Intel designs an improved 12VHPWR Plug Connector

[DuckDodgers]

 

 

Quote

Intel is currently working on a design update for the male connector of the 12+4 pin ATX 12VHPWR power connector. This redesign aims to enhance the connection and retention of the connector, while also reducing the risk of potential fire hazards caused by improper mating or retention.

The updated male connector design will remain fully compatible with female connectors used in existing devices. The primary modification involves a switch from the current "three-dimple" joint to a "push-spring" type joint for the connector pins.

 

Sometimes a design can't be foolproof tested enough, until it "matures" in the open market and break some eggs. Still too much of the QA process is skipped over to meet the deadlines.

 

Source:

https://www.guru3d.com/news_story/intel_preps_improved_design_for_124_pin_atx_12vhpwr_power_connector_for_enhanced_safety.html

[Uttamattamakin]

I love this.  It always irked me that the answer was "user error" when the issue was a bad design.  

As I said in the start when you run more and more power through a smaller and smaller connector bad things can happen if you're not careful.  It's like channeling firehose pressure through a garden hose sized tube.  You might be able to do it but if things go wrong they go really wrong. 

Good on them for coming up with a better solution of the root problem bad, user unfriendly, design. 

[Poinkachu]

34 minutes ago, Uttamattamakin said:

I love this.  It always irked me that the answer was "user error" when the issue was a bad design.  

As I said in the start when you run more and more power through a smaller and smaller connector bad things can happen if you're not careful.  It's like channeling firehose pressure through a garden hose sized tube.  You might be able to do it but if things go wrong they go really wrong. 

Good on them for coming up with a better solution of the root problem bad, user unfriendly, design. 

IMHO it's a combination of both user error & bad design (be it the connector, or NVIDIA's FE GPU shroud design making it harder to see improper connection from the back of the card)

Although in FE's case, I lean more towards bad design than user error.

Years of being conditioned to seeing PCIe connectors from the back of the card played a part too, we hardly ever checked it from other angle.

 

As much as I want to yell "Yay" for the new idea, I don't know until it's properly abus tested. I'm no pro, but the spring design seem to use less materials (the corners is gone if what I'm seeing on the pic is right), question is does this make the terminal more flimsy or not, I guess.

 

My Logic so far :

It might prevent accidental "connector slipping out", mainly due to the new design having tighter connection (in case the clip isn't properly locked in the first place), but I doubt the new design will prevent failure due to improper connection if the user fail to make proper one in the first place.

 

But eh, at the very least there's someone trying to make it better I guess.
 

 

But yeah, this is just an observation from someone who never uses these connectors, so go easy on me.

 

[Forbidden Wafer]

17 minutes ago, Poinkachu said:

Years of being conditioned to seeing PCIe connectors from the back of the card played a part too, we hardly ever checked it from other angle.

From the videos I watched, it also didn't seem to click when correctly connected.
 

If it did, it would be more straightforward to just say: if you haven't heard a click, it isn't properly connected. If you can't make it click after applying normal pressure, look the for reseller to test it.

[pas008]

21 minutes ago, Poinkachu said:

IMHO it's a combination of both user error & bad design (be it the connector, or NVIDIA's FE GPU shroud design making it harder to see improper connection from the back of the card)

Although in FE's case, I lean more towards bad design than user error.

Years of being conditioned to seeing PCIe connectors from the back of the card played a part too, we hardly ever checked it from other angle.

 

As much as I want to yell "Yay" for the new idea, I don't know until it's properly abus tested. I'm no pro, but the spring design seem to use less materials (the corners is gone if what I'm seeing on the pic is right), question is does this make the terminal more flimsy or not, I guess.

 

My Logic so far :

It might prevent accidental "connector slipping out", mainly due to the new design having tighter connection (in case the clip isn't properly locked in the first place), but I doubt the new design will prevent failure due to improper connection if the user fail to make proper one in the first place.

 

But eh, at the very least there's someone trying to make it better I guess.
 

 

But yeah, this is just an observation from someone who never uses these connectors, so go easy on me.

 

did nvidia 12pin from 3k series have issues?

[Poinkachu]

33 minutes ago, Forbidden Wafer said:

From the videos I watched, it also didn't seem to click when correctly connected.
If it did, it would be more straightforward to just say: if you haven't heard a click, it isn't properly connected. If you can't make it click after applying normal pressure, look the for reseller to test it.

Eh well, since I don't own any RTX40, I can't say much about the click sound.
Maybe it's there but faint, maybe there isn't any. Maybe on some it made the sound, and not on some others. Idk

I have had a GPU where the PCIe connector locks securely but without a click sound, I recall it took me quite some effort to insert the PCIe connectors.

Maybe there was but faint and I couldn't hear it well due to me spouting some curses & swears due to having to put a lot of effort to plug the connectors.

(Seriously +2 pins can be so damn annoying sometimes, especially on cards that requires 3 PCIe connectors)

 

But yeah, I made it a habbit to try by pressing the clip after I insert the connector. Just in case.
12VHPWR though, I think mostly have the clip on the underside, which is one of the reason why I said I lean more towards bad design.

 

28 minutes ago, pas008 said:

did nvidia 12pin from 3k series have issues?

I don't think I can recall hearing / reading news about massive problem with it. CMIIW

Those are different kind anyway, it doesn't have the 4 small pins housing.

[pas008]

21 minutes ago, Poinkachu said:

 

I don't think I can recall hearing / reading news about massive problem with it. CMIIW

Those are different kind anyway, it doesn't have the 4 small pins housing.

i believe the pinout is the same for the 12 just no sensing pins

[Brooksie359]

1 hour ago, Poinkachu said:

IMHO it's a combination of both user error & bad design (be it the connector, or NVIDIA's FE GPU shroud design making it harder to see improper connection from the back of the card)

Although in FE's case, I lean more towards bad design than user error.

Years of being conditioned to seeing PCIe connectors from the back of the card played a part too, we hardly ever checked it from other angle.

 

As much as I want to yell "Yay" for the new idea, I don't know until it's properly abus tested. I'm no pro, but the spring design seem to use less materials (the corners is gone if what I'm seeing on the pic is right), question is does this make the terminal more flimsy or not, I guess.

 

My Logic so far :

It might prevent accidental "connector slipping out", mainly due to the new design having tighter connection (in case the clip isn't properly locked in the first place), but I doubt the new design will prevent failure due to improper connection if the user fail to make proper one in the first place.

 

But eh, at the very least there's someone trying to make it better I guess.
 

 

But yeah, this is just an observation from someone who never uses these connectors, so go easy on me.

 

That is just a bad excuse for bad design. Things should be built with user error in mind especially when it comes to things like starting on fire. I shouldn't have to worry about things starting on fire or exploding because something isn't plugged in all of the way. I mean we have detectors on garage doors to prevent accidents from user error and other types of safety measures. There is 0 excuse to have a design that can result in the start of a fire if it isn't plugged in all of the way. 

[pas008]

1 hour ago, Brooksie359 said:

That is just a bad excuse for bad design. Things should be built with user error in mind especially when it comes to things like starting on fire. I shouldn't have to worry about things starting on fire or exploding because something isn't plugged in all of the way. I mean we have detectors on garage doors to prevent accidents from user error and other types of safety measures. There is 0 excuse to have a design that can result in the start of a fire if it isn't plugged in all of the way. 

4 pin molex?

[Kisai]

2 hours ago, Brooksie359 said:

That is just a bad excuse for bad design. Things should be built with user error in mind especially when it comes to things like starting on fire. 

There's "built with error in mind" and then there is "built with the incompetency of the average weaponized-lawsuit American"

 

It has to literately be idiot-proof. One of those ways to ensure that happens is by over-engineering the connector and having fuses. In fact the way that could have been prevented would be by having resettable fuses on the PSU side to prevent single-wire over-current. Or even dumb in-line fuses on modular cables so that they all blow sequentially before the GPU connector overheats. 

 

Or you know, just having a thermal diode under the connector so that if it crosses the expected thermal limit, it tells the GPU to dial back power until the thermals drop, and send an error that the GPU is being safety throttled.

 

The only reason anything melted is because the connector was under-engineered. Even previous 6/8 pin connectors sometimes melted due to user error, but it was never large quantities of cards, it tended to be people who were doing something stupid, like crypto mining while not using cables that can support that power.

[Brooksie359]

22 minutes ago, Kisai said:

There's "built with error in mind" and then there is "built with the incompetency of the average weaponized-lawsuit American"

 

It has to literately be idiot-proof. One of those ways to ensure that happens is by over-engineering the connector and having fuses. In fact the way that could have been prevented would be by having resettable fuses on the PSU side to prevent single-wire over-current. Or even dumb in-line fuses on modular cables so that they all blow sequentially before the GPU connector overheats. 

 

Or you know, just having a thermal diode under the connector so that if it crosses the expected thermal limit, it tells the GPU to dial back power until the thermals drop, and send an error that the GPU is being safety throttled.

 

The only reason anything melted is because the connector was under-engineered. Even previous 6/8 pin connectors sometimes melted due to user error, but it was never large quantities of cards, it tended to be people who were doing something stupid, like crypto mining while not using cables that can support that power.

I mean not bring fully plugged in is not idiot proof tbh. I feel like anybody could make that mistake or even the cable could be dislodged if you are moving around the pc. I mean if you had to move places and the cable gets dislodged when moving to your new place it would suck if the things sets on fire. It should be designed with things like that in mind for sure. Also unfortunately idiot proof is not only meant for adults but kids too. I mean kids do stupid things so having overengineered is always good. I would assume most adults wouldn't get caught in a garage door closing but you bet a kid would try and go through a closing garage door. Anyways I am not against things being idiot proof as most people can be idiots at times especially under the influence. Everything seems like a good idea after drinking a few beers. 

[Senzelian]

So what exactly was the problem with this again? 

 

[HenrySalayne]

5 hours ago, Uttamattamakin said:

I love this.  It always irked me that the answer was "user error" when the issue was a bad design.  

As I said in the start when you run more and more power through a smaller and smaller connector bad things can happen if you're not careful.  It's like channeling firehose pressure through a garden hose sized tube.  You might be able to do it but if things go wrong they go really wrong. 

Good on them for coming up with a better solution of the root problem bad, user unfriendly, design. 

I have used this connector and it's not a bad design. In comparison to other connectors (24P on motherboards) it's easier to get into the socket and the click is not the most pronounced but you have enough feedback (if you care) to make sure it's seated correctly.

The "oversight" is the power density you mentioned. Users were not fully seating connectors for decades, but the consequences were mostly not severe enough to stand out and be noticed by the public and the manufacturers. The 12VHPWR connector just brought a wide-spread malpractice to the surface.

And a part of the problem might be the stiff cable. I "pre-shaped" mine with a hot air gun or I would had to squeeze the cable like crazy to make it fit.

[HenrySalayne]

4 hours ago, Brooksie359 said:

That is just a bad excuse for bad design. Things should be built with user error in mind especially when it comes to things like starting on fire. I shouldn't have to worry about things starting on fire or exploding because something isn't plugged in all of the way. I mean we have detectors on garage doors to prevent accidents from user error and other types of safety measures. There is 0 excuse to have a design that can result in the start of a fire if it isn't plugged in all of the way. 

1 hour ago, Brooksie359 said:

I mean not bring fully plugged in is not idiot proof tbh. I feel like anybody could make that mistake or even the cable could be dislodged if you are moving around the pc. I mean if you had to move places and the cable gets dislodged when moving to your new place it would suck if the things sets on fire. It should be designed with things like that in mind for sure. Also unfortunately idiot proof is not only meant for adults but kids too. I mean kids do stupid things so having overengineered is always good. I would assume most adults wouldn't get caught in a garage door closing but you bet a kid would try and go through a closing garage door. Anyways I am not against things being idiot proof as most people can be idiots at times especially under the influence. Everything seems like a good idea after drinking a few beers. 

I mean there are countries on this planet using plugs that can electrocute the user if they are not plugged in all the way. And they are in use several times a day.

On the other side, the 12VHPWR connector has to be plugged in once in maybe 3 years? I would argue you can ask users to be a little bit more careful in this case.

[Shimejii]

14 minutes ago, Senzelian said:

So what exactly was the problem with this again? 

 

I mean its a lot of plugs, They tried to solve the problem of not needing as much copper wiring, lessen the cost and the amount of cable routing. It makes sense, it was just a poorly done execution for not having the retention part. You always have to design these plugs for the stupidest people, because if there is a way to mess it up they will. Just because they have a lot of money does not mean they are logical or know what they are doing. 

[Senzelian]

1 minute ago, Shimejii said:

They tried to solve the problem of not needing as much copper wiring, lessen the cost and the amount of cable routing. It makes sense

To me this doesn't make sense, cause you'd think a given amount of power always requires the same amount of mass in copper wiring to transfer it from A to B.

[leadeater]

24 minutes ago, Senzelian said:

So what exactly was the problem with this again? 

 

Nothing, just get a single cable with two 8-Pins and then get two splitters to make four 8-Pins and you're golden 

[Uttamattamakin]

14 minutes ago, HenrySalayne said:

I have used this connector and it's not a bad design. In comparison to other connectors (24P on motherboards) it's easier to get into the socket and the click is not the most pronounced but you have enough feedback (if you care) to make sure it's seated correctly.

Yes this is one of those things.  Like if there was a loud satisfying CLICK ... that's what I'd be listening for and feeling for.  That's what just about every other connector does.  It takes a lot of force to insert but there is some clear feedback that you're good. 

 

14 minutes ago, HenrySalayne said:

The "oversight" is the power density you mentioned. Users were not fully seating connectors for decades, but the consequences were mostly not severe enough to stand out and be noticed by the public and the manufacturers. The 12VHPWR connector just brought a wide-spread malpractice to the surface.

Yes this is what I was driving at back when the news broke.  As we channel more power into a component the more dangerous it gets to let ordinary users or even ones able to build their own PC mess with it.  You know?  It's like the reason we don't mess around inside a power supply.   I hope Intel's fix is really a fix. 

 

14 minutes ago, HenrySalayne said:

And a part of the problem might be the stiff cable. I "pre-shaped" mine with a hot air gun or I would had to squeeze the cable like crazy to make it fit.

Yeah that's easy everyone should've thought to do that. 

 

7 minutes ago, Senzelian said:

To me this doesn't make sense, cause you'd think a given amount of power always requires the same amount of mass in copper wiring to transfer it from A to B.

Sort of... if you don't mind a lot more resistance in the wire as the cross section goes down the resistance will go up.  Just as a narrow river flows faster, and more turbulently,  than a wide one.   

3 minutes ago, leadeater said:

Nothing, just get a single cable with two 8-Pins and then get two splitters to make four 8-Pins and you're golden 

Don't forget the water cooling we'll need for the power cables themselves soon. 

[HenrySalayne]

20 minutes ago, Senzelian said:

To me this doesn't make sense, cause you'd think a given amount of power always requires the same amount of mass in copper wiring to transfer it from A to B.

The 12VHPWR uses 6 wires (12 in total) to carry up to 600W at 12V. So in total it's 50A and 8.3A per wire.

It's recommended to get a maximum current density of around The rule of thumb says 6A per mm² in a copper conductor is fine. So these wires would need roundabout a cross-section of 1.38 mm². They are using AWG16 (1.31 mm²) for the 12VHPWR cables, so this should be ok.

On a nano-scale (like inside the CPU), the copper wires have to bear several hundred Amperes per mm². The physics of wire gauge are not as straight forward as it may seem.

 

[Godlygamer23]

9 minutes ago, HenrySalayne said:

The 12VHPWR uses 6 wires (12 in total) to carry up to 600W at 12V. So in total it's 50A and 8.3A per wire.

It's recommended to get a maximum current density of around 6A per mm² in a copper conductor. So these wires would need roundabout a cross-section of 1.38 mm². They are using AWG16 (1.31 mm²) for the 12VHPWR cables, so this should be fine.

On a nano-scale (like inside the CPU), the copper wires have to bear several hundred Amperes per mm². The physics of wire gauge are not as straight forward as it may seem.

In the case of CPUs though, most of that energy is converted to heat(except for a VERY small amount), and coolers are present to pull that heat away. It's not about the current or voltage per se. It's about how much heat is being put out. You need your wires to stay cool enough in the given operating environment so they don't get so hot they melt the connector, or ignite the insulation. In CPUs, those wires are technically there, but obviously significantly smaller than the wires used in a cable, but like I said, they have large heatsinks mounted to them, or blocks for water cooling. Cables don't have that luxury. 

[ouroesa]

I still think the problem is not the connector per se but the amount of power they are delivering through those teeny plugs/wires. They flew too close to the sun with that design and now the wax is melting. I'm still of the opinion that the cards are simply too power hungry and that's why we needed so many 8pin connectors on a single card that it started looking stupid and nobody stopped to think why it looked stupid. It looked stupid because it was absurd that a consume class GPU was pulling that much power!

[Kisai]

16 minutes ago, ouroesa said:

I still think the problem is not the connector per se but the amount of power they are delivering through those teeny plugs/wires. They flew too close to the sun with that design and now the wax is melting. I'm still of the opinion that the cards are simply too power hungry and that's why we needed so many 8pin connectors on a single card that it started looking stupid and nobody stopped to think why it looked stupid. It looked stupid because it was absurd that a consume class GPU was pulling that much power!

A desktop PC should not have a 600w GPU. Those belong in a HEDT, and AMD/Intel have been neglecting the HEDT market, since people once bought up to 4 GPU's, just because they could. You can't even do that today because nobody in the US could operate it. 1 CPU + 1 GPU should not cross 900w.  That's space-heater territory.

 

 

[Brooksie359]

4 hours ago, HenrySalayne said:

I mean there are countries on this planet using plugs that can electrocute the user if they are not plugged in all the way. And they are in use several times a day.

On the other side, the 12VHPWR connector has to be plugged in once in maybe 3 years? I would argue you can ask users to be a little bit more careful in this case.

I would argue they should replace the plug type if you can easily be electrocuted if not plugged in all of the way. Also its not like the plug can't come lose and then you have the thing start on fire for no fault of your own. 

[Poinkachu]

6 hours ago, Brooksie359 said:

I mean not bring fully plugged in is not idiot proof tbh. I feel like anybody could make that mistake or even the cable could be dislodged if you are moving around the pc. I mean if you had to move places and the cable gets dislodged when moving to your new place it would suck if the things sets on fire. It should be designed with things like that in mind for sure. Also unfortunately idiot proof is not only meant for adults but kids too. I mean kids do stupid things so having overengineered is always good. I would assume most adults wouldn't get caught in a garage door closing but you bet a kid would try and go through a closing garage door. Anyways I am not against things being idiot proof as most people can be idiots at times especially under the influence. Everything seems like a good idea after drinking a few beers. 

Well... there's always soldering the 12VHPWR side directly onto the PCB. But I doubt lot of users gonna like it, especially with how bulky & ugly those adapters are, which looks even uglier after some PSU manufacturers & Cablemod released their 12VHPWR.

 

There's also the method like old time VGA / DVI connector, having mounting screws on their side. Although based on my personal experience, I hardly ever seen someone uses them (the screws), especially after multiple times of replugging in a relatively short time.

 

I get your point, but yeah.... good design is only as good as the user who use the thing.

I mean, we can enclose a PC in a thick seamless metal box, but if someone is idiot enough they'll find ways to destroy it.

 

If someone moves their PC around so much there's possibility of stuffs dislodging, or if someone brings their PC to a new place, due dilligence is to check wether or not everything is fine before powering up.

 

If it's a kid, then their parents shouldn't let them mess with a dangerous thing. In fact, they shouldn't be let around potentially dangerous thing without supervision.

I mean, sensors or other mechanical safety measures can be defective, or break at some point.

 

As much as I want to put all the blame on companies the reality is :

At the end of the day, it is our own responsibility to check and/or research the things we or our loved ones are about to use so as not to cause grief / disaster. Or you know, to atleast minimize the chance of it happening.

Of course, vendor can design things to make it more convenient for users to spot mistakes / possible-future-disaster.

 

So no, I'm not saying it's all entirely user's fault, because like I've said before, 4090's shroud is quite well at hiding the terminal, making it harder to spot mistake.

Though I can't say it's fully companies fault either, because this is something that can be observed & checked without disassembly or extensive knowledge.

At the very least for this particular topic.

 

But yeah, idk, I've lived in a "jank & cheaping out" country for so long, that even if something is built truly idiotproof, I'd still research & do proper check and recheck as much as I can about it in the end.

Out of habbit, a bit paranoia, & some skepticism :x

[Kisai]

8 minutes ago, Poinkachu said:

 

There's also the method like old time VGA / DVI connector, having mounting screws on their side. Although based on my personal experience, I hardly ever seen someone uses them (the screws), especially after multiple times of replugging in a relatively short time.

 

Honestly, they should bring back "thumb screw tension" for everything. We won't see it on GPU's because the space is already too tight, but realistically, that would have solved the user error potential by ensuring that the screw grounds the bottom of the screw shaft to signal it's been tightened. No signal, no power.

 

But that horse left the barn already, so we're going to be stuck with a generation of GPU's and PSU's that are potential fire hazards.