I'm looking forward to the H++ Connectors. I don't think we can get our hands on them as of now. We'll upgrade and fix every card with a 12VHPWR connector.. Done. Everyone is happy.
Do you have a mailing list available to 4090 owners to let us know if/when you have them available? Also, how much would something like that cost? (Of course, not knowing the cost of the part makes it harder to answer that, simply wondering about average labour charges.)
@@triadwarfare What if someone has a brand new never opened 4090, they open the box, find the old plug, and won't use the card until it has the new plug on it? Is there some way to get a brand new card that has the 12vHPWR replaced with a brand new card that has the 12v 2x6, or is there a way to get the card's connector replaced with the "right" one under warranty? EDIT: Reason I'm asking is because I'm almost certainly going to be buying a card online, and I don't want to have to send 20 of them back because they all have the old plug on them.
All the current 4090's need to be retrofitted with the new connector. If Nvidia is putting these updated connectors into all of their new batches of GPUs, then there was clearly a problem with the original design.
Its people not plugging the cable in correctly also, and physically damaging things with too acute cable bends. Its a crap connector design that has far outlived its safe usefulness by now.
@@zybchexcept at this point it's not user error when plugging in. Have you missed the videos on 4090s melting the plug and cable mod adapters when the cable is fully seated? It's an internal flaw in the plug for the higher power draw 4090 cards.
Double-up on the retention clips. 1 "hook" part of the clip for each "6-pack" grouping of pins. And the release tab is connected to both of those hook parts, obviously.
Internal emails leaked from PCI SIG before the 40 series launch stated that Nvidia knew the cables were melting on both the PSU end of the cable and the GPU end of the 12vpwr connectors. They should have put the brakes on and did more testing but they released the cards anyway.
Another super simple thing I saw some manufacturers do is paint the end of the plug in a different color (e.g. bright yellow). So if you plug it in and see yellow, it's not seated right.
Even just moving it to both sides instead of one on top would probably help, since the current one doesn't do much for the connector being a bit off-angle. Or even go back to the VGA/DVI style screw-downs if you want to be really sure.
They are clearly keeping all changes backwards compatible with pre-existing HW and cables to make all mix-n-match configurations functional. Modifying retention mechanism would loose that.
@@hydrargyrumnight Which doesn't make sense, they could have just made a total mod when they came up with the spec to begin with instead of just using the same design with smaller pins.
I mentioned this before. Ford Motor Company had this same exact problem with some cars in the 90's, my Mustang being one of them. The connector(never found out which exact one), under the steering column had to be changed due to the fact that the pins were too close and the wires were too small to handle the CURRENT(AMP) going through them. It caused a fire hazard. Credit to them for creating a recall as fast as they did.
It really sucks that this has been an issue for a very long time. The sense pins not being the shortest pins in the connector, was a major oversight! I feel they should also upgrade the clip tolerances to eliminate/minimize the connector wobble. Im soo glad i got my 3080ti strix card last summer, instead of waiting for 40 series! Although, it was pretty pricey...
also bought 3080Ti like half a year ago, very happy with it for it's price ($600) but not happy for the power draw. Looking forward to getting one of 4080 when they cost ±900
@@TheExileFox GN did not help by concluding user error had something to do with it. They created a lot of stooges that are still blaming the user despite proof that connectors can disconnect on their own even when clipped due to the poor connector design.
"It really sucks that this has been an issue for a very long time" Tell me about it. Noobs not properly seating power cable and using 3rd party adapters having melting and burning issues have been happening since GPUs started having their own power cables instead of MOBO power.
@@camotech1314 Because some people have that sort of money to waste? Because what happens at the high end will filter through to the ordinary user? Because some people with different priorities might not consider it a waste at all?
There's also another aspect to wire gauge. The thinner the wire, the HIGHER the resistance over the length of that wire, the thicker the wire, the LOWER that resistance is. Resistance in a current load creates heat and that heat will show up at the connector before it shows up in the wire.
Every time I see this stuff I'm so happy with my ancient simply plug and play 3080. And I have the cables pulled hard 90 degrees to the right and wrapped around the side of the card to the back.
@@laserbeansprouts Ok, but it's a little dusty. It's from the pre-covidian era. Non-LHR first run architecture. For like 3 minutes I had the second fastest GPU in the world.
The term used for cables moving due to heat changes is called thermal creep and is a major issue. Most higher end systems use some type of retention system (wire brace, velcro, etc) to hold connection in place
The old pcie 6 and 8 pin connectors have a clip integral to the connectors that accomplishes that task quite well, but the clip on the 12vhp connector is too long, which leaves room for the connector to unseat itself before the clip stops it from loosening further.
Would a few strategically placed blobs of high temperature hot-melt glue solve the problem? I know that they use that trick to hold the plug for the PSU internal cooling fan into the socket inside most PC PSUs these days, because it gets rid of the thermal creep problem. The other advantage is that it doesn't prevent service, if you need to service the card or remove it from the system for other reasons, just un-stick the hot melt glue and remove the connector as normal. One extra step but it solves a lot of problems.
Jay, I started building PCs back in the early 1990s. Loose connectors were never a problem back then. Most of our devices, like monitors and printers, used cables to attach them to the ports on the PC and thumb screws to secure the plugs to the PC. You knew the connection was secure and would not separate from the PC. As I remember things, our MOLEX connectors used two retainer clips. To paraphrase an old anti-acid commercial, "Click-Click. Oh, what a relief it is."
@@superjimse Likewise DVI and, my personal favourite, 13W3 (I use a lot of SGIs). SCSI also had good connectors, though SAS changed to become a bit more flimsy.
@@administrator4728 Thank you for your reply. I always read them so I can get better incite with my comments. I have a new 4090 GPU, and this connector problem disturbs me. I'm getting older, and sometimes my memory fails me, so I might be wrong about the two MOLEX connection clips.
@@chrissyboy7047 Thank you for your reply. I feel that the older-style connectors were indeed better. The PC component designers seem to like to take two steps forwards and one step backward.
The power limit increase kinda suggests they're getting ready for a variant with higher power draw .. which would be expected of something like a 4090Ti 😉
It's great that they admitted fault but with the pressure from the media and health and safety issues they had little to no choice. This is the R&D that should have been done before the release of the product and due to the safety issue the companies need to do a recall. Great video Jay
Jay, if you can do that sense pin loosening thing as a follow up, that'd be valuable. If you could get a card going black screen and 100% fan on camera, it'd probably help some folks troubleshoot their cards better.
I had this exact thing happen to me last week. Changed my power cable routing from the top of my 4070ti to the bottom and got "No signal detected" and instant 100% gpu fan speed when i turned on the system. Had to wiggle the wires going into the plug to fix it. I never unplugged the adapter from the gpu, and where it plugs I to the gpu was tight. These adapters are so fragile that I would like to see a tsb with new adapters sent out.
I randomly on a daily basis, encountered black screens with fans full speed and audio still playing. I pushed all my connections in tight and it's been stable so far for 2 days.
I also encountered black screen and 100% fan speed issues on a daily basis after getting the cablemod's 12VHPWR cable even though i have checked that the connectors are all the way in and reseated the plug multiple times, but the black screen 100% fan speed issues would still happen frequently and randomly, but back then i didnt realise the cable was the issue, until i decided to go back to the 12VHPWR adaptor that came with the gpu and realise that the issue is gone. Decided to RMA the cable with cablemod and they didnt ask a second question and replaced the cable immediately, which was nice. But even with the new cable it still happen again for 1 or 2 times after using it for about 2months now, which was much better than previously where it was happening almost everyday, but still not optimal.
@@flamebullet I was having this problem and I uninstalled and reinstalled the graphics drivers and it seems to have fixed the issue. I only did it a couple days ago though so I’m not sure if it’s completely fixed but it ran for 9 hours yesterday with no problems even though before it was happening like every hour.
@@sotempermental247 i have already ddu the drivers before switching the cables and it seems to work fine for a while and i thought i fixed it until it happens again after a few days. Other things i tried including reverting my overclock since i thought maybe it was unstable(it was just a slight overclock, not aggressive), and even underclock the card.
I put Nyogel dielectric grease on my 12VHPWR to help combat terminal fretting due to inherent case vibrations from fans and pumps. Also built a connector clamp block to eliminate any possibility of the connector being wiggled crooked.
🪄✨ Made with SummarizeYT app 0:00 - The 12VHPWR plug is finally getting an update and a name change to 12V-2x6. 0:23 - The update aims to prevent melting and improve the reliability of the plug. 1:20 - The changes come from the collaboration of various board manufacturers, not from Nvidia or Intel. 2:32 - The new specification includes changes such as power draw, temperature limits, and minimum force required to pull the plug. 4:02 - Different manufacturers and suppliers use different pin types and wire gauges, leading to variations in the adapters. 5:01 - The new specification ensures compatibility regardless of wire gauge. 6:43 - There are two types of pins in the market: Astron and NTK. NTK offers better contact and greater contact area. 8:02 - The standardized spec recommends the use of NTK pins instead of Astron. 8:27 - The new spec increases the power rating to 675 Watts, allowing for better power delivery. 9:00 - The spec also allows board partners to draw a full 75 Watts from the PCI Express slot. 9:35 - The new 12V-2x6 plug has more contact surface area compared to the Astron 3 dimple plug. 9:39 - Changes to the terminal side of the plug have improved contact points and decreased the chances of poor contact and melting. 11:01 - The new specification includes a minimum of 150 watt sense pin for better communication between the graphics card and the power supply. 12:10 - The new plug design includes a safety mechanism that disconnects the plug before it can cause melting or poor contact. 13:16 - The plug now has a minimum wire length requirement to prevent tension on the pins inside the plug. 14:00 - Loose plugs or damaged sense wires can cause black screen issues and 100% fan speed. 15:00 - Using NTK pins on all adapters is recommended to ensure reliability and prevent melting. 17:06 - Users with three-dimple plugs should consider exchanging them for NTK plugs to improve safety. 17:28 - Repair services may start offering plug upgrades for existing cards to prevent future issues. 19:02 - Using NTK pins was discouraged in the past due to higher costs, but it provides better reliability.
Same user error issues even with the new connector. Properly seat your power cables, buy a case that fits, do not bend cables, do not use 3rd party adapters, have a professional who knows what they are doing building rigs and you will have a %99.9 success rate according to the data
One reason the dimples may wear out is metal fatigue. Metal, including copper, expands as it heats up especially when under heavy loads. Then when it cools off, the metal will retract making a looser connection. Eventually those dimples will Stay loose. We had that problem with our deployment equipment in the desert. We started several fires not knowing why at first. Once we figured it out we had to retrofit connections and newer systems came with better connections.
That was an excellent job Jay, thanks for taking the time and putting the effort in to a subject that's not easy to explain. Well done 👍 and kudos for the edits.
My friend and I both have had the black screen issue with our 4090’s. I solved mine by removing the 90 degree angle connector from Cablemods. His is currently in RMA.
Can confirm, I was having the black screen & 100% fan speed issue and it was a loose cable mod connecter at the card. Visually it looked fine, but I reseated it and haven't had an issue in a week. So possibly some defective adapters for other people but I 100% think that a loose connection is the issue.
Same. Tried everything and nothing worked. Turned out to be the new replacement Corsair cable. Replaced it with the default cable that came with the card and not a single issue for about 5 months.
Properly seat your power cables, buy a case that fits, do not bend cables, do not use 3rd party adapters, have a professional who knows what they are doing building rigs and you will have a %99.9 success rate according to the data
The problem is NOT in the cables. The problem is in a bad 12 V standard. Communication centers and data centers have been using a 48 V standard for decades (I worked with it in 1972). That would reduce the current by a factor of 4, stopping the cables from melting. One reason they chose 12 V is that the 1920 buck converter used in VRMs cannot do a 48V to 1 V reduction. You have to use a 2023 design such as the Ćuk-buck2 converter. Unfortunately, it is cheaper, smaller, more efficient, and provides a better transient response. Adopting it would require the "engineers" to actually learn some of the last 100 years of improvements in power conversion technology.
I was wondering why they didn't just throw a couple of sense pins on a wholly different connector like an XC90 (rated for up to 90 amps). 55 amps is light work for that connector.
I sure wish that WWII game had existed decades ago... My grandpa, rest his soul, would have LOVE that WWII game! He had filled his library full of books about the warships from that era. He loved opening them up and talking about them. I can think of nothing he would have enjoyed more than that game!
It's about time! My Strix 4090 got burnt down even though i had my pretty connector connected correctly. It destroyed the 180° cable mod adapter too. It's a garbage connector.
As someone with a original 3080 I hope the 5080 series fixes all the 40 series cards problems (Size,power consumption/temps, PRICE, and connectors!) Big reasons why I never "upgraded"
The size can be great, cool and quiet running cards! , the 40 series is very power efficient already, temps are great, PRICE is crap and the connector needs a fix
It will fix most of those problems, But it will not fix price. EDIT: Oh and you can also forget about the top-of-the-line cards using any less power than "all that the connectors can give them" that's the whole reason they're able to perform better than ever is that they're drawing more power than ever. Efficiency gains need smaller transistors (that's a ways out), or a large change in micro-architecture with an eye towards power efficiency rather than raw performance (not happening, or if you do want that, go buy an AMD card instead).
@@44R0Ndin Hmm? The 4090 i rarely see draw even 400w but Cyberpunk 2077 with patch tracing, 4K everything maxed (No DLSS and Frame gen) will push it to 500. You can also limit it around 60 - 70% and lose barely any performance. RDNA3 is not very efficient
I'm glad PCI-SIG is getting an update out, but I have some questions. I saw no mention of the X-axis rocking that caused the cable to become disjointed at the board connector. Wouldn't a three sided shroud around the wire connector prevent that in addition to the updated pin designs? Additionally, regarding the sense pins specifically, I had figured they would be cut back, but I'm curious as to why CARD_PWR_STABLE wasn't lengthened instead of being cut, since supposedly that's the lynchpin with regard to letting the PSU know to cut power to the GPU in case of overload or off-kilter power delivery circumstances. Any thoughts on these?
Just speculation here, but maybe a shortening of CARD_PWR_STABLE still serves the same purposes? If that pin becomes disconnected, then the power supply looses that signal and (hypothetically) still cuts power to the GPU? Personally, I don’t care what pins get lengthened or shortened, as long as the failsafes do their job somehow and prevent the GPU from burning up. I’m also a 30 series owner who isn’t worried about 40 series as the whole lineup has been a disaster.
@@StolenJoker84 CARD_PWR_STABLE is by default open. That is, there is no connection at all even when the sense pin of the same name is fully seated. This is by design. The GPU will short (close) the circuit on that pin if there is an overload or another power anomaly (in theory anyway). By lengthening this particular pin to full length, it will always be connected at the connector and thus the GPU will be able to perform its failsafe of shorting that pin in an abnormal power delivery scenario. The rest of the pins being shortered is something I had already figured should be done. The new update does not unfortunately cover the rocking present when too much strain is placed on the connector from cable positioning/stress. A three sided shroug (top/sides) that is flush with the board connector housing and going down far enough should prevent the cable from ever rocking, and at least in that fashion prevent the connector from entering into a situation where arcing occurs (this is what heats up the connector and melts it). I'm kind of disappointed to see that all these things were being addressed but not these two, especially after the GN videos clearly showed wear lines that weren't parallel to the end of the connectors, but lopsided from the connector being both loose *and* pulled to one side.
Properly seat your power cables, buy a case that fits, do not bend cables, do not use 3rd party adapters, have a professional who knows what they are doing building rigs and you will have a %99.9 success rate according to the data
The 30C above ambient will not refer to the ambient in your room but the ambient in the case in which it is fitted. Since most plastics start softening at around 100C I would prefer to see an absolute temperature specification and not a relative
Hey Jay I was a big time RC car and truck user when I was a kid anyone that was in the late 80s to 90s prior to Lipo were forced to use Nicad batteries and the higher the volts and amps lead to massive battery fires and shorts this reminds me of back in the day, one thing that came to mind was the connection between speed controller and battery the connection that had the split in the receiving pin or female side was found to be the culprit in the loss of connector when the plugs were connected and the power was surging through it any sort of gap or misalignment of the fonnection resulted in an arc similar to the idea of a spark plug, In this scenario it can be just as detrimental as it creates heat and with heat then the connectors with the split would open up creating more arcs all of the gpu plugs are reminders of the old Tamiya plug styles or Molex that as a racer we would clip off in favor of the deans blade connectors back in the day, im curious if any of this holds true in these current connectors on gpus now adays
I bought a 12vhpwr cable and 90 degree adapter from cablemods, I would love to see an "upgrade plan" where you send your cables/adapter and for a reasonable price they would upgrade it to the new 12v-2x6 connector.
@@AndreasRP. You don't need new card, the old and new connectors would work with each other. There are the benefits of having the new plugs on the cards, but just the ntk upgrade would make it much safer.
@@AndreasRP. considering that companies made new cables, adapters, extensions, PSUs etc that would have been the worse outcome for both companies and users. imagine you bought 200-300$ new PSU for the new connector so you could "future proof" it and use it for the next decade or so, then at the next generation GPUs, that connector on the PSU becomes completely obsolete.
They should introduce lockable cables like the ones used in cars for everything. A lever latch that will only fully close if fully connected. No point for those half-arsed solutions.
So happy I went with a 4070 Ti. Not a single report on this card regarding melting cables / plugs. Been working like an absolute charm with my CableMod 12VHPWR cable for half a year now. :)
I have another theory as to what may potentially be causing it as well. If the pin is bent slightly (so it is no longer centered within the plastic housing), this will force the connector contacts apart. While I have not had this happen with a 16 pin, I have had it happen with an 8-pin. It didn't cause a fire or anything to melt, but it caused the system to continuously bluescreen. I had to discard that power supply cable since the contacts inside one of the connectors was forced open by the bent pin on the GPU connector. I also had to re-center the pin with needle-nose pliers to avoid the same fate happening to my other PSU cable.
Jay You just saved me a fortune, I was just about to pull the trigger on a 4090, I was going to order one at the end of this week, but now I’m going to wait, great video thank you 👍
"now I’m going to wait" how long are we you gonna wait? weeks, a month, 6 months? that's what we need to know. Do we need a new cable too for our 1000 watt PSUs?
I bought a used 3090 Strix because it has 3X 8 pin connectors. I looked at the 4090 but every one I tried had coil whine and some of them were bad. I would have bought a Red Devil 7900 XTX but they were hard to get then. I am going to build a new rig in a couple years. I might put an AMD card in it.
This is the first time in a long time my strategy of skipping a generation for video card updates has saved me some trouble. Thanks Mr.3090....you may be a little slower at 4k, but we're doing just fine and also not on fire.
I don't think these changes being made are going to cut it. 4090's are burning out at the power plug like Canadian wildfire, and 4080's that draw 320-350W are also still burning out, just not quite as rampantly as 4090s. 4070's however are not, at only 200-250W draw. I have said from the beginning when I first looked at the specs and total contact area that with the total contact area of all pins and terminals that 250-300W PER connector was likely the limit of the plug. It is only around 45% more contact area than the ATX-2.0 8/6+2 pin connector, and only ~70% more than the ATX 6pin, which translates into the same percentage of increased safe transfer capacity. Contact area determines safe wattage via ohms, or impedance. Impedance translates into heat when you try pushing more power through due to natural resistance. At least the change in terminal style from minimal nub contacts to the spring style bars takes that into mind, but it is still not likely enough. I am hypothesizing that they are figuring it at double the 6 pin terminal capability due to twice the pins. But I have also seen 4 pin Molex connectors carry more current that smaller 8 and 12 pin connectors due to surface area impedance. Total safe transfer capacity is directly proportionate to total surface contact area. I think the ACTUAL solution is to tone down the wattage per connector and increase the number of connectors. No matter how in depth and complex the project you are working on, you should NEVER neglect or forget the basics. Even the most complex designs and standards start with the basics, because if they don't start with the basics the basics get forgotten and failure becomes a forgone conclusion.
I loved the segment on this that Steve from GamersNexus made in his Weekly News Recap. "I was looking at the changes and thinking, 'This looks familiar...'." The changes were basically exactly what he asked for. 😂 Thanks for the coverage, Jay.
GN is the source of the victim blaming tho. With all their testing, they still erroneously blamed the user in the begining. Sadly, a sign that they are starting to pander to advertisers. GN was the main source of the bad info despite any good info they had.
In addition to heat, I suspect all the vibration from fans and pumps eventually shakes some cables looser. The combination would certainly have that affect.
Honestly, I'm thinking it's still best to wait and see, or if you really want a 4090 go get something that isn't that high on the tier list in terms of power limit setting. I recall some AIBs are set to use the whole 650w while some are fine with 450w as the power limit. Galax's HOF even uses 2x 12VHPWR with an LED light display that tells you if your card is only able to use half the power due to not being plugged properly. Here in Asia, I can find a lot of those L-shaped 12VHPWR connectors that look exactly like what Jay showed before and there is one thing that is labeled the same across all of them from what I have seen. All of them say "rated for 450w". Now I wonder if that's a possible cause of issue if you, say, use one of these connectors on an ASUS Strix RTX 4090 which is said to be set to have the full 650w power limit, and then plug the 12VHPWR connector properly as a whole. Will the connector cause an issue? I can only guess.
This is why I went with the MSI 4090 Gaming X Trio. They power limit them for thermals just a happy accident. That It also keeps the damn thing from melting as well.
@@lostinny690 Yeah, I actually heard about that for that AIB. Apparently the Gaming X Trio still uses the "signature" Gaming X Trio shroud format with the fans and all from the previous Gen. so in order to tame the temps, they had to deal with the power limit. The result was a 4090 that was no less powerful than the others, but is still well balanced with cooling for its preset power limits. Definitely a happy accident.
@@sv0g391 I don't think they had to power limit those cards as they don't draw as much, nor do they get as hot. Also, to my knowledge, no 4080 or 4070ti has ever had an issue with the melting plug.
Thanks for covering this Jay. There's no reason the connectors aren't made to be 90 degree in the first place given most people need to bend the crap outta the cables to manage them in cases.
Kind of random, but I always enjoy your videos. You guys all do a great job with this channel keeping things fun and information rich. Keep up the great work and know that your hard work is greatly appreciated!
Honestly this issue pushed me away from the 4090. I know there's "bigger reasons" I shouldn't want one but gaw damnit I can afford it and I wanna use the types of graphics it's makes possible
Also why I shied away from the 4090 and ended up with an XTX. I sometimes still "regret" not getting the 4090, but I know I'd be more upset if the card melted b/c of that adapter.
I think the primary problem with the 12VHWPR connector is they shrunk the 2x6 layout to match the width of a PCIE 8-pin connector, and of these little problems cropped up and combined together to create the failures. Smaller pins means less surface contact, the 3 dimple situation for example. There should always be a landscape cost for higher power draw. The larger chambers can also accommodate larger wires as well and better connections across the overall connector. There was no valid reason the connector needed to be so small.
As a engineer I will tell you that we are always on a tight budget of time and money. Even if we tell the higher ups this won't work because (insert Technical reason) and we need to (insert better but more expensive option) keep it safe. we will get overuled, they don't understand what we are saying and won't comply. All they want to see is numbers going up so they can afford they're 3rd car. And even if it destroys the company 1 year later it won't matter they will just go and destroy the next company they already made there money.
@@leroyjenkins0736 I always blame the bean counters over blaming the engineers. Engineers naturally want to make something that lasts a long time and has a lot of features. But that's "too expensive" they say. And then in the same breath they charge about $2000 USD for a 4090 that hey guess what we saved $0.01 by using this cheaper power connector to carry 600 watts, so now it melts instead of providing years of trouble-free operation. Fucking bean-counters. Every time. That's even what killed both space shuttles (and their crews). The engineers with Challenger even SAID "Don't launch now, this is outside the launch constraints we agreed on when we designed the thing" and yet they got shouted out of mission control. Yeah that went well, we lost 7 good people because of that whole mess of stupidity.
Great indepth video and promising signs ahead, sometimes it sucks being the early adopter of new generation of stuff, glad I've been holding off buying a new gpu, guess will wait to see how the new batch of cables / gpu's go once its all released.
Had my MSI 4090 since February with the EK waterblock and active backplate and no issues yet however I wouldn't mind paying for the upgrade if it means keeping my 4090 safe.
The problem with the 12vhpwr connection is the fact that you are pushing 16awg wiring to it's limit and the pins are smaller than 16awg which makes them heat up. If you notice in the new drawing the pins are larger than the original to give more area for current flow. They cannot have a smaller area/contact point than 16awg stranded wire. I made this observation when you made the first video about this. No one is talking about the connector pin size but they increased it for a reason. THAT IS THE PROBLEM but they will never point it out because it makes them look bad.
I always thought it was madness to take what was basically 2x 8 pin connectors which would have carried a total of 300W between them, double the power delivery and shrink the size of the contacts/ connectors at the same time....or were my concerns too simplistic and common sense-like??
4080's have less issue been some but not on scale of 4090.I just went with a 4080 myself until this issue is done but my 4080 serving me well so keep till the 5000 series or see if AMD actually steps it this time with the 8000 series
Few things that is concering. 30 degrees C delta at presumed max operating power for me seems like a lot. Because this would mean the safety factor is very close to 1. As soon as the power that goes through the plug exceeds its max then the tempature increases exceptionally. Higher temp higher resistance more resistance becomes more tempature. What i would like to see if it was 5 degrees max, and a safety factor of 2 or more. But to cut cost they probably went with a safety factor of 1.3 or pretty low.
they should put clips similar to like what you would see on a guitar case, where one side is a loop and lever and the other side is a stud and the loop pulls the latch closed as you flip the lever
So the CableMod 180 adapter "may" have the Astron connector? I may need to check on that. Honestly, anything with this amperage rating should have at least a bisected male and a solid circular terminal. Think XT90 with an 8ga. crammed in it.
I dont understand why the GPUs dont use bullet connectors like in the XT60 plug for lipo batteries that is used on rc cars with high current drain. The 3.5mm bullet connector in the XT60 plug can handle 60A at 12V. If they had got 2 connectors for 12V, 2 for ground and 4 for the sense wires, it would be less chance to fail than the normal 8-pin connector.
Based on a couple videos on the new plug, it was never mentioned if the retaining clip was adapted or widened to limit the amount of wiggle of the previous version.
I knew a builder that had a "key" set that would release the pins in Molex connectors, and he would add a drop of Silver solder to each pin and then reinsert them in the connector. When I asked him about that he told me that "People have a habit of pulling on wires to disconnect components, and this helps save the connector."
I figured this out in the beginning. What took them so long? They deviated from the standard Molex plug and connectors and screwed it up. I have been in this industry since the late 80's and this bugs me when an obvious flaw is designed into equipment.
18 gauge is definitely inadequate, 10 amps max on 16, and 18 is thinner. but as he said the pins can be part of it too. most people think current runs through the wire, it does not. it runs on the outside surface of the wire and the pin, if you don't have enough surface area contact at a joint ( plug) it can heat up, also if there isn't enough friction, you will also get heat build up.
Nice hat, bro. Wear that around Morro Bay. You can walk right into the Yacht Club. Well, the gate's never locked, so you can walk right in anyway. They usually have a free hamburger BBQ every Friday afternoon, FYI.
Got the angled v1.1 adapter in the NICK of time just checked my previous adapter, started melting luckily the gpu connector itself is fine save for the top right connector now being dark grey
As long as we are monitoring sense wires in the connector, why not add a thermistor, that could actually detect if the connector temperature rise becomes unacceptable? Seems to me it wouldn't add much cost.
What's wrong with using the XT60 connector? It solves every problem on the 4090 when it comes to power draw. It seats in better, it feels better when it's seated, there's no ambiguity when it is in; no more 'is it in fully?' and you can see it's seated in. Plus, it can carry 60 amps and above at these voltages.
Jay, My son and I went to Microcenter over 4th of July Saturday and purchased two ASUS Strix OC 4080 cards. At the same time we ordered two of the Cablemod right angle adapters that arrived the next day. So on Sunday, just as the adapters arrived, my son notified me that he just had a meltdown on his 12v cable. we were OCD ( as you mentioned in this video) about making them secure and ensuring they were not bent. PS was a EVGA 1000w Supernova G2 fyi. We took the card back to Microcenter and they replaced it no questions asked and the manager was aware of the issue. Anyway, we both are using the adapter, pre-bent cable running straight down and then elbowed to run rearword and made sure there is zero stress on the plug. Very odd that it happened on a 4080 though… we have pics if you want to see…let me know.
Jay - great video - I Can 100% confirm your point around the 15 min mark; re: wiggling the connector to get the black screen and 100% fan ramp; This is continually happening with my Gigabyte OC 4090; This happens both randomly and with a physical intervention on the plug.
Why am I watching a video about pins & cables for GPU's I cannot afford and don't have I don't know - however, my brain loves information and Jay video's are informative and addictive. Thanx for all the content - I'm not sure knowing this for talk at a party will make me friends though. Love it all ❤
so glad i made cables for a living and do quality for them now and made my own when there were issues. it uses general molex style pins and connectors. i really just used spare ntk stuff from work. i really hope they actually fix things for the public consumer.
You know, since you do this for a living, you could have (and probably still can) potentially start up your own business making cables for this application that are "actually worth buying" NTK stuff. I've literally never seen any connector in an automobile use that "3 bumps" connection method, and with this whole mess with the +12vHPWR connector's issues with melting, I have a pretty good guess why I've never seen it in the automotive world.
@@44R0Ndin I could make custom cables but i would never in my life make cables i could not 100% guarantee, and with this plug you cant at the moment above 450watts.
AMD was SOO smart in avoiding using this connection right off the bat. They were smart to think ".. lets see if this actually works first". This is such a costly mistake for 4090 owners given how freaking expensive those cards are.
I'm looking forward to the H++ Connectors. I don't think we can get our hands on them as of now. We'll upgrade and fix every card with a 12VHPWR connector.. Done. Everyone is happy.
Maybe you can do that colab with Jay and get him to fix some for you 😉
Do you have a mailing list available to 4090 owners to let us know if/when you have them available? Also, how much would something like that cost? (Of course, not knowing the cost of the part makes it harder to answer that, simply wondering about average labour charges.)
@@EvoTorrance What about warranty? Would it be void if someone changed plug?
@@ultrasaiyan4283the ones who come to him are the ones who are denied warranty.
@@triadwarfare What if someone has a brand new never opened 4090, they open the box, find the old plug, and won't use the card until it has the new plug on it? Is there some way to get a brand new card that has the 12vHPWR replaced with a brand new card that has the 12v 2x6, or is there a way to get the card's connector replaced with the "right" one under warranty?
EDIT: Reason I'm asking is because I'm almost certainly going to be buying a card online, and I don't want to have to send 20 of them back because they all have the old plug on them.
All the current 4090's need to be retrofitted with the new connector. If Nvidia is putting these updated connectors into all of their new batches of GPUs, then there was clearly a problem with the original design.
its mostly the male end not the female
It's the cable, not the graphics card.
Its people not plugging the cable in correctly also, and physically damaging things with too acute cable bends. Its a crap connector design that has far outlived its safe usefulness by now.
The new revision is backwards compatible with existing 12VHPWR.
@@zybchexcept at this point it's not user error when plugging in. Have you missed the videos on 4090s melting the plug and cable mod adapters when the cable is fully seated? It's an internal flaw in the plug for the higher power draw 4090 cards.
The changes all sound great, but there's one more change I'd like to see... A wider retention clip to help eliminate side-to-side wiggle!
I made my own connector position assurance clamp block that bolts to an EK waterblock because that clip is so poor
Or just jump up to an automotive retention setup. The kind with the metal clip that manufacturers use for fuel injectors
@@thespacedpirate That would be hilarious!
Or we could use screws like on VGA, DVI etc. Those screws won't wiggle loose
Double-up on the retention clips. 1 "hook" part of the clip for each "6-pack" grouping of pins.
And the release tab is connected to both of those hook parts, obviously.
Strange the sense pins weren't already designed to be a safety. Great change though. Crazy how much power these cards can draw.
Right!?!
Never seen my 4090 draw more than 400 in normal usage
To me that's still pulling alotta power through those tiny cables/pins.
Internal emails leaked from PCI SIG before the 40 series launch stated that Nvidia knew the cables were melting on both the PSU end of the cable and the GPU end of the 12vpwr
connectors. They should have put the brakes on and did more testing but they released the cards anyway.
@@davidbrennan5 because $$$. I'm sure its only being addressed because its so widely known
Another super simple thing I saw some manufacturers do is paint the end of the plug in a different color (e.g. bright yellow). So if you plug it in and see yellow, it's not seated right.
meh it's fine so what if your card burns it's fine🤣
They should really look at redesigning the retention mechanism for the Pins/Terminals as well.
The little PCI clip dosen't seem to be cutting it.
Even just moving it to both sides instead of one on top would probably help, since the current one doesn't do much for the connector being a bit off-angle.
Or even go back to the VGA/DVI style screw-downs if you want to be really sure.
They are clearly keeping all changes backwards compatible with pre-existing HW and cables to make all mix-n-match configurations functional.
Modifying retention mechanism would loose that.
@@hydrargyrumnight Which doesn't make sense, they could have just made a total mod when they came up with the spec to begin with instead of just using the same design with smaller pins.
That clip can break too...
Should go to 2 mounting screws.
I mentioned this before. Ford Motor Company had this same exact problem with some cars in the 90's, my Mustang being one of them. The connector(never found out which exact one), under the steering column had to be changed due to the fact that the pins were too close and the wires were too small to handle the CURRENT(AMP) going through them. It caused a fire hazard. Credit to them for creating a recall as fast as they did.
Phil's editing kills me 🤣 I love that flame from the side of the screen talking about the 4090 on the bench lol
I was expecting an ifixit ad to just start.
It really sucks that this has been an issue for a very long time. The sense pins not being the shortest pins in the connector, was a major oversight! I feel they should also upgrade the clip tolerances to eliminate/minimize the connector wobble. Im soo glad i got my 3080ti strix card last summer, instead of waiting for 40 series! Although, it was pretty pricey...
you spend $1k for a 3080ti? that money could easily get you a 4080 at 16gb of vram now... not worth the money
Not an oversight. Carelessness or incompetence. You choose which one.
also bought 3080Ti like half a year ago, very happy with it for it's price ($600) but not happy for the power draw. Looking forward to getting one of 4080 when they cost ±900
@@TheExileFox GN did not help by concluding user error had something to do with it. They created a lot of stooges that are still blaming the user despite proof that connectors can disconnect on their own even when clipped due to the poor connector design.
"It really sucks that this has been an issue for a very long time" Tell me about it. Noobs not properly seating power cable and using 3rd party adapters having melting and burning issues have been happening since GPUs started having their own power cables instead of MOBO power.
I've heard that the there is a 8-pin connector that works very well.
100% This one will melt as well.
No they wont... Just shamefull nvidea beber admitted this to BE a real mistake on their behalf
@@Saiohleetit's beber baby
@@Saiohleet You be the guinea pig. I will buy cards with the 8 pin connectors.
Why even bother with these cards, unless your job in life is to waste money.
@@camotech1314 Because some people have that sort of money to waste? Because what happens at the high end will filter through to the ordinary user? Because some people with different priorities might not consider it a waste at all?
"The 4090 is over on the bench, hopefully not melting..." *flames from Phil* - Just brilliant.
There's also another aspect to wire gauge. The thinner the wire, the HIGHER the resistance over the length of that wire, the thicker the wire, the LOWER that resistance is. Resistance in a current load creates heat and that heat will show up at the connector before it shows up in the wire.
who new it was a major issue to replace the 6 pin and 8 pin power connector with a single power connector on the gpu to take all that power🤔🤔
Why............???
Every time I see this stuff I'm so happy with my ancient simply plug and play 3080.
And I have the cables pulled hard 90 degrees to the right and wrapped around the side of the card to the back.
@@laserbeansprouts Ok, but it's a little dusty. It's from the pre-covidian era. Non-LHR first run architecture. For like 3 minutes I had the second fastest GPU in the world.
The term used for cables moving due to heat changes is called thermal creep and is a major issue. Most higher end systems use some type of retention system (wire brace, velcro, etc) to hold connection in place
The old pcie 6 and 8 pin connectors have a clip integral to the connectors that accomplishes that task quite well, but the clip on the 12vhp connector is too long, which leaves room for the connector to unseat itself before the clip stops it from loosening further.
Would a few strategically placed blobs of high temperature hot-melt glue solve the problem? I know that they use that trick to hold the plug for the PSU internal cooling fan into the socket inside most PC PSUs these days, because it gets rid of the thermal creep problem.
The other advantage is that it doesn't prevent service, if you need to service the card or remove it from the system for other reasons, just un-stick the hot melt glue and remove the connector as normal.
One extra step but it solves a lot of problems.
Jay, I started building PCs back in the early 1990s. Loose connectors were never a problem back then. Most of our devices, like monitors and printers, used cables to attach them to the ports on the PC and thumb screws to secure the plugs to the PC. You knew the connection was secure and would not separate from the PC. As I remember things, our MOLEX connectors used two retainer clips. To paraphrase an old anti-acid commercial, "Click-Click. Oh, what a relief it is."
god i love vga connectors with thumb screws they have solid connectors.
@@superjimse Likewise DVI and, my personal favourite, 13W3 (I use a lot of SGIs). SCSI also had good connectors, though SAS changed to become a bit more flimsy.
And lets not forget the trusty old parrallel ports for our dot matrix printers 😂
@@administrator4728 Thank you for your reply. I always read them so I can get better incite with my comments. I have a new 4090 GPU, and this connector problem disturbs me. I'm getting older, and sometimes my memory fails me, so I might be wrong about the two MOLEX connection clips.
@@chrissyboy7047 Thank you for your reply. I feel that the older-style connectors were indeed better. The PC component designers seem to like to take two steps forwards and one step backward.
The power limit increase kinda suggests they're getting ready for a variant with higher power draw .. which would be expected of something like a 4090Ti 😉
This wasnt nvidia.
It's great that they admitted fault but with the pressure from the media and health and safety issues they had little to no choice. This is the R&D that should have been done before the release of the product and due to the safety issue the companies need to do a recall. Great video Jay
Jay, if you can do that sense pin loosening thing as a follow up, that'd be valuable. If you could get a card going black screen and 100% fan on camera, it'd probably help some folks troubleshoot their cards better.
I had this exact thing happen to me last week. Changed my power cable routing from the top of my 4070ti to the bottom and got "No signal detected" and instant 100% gpu fan speed when i turned on the system. Had to wiggle the wires going into the plug to fix it. I never unplugged the adapter from the gpu, and where it plugs I to the gpu was tight. These adapters are so fragile that I would like to see a tsb with new adapters sent out.
I randomly on a daily basis, encountered black screens with fans full speed and audio still playing. I pushed all my connections in tight and it's been stable so far for 2 days.
I also encountered black screen and 100% fan speed issues on a daily basis after getting the cablemod's 12VHPWR cable even though i have checked that the connectors are all the way in and reseated the plug multiple times, but the black screen 100% fan speed issues would still happen frequently and randomly, but back then i didnt realise the cable was the issue, until i decided to go back to the 12VHPWR adaptor that came with the gpu and realise that the issue is gone.
Decided to RMA the cable with cablemod and they didnt ask a second question and replaced the cable immediately, which was nice.
But even with the new cable it still happen again for 1 or 2 times after using it for about 2months now, which was much better than previously where it was happening almost everyday, but still not optimal.
@@flamebullet I was having this problem and I uninstalled and reinstalled the graphics drivers and it seems to have fixed the issue. I only did it a couple days ago though so I’m not sure if it’s completely fixed but it ran for 9 hours yesterday with no problems even though before it was happening like every hour.
@@sotempermental247 i have already ddu the drivers before switching the cables and it seems to work fine for a while and i thought i fixed it until it happens again after a few days. Other things i tried including reverting my overclock since i thought maybe it was unstable(it was just a slight overclock, not aggressive), and even underclock the card.
There really needs to be a formal recall on these things, whether someone's current GPU has already fried or not.
I put Nyogel dielectric grease on my 12VHPWR to help combat terminal fretting due to inherent case vibrations from fans and pumps.
Also built a connector clamp block to eliminate any possibility of the connector being wiggled crooked.
Links or pics?? Would like to see this sounds like a good idea !!
Send pics please
@@timlaw4720 I posted a video on my channel about it
🪄✨ Made with SummarizeYT app
0:00 - The 12VHPWR plug is finally getting an update and a name change to 12V-2x6.
0:23 - The update aims to prevent melting and improve the reliability of the plug.
1:20 - The changes come from the collaboration of various board manufacturers, not from Nvidia or Intel.
2:32 - The new specification includes changes such as power draw, temperature limits, and minimum force required to pull the plug.
4:02 - Different manufacturers and suppliers use different pin types and wire gauges, leading to variations in the adapters.
5:01 - The new specification ensures compatibility regardless of wire gauge.
6:43 - There are two types of pins in the market: Astron and NTK. NTK offers better contact and greater contact area.
8:02 - The standardized spec recommends the use of NTK pins instead of Astron.
8:27 - The new spec increases the power rating to 675 Watts, allowing for better power delivery.
9:00 - The spec also allows board partners to draw a full 75 Watts from the PCI Express slot.
9:35 - The new 12V-2x6 plug has more contact surface area compared to the Astron 3 dimple plug.
9:39 - Changes to the terminal side of the plug have improved contact points and decreased the chances of poor contact and melting.
11:01 - The new specification includes a minimum of 150 watt sense pin for better communication between the graphics card and the power supply.
12:10 - The new plug design includes a safety mechanism that disconnects the plug before it can cause melting or poor contact.
13:16 - The plug now has a minimum wire length requirement to prevent tension on the pins inside the plug.
14:00 - Loose plugs or damaged sense wires can cause black screen issues and 100% fan speed.
15:00 - Using NTK pins on all adapters is recommended to ensure reliability and prevent melting.
17:06 - Users with three-dimple plugs should consider exchanging them for NTK plugs to improve safety.
17:28 - Repair services may start offering plug upgrades for existing cards to prevent future issues.
19:02 - Using NTK pins was discouraged in the past due to higher costs, but it provides better reliability.
The new connectors are now on current 4090s from a couple of months! Also the current cables are compatible with the new connector!
Yes, saw a conformation on Reddit today. You have breaking news! Good job sir!
How do you know if the card has new connectors?
Same user error issues even with the new connector. Properly seat your power cables, buy a case that fits, do not bend cables, do not use 3rd party adapters, have a professional who knows what they are doing building rigs and you will have a %99.9 success rate according to the data
One reason the dimples may wear out is metal fatigue. Metal, including copper, expands as it heats up especially when under heavy loads. Then when it cools off, the metal will retract making a looser connection. Eventually those dimples will Stay loose. We had that problem with our deployment equipment in the desert. We started several fires not knowing why at first. Once we figured it out we had to retrofit connections and newer systems came with better connections.
That was an excellent job Jay, thanks for taking the time and putting the effort in to a subject that's not easy to explain. Well done 👍 and kudos for the edits.
are you playing world of warships?🤣
My friend and I both have had the black screen issue with our 4090’s. I solved mine by removing the 90 degree angle connector from Cablemods. His is currently in RMA.
It's what he said, it was probably not sitted properly.
Can confirm, I was having the black screen & 100% fan speed issue and it was a loose cable mod connecter at the card. Visually it looked fine, but I reseated it and haven't had an issue in a week. So possibly some defective adapters for other people but I 100% think that a loose connection is the issue.
Exactly same thing happened to me.
Same. Tried everything and nothing worked. Turned out to be the new replacement Corsair cable. Replaced it with the default cable that came with the card and not a single issue for about 5 months.
Properly seat your power cables, buy a case that fits, do not bend cables, do not use 3rd party adapters, have a professional who knows what they are doing building rigs and you will have a %99.9 success rate according to the data
The problem is NOT in the cables. The problem is in a bad 12 V standard. Communication centers and data centers have been using a 48 V standard for decades (I worked with it in 1972). That would reduce the current by a factor of 4, stopping the cables from melting.
One reason they chose 12 V is that the 1920 buck converter used in VRMs cannot do a 48V to 1 V reduction. You have to use a 2023 design such as the Ćuk-buck2 converter. Unfortunately, it is cheaper, smaller, more efficient, and provides a better transient response. Adopting it would require the "engineers" to actually learn some of the last 100 years of improvements in power conversion technology.
I feel like this plug should work kind of like how an old VGA plug works and you actually tension it tight
I was wondering why they didn't just throw a couple of sense pins on a wholly different connector like an XC90 (rated for up to 90 amps).
55 amps is light work for that connector.
2:08 things are built the way they should. Engineers just did not think that .01% of users would be incapable of pluging a connector all the way in
not often you allow the big F bomb to be in uncensored jay, great video. Keep it up, team!
I sure wish that WWII game had existed decades ago... My grandpa, rest his soul, would have LOVE that WWII game! He had filled his library full of books about the warships from that era. He loved opening them up and talking about them. I can think of nothing he would have enjoyed more than that game!
It would be really nice if there was a threaded connector or some way to keep the adapter secure.
are you playing world of warships?
if not you should then you can sink jays battleship🤣
Threaded........how. Do you even know what this connector looks like?
It's about time! My Strix 4090 got burnt down even though i had my pretty connector connected correctly. It destroyed the 180° cable mod adapter too. It's a garbage connector.
As someone with a original 3080 I hope the 5080 series fixes all the 40 series cards problems (Size,power consumption/temps, PRICE, and connectors!) Big reasons why I never "upgraded"
The size can be great, cool and quiet running cards! , the 40 series is very power efficient already, temps are great, PRICE is crap and the connector needs a fix
It will fix most of those problems, But it will not fix price.
EDIT: Oh and you can also forget about the top-of-the-line cards using any less power than "all that the connectors can give them" that's the whole reason they're able to perform better than ever is that they're drawing more power than ever. Efficiency gains need smaller transistors (that's a ways out), or a large change in micro-architecture with an eye towards power efficiency rather than raw performance (not happening, or if you do want that, go buy an AMD card instead).
@@44R0Ndin Hmm? The 4090 i rarely see draw even 400w but Cyberpunk 2077 with patch tracing, 4K everything maxed (No DLSS and Frame gen) will push it to 500. You can also limit it around 60 - 70% and lose barely any performance. RDNA3 is not very efficient
Good to know they're doing something about it. My MSI Suprim X 4090 melted today, together with the cablemod adapter...
This is good content. You're channel is always good, Jay.
Thanks Jay, for taking the time to teach me about the birds and the bees.
I'm glad PCI-SIG is getting an update out, but I have some questions. I saw no mention of the X-axis rocking that caused the cable to become disjointed at the board connector. Wouldn't a three sided shroud around the wire connector prevent that in addition to the updated pin designs? Additionally, regarding the sense pins specifically, I had figured they would be cut back, but I'm curious as to why CARD_PWR_STABLE wasn't lengthened instead of being cut, since supposedly that's the lynchpin with regard to letting the PSU know to cut power to the GPU in case of overload or off-kilter power delivery circumstances. Any thoughts on these?
Just speculation here, but maybe a shortening of CARD_PWR_STABLE still serves the same purposes? If that pin becomes disconnected, then the power supply looses that signal and (hypothetically) still cuts power to the GPU?
Personally, I don’t care what pins get lengthened or shortened, as long as the failsafes do their job somehow and prevent the GPU from burning up. I’m also a 30 series owner who isn’t worried about 40 series as the whole lineup has been a disaster.
@@StolenJoker84 CARD_PWR_STABLE is by default open. That is, there is no connection at all even when the sense pin of the same name is fully seated. This is by design. The GPU will short (close) the circuit on that pin if there is an overload or another power anomaly (in theory anyway). By lengthening this particular pin to full length, it will always be connected at the connector and thus the GPU will be able to perform its failsafe of shorting that pin in an abnormal power delivery scenario. The rest of the pins being shortered is something I had already figured should be done.
The new update does not unfortunately cover the rocking present when too much strain is placed on the connector from cable positioning/stress. A three sided shroug (top/sides) that is flush with the board connector housing and going down far enough should prevent the cable from ever rocking, and at least in that fashion prevent the connector from entering into a situation where arcing occurs (this is what heats up the connector and melts it). I'm kind of disappointed to see that all these things were being addressed but not these two, especially after the GN videos clearly showed wear lines that weren't parallel to the end of the connectors, but lopsided from the connector being both loose *and* pulled to one side.
@@SquishedTia Ah. Ok. I wasn’t sure how that signal worked, so thank you for the explanation.
Properly seat your power cables, buy a case that fits, do not bend cables, do not use 3rd party adapters, have a professional who knows what they are doing building rigs and you will have a %99.9 success rate according to the data
The 30C above ambient will not refer to the ambient in your room but the ambient in the case in which it is fitted.
Since most plastics start softening at around 100C I would prefer to see an absolute temperature specification and not a relative
So if my 4090FE burns my house down, assuming I survive, I can take legal action against the leather clad bell-end???
Thanks for this update Jay! Cheers.
Man ! Jay ! I love how thoughtful and clear your presentations are...Good Work!
Hey Jay I was a big time RC car and truck user when I was a kid anyone that was in the late 80s to 90s prior to Lipo were forced to use Nicad batteries and the higher the volts and amps lead to massive battery fires and shorts this reminds me of back in the day, one thing that came to mind was the connection between speed controller and battery the connection that had the split in the receiving pin or female side was found to be the culprit in the loss of connector when the plugs were connected and the power was surging through it any sort of gap or misalignment of the fonnection resulted in an arc similar to the idea of a spark plug, In this scenario it can be just as detrimental as it creates heat and with heat then the connectors with the split would open up creating more arcs all of the gpu plugs are reminders of the old Tamiya plug styles or Molex that as a racer we would clip off in favor of the deans blade connectors back in the day, im curious if any of this holds true in these current connectors on gpus now adays
I bought a 12vhpwr cable and 90 degree adapter from cablemods, I would love to see an "upgrade plan" where you send your cables/adapter and for a reasonable price they would upgrade it to the new 12v-2x6 connector.
You would also need a new card to get the new adapter tho? So why would Cablemod do that?
@@AndreasRP. You don't need new card, the old and new connectors would work with each other. There are the benefits of having the new plugs on the cards, but just the ntk upgrade would make it much safer.
@@eliadbu I see, I thought they redesigned the adapter completely
@@AndreasRP. considering that companies made new cables, adapters, extensions, PSUs etc that would have been the worse outcome for both companies and users. imagine you bought 200-300$ new PSU for the new connector so you could "future proof" it and use it for the next decade or so, then at the next generation GPUs, that connector on the PSU becomes completely obsolete.
@@eliadbu That's happened before, doesn't anyone remember the AGP graphics card socket?
Thank you Jay, very informative
Glad I went with a 4080 😊
Same lol.. that thing eat like 300w on worse scenario... And perform very well
They should introduce lockable cables like the ones used in cars for everything. A lever latch that will only fully close if fully connected. No point for those half-arsed solutions.
So happy I went with a 4070 Ti. Not a single report on this card regarding melting cables / plugs.
Been working like an absolute charm with my CableMod 12VHPWR cable for half a year now. :)
I have another theory as to what may potentially be causing it as well. If the pin is bent slightly (so it is no longer centered within the plastic housing), this will force the connector contacts apart. While I have not had this happen with a 16 pin, I have had it happen with an 8-pin. It didn't cause a fire or anything to melt, but it caused the system to continuously bluescreen. I had to discard that power supply cable since the contacts inside one of the connectors was forced open by the bent pin on the GPU connector. I also had to re-center the pin with needle-nose pliers to avoid the same fate happening to my other PSU cable.
Jay You just saved me a fortune, I was just about to pull the trigger on a 4090, I was going to order one at the end of this week, but now I’m going to wait, great video thank you 👍
"now I’m going to wait" how long are we you gonna wait? weeks, a month, 6 months? that's what we need to know. Do we need a new cable too for our 1000 watt PSUs?
@@paulworster3683 I'm just gonna wait for 50 series! I'm just wait for the new 60 series, way better! I'm just gonna w-
I bought a used 3090 Strix because it has 3X 8 pin connectors. I looked at the 4090 but every one I tried had coil whine and some of them were bad. I would have bought a Red Devil 7900 XTX but they were hard to get then. I am going to build a new rig in a couple years. I might put an AMD card in it.
@@kristofvanderveen34 that’s ok I can keep my 3080Ti for another year.
@@paulworster3683 well I’ve got a 3080ti, so I guess I can wait until they sort this mess out.
the world of warships ad spot is the only one i watch all the way through just to see Admiral Jay
At 600+ watts, design the cards to plug into the wall itself and not the PSU. That would be common sense.
Probably the only video that explained it properly, thank you
This is the first time in a long time my strategy of skipping a generation for video card updates has saved me some trouble. Thanks Mr.3090....you may be a little slower at 4k, but we're doing just fine and also not on fire.
I don't think these changes being made are going to cut it. 4090's are burning out at the power plug like Canadian wildfire, and 4080's that draw 320-350W are also still burning out, just not quite as rampantly as 4090s. 4070's however are not, at only 200-250W draw. I have said from the beginning when I first looked at the specs and total contact area that with the total contact area of all pins and terminals that 250-300W PER connector was likely the limit of the plug. It is only around 45% more contact area than the ATX-2.0 8/6+2 pin connector, and only ~70% more than the ATX 6pin, which translates into the same percentage of increased safe transfer capacity. Contact area determines safe wattage via ohms, or impedance. Impedance translates into heat when you try pushing more power through due to natural resistance. At least the change in terminal style from minimal nub contacts to the spring style bars takes that into mind, but it is still not likely enough. I am hypothesizing that they are figuring it at double the 6 pin terminal capability due to twice the pins. But I have also seen 4 pin Molex connectors carry more current that smaller 8 and 12 pin connectors due to surface area impedance. Total safe transfer capacity is directly proportionate to total surface contact area. I think the ACTUAL solution is to tone down the wattage per connector and increase the number of connectors. No matter how in depth and complex the project you are working on, you should NEVER neglect or forget the basics. Even the most complex designs and standards start with the basics, because if they don't start with the basics the basics get forgotten and failure becomes a forgone conclusion.
I loved the segment on this that Steve from GamersNexus made in his Weekly News Recap. "I was looking at the changes and thinking, 'This looks familiar...'." The changes were basically exactly what he asked for. 😂 Thanks for the coverage, Jay.
GN is the source of the victim blaming tho. With all their testing, they still erroneously blamed the user in the begining. Sadly, a sign that they are starting to pander to advertisers. GN was the main source of the bad info despite any good info they had.
In addition to heat, I suspect all the vibration from fans and pumps eventually shakes some cables looser. The combination would certainly have that affect.
Would vibration be caused by your computer being placed on an uneven surface?
Honestly, I'm thinking it's still best to wait and see, or if you really want a 4090 go get something that isn't that high on the tier list in terms of power limit setting. I recall some AIBs are set to use the whole 650w while some are fine with 450w as the power limit. Galax's HOF even uses 2x 12VHPWR with an LED light display that tells you if your card is only able to use half the power due to not being plugged properly. Here in Asia, I can find a lot of those L-shaped 12VHPWR connectors that look exactly like what Jay showed before and there is one thing that is labeled the same across all of them from what I have seen. All of them say "rated for 450w". Now I wonder if that's a possible cause of issue if you, say, use one of these connectors on an ASUS Strix RTX 4090 which is said to be set to have the full 650w power limit, and then plug the 12VHPWR connector properly as a whole. Will the connector cause an issue? I can only guess.
This is why I went with the MSI 4090 Gaming X Trio. They power limit them for thermals just a happy accident. That It also keeps the damn thing from melting as well.
@@lostinny690 Yeah, I actually heard about that for that AIB. Apparently the Gaming X Trio still uses the "signature" Gaming X Trio shroud format with the fans and all from the previous Gen. so in order to tame the temps, they had to deal with the power limit. The result was a 4090 that was no less powerful than the others, but is still well balanced with cooling for its preset power limits. Definitely a happy accident.
Does the same apply for the gaming x trio 4080 and 4070ti?
@@sv0g391 I don't think they had to power limit those cards as they don't draw as much, nor do they get as hot. Also, to my knowledge, no 4080 or 4070ti has ever had an issue with the melting plug.
Thanks for covering this Jay.
There's no reason the connectors aren't made to be 90 degree in the first place given most people need to bend the crap outta the cables to manage them in cases.
Kind of random, but I always enjoy your videos. You guys all do a great job with this channel keeping things fun and information rich. Keep up the great work and know that your hard work is greatly appreciated!
so the cards are burning and they are increasing the power makes sense to me🤣
Great user technical review, THANK YOU for your 2 cents.
That's actually a good thing that I couldn't afford the RTX 4090 last year. Should be ripe by the end of this year
5000 series is coming out in 2025
@@AngryChineseWoman I've waited enough.
The failure rate is still relatively low, at 0.05% to 0.1% based on the data collected from board partners, cable makers and GN's viewers.
Honestly this issue pushed me away from the 4090. I know there's "bigger reasons" I shouldn't want one but gaw damnit I can afford it and I wanna use the types of graphics it's makes possible
Radeon Pro 7900 😈
it wasnt a big deal tbh, i have my 4090 and i find really dificult to make that user error
Also why I shied away from the 4090 and ended up with an XTX. I sometimes still "regret" not getting the 4090, but I know I'd be more upset if the card melted b/c of that adapter.
@@IvanScore4life do you think 3rd party adapters would be safe enough to use? I'm gonna need a 90° adapter to use in my case
@Marzsala I am a sim racer and I've noticed that card seems to do really good with those physics, but I've heard the amd cards are unreliable?
I think the primary problem with the 12VHWPR connector is they shrunk the 2x6 layout to match the width of a PCIE 8-pin connector, and of these little problems cropped up and combined together to create the failures. Smaller pins means less surface contact, the 3 dimple situation for example. There should always be a landscape cost for higher power draw. The larger chambers can also accommodate larger wires as well and better connections across the overall connector. There was no valid reason the connector needed to be so small.
Whoever was responsible for this mess needs to pay for GPUs to get an upgraded power plug
As a engineer I will tell you that we are always on a tight budget of time and money. Even if we tell the higher ups this won't work because (insert Technical reason) and we need to (insert better but more expensive option) keep it safe. we will get overuled, they don't understand what we are saying and won't comply. All they want to see is numbers going up so they can afford they're 3rd car. And even if it destroys the company 1 year later it won't matter they will just go and destroy the next company they already made there money.
@@leroyjenkins0736 I always blame the bean counters over blaming the engineers.
Engineers naturally want to make something that lasts a long time and has a lot of features.
But that's "too expensive" they say.
And then in the same breath they charge about $2000 USD for a 4090 that hey guess what we saved $0.01 by using this cheaper power connector to carry 600 watts, so now it melts instead of providing years of trouble-free operation.
Fucking bean-counters. Every time.
That's even what killed both space shuttles (and their crews).
The engineers with Challenger even SAID "Don't launch now, this is outside the launch constraints we agreed on when we designed the thing" and yet they got shouted out of mission control.
Yeah that went well, we lost 7 good people because of that whole mess of stupidity.
DAMN ! new crazy grip on those plugs, GJ !!
Great indepth video and promising signs ahead, sometimes it sucks being the early adopter of new generation of stuff, glad I've been holding off buying a new gpu, guess will wait to see how the new batch of cables / gpu's go once its all released.
Had my MSI 4090 since February with the EK waterblock and active backplate and no issues yet however I wouldn't mind paying for the upgrade if it means keeping my 4090 safe.
Same, had my 4090 TUF since release with no issue but till def get this
The problem with the 12vhpwr connection is the fact that you are pushing 16awg wiring to it's limit and the pins are smaller than 16awg which makes them heat up. If you notice in the new drawing the pins are larger than the original to give more area for current flow. They cannot have a smaller area/contact point than 16awg stranded wire. I made this observation when you made the first video about this. No one is talking about the connector pin size but they increased it for a reason. THAT IS THE PROBLEM but they will never point it out because it makes them look bad.
I always thought it was madness to take what was basically 2x 8 pin connectors which would have carried a total of 300W between them, double the power delivery and shrink the size of the contacts/ connectors at the same time....or were my concerns too simplistic and common sense-like??
@@darthvalkyr The issue is dumb engineers. 1/4 of the problems in the world today are stupid engineers and the other 3/4 is just lawyers.
I have a PNY 4080 and i haven't any issue with the connector. I am happy so far.
4080's have less issue been some but not on scale of 4090.I just went with a 4080 myself until this issue is done but my 4080 serving me well so keep till the 5000 series or see if AMD actually steps it this time with the 8000 series
@@johnnyboogalo4897 yeah the most cards have connector issues are the 4090... I am curious what Nvidia do with the next GPU series.
Did you watch the video? Lol. Mostly only 4090s are being affected, because they're the only cards that draw enough power to melt.
this is good but we shouldn't even be having this conversation
Jay looks pretty sharp as a rear admiral.
Few things that is concering. 30 degrees C delta at presumed max operating power for me seems like a lot. Because this would mean the safety factor is very close to 1. As soon as the power that goes through the plug exceeds its max then the tempature increases exceptionally. Higher temp higher resistance more resistance becomes more tempature.
What i would like to see if it was 5 degrees max, and a safety factor of 2 or more. But to cut cost they probably went with a safety factor of 1.3 or pretty low.
they should put clips similar to like what you would see on a guitar case, where one side is a loop and lever and the other side is a stud and the loop pulls the latch closed as you flip the lever
So the CableMod 180 adapter "may" have the Astron connector? I may need to check on that. Honestly, anything with this amperage rating should have at least a bisected male and a solid circular terminal. Think XT90 with an 8ga. crammed in it.
The fan vibration will cause a disconnect too.
I dont understand why the GPUs dont use bullet connectors like in the XT60 plug for lipo batteries that is used on rc cars with high current drain. The 3.5mm bullet connector in the XT60 plug can handle 60A at 12V. If they had got 2 connectors for 12V, 2 for ground and 4 for the sense wires, it would be less chance to fail than the normal 8-pin connector.
Based on a couple videos on the new plug, it was never mentioned if the retaining clip was adapted or widened to limit the amount of wiggle of the previous version.
That is good to know about the black screen and 100% fan speed been having some of that since getting my 4090.
Great video Jay, I felt like I was back in a big meeting at work...
I knew a builder that had a "key" set that would release the pins in Molex connectors, and he would add a drop of Silver solder to each pin and then reinsert them in the connector. When I asked him about that he told me that "People have a habit of pulling on wires to disconnect components, and this helps save the connector."
I figured this out in the beginning. What took them so long? They deviated from the standard Molex plug and connectors and screwed it up. I have been in this industry since the late 80's and this bugs me when an obvious flaw is designed into equipment.
18 gauge is definitely inadequate, 10 amps max on 16, and 18 is thinner. but as he said the pins can be part of it too. most people think current runs through the wire, it does not. it runs on the outside surface of the wire and the pin, if you don't have enough surface area contact at a joint ( plug) it can heat up, also if there isn't enough friction, you will also get heat build up.
Jayz, I think we need a new celebrity build or something like that
Nice hat, bro. Wear that around Morro Bay. You can walk right into the Yacht Club. Well, the gate's never locked, so you can walk right in anyway. They usually have a free hamburger BBQ every Friday afternoon, FYI.
Got the angled v1.1 adapter in the NICK of time just checked my previous adapter, started melting luckily the gpu connector itself is fine save for the top right connector now being dark grey
As long as we are monitoring sense wires in the connector, why not add a thermistor, that could actually detect if the connector temperature rise becomes unacceptable? Seems to me it wouldn't add much cost.
What's wrong with using the XT60 connector? It solves every problem on the 4090 when it comes to power draw. It seats in better, it feels better when it's seated, there's no ambiguity when it is in; no more 'is it in fully?' and you can see it's seated in. Plus, it can carry 60 amps and above at these voltages.
Jay, My son and I went to Microcenter over 4th of July Saturday and purchased two ASUS Strix OC 4080 cards. At the same time we ordered two of the Cablemod right angle adapters that arrived the next day. So on Sunday, just as the adapters arrived, my son notified me that he just had a meltdown on his 12v cable. we were OCD ( as you mentioned in this video) about making them secure and ensuring they were not bent. PS was a EVGA 1000w Supernova G2 fyi. We took the card back to Microcenter and they replaced it no questions asked and the manager was aware of the issue. Anyway, we both are using the adapter, pre-bent cable running straight down and then elbowed to run rearword and made sure there is zero stress on the plug. Very odd that it happened on a 4080 though… we have pics if you want to see…let me know.
Jay - great video - I Can 100% confirm your point around the 15 min mark; re: wiggling the connector to get the black screen and 100% fan ramp; This is continually happening with my Gigabyte OC 4090; This happens both randomly and with a physical intervention on the plug.
Why am I watching a video about pins & cables for GPU's I cannot afford and don't have I don't know - however, my brain loves information and Jay video's are informative and addictive.
Thanx for all the content - I'm not sure knowing this for talk at a party will make me friends though.
Love it all ❤
"Over there, hopefully not melting ..." *FWOOSH!!!* Well done, folks. Well done! 😄
Very well explained, thanks!
love the money stack above the box
That last line... Ooofff! I felt that!
so glad i made cables for a living and do quality for them now and made my own when there were issues. it uses general molex style pins and connectors. i really just used spare ntk stuff from work. i really hope they actually fix things for the public consumer.
You know, since you do this for a living, you could have (and probably still can) potentially start up your own business making cables for this application that are "actually worth buying" NTK stuff.
I've literally never seen any connector in an automobile use that "3 bumps" connection method, and with this whole mess with the +12vHPWR connector's issues with melting, I have a pretty good guess why I've never seen it in the automotive world.
@@44R0Ndin I could make custom cables but i would never in my life make cables i could not 100% guarantee, and with this plug you cant at the moment above 450watts.
AMD was SOO smart in avoiding using this connection right off the bat. They were smart to think ".. lets see if this actually works first". This is such a costly mistake for 4090 owners given how freaking expensive those cards are.
The only 2 ads I dont skip, World Of Warships and iFixit.