from the https://www.youtube.com/watch?v=rY0WxgSXdEE dept.
Another GeForce RTX 4090 16-pin Adapter Bites The Dust (Updated):
Update 10/26/2022 6:15 pm PT
Gigabyte has sent Redditor reggie gakil a replacement for his GeForce RTX 4090 Gaming OC. Next, the damaged graphics card and melted power adapter will make their way to Nvidia for inspection.
Meanwhile, a fifth GeForce RTX 4090 owner has come forward on Reddit to report a similar meltdown on his power adapter. He accidentally discovered it when disconnecting his Asus TUF Gaming GeForce RTX 4090 OC Edition to install his newly acquired Core i9-13900K Raptor Lake processor.
The Redditor's photograph showed that some pins had experienced excessive heat. The corner terminals look slightly backed out, whereas one of the ports has already started to melt. The user shared another photograph of his system, and the power adapter appears to be seated properly, and the cable isn't overly bent, which discards a bad installation.
Original Article:
[...] Previously, two GeForce RTX 4090 owners have reported experiencing 16-pin power adapter meltdowns. Both owned custom models, with the first owner rocking a Gigabyte GeForce RTX 4090 Gaming OC and the second user with an Asus TUF Gaming GeForce RTX 4090 OC Edition. The third report comes from a Facebook user Charlie Woods (then tweeted via WCCF's Hassan Mujtaba), who coincidentally has an Asus TUF Gaming GeForce RTX 4090. According to the owner's recount of the facts, he was benchmarking when he smelled the smoke from the power adapter. Fortunately, he could pull the connector out before it did any damage to the graphics card.
Nvidia launched an investigation into the matter when the first case surfaced, and with good reason, since meltdowns are starting to become widespread. The user feedback shows that the problem may reside with the design of the 12VHPWR power connector. Bending the cables too close to the connector seemingly causes some terminals to loosen up, leading to uneven mating. In addition, it unbalances the load across the other terminals. PCI-SIG documented the potential thermal variance issue long before the 12VHPWR power connector debuted on the GeForce RTX 4090. Therefore, it certainly comes as a shock that the problem is still present in the finished product.
[...] PCI-SIG's tests revealed that the power connector was overheating at the mating point. The problem affected both rows of pins. PCI-SIG performed the tests at an ambient temperature of 26 degrees Celsius and took readings from the hot spots at 2.5 hours. The end of the cable connected to the power supply showed readings between 51.3 to 52, whereas the end to the graphics card peaked at 150.2. PCI-SIG observed melting between 10 to 30 hours. The unbalanced current resulted from the resistance variation between the different pins. The standards body noted that bending led to high resistance in the other pins, causing the current to transfer to the lowest resistance.
When excessively bending the cable, the loss of mating contact can happen to 6-pin or 8-pin PCIe power connectors. It's nothing new. However, the problem seems more prevalent with the 12VHPWR power connector, so users should pay extra attention to the installation. Custom cable manufacturers, such as CableMod, recommend a minimum distance of 35mm from the connector before bending the cable. The company also sells a 90-degree power adapter to mitigate the cable bending problem.
[...] GeForce RTX 4090 owners should probably recheck their graphics card's installation and avoid bending the cable for the 16-pin power adapter where possible. Hopefully, Nvidia will get to the bottom of the issue quickly and provide consumers with a solution if needed.
(Score: 4, Informative) by drussell on Friday October 28 2022, @02:16PM (1 child)
No, it doesn't... At least not here, anyway...
(Score: 3, Informative) by martyb on Friday October 28 2022, @03:05PM
Wit is intellect, dancing.
(Score: 1) by Runaway1956 on Friday October 28 2022, @02:32PM (4 children)
That's rather funny. Unless you start out purchasing a huge mega-cabinet, you have little choice in whether you're going to bend a cable, or oftentimes, where. The 780 only fit like *THIS* and the 2070 only fits like *THAT*. There isn't a lot of room to play in, even in a full sized tower. I can't imagine trying to stuff a gaming card into anything smaller. The older 4-pin and 8-pin cables aren't easy to route, and a 12 pin is going to be bloody hell. To avoid bends where you don't want them, the cable is going to have to be equipped with better connectors, sold with a 90 degree bend in just the right place, or two 45 degree bends in just the right place, as the case may be.
I haven't even mentioned the fact that opening a case after several days, or a few weeks, may reveal that the cable has taken on a bend that you didn't intentionally create!
Abortion is the number one killed of children in the United States.
(Score: 2) by looorg on Friday October 28 2022, @05:39PM (2 children)
Can't the electrons flow past sharp corners?
(Score: 2, Informative) by Runaway1956 on Friday October 28 2022, @07:50PM
The problem isn't the electron flow, so much as the wire being deformed, and changing the resistance of the wire. You never see house wiring, or industrial wiring being bent sharply. In general, wire is pulled through conduit to protect the wire from being bent like that. Worse, in this case, is the fact that the bend is being made in the immediate vicinity of the plug, and the solder joints at the plug are apparently being damaged. Someone posted earlier that the plugs in question were shoddily built, and they won't tolerate any strain. They may even melt down if no strain is ever put on them.
But, yeah, the electrons are going to flow, until they can no longer find a path to ground. And, that is precisely the problem. Increased resistance causes heating, which causes the entire plug to melt down, possibly catching fire, and burning up all the expensive gear in the vicinity - if not your home.
Abortion is the number one killed of children in the United States.
(Score: 0) by Anonymous Coward on Friday October 28 2022, @10:56PM
Many do, but due to the magnetic field that results from angular acceleration of the electrons, some of them are deflected and become scattered beta particles. Of course they then can raise havoc with the electronics, and also cause micro welds, causing some things to be infuriatingly stuck together.
(Score: 3, Touché) by driverless on Saturday October 29 2022, @07:48AM
"You're bending it wrong".
(Score: 5, Informative) by EvilSS on Friday October 28 2022, @02:39PM
(Score: 2) by Freeman on Friday October 28 2022, @03:33PM
Sometimes, being on the cutting edge. Really hurts.
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2) by fraxinus-tree on Friday October 28 2022, @04:27PM (6 children)
Sorry, people, 12V is tooo 1960-y. Earlier cars were 6V and 6V sucked big time when it came to crank the engine by electricity. 24V or 36V looks like a good temporary solution if the power draw of these devices keeps increasing.
(Score: 4, Insightful) by RS3 on Saturday October 29 2022, @12:30AM (5 children)
Maybe you or someone else knows more about this, but AFAIK, the 12V goes into switching regulators that drop it to 5, 3.3, 2.5, 1.6 (or whatever VCore is), and lower voltages. The higher the incoming voltage is, the more you have to drop. Which would be okay, except switching regulators aren't perfect, so you'll dissipate (lose) more power (watts) if you start with higher voltage. Inherent switching losses (charge storage time) would be worse due to greater volts that need to be switched. It's a good thought though.
Really, these currents aren't that big of a deal. I'm speculating, but from what little I've seen, it's more about inexperienced designers, who are okay with tiny currents in high-speed digital circuits, but not so good with power stuff. Maybe also not being checked by someone with better experience, esp. power / connector stuff.
But it's not just nVidia- look how many laptops fail because the power jack comes loose. Electrically okay, but mechanically someone is an idiot to think simple (and lead-free) solder is enough to hold something (connectors, switches, controls, etc.) that's getting tugged on over and over. And, they might be okay if someone would hand solder the connector and use a good fillet of solder, rather than relying on the scant solder from the solder paste / reflow oven process.
It'd be awesome if a laptop maker would put a little trap door so people could get in and resolder the connector without having to completely disassemble the entire laptop. Maybe I'll cut one next time...
(Score: 2) by driverless on Saturday October 29 2022, @07:51AM (1 child)
In particular if you're running 12V over dozens of meters and expecting to get 12V at the other end then you'll be in trouble, but in this case it's tens of cm and they're running it into buck converters that don't care much about the exact input voltage.
(Score: 2) by RS3 on Saturday October 29 2022, @08:24AM
I agree in part. Maybe you know more about it, but AFAIK, the greater the voltage (potential) difference between the on and off states, the more work the switching transistors have to do, and the more I^2 * R losses you have in the transistors. No? In fact, isn't that why CPUs use such low Vcore- to get faster clock speeds by reducing switching times and heat generated?
Oh, and maybe you don't know but many power supplies and voltage regulators have Vsense inputs- you run a separate pair of wires from the PS to the load and the PS compensates for wire resistive loss (up to a point of course).
(Score: 3, Informative) by fraxinus-tree on Saturday October 29 2022, @06:12PM (2 children)
I know very well what you are talking about. These buck converters are not switching-loss dominated in the first place. And if they were, one could go from buck converters to transformer-based converters.
Whatever you do with modern (e.g. post-1900, yes, 1900) power technology, you are in trouble if your amperes are more than your volts. Standards, compatibility, safety, inertia and other factors can pretty much force you into the unpleasant ampere-intensive area, but be prepared for things like in The Fancy Article.
p.s. I am European. I was always impressed by the gross inefficiency of the USA consumer 120V power grid. And I am even more impressed by 12V-wired US trucks and buses.
(Score: 2) by RS3 on Saturday October 29 2022, @09:15PM (1 child)
I don't quite understand your meaning, but I'll take a guess and say: I = C * dV/dT. So, the greater the voltage change, the greater the I and therefor greater I^2 * R losses. Maybe some very new transistors could do the job with higher Vin and not generate a lot of waste heat, I don't know.
But then you have the problem of PC power supplies don't supply more than 12V, so you'd have to have new / different power supplies to power the super hungry nVidia cards. I'm sure it could be done, and maybe that will be the answer if this trend continues.
Technically correct, but when you design a product you have to consider many more things, like physical size and cost. A transformer which can handle the needed power would be fairly large and costly.
Frankly I don't think any of this is necessary, just get people to do good correct design and manufacturing. It ain't rocket-science, it's just cost-cutting and rushing-to-market gone too far, as we see everywhere.
Regarding our 120VAC, I've had this discussion before, here and on /., and I'm surprised at how long, far, and deep this misinformation has gone outside of the US. Time to spread the truth, okay?
1) I've been working in a food processing factory and almost all of our motors are 480VAC 3-phase. Even smallish (sub kilowatt ones). 480 3 phase is pretty common in industrial, some commercial equipment. At some point the big stuff gets much higher voltages, depending on how big. I don't have anything above 480.
2) Residential (houses) are _always_ fed by 240V "split-phase", meaning the utility transformer's 240VAC secondary is center-tapped, and the center tap is grounded (earthed). Larger appliances like my water well pump, dryer, stove, oven, hot water heater, etc., ARE 240V. Room outlets are generally 120V because it's much safer for humans. Circuit "feeders" are typically 15 or 20 amp circuit breakered. It is possible, and sometimes done, to have a 240V outlet for some things like quite large window air conditioners and a few other things, but it's quite normal in the US to have 240 where needed.
As long as I can remember, using higher voltages in cars has been talked about, and some are (finally) doing it. You need to consider the many factors that go into decisions. Without digging into history, cars became 12V probably 70 years ago. There's a huge legacy of electrical things in cars that are used from year to year and model to model. Many parts are even used by competing brands. My 2002 Chevy has some sensors that were also used in Toyotas, for example.
I'm in favor of higher voltages in cars, but again, in engineering (I'm an EE) there are many many factors to consider. Going from 12 to 24 or 48 might not seem like a lot, but better insulation will be required, everywhere in everything. Batteries will cost more. All electrical things will be quite different. All connectors should be changed so that no 12V things get accidentally plugged in to 24 or 36 or 48. I'm sure I'll think of more considerations.
(Score: 2) by fraxinus-tree on Sunday October 30 2022, @09:55AM
Really an EE? Then you know that whatever switching device (be it a transistor in a buck converter) loses energy not only by switching, but also by just conducting current. Depending on the use case, the pure resistive loss can dominate the switching loss and this is exactly what happens in the cpu-powering buck converters (specs like 0.9V / 100A are not uncommon there). A lot of them are actually transformer-based (an inductor and a transformer are quite similar on a motherboard) and not a pure buck.
In regard to the industrial power: The industry has a lot of options, 3-phase 660VAC here are quite common as well. I know very well that you can have 240V in residental wiring in the US and this is generally good. On the other hand, here I have 3-phase 230V (400V between phases) in my basement. It is 110V what is wrong. No, it is not safer - the high currents make all contacts a pain in the butt and the high current/low voltage combination makes breakers less reliable. It is "schuko" socket that is safer in regard to humans touching live wires. 110 also uses tons and tons of extra copper. Did I say copper? Half of it is Al with its own can of worms. We don't use Al indoors here.
In regard to cars: European trucks of ~5 ton payload and above and all busses are 24V. So is e.g. any military transport and machinery all over NATO, including cars. These were 24V from ~1960s or 1970s on. (Soviet military machinery is 24V as well.) So it is possible, isn't it? Batteries actually cost less. Cabling costs and weights less. Bulbs cost the same. Short circuit fires happen less. Headlights and taillights run brighter. Yes, you take care not to plug 12V-only device in a 24V car, but this is almost a non-issue because most car-powered gear (chargers, GPS, etc...) is universal 12/24 these days.
(Score: 2) by RamiK on Friday October 28 2022, @06:48PM
~ Jensen Huang musings on Dylan Thomas
compiling...