The 12VHPWR adapter cable is DANGEROUS! But NVIDIA doesn't agree...
Vložit
- čas přidán 23. 10. 2022
- Get your JayzTwoCents Merch Here! - www.jayztwocents.com
○○○○○○ Items featured in this video available at Amazon ○○○○○○
► Amazon US - bit.ly/1meybOF
► Amazon UK - amzn.to/Zx813L
► Amazon Canada - amzn.to/1tl6vc6
••• Follow me on your favorite Social Media! •••
Facebook: / jayztwocents
Twitter: / jayztwocents
Instagram: / jayztwocents
SUBSCRIBE! bit.ly/sub2JayzTwoCents - Věda a technologie
Test the cable you bent, and use a Thermal Imaging Camera to see if it gets hot (or even melts).
Yeah, I pretty much already resigned to the fact that Im going to have to do this.
@@Jayztwocents You will have my view! And my like!
@@Jayztwocents waiting for the vid now. Be safe a fire extinguisher handy! You maniac haha
These cards should come bundled with a high quality surge protector now. 600w through these tiny gauge cables…… add to that a connector that was designed and built with anything but safety in mind…….
Or just wait a few days for more cases of this
Intel not using the very plug that they helped design is absolutely hilarious
It wasn't ready honestly for mass use. Besides their graphics cards aren't using power too much
Giving me Apple and USB C vibes.
It really cracks me up with this theory. 🤣
their cards don't need anywhere 600 watts
😂😂😂😂
The more problems there are, the more the EVGA breakup seemed to be a 200iq move. Dude if NVIDIA has to spend money redesigning every 4090, EVGA is going to be LAUGHING at them. Oh my god this is too good.
They'll probably just make a new adapter with a right angle plug.
All they have to do is not use a plastic that’s in a ton of computer parts that starts deforming at only 105C. ABS plastic shouldn’t be in computers anymore
Manufacturers should have freedom of choice in power connectors
I mean, I have yet to see a report of this happening on Founders Edition cards. So far all the legitimate sources say it’s all reports from AIB cards. Which include non-Nvidia cables.
Also, EVGA isn’t going to be laughing anytime soon, their mobos aren’t really relevant and that’d be a hard game to break into anyways with MSI and Asus in the mid and high end with Asrock (literally just Asus) covering the budget end, and while their PSUs are good, there’s better ones at similar price points usually (for example Corsair, Seasonic, or Asus’ ROG strix line). Their AIO coolers are also very mediocre. Like…what are they going to make that’s relevant to keep the company alive? Switching to AMD in GPUs would’ve increased their chances but it’s not like RDNA is relevant in any meaningful way.
Weird comment.
8.5 per pin would equate to around 600W (612W to be exact) That's pretty close to the spec, maybe not much headroom there... (Before anyone asks how I worked that out, it's 8.5A x 6 pins (because the current path flows in through 6 and out through the other 6, so that's how many you count) x 12V (Watts = Amps x Volts) So 6 pins * 8.5 Amps = 51 Amps total, multiply that by 12 Volts and you've got your 612 Watts. The problem I think is that they're running the connector so close to it's limit that unless ALL 12 PINS make a really good contact, the current will be shared unevenly and may put extra load on the other pins both heating the pins that don't contact properly due to resistive heating and also overloading the others that take the additional load.
Best way to use 4090 is to just decrease power target to 80, significant boost to efficiency and eliminates majority of the risk from connector.
I think connector is great when used at or below 450, problems arise when it is being pushed to 600.
@@erenbalatkan5945 but then it is better to stay on older 30xx series and push it to the max and having nothing to worry about lol. seems very counter intuitive
@@erenbalatkan5945 Under normal usage the 4090 rarely ever exceeds 450 watts anyway, so this isn't much of a problem.
Tfw youtube comments sections understood Nvidia products better than any of the higher ups assume to
So basically. People are right, in that the PSU should send more volts than just 12 to the GPU. But what PSUs can do that? None of them.
I’ve seen this 100 times in the comments, with just more “Nvidia bad” fluff.
Even with all of the red flags, Nvidia's refusal to accept that this is a legitimate problem has always been baffling to me
it would cost them money. money is more important than customers
@@pagannova3621 Aren't they already losing money anyways?
It’s just like car companies. Until the cost of settlements becomes higher than the cost of recall, people can die for all they care.
Even with all the red flags, people will still continue to buy from them. Look at when it was announced about that amazon MMO wrecking some 3090's I think it was... did it stop people from buying the game after the face 'just to see'. Nope
Just like how the 4080 12g was "Confusing" people. Kinda feels like there blaming us for them trying to rip people off.
I am certain Nvidia marketing department is super thrilled to see that branded melted connector.
Maybe this is karma for all the push nvidia has done in the past to push for software developers(on the 3d industry end) to use software that is nvidia specific. Hopefully nobody gets hurt in this and nobody's houses burn down, BUT hopefully this pulls nvidia's absolute monopoly in the market back a bit and causes their footing to slip so that more software is gpu agnostic and the "industry standard" no longer has to include an nvidia card to get the job done. There are entirely too many scenarios in the industry where if you don't have an nvidia card you quite simply can't use a particular program in a production workflow environment. Its been this way for a long time too. HOPEFULLY this cuts them deep enough that they have to take a back seat for a minute so that their hold lessens on the software developers who can open their workstations to be more compatible with amd and the new intel cards and any other gpu player that might enter the market. I say this because i work in the 3d graphics industry and I have ALWAYS been an amd fan long before they were amd, I always liked their architecture for their gpus and they make really great, and more importantly, really stable cards in my humble opinion. And while I do currently own two amd gpus, both of which are quite nice, I can't use them in my typical workday. I have to use my nvidia machine because i use too much software that relies on nvidia specific stuff and if I use amd I have to do the calculations with my cpu which is job suicide because it is so much slower. And the saddest part is, my amd cards are techically faster and better than my nvidia but theyy might as well be paper weights in the industry. Especially when you consider than in any given day I have to use 6 different software programs, and no less than half have an nvidia specific pipeline. (and the ones who don't often use something like opengl or similar which isn't always as up-to-date and causes its own issues)
@@bryanharrison3889 Hopefully. I'm just a gamer anyway. AmD will be my next card.
Is it at all possible that JonnyGURU was just being sarcastic here? Just a quick glance strikes me as pure sarcasm.
Ow the PR team is sending dms to people with burned connectors.
@@zkilla4611 AMD is using the same connector on their new cards. PCI-SIG's standard is the main problem.
your being to kind. You raised this issue before anyone. Credit to you..keep up the good work
I put together a video editing PC at work with a 4090. That was my first time seeing that connector. The 4090 needs 4 of the traditional GPU power supply cables/connectors. Being handy with electronics I know that electric current delivery has everything to do with the gauge of copper wire. I also know that using connectors limit the amount of current that can pass through. You have to overspec the connector to handle more current than the current limit of the wire. Typically those connectors are much larger than the wire to allow more surface contact area on the pins.
It was weird seeing these 4 large power cables go into this relatively small GPU connector. And the 4090 is almost twice the power consumption of previous generations, so that makes it even more concerning.
I’m mostly confused why they cared to make the connector so small when there is plenty of room to just keep doing what they have been doing for years.
Hey Jay, great video and we had to expect this, I have been working with 12V DC circuits for the last 17 years, formed as a car mechanic and now operating the service department of a Semi-Trailer dealer and I simply can't understand how the PCI specification can rely on these tiny connectors for powering 600W, that's 50 AMPS at 12V! You should look at the specification sheet of Anderson Powerpole PP15/45 (I looked at document 2020-0055 DS-PP1545 REV C7). Those connectors are rated for a maximum of 45 amps and are WAY stronger than what PCI/Nvidia uses. Also, those ratings are with 10 AWG wires, that's 6MM (or 1/4") of copper! The PC industry does not seem to take into account for "what can go wrong will go wrong"...
powerpoles was my thought as well.
I would just like to gloat, I predicted serious safety problems the moment that 600W spec leaked.
Totally agree bud.
But that’s where multiple pins are supposed to come into play being 50a/12 pins is a bit over 4A per pin.
Now if they’re not spreading the load across all of the pins properly you’re going to have issues but a single 16 awg conductor can take 13 amps on its own so on paper it seems doable if the connector is rated for 8.5A per pin.
Stress points on crimps like he points out is definitely something that can come into play and he’s not wrong in thinking that as a starting point IMO
@@Mr.Morden Who are you gloating to?
Putting that much power through such a small connector immediately made me think that’s gonna be a problem. Can’t wait to see the prebuilt thermal volcanos go up spectacularly.
I think they should have looked into RC model parts for some suitable connectors.
The 3090ti variant of those connector had some genius because of the slightly put inwards tendencie... but on the 4090?... Time for "hard"wire-modding the splitters on the pcb of the card!
When I used to race 1/10 touring on a xxxs losi G+ I was running 6 sub c 3300mah nimhs that we had soldered together with bars made from melted down 24k jewelry and we soldered our leads from the speed controller to the battery packs every race. No connectors back in 01-02 for me.
Honestly agreed, even doing 500w clean amplifiers in cars don’t use thin cables, so to me using those tiny pins is baffling
Those molex terminals, if quality parts, are perfectly fine at those currents. So the concept, on paper, is fine. The issue is that they have a lot less wiggle room than the old connectors - mainly because the ATX spec had a 2x safety margin. So less than perfect (read cheap parts) still wouldn't melt stuff.
I agree with your stance on this situation. You encountered an issue and concern and you're just giving everyone a heads up and warning. Now users can be mindful about this and hopefully take care.
When did Jay encounter an issue exactly?
Oh boy, thanks Jay and everyone for the heads up!
Going to check my 4090 right now as I bent the hell out of that connector to get it to fit. I have been some plastic smells coming from the computer since the card was put in. But it was a full rebuild with a new PSU so just thought it was stuff bedding in.
I don't want to look!
So far I have been very impressed with the card, GPU temps are normally under 60c, never seen it hit 65c. 120hz 4k gaming with everything on max. Mostly Vermintide 2 and Metro Exodus EE.
I'll report back the good or bad news, plus the computer components.
EDIT: looks like I caught it in time. The "special" power cable was bent as much as i possibly could be to fit the side window on. I had bent it under the card though as that was neater.
I checked the female connectors in the cable. The ones that had to go the longest distance, outside of the curve, had depressed in the connector around 0.25 to 0.5 mm.
I straightened the cable and the depressed connectors did bounce back a bit, but not much. I am now running with the side off the case.
Now I need a new case with vertical GPU mount and a new special power connector.
It really does make you wonder why they didn't include a 90° connector to start with. They had to know from their own testing that it was going to be a problem.
bending cables and connectors costs money so go figure
Absolutely. That seems to be the easiest and most safe solution for now, but comes at additional cost on the customer's side that's not really optional if fire hazard is a possibility.
Nvidia and the AIBs that shipped the Nvidia adaptor should send their customers a free 90° adaptor.
Also, AMD please take notes on this to avoid a potential crash course.
@@thelegendaryklobb2879 For the price of the 4090, they should send at least 3, lol. Especially those trash plugs that were rated for like 30 inserts.
@Carl Gunderson Translation for non-Americans please
What do u mean ,quote on quote "it all just works man"
I was working in a server centre on Friday upgrading their incoming supplies and talking to the server techs they're dreading the new spec rolling into professional and server grade hardware because it's even more of a confined space and bigger costs if something goes wrong. You know a spec is flawed in any industry when the people that are doing day-to-day maintenance consider it a step backwards/ danger.
No changes ever! Just me on my cellphone eating Chik-Fil-A before I fall asleep!
When the servers catch fire because of this, and burns down the building, then they might learn the lesson.... Maybe.
@@NorseGraphic Datacenters have much better protections than any home.
I honestly like the look of the symmetry that 3x 8pin and the 24pin has. but i also like that they are trying new stuff. but they should have waited for the ATX3.0 PSUs to give more time for testing
Actually about the bending, Igor from Igors lab said, that when you plug in the cable FIRST and then bend it (which is, of course, easier, because it is fixed on one side now) the whole pressure goes to the connector. And the pins (card side), which are quite thick, can then bend/open the female part of the connector (cable side) just a little bit, which also leads to less contact and thereby to a higher resistance (=higher temp). This and the fact, that the outer cables are only soldered to one little metal plate, which breaks off more easily than the middle cables, which are soldered to two plates at once. And the connection from the little plates to the 'female' part of the connector is REALLY thin.
I love that an industry standard was pushed through that has such specific caveats. 100% safe and no issues...I mean, as long as you don't bend the cable within 35mm of the connector; which we realize is impossible to do in the majority of cases due to the size of the card. That's a consumer problem, though, I suppose.
That diagram from Cablemod is EXTREMELY concerning. Kind of bonkers that Nvidia is dismissing this out of hand. Do they really expect people to always have a 35mm relief before the bend? Do they not realize that many people building PCs basically just jam their cables in their cases and call it a day?
Exactly. That responsibility certainly wouldn't fall on the shoulders of the R&D team that spent years developing this pile of shit or the company that opted for cheaper connectors in lieu of a system that worked perfectly. Obviously this is the consumer's fault/issue/problem and responsibility.
Here's the thing though right, I have the ultimate solution and the tech industry isn't even ready, it's never been done before... What if we, and hear me out, used 2 connectors...
The 12vhpwr has 2x as many +12v pins as an 8 pin power cable (6v3) and 1.5x as many grounds (6v4). So what if instead of pushing twice as much power per connection we simply didn't.
Oh noo, the newly bridge is fall and eat many life in The process. Well that's the bridge user fault, i suppose
Cablemod now has a 90 degree adapter for 12VHPWR. I have a feeling those are going to be very popular. They should really come in the box for any card using them.
VHPWR = Very Hot Power
why is it 12 hp? 786 watts is 1 horsepower. Are you saying that all 12 leads have full power going through them? I didnt know that. Thats a shitload of electricity. My houses AC has only a half to 3/4 horsepower. Thats Krazy!
It's Voltage High Power
@@richiebricker Um, 12VHPWR -> 12*Volts* High PoWeR?
It's so high it can see tomorrow
amd is NOT using this connector on RX 7000 cards
Make a followup to this now that Steve on GN has done some real testing and analysis rather than just throwing out theories.
Igor and this dude wil never do that because they just care for rush clicks. Not because they want to find out about the truth.
Also, remember that your fire extinguisher needs to be rated for electric fires.
smoke alarm too otherwise it's pointless!
Should fit a fire suppressant system inside the case.
ABC class car fire ext. works
Also make sure the fire extinguisher is actually in service period haha
I'm already feeling sorry for the poor sod that let's rip with a powder extinguisher. That's some next level mess to clean up after 😂
I have a degree in electrical engineering, and although my area was more on the IC side of things, 600W at 12V which is 50A of current through that small of a detachable plug is just plain crazy. Think of your NEMA 14-30 Dryer plug, that is rated for 30A.
Maybe its time to raise the supply voltage in a PC now that parts are consuming so much more power. 24V or 36V will reduce thermal loads on cables and connectors a lot and still be safe voltages and the insulation requirements are not gonna be different compared to 12V, and the PSUs use DC-DC converting already so making new versions should not be too hard. There will be a period of pain of compatibility for sure but it will be worth it in the long run.
Heavy duty commercial automotive guys like me are laughing at how stupid this is too. I've seen OEMs pull this kind of stupid shit too with Halogen light circuits. Where the pins burn up in the breakout harness connectors after the copper oxidizes from sustained electrical load from years of use, increasing the overall circuit resistance. The designed circuit load rating for the gauge wire and size pins are no longer within spec, and the connector pins are the first fail point, because they're the smallest pieces of copper in the circuit.
I agree, I worked with 24vdc systems and the amp draw as you stated is alot lot less. 50a on that small of a pin would never pass in aircraft world. This is truly an expensive design mistake.
Moving to 24v or 36v is a great idea. You just make new connectors that don't fit the current stuff and problem solved. It'll start with commercial adoption and move down to retail. But the current draw now is becoming insane for the cheaply made PSU we're all being sold these days.
What I can’t believe is that an engineer said they just replaced burnt cables. I’m also EE and have designed MANY boards with high power interfaces. This connector doesn’t seem derated very well and with POOR strain relief.
Any burnt ANYTHING during verification testing led to an automatic halt to testing as it would be considered a safety issue.
Sad!!
I am also an electrical engineer. Another solution would be for everyone involved to see reason and not think 600 watts consumed by a single PC component is anything less than insane. This is the result of expecting and pushing perpetual growth in specs. We can no longer make the same gains we used to so now we dump more and more power into bigger and bigger components. We knew 10 years ago we were about to be pushing right up against the limits of physics. We couldn't get smaller and more efficient anymore, only way to go is up and out.
We are Teclab and made the tests (yoi called us Galax).
We run more than than 25, considering more than 12 minutes overload (rising 300W per step), and reached 127 Celsius.
Neither connector or cable melted, and it's still working.
Then we installed bad mounting (bad assembly, both horizontal and vertical, then temperature increasing over the time, what can cause damage and melt.
About 150 min testing alive.
Thanks
Oh damn. This really is an issue. Thank you for doubling-down on this and calling it out early. It was because of you I skipped out on the 40 series--the risk of fire is just not worth it.
600w at 12v is 50 amps. 50 AMPS. That’s an insane amount of power for 6 little cables to carry. (Uni direction)
That's 8.33A per pin, if all the pins carry load equally. If those pins are rated to 8.5A then potentially it could be sitting right on it's load rating for hours at a time.
18AWG wire can handle 19A per wire, so almost 120A total. The issue is the pins in the connector not making proper contact, raising the resistance and thus heat generation.
@@SixDasher They shouldn't try to make a high load connector, small and dainty and possibly easy to break, or not connect properly. The current PCIe connectors are chunky and safer.
@@SixDasher 18guage wire is rated for 14 amps at 90c. This can vary. For example, at 12v it can only handle 10amps to be at an acceptable temp. Which is too close imo to the 8.33 rating per wire/pin.
@@matthewfriemoth2502 The 12VHPWR spec requires 16AWG wires for all +12V and Gnd
Nvidia needs to ship the straight connector version, as well as 90 degree connectors in both axis (bending back towards the backplate, and bending towards the back of the card). Building the turn into the rigid connector and using it appropriately is the only safe way I can see this moving forward.
btw some cards have inverted connector so we should have two variants of 90 degree connectors for both axis. 1 straight connector, 2 connectors for cable going up for possible vertical mount (regular and inverted), 2 connectors for cable going down (regular and inverted), and possibly 2 connectors for cables going along the side of gpu (regular and inverted). I'm LOVING this future-proof design made by the most brilliant people of hardware engineering! (sarcastic claps)
I came reading the comments before suggesting this and I'll bet cable mod and others are doing this already. If not trade mark pending..... Kindjie & Davo power adaptors Pty Ltd.
You mean re-invent the 90degree connector from EVGA... they have an adjustable power rail for the 10/20 series cards...love mine used for both series over the last 5 years...
Absolutely. And they need to provide one free of charge to everyone who purchased a card with this connector on it.
@@Tink-GB EVGA makes a good one. I don't know who made it first, but I'm using some decent no-name 180º 8-pin connectors from China for my 3090s. Obviously not needed, but they look clean and reduce cable strain.
I remember using similar connectors for IDE and Molex connectors in my cramped 486 in the 90s! The technology exists, dammit!
I dunno how the "smarts" in these new connectors work, but I bet a nice black PCB with straight through runs from one connector to another would work fine.😂
In the car audio world, the more power and amperage you push through a device such as a subwoofer; you need thicker wires to supply sufficient power current efficiently. This can also apply to a high powered graphics card.
Nice to hear your thoughts and detective work on this case, Jay. I am waiting for CableMod's vertical adapter for when I pick up a 4090 FE personally. But even if not, do you know if you'd have the 35mm+ clearance in your SMA-8 build? Obviously not a mass consumer case for numerous reasons but yeah we'll just have to see if this is an isolated incident or a growing problem.
i love that CableMod is putting out a really stout 90 degree 12hpwr connector so your cable drops back along the surface of the card instead of pushing against cases and whatnot. i feel like those are going to sell really well
Not to mention it looks way better than a spaghetti monster coming off your gpu lol
in light of this R&D debacle, those connectors should be included free with each already overpriced card.
@@bryanharrison3889 While I agree in principal, your already a whale for buying a 4090 in the first place.
@@jonbeckham6463 Or maybe some of us just saved up... I for one am still using my 970.
@@k9px retired the 390 nitro for this
They should make 90 degree connectors that have the plastic also hold everything together for a longer distance, so when the plastic ends it is already at the point it is ok to bend.
sic, they’ll be a big seller
Cable mod has one after some quick googling for the 12 pin
While this is the right idea for the problem, its still a Solution to a problem, which didnt exist until nvidia went hurrdurr on the totally fine standard connectors.
It's stupid you have to watch so much on the bending of the cables, in an already cramped space because the card it self takes up most of the case.
Maybe the card developers should rethink there designs. No more "MoRe pOwEr = MORE POWER" and more "More efficiency = More Power" would serve us all better.
I'll never understand while every new cpu and gpu generation becomes more efficient the manufactures see the need to ramp up the power to squeeze the last 2 percent out of it to win some benchmarks.
@@CeLLeGER i mean you dont need to use the extra power, pretty sure they are more efficent than 30 series too.
@@Neishy4AGTE Though the main issue is that new CPU's and GPU's kinda fully overclocked out of the box. Sacrificing the last 5% of performance will reduce power consumption by about 30% or so. This lowered power consumption is more safe, cool, and quiet to run. Often enough performance is more consistent too if you no longer have a temperature bottleneck.
hahah I'll never stop watching this channel! The delivery is everything!
JAY! If you bent the pins, you might be able to put the connector under some magnification to see if the pins moved in their housing. You'll need a bright light, and maybe a microscope, like Lewis has. (Maybe. You might be able to get away with a decent magnifying glass.)
It might also be a good idea to purchase a couple of cables and do before/after shots. Lot cheaper than sacrificing a 4090.
Thank you Jay for putting consumer safety as top priority and not worrying about likes, dislikes, etc. This is the info that NEEDS to be shared throughout the industry and with builders and enthusiasts. Thank you for what you do.
If this was NZXT we wouldn't be hearing so much lol
who is giving him dislikes for this lmao? The likes are overwhelmingly in his favor. He isn't martyring himself over this.
@@_BangDroid_ Do you really think that if it was nzxt jay wouldn't bring up a potential fire hazard? lol. Ok.
@@_BangDroid_ He has videos about the NZXT H1 fire hazard, he reported on it the same as everyone else.
Whenever I see higher power requirements and see the connections get smaller, I start to question things. I had a feeling, and I’m glad I wasn’t the only one who was suspicious.
Eyup, really scares me, the Type C cable
Just wait for 1000W via Qi. 😎
Yeah, its like the manufactures like Nvidia, AMD and everyone else are trying to ignore fundamental physics or at least try to sell it as a "non-problem".
But, oh wondorous wonder, it is a problem which they then want to try to sweep under the carpet or still try to tell everyone "Hey, he's just stering up fear, don't worry - buy our damned overpriced products"....
@@finestPlugins That's called an induction stove top. 😅
@@ilovefunnyamv2nd This is true, but on the other hand I have yet to see a single USB-C cable or port fail on multiple devices. Compare that with Micro/Mini USB, where the port or cable almost always fail after so many insertions and require constant wiggling to make a solid connection. The connector itself is equally as important for safe power load as the gauge of the wire, and USB-C is simply a damn good connector.
When Jay gave that look in the beginning I lost it. Rolling on the floor laughing. Proof in the pudding
great, now we gonna see 2x12 pins connectors. Every day that passes i got more surprised of how smart EVGA is. Imagine developing something what uses the new and unusual USB type X connector, *and use 2 of them* to work...
I wondered about these connectors back when I first saw them. I'm an auto technician with training in hybrids and electric cars and I see melted pins in plastic connectors quite often. Knowing what I know about the theories of connector design and the wattage the new 12 pin is supposed to carry, I was fairly sure they were on the limit of something that small.
Imagine running into a bent hs-can circuit from the factory 😂 or better yet, a bent hv wire assembly from the hv battery bwahahaha
@Luka Tim Pascal dropped with effort. 2000 series was a scam, 3000 was what 2000 should have been. They're scrambling and alientating >.>
Totally agree. I have some connector/wires experience and when I saw the connector I got a bit concerned. It makes me wonder just how close to the edge they were getting. When Jay put it near the case and showed how little area there was to bend I got really concerned because bending is terrible, not just for the connector but the wires in general, especially at these higher voltages.
Based on the CableMod email ( @5:10 ) and the PCI Sig slide ( @3:13 ) it looks like everyone from the sub-tier suppliers all the way up to Nvidia (and I wouldn't be surprised if some big-wigs at Nvidia were shown these issues) knew that it was a potentially massive issue. To have the nerve to send Jay that smarmy, Emoticon-ending email just reflects badly on their team, especially when they obviously had many concerns themselves internally.
Don't forget that the cables come pre-bent for packaging purposes
The current rating for a pin(s) is directly dependent on the amount of surface area engagement, if you reduce that surface area without reducing current they will heat up.
That first email feels like a "don't worry, more power through the same connector can't be an issue" and that's super concerning to me. I'm really happy I can avoid using it for a while with my EVGA 3080 ti. This connector really needs a couple more iterations before I'd feel comfortable with it. They should have just made a 16 pin like you said.
Don't give up Jay, this is not a joke, people can lose their homes and / or their lives using this crap. I am surprised that the FCC / FTC are not looking at it yet, maybe we need to talk to GN to push it further (they have experience in these matters).
Thanks for making people aware.
Agreed, this needs to be investigated further.
This maybe an isolated case but if it becomes an issue to several users, FCC/FTC will come in. This could be a fire hazard.
Most likely some people need to die first in a fire caused by this issue before these arrogant companies like Nvidia and Corsair will start putting SAFETY BEFORE PROFIT AND PR BULLSHIT... ☹☹☹
It’s funny how 1080 ti managed to come up every now and then on almost every channel. That card was such a legend.
got a used one for $450 after 2000 launched... sooooo good
nvidia should stick with GTX
@@DogeRider985 🤣
I'm still running one and have no problem with it
@@michaelingram9354 I have 1 too but upgraded to a 3080 now
This plastic should be the same material m2 ssd connectors are made of. I remember using my BGA station to experiment, and this plastic melts at about 250 degrees of Celsius that can withstand soldering lead free solder which melts at 217 degrees, so this is really a sturdy plastic. Those connections are more like welding, rather than power delivery.
Very fair and balanced content. Well done.
If you have legitimate concerns regaurding safety of power connectors, it's better to say now and help prevent a serious accident. You done good here Jay. keep up the good work that you do.
The more the 40 series is discussed, the more EVGA’s decision to part ways from NVIDIA makes sense
Yep, they saw this shitstorm coming 10 miles away and got out of the way. Good for them.
Thanks for pointing it out, must’ve listened to their internal engineers.
I believe they would use 2x 8pin plus some 4pin for data, and reduction cable for it eventually
The wall of EVGA cards behind Jay is perfect!
I guess if the new 12 pin adapter is mandated from Nvidia and evga engineers calculated that it will melt the. I guess them opting out is valid since they said something along the lines of "nvidias requirements became impossible to fulfill" iirc
As someone who hasn't been in the loop about PC components for a minutes it's a good thing I started looking.
Curious to know if this will also have a chance of happening with an actual pci5 powersupply that uses 1 cable
You're not fear mongering at all, even if that particular burnt connector, turns out to be some strange corner case. The 12VHPWR connection is new, and is essentially being beta tested right now. We're all still learning about it. So, I hope you continue to bring us updates, as the latest information about the connection unfolds.
@Optaunix lay off the weed bruh.
The card was vertical with the cable only bending sideways due to it own weight and route of the cable. It was not over flexed at all. Cable mod attached a picture showing the cable must only bend downward and at the 35mm distance from the black connector. If that's makes sense. Over clocking hardware or whatever his name is did a vid with more info and pics.
@@Ryan-re1rs so what do u think cause this to happened ?
We shouldn't be learning on customers. This is not some fringe knowledge. Dangers of passing high current through cables is well known. The science is there, they just chose to ignore it.
The connector is 12V, right? The card is rated at 450W? 450/12 and you have 37.5A of current, at 600W it's 50A. It's very hard to believe they didn't realize that before final product.
@@JorgeForge I agree with you wholeheartedly. Just to be clear though, I only mentioned that they *are* beta testing it on us, not that I agree with it or think that it's okay, by any means. It is, what it is, though.
Hello,
I've worked with plenty of electronics, including plenty of high power systems. I agree with you that it is absolutely a contact issue. Despite the smaller connector, current should still not be a problem, cables of that thickness can handle well over 600 watts. However, poor contact as a result of stress can absolutely cause overheating.
So if i buy MSI 4090 the cable MUST be as straight as possible?
It's most likely the bending of the cables near the connector. In your test, you pre-bent it which is different to what would happen normally.
Realistically, the force is constantly applied onto the cable by the side panel, which may cause the pins to break contact or be misaligned, and it might bend the pins. These factors could cause higher current draw at the misaligned pin, or cause electrical arcs to bridge across the pins, and or depending on the internals, might cause another pin to draw more current.
This would most likely be the culprit I'd reckon.
thanks Jay, lovely video. saves tons of headaches. Maybe EVGA saw that incoming and preffered to just microwave some popcorns while seeing it on real time?
I laughed out loud at the Intel comment about them helping design the new cable and not using it themselves to burn up the competition
Same here dude. Same time and laughed loudly.
Intel be like: who's running hot now, bitch?!
@@Technatorium Also laughed loudly
That's something Intel would totally do, design a crap connector and not use it, not the most ethical company
You’re 100% right, that is a connection failure issue. If that was a too much current/heat problem, that card would have melted down and suffered a way worse “catastrophic” failure.
Which, according to sources for channels like Moore's Law Is Dead, was the case when they were testing a 40-series Titan prototype in the Nvidia labs, going with a 600W+ firmware profile. On some occasions, the connector would melt, maybe ignite. On others, the card itself would catch fire. Needless to say, there won't be a Titan GPU this generation for the foreseeable future.
I should add that it's why the 4090 is limited to 450W right out of the box. One could experiment with overclocking by making sure that all 8-pin slots are occupied and feeding power so that they have the 33% voltage headroom.
@@SelecaoOfMidas It's honestly pathetic that GPU manufacturers push power limits to boost performence...
PCIe spec should have power consumption standards like 5, 15, 30, 70, 150W and so on that would need to be followed.
This would force companies to truly innovate their designs, instead of just nearly frying every card and home breaker just to squeeze some FPS...
@@shapelessed I don't think we should've ever exceeded 200-250 watts at most. I was fine with 150 watts really.
Thanks for sharing this Jay. Now I'm for sure going to wait to get a card until they have this issue solved.
13:20 this opened my eyes now, you said that very well Jay
nvidia trying to beat NZXT and Gigabyte for the title of "how many PC's can we catch fire before someone notices"
You Forgot Asus Motherboards
@@danielkn196They at least identified the error
@@Sonyoucef 🤝
Looking forward to the GamersNexus montage
It's not just Nvidia's fault, even if it were nice.
Intel and PCIe speks are the core problem.
Nvidia has reported that there are problems.
I guarantee this was caught by numerous engineers throughout testing. Usually this kind of issue is because management overrides their advice.
engineer: “No. No. No. NO! For the love of all that is holy, NO!”
management, upchain: “My engineers say maybe.”
Getting RROD vibes. Engineers told them too.
Quality testers and safety testers: Sir this is a fire hazard we should change it.
Higher ups: Hmmm that'll cost us money and we love money so no,send it like this and let's pray to the divinities above we can blame the clients.
I mean Jay literally showed a screenshot of them (nvidia engineers) reporting the issue
@@SHAdow98V “Will it cost us more to resolve customer complaints, or to prevent the issue?” I dunno, if we start a fire and someone dies that’s really bad. “So it will cost more to prevent the issue, ship it.” 😐
Thanks for the PCA Jay.. makes me have a bit of buyers remorse about the FE 4090 showing up tomorrow.. I did contact Seasonic to get their cable made for the 4090s, hope it is built better somehow.
I would think they had to see some of this during testing. Easy solution would be to make the plug reinforcement a bit more rigid and include a 90-degree angle connector on the adapter coming from power supply cables. Then no need to bend the cable just to clear side panels. The connector and pins seem a bit fragile considering how much power they can pull.
Damn, good job EVGA seeing the problem, being told not to worry about it, and telling Nvidia that enough is enough.
I was thinking the same thing. EVGA abandoned ship at the perfect time. I truly hope that they will join up with AMD. Might not be this gen, but maybe next gen. I know they said they don’t plan to do so. But things can change. I hope to see them selling GPUs again.
@BainWrangler Might wanna get that RMA’d. I’ve only ever used EVGA PSUs and I’ve never had any issues. Could be a bad unit you have. They have awesome customer service and most of their PSUs have a 10 year warranty so I would give them a call.
@BainWrangler I have always used EVGA P2/G3 PSU's 1k+ watt units and never had a problem at all with coil whine at all its incredibly quiet and well within your warranty to RMA it.
@BainWrangler Is coil whine the same as cap whine? Hot glue is often used to tame coil whine.
I think it's happened to what? Two people so far? The sky is falling...
I've been an Aircraft electrician/avionics technician for 10+ years and can safely say that this makes perfect sense. Stress on pin type connectors that compromises pin surface contact can absolutely cause overheating. Scary stuff.
If you really are then you would know that any connector that's not a cannon plug has the same risk.. Do what's so scary?
@@ChuckTownRC51 you must work for nvidia
@@ChuckTownRC51 lol cannon plugs have the same risk. If you really are one then you should agree that Nvidia introducing a product that makes this issue more eminent is a bad thing. The card is too big for that wimpy connector/harness combo.
IMO they would have been better off with two 1/4 inch bolt on terminals and eyelet connectors over 10awg wire to the old pcie connectors......
@@ChuckTownRC51 you can 100% fail to put a cannon plug on properly, typically a large plug with lots of keyways to line up. I saw it happen, caused IFE and immediate impound of a airplane. This was a military jet not civillian. Point is, unless you make sure its seated properly any plug config can do this. Ever see burned pins on a cannon plug, I have, bad seating resulted in heat concentration. Cannon plugs just dont typically melt.
that type of issue is caused by a not correctly seated connector, all the current is flowing through a smaller area rather than the complete contact surface causing it to heat up dramatically at the point of contact. i've seen a similar issue on a car once, a blade fuse wiggled loose but still made enough contact for the lights that it was the fuse for to work, as not all the blade was making contact the connector the points of contact became incredibly hot burning the plastic of the fuse box it was in, replaced the fuse and made sure it was seated correctly and no more issues, customer didn't want to replace the fuse box for a relatively small area of damage (can't blame them it was an expensive part which would require a fair bit of labour to replace) so we cleaned it up as best we could. think the filament of a traditional light bulb, it gets bright due to it being so thin, if the filament was thicker it would not get as bright or would require more juice to get bright, now wrap the filament in plastic and expose to air, hey presto, fire. how the contact on the GPU became an issue is anyone's guess though.
"It all just works man"
would love this to be on a t-shirt merch soon.
Waiting for Jay's next build with built in fire suppression. It should make R2D2 sounds as it puts out any fire.
He'll be taking water cooling to a whole other level!
Sounds sort of fun.
Trouble is you do not wanna share a room with something putting out a computer fire. Not because there is like water and risk of shock, but for a fire suppression to have any chance of fitting in a computer build it will have to be a Synthetic Gas Fire Suppression. That is a system that stores a liquid, which it turns into gas and puts out fire by replacing the oxygen effectively chocking the fire and hopefully saving the gear. Trouble is of course us humans also like oxygen.
i don't see why they have to use such tiny pins for such a chunky power draw, they are insane :D
Frankly the smaller connectors make the product seem a little pathetic. It's like drip-feeding a pot of soup with water. My card has 3 full 8-pin connectors and it looks like a turbocharged faucet because of it.
if it wants 600 watts, it might as well plug into the mains, and regulate its own power on the card, i wouldn't be too fussed running 2 power leads to the back of my pc
I'm getting really tired of big companies that think they can just brush off safety concerns.
And REALLY happy that CZcamsrs like you and Gamers Nexus are calling them out WHEN they do fk up like that.
Information sharing is critical in IT. GREAT to know!! Merci
Knowing most users have to bend the cable somehow anyway, the "easy fix" would be a redesign of the adaptor cable so the metal connections are stuck in a 90 degrees plastic housing. It would eliminates the "first bend" issue, make sure the pins cannot be pulled off the housing and the product would look better anyway.
A better solution is a recall and a redesign of the cards themselves to make a new standard with bigger pins or use normal 8 pins connectors. But they will not do it, since it is way too expensive. I would have many more questions on that connector testing... I hope some regulations apply here.
Or just buy a 3.0 psu
They could simply offer a free change of the connector. Like people ship their graphics cards back to NVIDIA, and other manufacturers, and then they take off the current connectors, and change it into something else.
If this isn't possible, then they can simply redesign the PCB while keeping the rest of the parts the same, and just charge people $20 or something, or do it for free and take the L
If Nvidia wasn't being cheap, they could have made the cable longer and even make the connector a right-angle connector to avoid this.
It's not the cable length, but the bend required right at the plug due to case side panel distances. Agreed on a right angle connector, and sounds like Cable a Mod is working on one.
I'm not sure why right angle GPU connectors aren't the norm. Oh wait, right, money. Fucking companies. Who the fuck doesn't bend their GPU cable?
the problem isn't the adapter, it's the card. the card shouldn't need a right angle adapter, it should be built in such a way that the cable can plug in and easily accommodate 35 mm of space without bending afterwards, ideally naturally extending 35 mm without bending. that could easily be done. a connector near the front of the card, perpendicular to the board and facing backwards would allow the power cable to run parallel to the card as it came out. clips that held the cable 4cm from the connector would be sure a good fit.
@@edwardallenthree Nah, you don't design the card around the adapter its the other way around. All they had to do was design a better adapter. Much simpler, much cheaper.
charge you 2k+ for a card and skimp on the adapter ffs. I ain't buying this shit.
this is why i listen to JAYZ
AMD's 7000 series RND3 will NOT have this power connector.
Based on the gauge of the wire, the theoretical maximum before failure is around 57 amps. The connector is currently specced for 50 amps at 12 volts (volts x amps = watts), so there's already not a lot of margin for error, just over 1 amp per voltage carrying pin. If there's some kind of failure in the connection integrity inside the connector, on only a few pins, the resistance of those pins will rise high enough to force more power through the pins that are connected properly, and given the 1 amp worth of margin of error, you're basically guaranteed to exceed the power carrying capacity of the pins. If you manage to do this across ALL the power carrying pins, you're effectively turning your power connector into a resistive heating element, and this is the result. I was saying this would happen back when the 30 series launched, it was already an issue back then, the power draw just wasn't high enough yet to see these kinds of outcomes. This connector would work IF it was only replacing 2 8 pins, and even then I'd be concerned about the safety profile present. As buildzoid noted, there's a MASSIVE safety margin on the old 8 pin connector, and it's there for a reason as you can see. Normally for anything this critical, you want a safety margin as high as 2 (essentially the connector can handle double what it's rated for), to prevent just this kind of issue. The new connector has a safety factor closer to 1.1, which is fucking nothing. This is irresponsible at best, whoever thought this was a good idea needs to be fired and go back to college, this is covered not even just in EE 101, but we also went over all of this stuff when I took my A+ certification, which is a 1 quarter certification at a community college (or you can study online and just take the cert when you're ready). Literally the most irresponsible thing I've ever seen, and the writing was on the wall the second they announced it on 30 series
We all know an engineer didn't make this decision.
Sounds like Nvidia, Intel, and the PCIE spec group has designed somthing that is GUARANTEED to fail. Cause in Real World application your likely to end up with at least a few connectors not 100% in, with how big the card is and how tight the connector bend would be required to be. So this is a design failure from step 1 it sounds like. And Ill admit Im NOT a engineer, but having THAT tight a margin for error is TOO fucking risky
That large margin is really important once you start thinking about things like insertion/removal cycles, surface oxidation, etc etc. I wouldn't feel comfortable using a connector like that again if it started to melt without being able to see that the metal surface is fully intact and not oxidized; it's extremely common for the same issue to happen again on a connector that already experienced it because it's likely operating at a deficit afterward just due to seeing the high temperature and off-gassing from overheated plastic.
@@big0bad0brad kinda like a good old light switch or power socket after a while. For EU 230 Volts it sometimes is enough to put 10-12 Amps through'em, much bigger and much farer displaced connectors...
I think there is some missing information here. While NVidea spec'd at 9.5 amps the MicroFit line has a standard terminal rated at 8.5 amps and a RMF line rated at 10.5 amps. Interesting from the video was the mention of 40 mating cycles. The 8.5 amp line is rated at 40 mating cycles while the 10.5 amp RMF line is rated at 25 mating cycles. There is an option for lubrication to the terminals to extend the 8.5 amp line to 250 mating cycles. But somewhere the data provided just doesn't match. If they used the 8.5 amp line then thats far less amps then the publication of 9.5 amps and if they used the RMF line then they downplayed the 10.5 amps but the mating cycles doesn't match. All that assuming 18 gauge terminals.
I love how there were so many people voicing concerns and no one bothered to do extra testing to be sure that it handles being moved around a bunch. Like you know you'd do with cable management 🤦
Just replace burned connectors and keep testing :)
The worst part is they 100% did, as they are rated for a very short number of insertions. They were tested for this shit, were shown to be completely unacceptable by any rational standard, and were shipped anyway.
they not even bother to extra testing on vaccine (which everyone will using) why bother extra testing on some pc part? LOL
the user will be the beta tester lol
@@drackar Don't confuse this with the disconnect cycle story that got blown out of proportion. That's the standard rating from Molex which is exactly the same as all the other PCIE connectors in the same family that we're used to. The bending concern is completely separate and could just as easily happen on the first connection.
Not really, we are going from having an over specced cable with plenty of headroom, to having an under specced cable with no headroom at all. That means where external factors like strain on the connector are concerned, losing 50% of your current carrying capacity before was fine as the connector was still in spec, whereas now losing 50% of your current carrying capacity can result in melted connectors.
a good reason to wait for the next gen when they roll back to the older connectors
🤣 that email got me.
Fantastic. Can't wait for the thin cable adapters from Amazon. Fire hazard.
There's nothing better than cheap under gaged, CCA wire adapters from Amazon!
@@cup_and_cone i like my cheap fire hazard cables directly from nvidia
I bought some Xbox360 power adapter for a weird project I had to reuse an old Power Supply so I cut the adapter to solder my own instead of cutting the Xbox cable and GOD!!! that was weak, looked like 20 awg wires to push north of 200 watts, was affraid to use it. The new 12 pin does trouble me as much... even more if used with any type of extension cables...
@@alexandreapril5694 how did you even compare "Xbox360 power adapter" to a legit power supply..
@@0Blueaura it was to power LED lights, not a computer. For your information I had the same strange feeling when connecting my Cablemods direct fit cables instead of the original Corsair on my RM850X...
Someone needs to come out with a rigid 90 connector that allows for the cables to re routed in a non stressed manner.
Hard to believe this isn't already a thing. Seems like a very simple solution and product to make.
As someone old enough to have used the 4 pin ketchup and mustard cable Molex connector typically used on older mechanical hard drives I can testify that the pins inside of thsse things move around so much that you can't even plug it in without fidgeting with the cables leading into it. There's also the possibility that some sketchy Chinese supplier didn't supply what they claimed they were supplying, whether that be metals/plastics or the physical size.
@@Mr.Morden it's sad Nvidia didn't just make their connector like cable mods. And instead are beta testing on their top of the line consumers
This is the solution in my head. Nvidia should just supply every owner with various 9 degree connectors. Maybe 1 going down and 1 to the right so they're not torqued. Call it a day.
Whelp statement answered. They come out Oct 31st. The AIB partners should be sending these out for the price they charge. Hell stick the ROG logo on it and you should get 40 bucks
At this point I'm left wondering why, since they still added extra pins for onboard diagnostics, they didn't bother putting bar styled contacts, like you see on server power supplies. At that point you still have to make a locking mechanism to secure contact but, even if it isn't perfect, you still have plenty of surface area to have current flow through.
Is it maybe they foresee a case of multiple 12V rails feeding a single device? I don't know
Jay has been killing it lately
I can't imagine paying $1600 only for it to fail catastrophically, now he's got to try and get it warrantied and hope the next one he gets doesn't fail too.
No canted cables in future builds 😂 people will be building structural skeletons to keep cable straight lol
he said multiple times that the card works
@@Dreyden12274 Would you trust it not cause a fire? I wouldn’t. So yeah maybe the card works but is it usable and trustworthy? @Sizukun
@@Dreyden12274 Doesn't matter if the card works. With that degree of damage there's no way I would want to continue to use that card if I could get it replaced under warranty. IF being the operative word as I wouldn't be surprised if the GPU company comes back and says it isn't covered as the issue was user created.
@@Dreyden12274 Ya but we don't know if it'll continue to work or catch fire because some of those pins look like they have some plastic melted onto them potentially reducing contact and causing this problem again
Well, I’m sure many seen the Split in the road coming, but thank you for holding the line. Many of the enthusiast, like myself, rely on the right information when it comes to our builds. Keep up the good work and never switch lanes. 🤙
Nvidia truly messed up big time with this connector plug. Not only is it in a horrible spot. Nearly every common PC case can’t even fit this card in horizontally, but now we are having a major potential of a possible recall on every 4090 sold…… lol. EVGA bailed at the right moment. Nvidia is turning into a joke.
This is your reminder that when someone from corporate says "it all just works," it absolutely does not "just work" and an engineer somewhere is sitting there saying "I told you so."
If a pin gets hot enough to melt the housing there's a good chance the metal has changed either in temper, or plating integrity. That can cause more resistive snowballing.
Maybe bad crimping can cause it too? Still think this plug is way too small to carry up to 600w. Hell, the 8-pin cables aren't optimal, but at least they spread power a lot more...
@@GrimpakTheMook it would have to be a really bad crimp, like very little contact which should never make it past their deficiency process. Like borderline breaking off at anytime bad crimp.
Those pins aren't tempered and how they're affected depends on exactly how hot they got and the plating used. Assuming the housing is ABS and the plating is tin, you have around 30 degrees of leeway. Not a lot, but the plastic will become soft and look "melted" way before that while the plating will not budge.
Is this why EVGA dropped out of Nvidia gpus business? I heard they burned gpus before.
@@fynkozari9271 it's not just this.
It all just works. Sounds like something Todd Howard said about his games. But video games can't catch on fire. This is a serious fire hazard. 🔥
16 times the wattage
A few phrases to run away from: "We know what to expect" "Trust me, I know what I'm doing", and "It all just works."
See that fire, you can touch it.
The fire is a "feature!"
The heat generated in wires or connectors is current (Amps) squared multiplied by resistance (Ohms). P(loss)=I²R. 600W going through the connector instead of 300W means 4x the temperature rise in an area close to a heatsink that could be above the ambient spec of the connector. It's not a problem in "ideal" conditions until practical realities are applied. Would be better to plan for more common worst case scenarios. AMD may have already decided to delay using the 12pin connector.
Go Get em Jay, I Might Buy some merch. to help you out over this insightfull review
.
6:33
"The connector is fine as long as the companies card is not mounted in a case. We've tested it under normal use circumstances."
-Nvidia QA, interviewed in the Nvidia test lab, next to the gpu on a test rig.
If this does start causing actual fires, there could be massive lawsuits potentially involving nearly every major player in the industry -NVidia, AMD, Intel (as a developer of the connector,) board partners, and PSU manufacturers could very easily find themselves in a legal and PR nightmare
Kudos to EVGA for running away from this cluster as quickly as they could.
@@andrewshaw9840 They're in the PSU market still so they can still be effected
ok I'll mention this first I work for cable assemblies company, we build every thing to IPC class 3 for those of you who understand. looking at what i could find of the amphenol spec for the connector the spec of the wire gauge and the connectors looks ok. but if you bend the wire at the back of the crimp with out any strain relive that is what happens (on any connector) it's just at such a high power rating your tolerance for the damage that can be caused doing this is a lot smaller. they should not expect the end user to treat this cable any differently than any other cable in their systems really and current atx case sizes really don't help for the bend radius needed for this cable.
Search for "Gardner Bender Ox-Gard OX-100". Apply a small amount to the inside of the connector and hopefully those "bad or intermittent connections" will be resolved.
This is EXACTLY the kind of thing that starts small and slowly becomes a crisis; but it won't happen for most people during the first few months or even years. With even a nominal horizontal force on the plug, the thin metal parts in these connectors which directly contact the pins will fatigue over time and split open at the seams. It's like the Xbox 360 "ring of death" situation where they gradually self-destructed. NVIDIA needs to get ahead of this and send people proper adapters with robust, possibly angled connectors and not to just be wrapping them in thick tape. EDIT: If you own one, make sure to watch Buildzoid's video "The 12VHPWR connector sucks" on his channel Actually Hardcore Overclocking.
Yes, this!
On every contact interface between metals, corrosion will gradually increase which will increase the resistance which will increase the heat which will speed up corrosion, and unless that is taken into account with overspec, it will start to melt and burn.
Yes, the wrapping is a really bad idea. It only increases the leverage of the cable on the connector. If they keep the connector, they will likely have to supply a selection of angled connectors with highly flexible cables to reduce forces on the connector.
Yeah but the ring of death won’t burn your house ….
@@Echristoffe Oh, I just meant how it's the kind of thing that will creep up over time like with the first 360s. I edited it to make it more clear. But yeah, this could get bad if people out there aren't aware of what's going on. Hopefully the word has gotten out.
@@ThisSteveGuy seriously after seeing the second video… I don’t know … maybe a bad batch ? Maybe some weak female connector ?
It feel weird that they have pushed a standard which is only used by Nvidia … AMD and intel are not using it for their latest gen …
I'm disappointed that this video wasn't sponsored by iFixit. The explosion laden ad would have been fitting! 🤣
Heat from the card,, the current in the wire, and the temp inside the case combined it something too consider for the cause.
I like how the Kingpin card has the plug on the rear (front?) of the card.) Why this isn't the standard for all cards is beyond me. Makes for much better cable management. Server cards are like this.
@JayzTwoCents The only solution I see is one we used in Car Stereo High Amp bends...a 90 degree block that relieves the strain and positions the cable more favorably. you can even connect 2 90s together and have a 180 so your wires are hidden. bending cables at the point of termination is always bad. Good video!
Yes and no: every single connection adds some impedance. Therefore the more you do, the more you add. Why every normal electrician will never say it's OK to use 15 prolongation power filters in a row even if you only connect to the last one and within the specced current?
@@andyshtroymish4997 yes just like 90 degree turns of cable add resistance to electrical flow. This is true in electrical theory, not as noticeable in these applications. Car stereo Amps pull so much power that given the quality of wire and size, it is not an issue. What is though is having a 00gauge wire pull out of a distro block and ground out lol. The same kind of 90 degree blocks used on the 24 pin and EPS as well as 8pin PCIe connectors would suffice. Lesser of two evils
The pins look the same size as normal board headers, those are usually rated for around 1 amp. 600W / 12 / 6 (for 6 12v connections) is 8.333A. I wonder what kind of testing they did, I hope it wasn't "Trust when the connector manufacturer told us 'don't worry about it, it will be fine!'"
standard current rating is 8 amps continuous per millimeter squared (13 peak) This connectors are so far past safe it's ridiculous
As a semi layman, I’d love a link to the connector you use, I’m worried about using my default one now. Some better version of a cable is cheap insurance against a fire. Thanks for the awesome video Jay.
Woah, great video mate!