to pull the two sides of the connectors together very tightly and lock them together so they do not move. On those video cards something like a zip tie might not be possible, but something like a tie might need to be part of the connector, or added to the connector to lock the socket and the plug together so they do not move apart.
I believe it's Nvidia's fault for making a connector that puts out THAT MUCH WATTAGE THAT SMALL. Any half Bates electrical engineer knows that the more current you're pushing through a wire the thicker it needs to be. The cables on these connectors are thinner than the wires I use to put my slime trackers together. That's pathetic
It's funny that people still defend Nvidia with all the evidence presented. Alex is so honest that from a business perspective, he's making money from repairing customer's 4000 series GPUs but he's not blaming user error for burnt connectors. Nvidia pays influencers to redirect blame away from the manufacturer and yet Alex is putting the spotlight on who is ultimately responsible for this issue.
The 12VHPWR connector was always a bad idea. 50A on 6 tiny pins doesnt work. (6 pins because you need +12 and ground @ 12 v = 50 A or 9 A per pin). That is more current than a home water heater/cloths drier, look at the connectors on those.
Looks like you have some more data points to further investigate this problem. You and Steve should get together and see if you can come to a mutual conclusion. :) I'd totally watch you teach Jay and gang how to desolder/solder! Thanks for the content and keep up the all the good work both on the bench and behind the camera!
I'm watching your videos for 5to6 years now and I agree with you Alex what you have said makes sense to my ear I saw the connector was plugged in till to the end you have tried to separate the connector 90 degrees for 5 minutes and some people what they are trying to say here is that 1+1=11😅 for me the factory must to take the responsibility for what they have done. Alex keep going like always 💪 and thank you for sharing you experience with us all those years❤❤❤❤
As a electrical specialist you can calculate consumption and size consumption if matching with conductors. If you have 500w consumption of video card and wires and connectors are at limit, every slight variation will cause burning connections. So the issue is in the middle, the connectors have either to big margins of contacts. There is also a fault of video card producers, they should make maybe bigger thicker male pins of connectors and will reduce margins of bad contact. One thing is for sure, that is a issue with bad connection, or terminals doesn't match with power delivery, they should make them thicker to support all that flow of amps.
I agree 100% with what you said, but I would say that the user not fully engaging/seating card and connector is a factor simply because it either accelerates the failure or creates a different type of failure than what you've been repairing. Still on the manufacturer for essentially causing the issue though, just to be clear.
I tend to trust service technicians, not video makers/"reviewers". Good job NFIX, the connector is not ideal for this power, they gotta test how much power goes through each pin etc. Nvidia should have resolved this long ago and accept cards for repair under warranty. So it's strange to see so many cards at 3rd party repairs.
At least he says "we could be completely wrong", but yeah, I agree that Steve is too ready to blame "user error". Since you had the cable, examined it closely, and are an expert, I would agree with you!
This problem falls directly into Nvidia's lap. If Nvidia thinks it is user error, then the user should not be allowed to install their 2 to 3 thousand dollar cards. It should come with a pre-installed cable.
The 4090 is a flawed design,even a hobby tinkerer like me can see that,not fit for purpose.Can it be rectified? Absolutely,even a stupid noob like me with mere mechanical engineering background understands that.Manufacturers and designers need to take responsibility for the products they put out instead of blameshifting and deflecting it on to the end user.Well done Northridge Fix for explaining water to the thirsty.Never stop
agree, NVidia should have not implemented this change al together, AMD was able to develop the 7900XT and XTX yet use the traditional connectors while still offering competitive performance, as a builder and a person who has a system integration business, this is a huge problem for us, the original cable extension that comes with the 40 series cards is just bad, it ruins the whole echo system, working with these cables on our systems is really annoying, making it look good is super difficult and will almost definitely need a custom cable which is also annoying, and now seeing these issues that impact sales, warranty claims, and client concern and worry, NVidia really pulled a number on all of us.
While I, and hopefully the entire userbase that follows Jay Steve or Alex, certainly appreciate the work each one of them devotes to investigative journalism backed by proven and professional techniques as well as decades of experience, certainly do not appreciate or advocate for any of them to work against each other and basically execute a tech pissing contest about Nvidia's melting 4090 power connectors. Grow up boys. Work together. Stop throwing rocks at each other. You're making us, the followers, the subscribers, the bread and butter of a chunk of your income, cannon fodder for this silly bickering. Work it out. Post something collaborative. We're not appreciative of tabloid tech news and updates.
I don't think anybody is against anybody. I made the video from my point of view and the way I see it. Opinions can differ. If all people shared similar opinions, the world would stop and no further innovations or advancements would be made. Looking at this from 3 points of view is not enough. That's where the viewer comes in. I mentioned at the end of the video, |let me know what you think, do you agree with GN, Jay or Me?" The topic is left neutral and open to further investigation. The way i see it, the issue remains an Nvidia and not a user problem. What I said should be taken with an open mind and not as someone attacking anyone.
It seems like you have almost the same opinion just you place all the blame on the 4090 having a design flaw connector, and Steve is saying it's both user error and a design flaw. Either way the conclusion is the same that the 4090 12vhpwr connector is designed wrong and needs to be fixed
After reading your comment, it is clear that you are too ignorant to realize how ignorant your comment was showing you'd be a hard one to teach, and I cannot be your teacher, but I will hope that you will read your one-sided very simple comment once more, despite how hard it may be, and realize it has only collapsed on itself due to you crying heavily over the page you are writing on, smearing the letters as you try to make an argument about how men shouldn't argue when it comes to different opinions like the one you just gave. Also while speaking on everyone else's opinions as if you know all. Broaden your horizon. Speak for yourself. Not everyone else 👌
Endorphin, how you came to that conclusion after watching this video, I have no idea. Quite honestly it doesn't even sound like you watched the video. None of the names you mentioned have been fighting or arguing against each other with regards to the 4090 connector. They simply came to their OWN conclusions. Obviously with opposing viewpoints someone will be correct and someone won't. To the community however the fact that all these techtubers are willing to go to great lengths to showcase the issue, only benefits us the consumer.
@@NorthridgeFix Well said
Agree with you Alex, if the connector misaligns so easily, the manufacturer should take responsibility.
which cable, the 3rd party cable ?, how many of these melted while using the standard cable, if I was looking for a failure point I would be testing the 3rd party cable to see if the materials used are up to the job i.e. can carry the current without heating up, when conductors heat up they increase in resistance to the current which them causes more heat and eventually total failure
@@geoffh4578 the problem is its not the third or 5th party cable that is melting.... its the actual connector that is melting.
If you know a car is going to burn down when you start it, would you still buy it?
If you buy one, knowing there is a problem, then the only fault is yours
@Geoff H They had this problem from the start, when only 1st party cables were available. It isn't the cable, it is the connection at the pins. This has all been well documented at this point.
@Kamelt Oefan What if only 1 in 1000 burns when you start it? Is it your fault then too?
gtx series: nothing bad happened with cable connectors
rtx 2000: nothing bad happened with cable connectors
rtx 3000: nothing bad happened with cable connectors
rtx 4000: cable connectors melting
users: - oh no, I forgot how to connect cables
lmao ... right on point bruh
Or like 20 years of general PCI-E power connectors no problem - New proprietary connector comes out and has issues - must be user error lmao
Exactly
hahaha
The thing is also... There's wayyyy more people now building their own pc's than before! And let's not forget the crypto miners niche also... So, more people can equal to more mistakes/accidents. It's important to not to jump to conclusions and investigate FIRST
Whatever ends up being the case, these cards cost WAY too much money for them to be damaged permanently by slight disconnect. Nvidia should fix this issue and take responsibility.
Completely agree
No wonder GPU sales are low.
It's not just about way too much....as current market stands, the flagship GPUs now cost thousands of dollars ....on top of that, this users bought this flagships just its to melts because supposedly user error....we gave away too much money to this corporate to treat the consumer like shit!
gamersnexus is such a fkin simp Xray and Lab tests to prove the user is to blame the mental gymnasitcs on this guy wow !!! nvidea should hire him
@@triadwarfare Probably has more to do with the video card gouging that only recently has cooled down left a very bad taste in a lot of peoples mouths. People are now just plain refusing to upgrade until they absolutely have to.
You are absolutely right, people didn't suddenly forget how to plug in a power connector, and ONLY on 4000 series gpu's.
Frankly, it's a disgusting practice when any company refuses to accept responsibility for their products and instead tries to gaslight their customers.
Especially a trillion dollar company
I’m on a MSI Ventus 3x 4090 with the 90degree cable mod adapter and Cable mod PSU cable. I’m also an independent IT consultant and PC builder who has built thousands of PC’s in my time. I have been keeping an eye on my system since the 4090 was installed. Non of my cables are bent or under tension I’m using an AIO and 140mm fans so little vibration is in the system. The GPU is power restricted to 80%. About once a week I check the cable is seated home and about half the time it has worked it’s way out a small amount. Thinking it could be vibrations from the fans I replaced them all with Noctua ones. Still happens. From what I can tell it’s due to thermal expansion. With the connector repeatedly heating and cooling with use it’s causing the connector to work it’s way out then once it’s far enough out to increase the resistance it starts to melt or arc. This is totally a design flaw. All the cards should be recalled and the connectors replaced with larger more substantial ones before a fire is caused and someone dies.
this guy PCs
im pretty sure you are right
There's a reason why 12V was always supplied over mutiple cables, to spread the power evenly. Combining all that power into one socket on a power hungry card is a disaster waiting to happen.
This
Well with hwinfo you'll see atleast on my 3080. It's far from evenly. But I agree with you I like the fact that the power is running through 3 connectors instead of one.
Another vote for THIS.. And that's where actual proof could come out that it's a bad design. If under load the individual pins are measured for actual current draw, then there just might be a smoking gun (pardon the pun) in that some of them will show a very large imbalance in current from one to another.
Too little surface area to pass all the voltage and current causing this issue.
Intel made the spec, and they chose not to use it. Should have been nVidia's first clue, but they've proven unable or unwilling to "read the room" as of late. On just about everything to do with consumers anyway, they probably don't care because their AI/enterprise side is cashing checks.
You are totally right Alex. We never had these problems when we had the older style of connectors because they were oversized for the job. The connectors now are theoretically up to handle the power of the cards..
The only user error there could possibly be here is purchasing a 4090
The 4090 should be $799 max
@@HaloinfiniteEternal Fantastic observation
4080 melting tooo
If you look at it from another angle:
I bought 4090 with the money i collected for 2 years. And now i'm learning soldering and electronic repair while waiting for the delivery of my wannabe mini laboratory to arrive, soldering station, heatgun, thermal camera, microscope, consumables, the whole shabang. I'm always wanted to learn e. repair, but was very lazy about it.
Now i have the most powerful gaming GPU in the world (that works, hope to not jinx it), and soon skills and equipment to fix it if needed.
Thanks for a little push 4090!
😂
I have the feeling the connector itself is undersized for the 500+Watt. On my previous card there were 3 connectors on the GPU to run with 300-400W and now we have 1, which is way smaller, which runs 500+.
I think if you're suddenly seeing a massive uptick in melting GPU cables when neither the connector or the average skill and ability of the users changed then there really is no other alternative than it's the GPU deaign.
not to mention small problem of where it starts melting and overheating ...being from the card side always
pretty much thought the same thing..
Maybe it's the influx of new, highly demanding, poorly optimized games that started coming out not long ago ;)
People has built computers for like over 20 years now with GPU:s that need external power hooked up to them. And this seems to be first time that the connector melts. So in my eyes the reason is VERY obvious, the connector has to be too undersized for the application. It can't be that suddenly hundreds or even thousands of people don't know how to connect that connector right.
6 and 8 pin atx connections have not failed because of their design, ever, and ATX standards have been around about +/- 50 years now. Defects and mistakes happen, but this connector is failing year one, repeatedly.
Some of it was user error, sure, but it's all nVidia's fault that it happened in the first place.
@@x8jason8x ATX origins - 1995. I know when you are young the difference between 30 and 50 is indistinguishable but ATX is not even 30.
Well maybe there are thousands of people who don't know how to connect cable so Nvidia should change design, provide sufficient training on card installation or put requirement that only qualified electrician may install 4090 in your computer.
Other connectors have melted before, this is just blowing up more because they're new cards and the connector was causing controversy before the 40 series even released.
@@stanimir4197 I'm not young first of all. Second, I'm not talking about form factors. ATX power supply connections can be found in things as old as IBM Selectrics.
I know when you think you know everything, you tend to neckbeard it, but probably save the fedora for the real dance, pal.
I'm not usually a fan of podcasts, but I'm really interested in watching Jay, Alex and Steve discussing this issue together.
Steve is wrong on this one, he can't see past his own ignorance, gotta love the whole money he's getting from Nvidia.
@@redragongaming "The shill gambit, a type of ad-hominem and poisoning-the-well rhetorical move, occurs when one party dismisses another party's arguments by proclaiming them to be on the payroll of some agency."
@@Ryan_DeWitt Your comment doesn't disprove what i just said, it only enforces the facts and those are with me. You can't prove Steve isn't involved with Nvidia because he is. Of course he's gonna side with the corporation, doing the opposite could mean penalties to him. Once you're involved with the rules you gotta play by the rules, otherwise no one's going to like the hard cold truth. I'm happy to say Alex and Jay are on the right track on this one, corporations are still made up of people and they gotta respect the customer, not shift the blame from one another.
@@redragongaming Do you have proof Nvidia is paying him off to have his opinion? The burden of proof is on you since you are the one making the claim. It is one thing to say you disagree with his conclusion, a whole other to make a "he is a paid shill" accusation.
This may come as a shock, but people may disagree with you and that doesn't mean they are getting paid somehow to do so.
@@redragongaming Go back to a "how to argue 101" class then return to this comment section just to start deleting your below-the-grass-level-brow comments. The case of Hitchens's razor in full public view here. One can hope something was at least learned.
I like Steve voicing out many manufacturing design errors and fighting for it. But in this one am siding with Alex and Jayz, it's Nvidia's fault to use a fragile connector like that.
Gamers Nexus knows nothing. They always try to get attention and pretend to be smart.
@@realtuner3560 Great example of you're damned if you do, damned if you don't. Don't try to do advanced testing, "you know nothing, get some good equipment and people who know what they are doing." Get the good equipment and consult with people who know what they are doing "You are just pretending to be smart."
@@realtuner3560 Yeah man i figured out Steve doesn't actually knows so much as he pretends to. I even found a webpage with him talking about a problem with the PCI Express Downstream Switch ports that weren't working properly on a motherboard, after that he posted steps onto fixing the issue but surprise surprise these steps were not relevant, the so called steps seemed to be nothing but click bait, it's how i know he's most likely a sellout.
@@Ryan_DeWitt Let it to a certified lab and run a test procedures for connectors like the ones used in automotive industry. Period. You dont need to put a show. Get your test done under standards in a certified lab.
@@38911bytefree This isn't the auto industry. It is the gaming industry. That is what youtube is. Info with entertainment. Channels that post cold info with zero entertainment do not last long. As Alex likes to say, you get a lot of scientist in the youtube comment sections. Plus they do consult and test beyond their capabilities or equipment that they do want to run are sent off to these places.
It's amazing that the new connector made it to users. Even from a laymans point of view, the change from 2/3/4 8pin plugs, to a single plug would make me expect failures. Perhaps it could be executed properly, but it just seems undersized to achieve any sort of thermal isolation. There should be larger gauge wires and larger pins, at a minimum, but a more sophisticated latching mechanism would help as well. Software or sensor checks against the connection would be great.
Was super paranoid about this and made sure my connecter was fully seated numerous times. Literally not even 20 minutes ago while playing Diablo 4 I smelled a plastic burning smell from my not even a month old PC and knew where it was coming from. Opened up my case and the connecter is hot to the touch. There’s definitely some melting and I’m scared to pull it off.
Which 4090 do you have?
@@dudee789 it was the white strix OC
Agree with you Alex so after 20 years we have all forgotten how to connect a GPU
Nvidia needs to take responsibility for its own products.
Classic move to always blame to the customers. In indonesia is something burn they will blame the user psu 😀
We have never seen as many melted connectors as we have on this nvidia 4080 - 90 series. It is very rare to see any melted 8-pin connector of the millions of video cards that exist in the world. If we only take this percentage of failure that we saw in the last 4 months in this generation 40... I believe that would be enough to carry out a recall and review this type of connector. For me Alex your point of view is the best!
This!
I am ALL FOR a NRF and Jayz2cents collaboration! Bring it on!
I agree with Alex. Throughout my many years of tinkering with electronics, I've never had a connector melt because it was plugged in partially. Most of the time it just won't make an electrical connection, and a re-seat fixes it. The only time a near connection poses a problem is with high voltage/amperage where electricity may arc across a small gap. Repeated arcs will melt even metal as seen in arc welding. The only other time they melt is when I run too much amperage over a connection but that was me being lazy and knowingly pushing the envelope on underrated wire/connectors.
If the connector fails through regular usage, it is just plain bad design. If you need a degree in electrical engineering to properly plug in a connector, it's also bad design in any commercial equipment. Imagine if plugs of any home appliance would burn your house down if they were not absolutely seated 100%.
Alex you are absolutely right about this problem. It can not possibly be a user error. I love your work. Keep it up Thanks
I also think that NVIDIA's fault is more precisely the connector on the cards and the high consumption of 400+ W, a simple solution is to return the old connectors that we have used for the last 10+ years and which AMD cards still use. If I said something wrong or wrote I apologize in advance. Thanks for the great videos NorthridgeFix, I follow you regularly, have a great day. 👍❤💯🔔
Its really way too much watts for just one small connector...
AMD was going to use the same connector but then saw all the problems and was like "oh hell no" and immediately redesigned their boards. Intel *designed* the connector (and didn't fully test it or engineer it properly), not the standards board, and even they don't use the connector for their motherboards or GPUs. Nvidia just doesn't want to admit fault because recalls are expensive, and they have a company policy of never admitting fault even when they screw up, and attacking anyone that criticizes them.
@@radoslavsk8591 I don't know but the cable on my MSI A1000G PCIe5 power supply says Cleary 600W on it.(meaning it can support 600Ws) I think is a combination of some 4090 models, but also power supply models, maybe people are trying to overclock cards that cannot go pass 400Ws or near 600Ws.
and this is why you Frame limit and / or Undervolt the GPU. It's basically a must, now a days.
@@BeatmasterAC NO! STANDARD USER WILL NOT UNDERVOLT a GPU! Because they not added 2 or 3 normal connectors!
Great video Alex! The 3 of you should get together and talk, and maybe test some of those cards. That would make an incredible video. Steve is more protective, I think, because they have done extensive testing on that problem already. Jay is more on track with what you're thinking; you could also show Jay how to resolder pins on CPUs, the right way. Wonderful work Alex, as always!
A recall needs to be issued for the 4090 cards. NVidia has enough resources to do their own analysis. They should hire Alex to test before releasing their product. 4090 is a fire risk waiting to happen.
Exactly, and one I'm not willing to take.
If user error is THIS prone to happening, it's a bad design, period.
or the design is just at the limit. so theres a chances many defect in one batch production.
I.e. Not fit for purpose.
I'm with you, I trust someone who dedicates every day to fixing the poor work of video card manufacturers more than someone who receives gifts from video card manufacturers...
I think a part that Steve missed is also the broken power connectors that seem to be a problem with those cards. There seems to be more going on than just not pushing in the connector all the way. A 250 gpu backlog is pretty insane, good luck!
These cards are huge. In my case, the CPU AIO hoses weren't long enough to get around the card, and I couldn't mount the fans. It's possible people are trying to install these into cases that are too narrow and these issues happen when trying to install the side panel.
It might be problem with the card itself but I think the point about cablemod sending them was that cablemod was like a buffer for the cards and collected them from longer period of time and then sent them all to you in short period of time which made more cards coming in shorter period of time than if users had sent them as they failed themselves.
I'd say the video itself squashes this one as both the video here and from Steve at GN state eight cards from cable mod. Whereas this video showed a whole rack of cards from users waiting to be fixed. Those eight cards aren't making or breaking this numbers wise.
@@Stubbies2003 true, but that is a fact that only came out in this video so hard to blame GN or J2C for that when it was completely unknown info to them
@@Henrik_Holst Yeah Jay has already leaned towards something wrong with the card at this point. Steve didn't shut the door in the video either. He left it open for something that wasn't known at the time of making the vid.
I agree with you totally. I wonk on automotive electronics. GM and their micro connectors for years have been too light to take the amp load within the circuit. The small terminals in the connector will start to heat up due to the fact they are too small. A fuse is installed in a circuit to be the weak link in case of failure on that circuit. If you have a weak terminal at the connector that is too small, you have made the terminals into a fuse. Another weak link. They may not burn as quickly as a fuse but they will sit and simmer causing melting problems in the connector.
This makes so much sense too. They are just pushing too much power in a small area
There should be a class action lawsuit against nvidia over this mess.
If it is always the same pin that melts, perhaps it is something related to the power that is used on this specific line. it means that probably one power phase consumes more than the others when feeding the gpu.
it would be interesting to test the consumption of this line to find out if this is the problem
The fact the 3090 had the same connector and we didn't see large amounts of burnt and destroyed connectors means that there is something wrong about the 4090's.
+1. Was about to say the same thing.
It very well could have had the same issue. It is just that its maximum power draw wasn't enough to exceed the melting point of the connector.
4090's can be pushed to higher power maybe? it's a waste on such an efficient node and architecture, but it's possible nonetheless. There's always that one person that cranks every setting to the absolute limit and then runs some excruciatingly GPU intensive benchmark that maxes out every square mm of the GPU chip.
3090 cards didn’t use with modded bios up to a thousand watts of power
People pushing 1,000 watts of power through a small connector is why they melt then go boom!! No plastic separation of metal pins, then they touch each other… this is whY high wattage applications use big gauge wiring and big pins and a connector that is huge that separates pins so they never touch no matter how hot connector gets…
Alex, I follow Jay and GN also, but if possible, maybe you can continue this topic with them, as I do believe that you 3 will deliver a proper RCA. Again, thank you for all the work you put in!
i would include IgorsLab as well.
I completely agree with your explanation Alex. In my understanding, it's a real shame how engineers are not able to design this card to manage a electron flow. That is the very first lesson to learn for a proper project. Engineering sucks.
Why are you agreeing that the engineering of the cable and/or adapter is not at fault?
25+ years of building PC's.
Same deal, 4090 connector melted and dealing with getting it fixed...
Simply put, how is it that countless PC's over the last 25+ years have never had "Oh the user needs to plug it in fully" issue.
Yet, as if by magic once Nvidia put out a new first gen connector, all of a sudden it's the users problem?
Sounds like a great concept!
New Business Idea: I'm going to design a parachute where as it works 100% so long as you pull the rip cord at "just the right speed". Any faster or slower, you die. Remember... that's on YOU for not timing your pull correctly! Get it right user!
*sigh*
Not only that these connectors that are melting ARE connected tightly. It doesn't matter. If it's fully clipped in any amount of wiggle, bump or otherwise should not be an issue. Doesn't happen with this commonality with SATA connectors. Doesn't happen in mass with molex connectors. Sure, some of them do at times and can of course but no defined pattern like we're seeing here.
Nvidia improving their connector in a few ways should say it all. Why improve something that doesn't have an issue?
NVIDIA is weighing the possible loss of reputation against the cost of recall to see if they really need to do anything at all. They are still counting on the shiesty sales agents to confuse customers and the fanboys to continue to show motherly love towards NVIDIA and defend them to the death. They are waiting to see if those two factors will be enough to drown out the customer complaints. This is how large scale business with a huge fanbase make business decisions. It's a natural phenomena. ✌🏽
True
man this gpu series is seriously a joke better to stick to old gen and let nvidia stop being greedy.
The CPU is a beast, but nvidia is going to let some cheap ass component ruin years of R&D.
@@hgbugalou It isn't a beast though. The 4000 series is hilariously awful. The 4060 Ti is debuting to be in a power scale of the 3060 Ti/3070? Bitch what? Why is a NEWER card in literally the same power tier as the old generation?
The 4090 doesn't do anything the 3090/Ti didn't either other than arbitrary "we locked off DLSS 9001 for ONLY 4000 series, because raisons" nonsense.
It's an unsafe and largely unneeded mess the 4000 (de)generation.
let me get this straight. lets say a car manufacturer has this special lug nuts for securing the tires down, user can install the tires correctly, torque them down correctly but there is still a chance of the tires to wiggle a bit if you turn them too much and some tires end up falling off. would that be user error? NO. i dont know what Steve is smoking. the evidence of that 90 degree adapter melting while fully seated is all the evidence you need. this is on NVIDIA.
The problem is everyone is right, its a poorly designed system which invites error from the user (partial connection, pressure on cables) and design fault (material ingress, mechanical stress fault).
Hopefully NVIDIA never enters the electric car market ...
Agreed
Steve is smoking Nvidia's 💵JMO
Steve is just a shitty shill now. GamersNexus has 0 credibility going forward in literally any matter.
No one in their right mind would even think to blame the melting cable saga on “user error”, unless of course, you’re trying to protect your sponsorship deals. I mean come on, it’s a cable for gods sake, we all know how to plug one in and it’s insulting to be lectured otherwise.
That's because these days, it seems like everyone thinks everyone else is dumber than they are.
GOOD LUCK WITH THAT.
I'm sick of it all.
Steve really wanted to convince us all it was user error. To an absurd level even. And guess what the cable and the connector are all being replaced, so i guess he was absolutely wrong, or was he wrong? maybe he was just doing someone a favor, no recalls. Thanks Steve.
I think you're absolutely right, it's a problem Nvidia has to solve, it cannot be solved by anyone else. I see the problem in an overstress of contact area which is considerable smaller than in the older solutions with up to 3x 8-pin connctors. Even if you would use some clever current balancing circuit to even out the currents every one of the 6 power pins has to carry they are just f*cking small for the job. If you take Tyco MQS contacts as a comparison they are rated for 7,5 Amps per pin @ 80 degs C. Multiply this by 6 and you get a maximum wattage of around 540 W under optimal conditions and ONLY if every contact takes an even amount of the load.
I can not get how mad an engineer might be to think that stunt may go OK taking 3 to 4 8-pin power outlets and put them on 12 pins in a smaller form factor connector. This is absolutely bonkers, no need for doing so if your education in physics basic course at the university has left some marks.
Even if you put the female connector straight into the card and give it a firm press, there will be a problem coming, maybe from micro vibrations of the contacts, reducing the effective area for the current, perhaps pulling small sparks, getting warmer, thermal changes may lead to changes in the dimensions of the metal pins and contact tongues, perhaps even some micro wearout happens after some time. The video card is not a static thing, there are fans moving, they speed up and down, other moving parts in the computer, even the user loading a CD, pressing a button, plugging in USB devices and so on. If you push the contacts far to its physical limits without any need to do so beside marketing reasons ("it is MY CONNECTOR" - Gollum image suggested while reading this) you will get into trouble. The trouble is there. And the trouble is exactly this connector or better the way it is used: too much amperage for the contacts, way off any reasonable specification limits I as an electronics engineer and a long year developer of automotive electronics would consider to be safe in any way. This connector is not right for up to 600 watts of power. Not if you wanna be on the safe side.
I thought a lot lately of replacing my 3090 by a 4090. But hell no, this connector seemed wrong to me the moment I first saw it months ago. And I'm feeling confirmed. I will not buy any graphics card forcing me to use this for sure. There is a need to revert to the old connectors in my opinion. But I do not think Nvidia is ticking this way. Which is a shame.
Same. I was about to get myself a new 4090 but as I kept looking for the best brand I came to learn about this SIGNIFICANTLY DANGEROUS issue. People: this could set your god damn house on fire. It's no joke. I am not willing to risk it.
Jay agreed with you and so do I, if 4090 cards are coming in bulk melted on the pcb nvidia should recall them. Even if it was user error for 2000 the card should have built in protection. I have limited my card to 300w until this is resolved
Why? If it melts they will give you new card anyways, won’t they? Bth there are cases melting with powerlimiting anyways
Alex I am 100 percent in agreement with you that it is related to Nvidia! Too many issues that it can’t be user related! I have been in IT field since 1991 and built many pc and never have had a melting issue on connectors until 4090! So I am def going to say it is manufactured issue!
Mine just went up in smoke today. I was one of the people warning others to be careful with the connector late last year and was super cautious myself. Only ever plugged it in once and made sure it wasn’t bent excessively, etc.
Glad I watched this video. I under the full impression that this was determined to be user error. As long as you attach the cable correctly, you're good. Seems like that's not the case based on this video, or at least not the case all for all cards. I was going to wait for a price reduction and get a 4090 then, but I've decided the melting risk isn't worth it. I'll wait for either a redesigned 40-series card with a safer connector or just pass on the 40 series completely and see what they do with the 50 series. I'll also give AMD a serious look finally, as this may be the time.
Once Nvidia and their partners release a card with a redesigned power connector, that will be the time to buy - until then these cards are just disasters waiting to happen.
Exactly!
I will just buy the old generation 3090, even with heating memory problem compared to melting cables is an ice...
You probably won't have to wait long. The next time a 4090 comes in with a cablemod melted to it, send it to Steve for x-rays.
At 18:19 you can see that the clip is ON, but there is still a lot of wiggle room there that could allow the connector to disengage. I could see this inadvertently happening in the power cable routing back to the power supply. Maybe that is putting stress on the connector even though it is fully plugged in. Since Alex has seen melted connectors on the card with the Cablemod installed, that's where the plot thickens.
It can only disengage if its physically touched after its installed or its under extreme pressure which the user should obviously know if they installed it.. Through my experience I've installed hundreds of gpus and no matter how tight the wires were once things go on they normally don't move. This is only happening with 4000 series and that should be enough evidence by itself.
So all the other GPU's that user's buy they only do not plug the cable properly on the 4090 but they do on all the others, how strange. I still say it is a problem on the card, it has to be, who not all other crads coming in with burnt connecteors?
In the past I have joked with Alex saying he is a PC surgeon, the thing is I meant it too, he knows his stuff and for him to make a stance like this I'm with him all the way. To him electronics is second nature, he know a fault when he see it.
Oh, I don't know.... Could possibly have something to do with it uses much, more power than any other previous card out there. More power=More heat. Any card could have an issue but you wouldn't know it if the power draw wasn't enough to create enough heat to exceed the melting point of the plastic connector.
I’m with you Alex. Looking at the “burn line” as Steve calls it I would imagine that the end of the plug in the cable and where it makes contact with the pin in the socket on the GPU. I wonder if there is an issue with vibration burning these up, I had it at work once where we had to apply a connector lubricant to a molex type plug as the excessive current draw and vibration burnt out the pins
Thanks for this, Northridge. I was never happy with Steve's soft declaration that it's a user error -- when does it go from being "user error" to "poor manufacturer design" if you have THAT MANY "user errors"? But I agree with Jay that Nvidia will try to ride this out, until the 5090 comes out. We shouldn't let them, but there's too many "user error" parrots are out there.
I totally agree with you. This connection on a card that also likes to consume 500 W with overclocking is an absolute botch. 500 W at 12V is a good 40 amps. If the plug is not 100% seated, the contacts will inevitably get very hot. The manufacturers have saved there in the wrong place and instead of the usual 6 and 8 pin connections they have developed a new one that simply cannot handle this performance.
Greetings from Dresden / Germany
The plug is poorly designed, and yes it is a big flaw for it to melt regardless of if its plugged in all the way or not. Basically if there's any chance of melting due to improperly plugged then that should be redesigned in a way to prevent it ever happening as you said.
dude, saying bullshit like "regardless of if its plugged or not" is idiotic. Because if thats the case who cares if its plugged, or power supply is weak, or motherboard has some serious manufacturing flaw. I mean, people are idiots and even if they do something wrong they will blame it on others. If cable is the issue, i agree that there should be some type of protection, like card not drawing power at all and is always in idle. Some older cards if i remember correctly could work without power cable but at like 50% power or something. Good way for protection.
@@milosstojanovic4623 its not a cable issue, in most cases the connector on gpu melts not the cables
@@jaycqc8136 that does not matter, cable goes into connector, connector goes into the card. Not the point i and he wrote.
Maybe end of cables into the connector produce too much heat, maybe users OC the card too much. The point is it should have some kind of "protection" if CONNECTOR is not CONNECTED properly. Next time when you read try to understand the point properly.
I just think it’s too much power for 1 connector, I’m with NorthridgeFix and jay on this one
The problem is the 12V power supplied from the PC power supply. Its the current that heats the connector, not the voltage, so by halving the current by doubling the voltage you could solve this issue easily. When PC's first came out they only needed 12 and 5 volt power (with lower current -12 and -5 volt rails and over time +-3V). PC's nowadays need more power (i.e. amps) to work, but at 12 volts and 500 watts (rendering graphics with Blender for example), you will be drawing more than 40 amps!!! The chips on the card don't use 12 volts. They are stepping down the voltage to say 3.3V/1.2V etc. A lot of the graphics card PCB space is just the on-board power supply to make the lower voltages. If the card could take 24V you would need only need approximately half the current (i.e. amps) to run the card as the incoming voltage would still be stepped down to voltages that the chips need anyway. Of course, this would mean another standard of power supplies with specific outputs for GPU's, but it would make the issue of over heating connectors go away.
I think Steve is trying to save face...I think he's in error...the 4090 was probably drawing too much current at certain operating conditions which caused the melting.
I totally agree with you because if a product has such high chances to fail in hand of end-user then it's the designer's fault. Its been years since we are using PCI-E power connectors for graphics card but never had so much issues with their connectors, cable mods, cables, etc.
I'm not a super computer geek but I am gonna put some points I have in my mind-
1. Low melting point plastic used in connector
2. Female pins are not thick enough to handle that much watts
3. Play between in male and female pins inside connectors even when the plastic body of the connectors feels tight and firm
I would like to mention again I'm no computer geek. Just putting what I had in mind.
agree with you, so many cards are having the fault, when designing a product for use, you have to design it so user problems don't happen at all or very unlikely to happen
Alex, you should run this card under thermal camera and measure the max temperature under load. This max t is the problem, the plastic used to make the connector cannot take this max t. It simply melts around the pins delivering so much power.
but why the plastic not melting when soldering reflow 280c in factory
@@ariewijaya1679 soldering reflow in factory is done by machines in seconds.
When heat is generated during hours of playing, it causes the plastic to eventually melt and liquify around the scorching pins.
The pins heating up means resistance to the current flow, so bad connection! The connector is not suitable for the power being asked to be delivered!
@@NeoGhk but it took minutes to ramp up and ramp down to room temperature
@@ariewijaya1679 nop. those things are laser soldered, search for videos in YT you'll find it fast... and even if they weren't its easy for a electrical joint to heat way more than that after all thats how we melt steel together...
Get a 3 way Collab with Jay and Steve! All of you bring failed cards, and work together.
This cuts out any “maybe this, maybe that” from the analysis process, and each of you can share your personal strengths. Plus, GN has excellent connections to industry diagnostic tools.
2+2=4!!!! But some people don't want to understand or accept it. When you're right, you're right.
@NorthridgeFix
Hi Alex !
Thank you for your work and your explainations.
Isn't it possible to remove the connector and plug wires with sensors (temperature, amp, voltage) between the card and the connector to measure and understand what's happenning here ?
I don't understand why nobody tried to measure and find the real problem.
I'm deeply invested in this! Love your videos, Alex!
I'm interested to know whether anyone has watched and tested this card with a thermal camera to see where and when the heat starts building up in the connector. I believe that there is a huge current draw on those specific pins when the card is under high processing.
Didn't Gamers Nexus do this in their 12vhpwr deep-dive? I guess the heat camera is not close enough to figure out specific pins, or that they modified their cable too much.
it looks as if gamer nexus got paid by NV to keep it down ... see the tone of his delivery in his last melting card video
Just want to say first off that I appreciate all the work you all do. Northridge, Jay, Steve, etc. My opinion. NVIDIA needs to man up and accept that they f**ked up and issue a recall and fix all 4090's released to date and figure out why this is happening. imo, Id 86 that connector. Go back to the 8 pins. Id rather have 3-5 connectors that are PROVEN to be sturdy and reliable no matter the application/situation than use 1 connector that is more a fire hazard than anything else in my house. imo, they are trying to crank too much power into these cards. They rushed the 40 series by overpowering them and it is now showing. Now wonder EVGA left the game. They saw what was coming.
I'm pretty sure it's way more productive for players like NFix, GN and JayZ just to check resistance of each connector's pin rather than overthinking who to blame :D
Or even better - desolder a healthy pin from a new card and test it against max amps allowed AND designed by the manufacturer
On my 4090 I had to push much harder than I was used to, to properly seat the cable. I can totally see someone not pushing it all the way in. In fact there are documented examples of this happening. Doesn't mean I think all the failures are related to that, just that it's certainly a concern.
Long time viewer, I appreciate your videos and enjoy learning from you. I am very familiar with the whole 12VHPWR saga affecting users of 4090s. What I would like to point out is that in the case of the 4090 you received Cablemods connector still attached, you are stating for a fact that the cable had been properly seated and that it still melted. However, you don't actually know if the connector suffered prior damage/melting, before the user noticing some instability lets say, and then pushing it in fully. But by then, the damage to the connector was critical and then continued melting. Unless you have emperical evidence otherwise, I think you cannot jump to a conclusion that you know for a fact the connector was properly connected at all times, and that it's purely a connector issue. While it is possible that your assumption is correct, people should remember that correlation does not imply causation.
yap, cuz that makes sense! A guy that is aware of this problem and even buys an adapter to make sure that he isn't putting too much stress on the connector by bending the cables, most likely did that or maybe burned the connector on purpose and then, while still hot, inserted the adapter just to feed this circlejerk about melting connectors. Im curious about what is realy causing it to fail (most likely a number of factors and one if for sure the heat generated by the current) but the common thing is 4090's and their connectors so it's pretty obvious who should be responsible
Incompatible pins, obviously. If you run that much current through them, you gotta certify the matching plug's contact resistance to be below some safety limit when it's "plugged all the way in". Neither Nvidia nor Cablemod with their Chinese suppliers bothered to come up with a contact resistance certification, so they both should pay for the RMA.
As a product designer there is no debate here. A product needs to work for the intended user base. If there are excessive fails of one type, either the design or the user instruction is insufficient.
It happens sometimes in products, especially high-performance products. I am no PC expert but it looks to me this board uses a lot of power. This will always increase the risk of power related design challenges (such as avoiding overheating).
The cause of the fault is not the ultimate question. By definition any unexpected, non third-party related. unacceptably high fail rate demonstrates a product failure, especially on a high-end product.
It happens, it is not a crime, but in my opinion the manufacturer needs to address it.
I have never seen any bad things happening with PCI-E cables, except for the new Nvidia connector.
The 4090s need a three cable adapter that DOESN'T plug into the card's native plastic connector, but instead is soldered to the card in place of the original connector. Problem solved.
Any connector, plug or anything that connects into something else needs to be designed with the rigors of regular plugging, unplugging, cable bending and ideally withstand at least some lateral forced placed on it. Even though you're not meant to be repeatedly unplugging and replugging a GPU's power cables, they can sometimes require some force to connect properly.
Even if it was caused exclusively by cables not being seated correctly, then it's still mostly on nVidia. Throughout the years as I've built my PCs I learned that you often need to apply some force to connect GPU power connectors and ensure they are seated properly. There is more often than not a subtle felt feedback given by the locking lever latching onto the connect once it is seated. There's nothing special or unique about that, and that's the hallmark of good design.
nVidia's 4090 cables and connectors came with warnings not to bend the cable within a certain distance from the connector. I mean, what the fuck nVidia?! They're basically admitting that they've made a structurally weak cable, most likely exacerbated by their ridiculously dense connector.
Obviously the connector. New connection, new issues.
I have little faith in gamers nexus, and shared your views on his first tests.
He pretty much helped the sale of that card while nvidia were sinking.
Now almost everyone has the same issue.
No time for influencers.
I agree with a little of everything. I believe Jay and Alex are correct in that the problem bottom line lies with Nvidia.
I believe Steve is correct t in his assumption that, because of bad design, it enables errors the user may never be aware of. The cable gets seated, and it feels tight, but it can walk itself out just enough.
Like Jay mentioned, there is still give even when fully seated.
Also have to remember Steve is going off of the small sample size he tested and, he did say he was willing to do more testing but, still believes that in the end, it does lie with the user to insure everything is correct.
Yes and No about that, if the connection feels tight and secure, especially with a cable mod connector and it is still melting because even that has wiggle, it's because of bad design on Nvidia.
I'd love to watch the three of them sit down and discuss all of this together and do testing.
People put way too much importance on what Gamer Nexus says, as if those clowns are experts in the field lol. For the record, Steve has never held Nvidia accountable for the burning 4090 cables and AMD for their recent fried CPUs---it's always "user error" and "Asus' fault."
Mr you are doing a good job it is the fault of the manufacturer NVIDIA you have proven it
Apparently you don't understand what proof is. Alex gave his opinion on what he believes. It is certainly not proof.
@@Ryan_DeWitt Apparently Steve GN's finding was only an opinion too, by your logic.
@@nerodante1980 When did I say it was proof? It is their opinion on what they have seen.
Intel had concerns about the connector and recommended a spring design over dimples in a PC world article.
Thank you for standing up for customers of these gpus, even when it comes to you're UA-cam contacts
It becomes clear that Nvidia simply chose the wrong connector for this level of DC current flow. The connector is simply inadequate to carry the current required by the card.
I agree north ridge on this it's not user fault it's a defect and to be honest I agree there is far too much power going through that 12v connector there should be 3 other connectors to share that 600 watt load as seems to me 12v connector cannot cope with the power consumption
It would be interesting to see what happens if you hard wired matching connectors to the cards.
Having the larger gauge wires and higher amp rating connectors , may be the key, to success.
The wires have never been a problem, why would bigger ones make any difference?
The mod should consist on replacing the connector to a higher amp ones.
@@--_DJ_-- The wires are connected to a very small connector.I think I heard that the board draws 300 watts of power.Those connectors might ,on a good day handle 100 .
Also, look at tha 12 volt cables, if wires and connectors of that size were on the board. there would be no issues.
thanks Alex for taking a stand on this. i have been working on computers for forty years now and have never seen damage of this type. The other technicians need to at least take into consideration their view isn't the only view. Only a fool thinks they know everything, a wise man listens to other people.
You nailed it brother. Why it's so hard to see i don't understand.
Right!
I agree with the guy that is fixing the cards that the problem is or appears to a design issue with the OEM connector on the card. I have experienced issues with connectors of various types in the past and some of those connectors I have fixed by using a zip tie or multiple zip ties to pull the two sides of the connectors together very tightly and lock them together so they do not move. On those video cards something like a zip tie might not be possible, but something like a tie might need to be part of the connector, or added to the connector to lock the socket and the plug together so they do not move apart.
I believe it's Nvidia's fault for making a connector that puts out THAT MUCH WATTAGE THAT SMALL. Any half Bates electrical engineer knows that the more current you're pushing through a wire the thicker it needs to be. The cables on these connectors are thinner than the wires I use to put my slime trackers together. That's pathetic
It's funny that people still defend Nvidia with all the evidence presented. Alex is so honest that from a business perspective, he's making money from repairing customer's 4000 series GPUs but he's not blaming user error for burnt connectors. Nvidia pays influencers to redirect blame away from the manufacturer and yet Alex is putting the spotlight on who is ultimately responsible for this issue.
Now i have a doubt with NEXUS
You videos are awesome.... I appreciate your no BS straight to the point style
The 12VHPWR connector was always a bad idea. 50A on 6 tiny pins doesnt work. (6 pins because you need +12 and ground @ 12 v = 50 A or 9 A per pin). That is more current than a home water heater/cloths drier, look at the connectors on those.
i agree with you and jay. I think gamer Nexus needs re-evaluate what they said.
Looks like you have some more data points to further investigate this problem. You and Steve should get together and see if you can come to a mutual conclusion. :) I'd totally watch you teach Jay and gang how to desolder/solder! Thanks for the content and keep up the all the good work both on the bench and behind the camera!
Ya that dude HAS to be getting kickbacks from NVIDIA. Absolute nonsense from him."Sunk cost falacy" Excellent work Alex!
Are you making the Fallacy fallacy? In all seriousness, GN sharing a different opinion based on their first hand evidence, doesn't make them a shill.
I'm watching your videos for 5to6 years now and I agree with you Alex what you have said makes sense to my ear I saw the connector was plugged in till to the end you have tried to separate the connector 90 degrees for 5 minutes and some people what they are trying to say here is that 1+1=11😅 for me the factory must to take the responsibility for what they have done. Alex keep going like always 💪 and thank you for sharing you experience with us all those years❤❤❤❤
As a electrical specialist you can calculate consumption and size consumption if matching with conductors.
If you have 500w consumption of video card and wires and connectors are at limit, every slight variation will cause burning connections. So the issue is in the middle, the connectors have either to big margins of contacts.
There is also a fault of video card producers, they should make maybe bigger thicker male pins of connectors and will reduce margins of bad contact.
One thing is for sure, that is a issue with bad connection, or terminals doesn't match with power delivery, they should make them thicker to support all that flow of amps.
I agree 100% with what you said, but I would say that the user not fully engaging/seating card and connector is a factor simply because it either accelerates the failure or creates a different type of failure than what you've been repairing. Still on the manufacturer for essentially causing the issue though, just to be clear.
I tend to trust service technicians, not video makers/"reviewers". Good job NFIX, the connector is not ideal for this power, they gotta test how much power goes through each pin etc. Nvidia should have resolved this long ago and accept cards for repair under warranty. So it's strange to see so many cards at 3rd party repairs.
At least he says "we could be completely wrong", but yeah, I agree that Steve is too ready to blame "user error". Since you had the cable, examined it closely, and are an expert, I would agree with you!
This problem falls directly into Nvidia's lap. If Nvidia thinks it is user error, then the user should not be allowed to install their 2 to 3 thousand dollar cards. It should come with a pre-installed cable.
The 4090 is a flawed design,even a hobby tinkerer like me can see that,not fit for purpose.Can it be rectified? Absolutely,even a stupid noob like me with mere mechanical engineering background understands that.Manufacturers and designers need to take responsibility for the products they put out instead of blameshifting and deflecting it on to the end user.Well done Northridge Fix for explaining water to the thirsty.Never stop
agree, NVidia should have not implemented this change al together, AMD was able to develop the 7900XT and XTX yet use the traditional connectors while still offering competitive performance, as a builder and a person who has a system integration business, this is a huge problem for us, the original cable extension that comes with the 40 series cards is just bad, it ruins the whole echo system, working with these cables on our systems is really annoying, making it look good is super difficult and will almost definitely need a custom cable which is also annoying, and now seeing these issues that impact sales, warranty claims, and client concern and worry, NVidia really pulled a number on all of us.