Proposed new Power Connector

[AusWolf]

One compact plug instead of two huge plugs is a progress and I like progress. If this compact plug used robust tabs instead of tiny pins, if would be great.

It's not progress if it burns up.

I do not follow the names either, I just call it Nvidia 12 pin.

But they are different standards as far as I know, so you have to follow the names.

 

[BoggledBeagle]

It's not progress if it burns up.

If you use damaged or badly made connector and push a lot of power through it.

But they are different standards as far as I know, so you have to follow the names.

The second version just does not tolerate being partially unplugged, the underlying tech is still the same.

 

[AusWolf]

If you use damaged or badly made connector and push a lot of power through it.

Have you assessed that the reports came from users of damaged or badly made connectors? What is a "badly made connector"?

I recommend this thread, and this post specifically:

Unofficial 12V-2x6V Power Connector Melts NVIDIA GeForce RTX 5090

They uses 110V/220V, that's different. 5090 runs on 12V 50A, and the working temperature >60C. So 1x 6AWG, or 2x 10AWG Give it some safe room and the solution could be just scale up the original 6 pin design with 10 AWG wires and connectors Or maybe we shouldn't have 600 W GPUs? I know...

www.techpowerup.com

The second version just does not tolerate being partially unplugged, the underlying tech is still the same.

I still can't be asked to follow up on it. Sorry. I'm happy with my 8-pins, leave me be. "If it's not broken, don't fix it" is my motto.

 

[DirtyDingusMcgee]

I'm actually all for solid connectors that you have to attach some type of pigtail to like a terminal block.

If you smoke it, you screwed up.

 

[Vayra86]

There is nobody who has really given me a real reason why we ever had to abandon pcie 8 pin.

The supposed reason was... eurhm... power increases? It 'didn't look nice' to have 4x 8 pin next to each other? (Makes you wonder what magic happens in people's heads when there are instead 3 of those plugs... which has been the case for years and still is)

There is no size constraint on GPUs to not just have a 4th plug on them. Heck, get creative, stack two on top of each other and side by side, you'll still not exceed GPU height and you'll be using what, 2cm width over 12x6?

Its a complete clusterfuck that gets us nowhere, except into buying new equipment, adapters, and applying workarounds to make sure said new equipment keeps working properly. Do anything wrong? User error!

Keep it Stupid Simple.

 

[FoulOnWhite]

A massive 4 slot cooler that is near the limit of length, and they use a daft flimsy little 12 pin connector.

 

[BoggledBeagle]

A massive 4 slot cooler that is near the limit of length, and they use a daft flimsy little 12 pin connector.

True, there is enough space on the large cards for four 8 pin connectors and in most PC cases enough space to run a forearm thick bundle of wires to the heater.

Apart from the tiny pins, manufacturers refusing to make high quality components and users refusing not to abuse the cables, everything would be fine and people could use one or two elegant small connectors to power their heater cards.

 

[FoulOnWhite]

True, there is enough space on the large cards for four 8 pin connectors and in most PC cases enough space to run a forearm thick bundle of wires to the heater.

Apart from the tiny pins, manufacturers refusing to make high quality components and users refusing not to abuse the cables, everything would be fine and people could use one or two elegant small connectors to power their heater cards.

What does it matter if there is a thick cable in your rig to power your near 1kw GPU, it doesn't. If you want your rig to look elegant inside, don't buy a near 1kw GPU, simple.

 

[3valatzy]

I am not sure that two 8-pin connectors won't be sufficient to withstand up to 1000W power draw?
Are there tests which actually have shown that there was a need for a new connector?

If so, there are two options:
1. Set an upper TDP limit of 250W - I'm looking at you greedy black leather
2. Design a new connector with very thick pins, in order to be able to carry that current.

 

[qxp]

I think we should stick to 375 W max on a single GPU. 75 W from the slot, plus 2x 150 W from two 8-pins is more than adequate. There's no need to reinvent the wheel.

375W is already a lot. An old-style soldering iron for electronics is 30 W. This is enough power to melt solder and plastic. If anything, the comfort you have in the trusty 150W 8-pin connector proves that high-power, high-current connectors and cables can work very well if properly engineered.

 

[R-T-B]

Interesting ideas but do some believe nv even pays attention to these ideas?

honestly, no sane company would. The internet is known to be full of people who present as engineers but should never be trusted to act as one.

 

[qxp]

I'm actually all for solid connectors that you have to attach some type of pigtail to like a terminal block.


View attachment 384468
If you smoke it, you screwed up.

Actually it is not so easy to achieve controlled resistance with this block, which is why a lot of modern connectors and switches use blades. The current thinking is that a good connection is formed when you have direct metal to metal contact. This is most easily achieved if you press one piece of metal into another, such as when crimping, or for many disconnects you have one piece of metal sliding past another and removing oxidation.

When you screw things together you just press surfaces and you can end up with higher resistance when using plain copper wires. This does not matter for low currents, but you don't want to do things this way for high current connectors.

I am not sure that two 8-pin connectors won't be sufficient to withstand up to 1000W power draw?
Are there tests which actually have shown that there was a need for a new connector?

If so, there are two options:
1. Set an upper TDP limit of 250W - I'm looking at you greedy black leather
2. Design a new connector with very thick pins, in order to be able to carry that current.

They probably thought that having to plug 3 8-pins is not really that reasonable and it would make sense to have just one plug.

What do you think of the connector requirements in the beginning of this thread ?

A connector could look like this (but with 4 pins in the middle):

https://www.mouser.com/ProductDetail/Amphenol-FCI/51720-10100401AALF?qs=iiM6sTJtPEeBo51viwF%2F3A%3D%3D

 

[taka]

This old design almost never failed.

Even with heavy oxidation it can still do 600Amps

 

[chrcoluk]

There is nobody who has really given me a real reason why we ever had to abandon pcie 8 pin.

The supposed reason was... eurhm... power increases? It 'didn't look nice' to have 4x 8 pin next to each other? (Makes you wonder what magic happens in people's heads when there are instead 3 of those plugs... which has been the case for years and still is)

There is no size constraint on GPUs to not just have a 4th plug on them. Heck, get creative, stack two on top of each other and side by side, you'll still not exceed GPU height and you'll be using what, 2cm width over 12x6?

Its a complete clusterfuck that gets us nowhere, except into buying new equipment, adapters, and applying workarounds to make sure said new equipment keeps working properly. Do anything wrong? User error!

Keep it Stupid Simple.

Business reasons I can think off.
Reduce PCB space consumed by power connector. Aka cost.
Kick backs from PSU manufacturers selling PSUs for the new standard.
Sensor pins, which in theory would make the whole thing safer, only allowing power draw based on sensed capacity, in reality it hasnt worked though.

 

[Dragam1337]

I think we should decide if this thread is about fixing the current problem and is aimed in the short future, or if it is a visionary piece about how things should look in 10 years.

In the first case the discussion about 48V makes no sense.

True - what you proposed could fairly quickly be implemented with current psu's.

 

[Assimilator]

This will require a complete PSU overhaul. Every existing PSU will go obsolete. We are not ready for this.

12V being too low is the whole reason this is a problem to begin with. The right way to fix the problem is to address it at the producer (PSU), not consumer (GPU etc.), because trying to do it at the latter ends up with the same nonsensical type of designs that we currently have with 12VHPWR.

ATX being stuck on 12V is a problem that should have been fixed years ago. Like all problems, the longer it is allowed to linger, the worse it will become. Get industry on board, figure out a plan, inform the public, ship new components, done. Procrastination because "muh obsolete" is not the answer.

 

[Chrispy_]

There's nothing wrong with 12VHPWR or 12V-6X2.
The problem with them isn't the connector design or the wiring gauge, it's the stupid power rating they've been given.

To keep the same safety margins as EPS 12V or PCIe 6, 6+2, or 8-pin connectors using the old MinFit Jr, a 12V-6X2 should be rated at about 250W at most. It's 6 pairs of 12V just like two 8-pin connectors is 6 pairs of 12V, but since the connectors are smaller and closer together, I've downrated it from 300W to 250W.

Simple, honest truth backed by the laws of physics.

a 600W 12V-6X2 connector is as silly as a 375W 8-pin PCIe connector; Yes, it'll work, but it'll probably get uncomfortably warm and sometimes a bit melty/burny.

My solution to this in the last round of discussions for 12VHPWR was just make a new connector with the same MiniFit Jr we've always been using and to stick to the tried-and-tested 75W per pair that PCIe uses in the worst-case scenario. That means that a 600W card is going to need a 16-pin MiniFit Jr connector with 8 12v pairs. Motherboards these days take 24-pins + 8-pin + 8-pin connectors, so why is a mere 16-pins a problem for a GPU that has equal (if not greater) power delivery requirements?

 

[DirtyDingusMcgee]

Actually it is not so easy to achieve controlled resistance with this block, which is why a lot of modern connectors and switches use blades. The current thinking is that a good connection is formed when you have direct metal to metal contact. This is most easily achieved if you press one piece of metal into another, such as when crimping, or for many disconnects you have one piece of metal sliding past another and removing oxidation.

When you screw things together you just press surfaces and you can end up with higher resistance when using plain copper wires. This does not matter for low currents, but you don't want to do things this way for high current connectors.



They probably thought that having to plug 3 8-pins is not really that reasonable and it would make sense to have just one plug.

What do you think of the connector requirements in the beginning of this thread ?

A connector could look like this (but with 4 pins in the middle):

https://www.mouser.com/ProductDetail/Amphenol-FCI/51720-10100401AALF?qs=iiM6sTJtPEeBo51viwF%2F3A%3D%3D

I was joking....

 

[Ruru]

I just don't understand that what's wrong with traditional PCIe cables. And no, I don't accept saving space from the PCB as an excuse. remember some Nvidia cards having a soldered connector which was moved away from the PCB itself? Those worked fine.

 

[qxp]

I just don't understand that what's wrong with traditional PCIe cables. And no, I don't accept saving space from the PCB as an excuse. remember some Nvidia cards having a soldered connector which was moved away from the PCB itself? Those worked fine.

They did work, which is good, but they were a kludge solution. You are providing +12V using 3 wires instead of one. Why ?? No reason. The right way is to up the voltage and use one wire for positive and one for ground.

 

[eidairaman1]

What truly needs to happen is the gpus need to stop drawing more and more current

 

[qxp]

What truly needs to happen is the gpus need to stop drawing more and more current

Well, progress usually means using more energy in more efficient ways. What would be interesting to see if people start modding houses to provide higher amperage and higher voltage outlets.

 

[Shrek]

Don't think we want to go higher than 240V and most houses in the US have such voltages available.

 

[qxp]

Don't think we want to go higher than 240V and most houses in the US have such voltages available.

Well, I'd love to have 3-phase, which most houses don't have. I agree that 480V is too dangerous. But all houses I have seen have at most 20A 110V wall outlets and only a few 240V circuits for appliances.

 

[eidairaman1]

I'm actually all for solid connectors that you have to attach some type of pigtail to like a terminal block.


View attachment 384468
If you smoke it, you screwed up.

Binding posts, used plenty of them at AT&T from 2013-2014...

Well, progress usually means using more energy in more efficient ways. What would be interesting to see if people start modding houses to provide higher amperage and higher voltage outlets.

Current isn't efficient as there is still a loss called heat, different current levels require different gauges of wire to keep a fire from happening because there is natural resistance in conductors which is impedance, would you suggest using 0 gauge aluminum wire just to power one of these fire hazard gpus?

Don't think we want to go higher than 240V and most houses in the US have such voltages available.

208, 230, 240

This old design almost never failed.
View attachment 384490

Even with heavy oxidation it can still do 600Amps

Toyota/Honda? The oxidation from the acid eventually causes conductivity to cease, seen it plenty of times.