Monday, July 3rd 2023

12VHPWR Connector Said to be Replaced by 12V-2x6 Connector

According to Igor's Lab, who has gotten their hands on a PCI-SIG draft engineer change notice, it looks like the not entirely uncontroversial 12VHPWR connector won't be long lived. The PCI-SIG is getting ready to replace it with the 12V-2x6 connector, which will be part of the ATX 3.1 spec and the PCI Express 6.0 spec. The new connector doesn't appear to have any major physical changes though, but there have been mechanical modifications, such as the sense pins having been recessed further back, to make sure a proper contact is made before higher power outputs can be requested by the GPU. The good news is that at least in the draft spec, the 12V-2x6 connector will be backwards compatible with 12VHPWR connectors.

One of the bigger changes, at least when it comes to how much power the new connector can deliver, is that there will be new 150 and 300 Watt modes in addition to the 450 and 600 Watt modes for the sense pin detection. The 12V-2x6 connector is rated for at least 9.2 Amps per pin and the new connectors will carry a H++ logo, with th older 12VHPWR connectors getting a H+ logo. The PCI-SIG has also added stricter requirements when it comes to the cable design and quality, which should hopefully prevent some of the issues the 12VHPWR implementations have suffered from. We should find out more details once the PCI-SIG has finalised the 12V-2x6 connector specification. In the meantime, you can hit up the source link for more technical drawings and details.
Source: Igor's Lab
Add your own comment

137 Comments on 12VHPWR Connector Said to be Replaced by 12V-2x6 Connector

#51
R-T-B
R0H1TOf course, would be far cheaper hiring us :cool:
Burn baby, burn.
Posted on Reply
#52
R0H1T
Season 2 coming up!
Posted on Reply
#53
STSMiner
ZoneDymopretty amazing stuff, how hard is it to develop a plug....hard apparently.

Im still in favor of an update to the PCI-E slot so it can carry all the power needed.
So you want to move all the power connectors to the motherboard to provide power to the GPU via the PCI-E slot or another slot (ASUS prototype springs to mind here), great idea, melted connectors on the motherboard instead of the GPU / PSU.

Motherboards would need to be able to handle that power delivery, thicker copper traces will be needed to handle the wattage, PCI-E slot would need an overhaul for that power delivery......

Some motherboard manufactures have started doing this with prototypes for dummies to beta test, they have even started moving the connectors to the back side of the motherboard (your going to need a special PC case here btw), is it a good idea ? Time will tell .... and if we start seeing melted connectors etc.
Posted on Reply
#54
Vayra86
TheLostSwedeThey changed how it's being done over the sense pins.


The "modular" 6+2-pin connectors has this problem every time...
I have to agree the 6+2 pin is annoying as hell. Except nine out of ten times you're just doing it wrong; +2 on the wrong end or upside down for example :)

But then after nine times you've figured that out... and you just can't get the tiny bugger in there :D I like to use a creditcard or something to push it down until its level with the 6..
Posted on Reply
#55
Ruru
S.T.A.R.S.
I'd like to be one of those who paid hella lot from a 4090, had that connector melting and reading that it was a total failure. Oh wait, I have a Radeon..
Posted on Reply
#56
Wolphin
TheDeeGee8-pin can do 340 watts.
I had seen 288W was the maximum, but the 50W difference isn't too big of a deal, just shows the connector is even safer. And since each device should only pull 150W per connector, the only time it gets anywhere near that, is when you use a splitter (Don't think there aren't many people using those these days...). Even then, the connector with the load is not right against the GPU, and tends not to have the cabling stresses, and can easily be made sure it is well seated to prevent the high resistance, which is overheating the connector.
Posted on Reply
#57
onemanhitsquad
Of no concern...dont have a problem with either of mine
Posted on Reply
#58
BorisDG
Same here. I saw few videos where people are able to wiggle a lot the connector. I tried that on mine Strix and it won't budge. It's solid hard. No melting, nothing.
Posted on Reply
#59
STSMiner
onemanhitsquadOf no concern...dont have a problem with either of mine
If the cable (the one that came with it) that's powering RTX 4090 FE has the ASTRON female pins in it you have been extremely lucky so far, seen some with melted connectors, not seen any with the NTK female pins in the provided power cable (12VHPWR end).

The cable that came with the RTX 3090 FE will have the NTK female pins in it (12VHPWR end, 1st gen connector without the sense pins), so you should be fine there.
Posted on Reply
#60
BorisDG
STSMinerIf the cable (the one that came with it) that's powering RTX 4090 FE has the ASTRON female pins in it you have been extremely lucky so far, seen some with melted connectors, not seen any with the NTK female pins in the provided power cable (12VHPWR end).

The cable that came with the RTX 3090 FE will have the NTK female pins in it (12VHPWR end, 1st gen connector without the sense pins), so you should be fine there.
How I can see if I'm with ASTRON or NTK without removing the cable? Is it written somewhere or some other outer sign?
Posted on Reply
#61
STSMiner
BorisDGHow I can see if I'm with ASTRON or NTK without removing the cable? Is it written somewhere or some other outer sign?
You will need to disconnect the cable from the GPU and physically look to see what type of pin is used.

See here

and here
Posted on Reply
#62
BorisDG
STSMinerYou will need to disconnect the cable from the GPU and physically look to see what type of pin is used.

See here

and here
Thanks! I see the picture says stricktly "FE". I'm with Strix tho.
Posted on Reply
#63
STSMiner
RTX 4090 cards from most of the Nvidia board partners had / used the wrong female pins in the connector.

They should be using the NTK female pins.
Posted on Reply
#64
R-T-B
STSMinerRTX 4090 cards from most of the Nvidia board partners had / used the wrong female pins in the connector.

They should be using the NTK female pins.
Astron and NTK aren't types of pins that are or aren't compatible, they are manufacturers of the same connector in slightly different styles at their own factory. They both are female pins though.

Astron does a worse job by all accounts and I wouldn't like getting them either, but the stats are still very low on burnup even with the plethora of Astrons out there. If it fits tight and doesn't move I'd not worry much.
Posted on Reply
#65
LabRat 891
R-T-BWhy do that when we have so many "aparently qualified" engineers telling us what's wrong in the comment section?
Having actual experienced professionals publicize their research and results allows cross-referencing with the "baseless opinions" of the "aparently qualified".
The egotistical ignore they're wrong, but the genuinely inquisitive learn something new; something that otherwise would've taken years of dedicated work/study (or more).

The 5600X3D discussion on TPU is an example, IMHO:
Some of the conjecture from 'armchair experts' proved out correct, or *very* close to what was revealed just before and with the announcement.

Just because someone is 'un(der) qualified' doesn't change how accurate/inaccurate their opinions and statements are. (vis-a-vis for 'the well-qualified expert' that history often proves wrong).
Posted on Reply
#66
STSMiner
R-T-BAstron and NTK aren't types of pins that are or aren't compatible, they are manufacturers of the same connector in slightly different styles at their own factory. They both are female pins though.

Astron does a worse job by all accounts and I wouldn't like getting them either, but the stats are still very low on burnup even with the plethora of Astrons out there. If it fits tight and doesn't move I'd not worry much.
Third party companies should also be taking note here, those like CableMod with their 12VHPWR cables and those right angled adaptors, and the new WireView PMD (see below) that uses these connectors, they are all using the ASTRON female pins for their devices, this includes PSU manufactures (be Quiet for example).



As you can see in the image below, these are using the ASTRON female pins (pins with dimples).



If anyone buys one of these, I hope they don't end up with the connectors melting.

UPDATE - They may have updated these connectors to use the NTK female pins on the WireView PMD, will have to wait and see when this device get's reviewed / into end users hands for more info.

Posted on Reply
#67
csendesmark
usinameThere is fundamental problem with the 12/16/2x6 pin connector
You gotta have fancy connectors hopefully with LED lights
Posted on Reply
#68
R-T-B
LabRat 891Having actual experienced professionals publicize their research and results allows cross-referencing with the "baseless opinions" of the "aparently qualified".
That's really not what I was responding to here.
Posted on Reply
#69
LabRat 891
R-T-BThat's really not what I was responding to here.
'kay.
I'll openly admit I have no idea what you were talking about, then. (I'd bet the person you quoted is probably clueless too, but that's between you 2)
Posted on Reply
#70
dicobalt
That's 662W at 9.2A per pin. Is that acceptable for transient loads while the power connector is heat soaked with ambient heat off the GPU cooler and also heated by the current? Did they improve the plastics to ensure heat tolerance? Is the clip designed to be tighter and not wiggle left and right?
Posted on Reply
#71
STSMiner
The connector casing that holds the female pin should not really be a problem with the heat for the most part, it's the type of female pin that's being used that's the problem, it has two seams (top and bottom), this makes it weak with it's tension, it's not going to be a snug fit with the male pin, the female pin will open up too much vs the NTK female pin that only has one seam along with it's four spring contact points within it, they have a bigger surface area as well.



Shoving 50 - 110.4 watts (volts x amps = watts) through one pin via those dimples is the problem - it's going to generate heat, lots of it if only three or two or one of those dimples are making contact with the male pin.
Now you have a 2nd issue, the power delivery via that pin is not enough, so the next pin is going to be supplying the the extra power, now this pin is getting hot because of the issue with the first pin, now duplicate that with the next pin and the next pin and so on because they are not making full contact with all the dimples on the male pin (small contact point / surface area), your going to end up with this mess below.



Can user error be blamed here ?
Sure, people make mistakes, but most will have pushed this connector in all the way and still have melted connectors, even with those 3rd party cables and those right angled adapters (CableMod).

So, now that leads to here, where are the melted connectors using the NTK female pins ?
If you have one, I'd love to see it !
Posted on Reply
#72
wheresmycar
TheLostSwede....such as the sense pins having been recessed further back, to make sure a proper contact is made before higher power outputs can be requested by the GPU.
i think its utter madness for the initial implementation not to have had it spot-on the first time. "to make proper contact" what an oversight!

I hope next Gen cards don't have the power sucker right in the middle. Aesthetically speaking, i preferred those x2/x3 8-pin hookups with sleeved cables pouring out like the Niagara Falls. That was art! The 12-pin looks like a lonely fart.
Posted on Reply
#73
kapone32
Well that did not take long. lol
Posted on Reply
#74
bug
mechtech"which will be part of the ATX 3.1 spec"

the pc world.......it goes in spurts.........no updates for years, then a bunch of updates in a short time.

well hopefully this addresses all the issues......
It's not that uncommon, if you think about it. If you don't update in a while (and who wants a new ATX spec every 6 months or so), changes can pile up. And when you implement several changes at once, there's a chance not everything will work smoothly all the time, hance the need for a new revision.
Posted on Reply
#75
Scrizz
onemanhitsquadOf no concern...dont have a problem with either of mine
Yep, not a problem here.
Posted on Reply
Add your own comment
Jul 22nd, 2024 14:15 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts