Intel Readies ATX 12VHPWR Connector Revision to Address Improper-Contact Fire Hazards

[jonnyGURU]

Must be a slow news day. This is as old as old gets and yet today it's "reported" on a dozen websites.

 

[RegaeRevaeb]

This isn't needed, it's all customer fault who can properly insert the connector which is perfect.

I'm hoping you meant that ironically. Yes, end users hold a certain amount of responsibility to use products (i.e. plugs here) as directed as a reasonable person would. This now means plugging everything in fully -- with authority -- and not bending the cable until it's clear 'X' cm. But that's come after issues were raised.

The strong information campaign by industry manufacturers, for example, is evidence that overall responsibility doesn't fall fully with end users. And, I'd are the Intel revision is a product in large part born from industry procucer responsibility -- liability, if you haven't yet decoded what I've been getting at.

(edited for grammar)

 

[mechtech]

Damn, I just got my Seasonic Vertex PSU, which right now is obsolete already

Early beta adopter risk........................everyone knows it.....................

 

[enb141]

Early beta adopter risk........................everyone knows it.....................

Well beta was supposed to be last year, this PSU is from this year.

 

[mechtech]

Well beta was supposed to be last year, this PSU is from this year.

ya, unfortunately PSU's makers implement new specs and standards slower than monitor makers do.............................so beta testing psu is about 3 years after a new spec implemented

 

[enb141]

ya, unfortunately PSU's makers implement new specs and standards slower than monitor makers do.............................so beta testing psu is about 3 years after a new spec implemented

I was hopping 6 months for PSU but apparently is more closer to your guessing time.

 

[maxfly]

I was hopping 6 months for PSU but apparently is more closer to your guessing time.

A new revision isn't going to render your Vertex obsolete, not by any stretch. Your integrated 12vhpwr connector will continue to work as it should. You have nothing to worry about.

 

[enb141]

A new revision isn't going to render your Vertex obsolete, not by any stretch. Your integrated 12vhpwr connector will continue to work as it should. You have nothing to worry about.

Well, depends, if for example, Nvidia decides to use something else when they release their 5XXX series cards, then it will be obsolete.

 

[maxfly]

Well, depends, if for example, Nvidia decides to use something else when they release their 5XXX series cards, then it will be obsolete.

That's a non-issue. A revision isn't a rebuild.. They aren't going to rebuild the entire unit to fix a single flaw.

 

[PapaTaipei]

600 watts to a single GPU is nuts.

 

[Chrispy_]

I like how Intel is trying to fix a fire hazard by creating a revision that will no doubt be very hard for end users to distinguish from the original fire-hazard version.

Do you know what also isn't a fire hazard?
The original 8-pin Mini-fit Jr connectors that we've been using without any problems for the last 19 years.

 

[enb141]

That's a non-issue. A revision isn't a rebuild.. They aren't going to rebuild the entire unit to fix a single flaw.

Then why they decided to create a new connector?

My guess is that this connector will be like the AGP, a short lived, replaced by PCIe.

 

[Chrispy_]

My guess is that this connector will be like the AGP, a short lived, replaced by PCIe.

If anything is going to replace the 12+4 pin HPWR connector, it'll probably be multiple 8-pin EPS (ATX12V) connector used by motherboards for CPUs. The connector and wires are rated for 336W (7A per wire pair) so a couple of those will do any current GPU without difficulty, using just one will power a 400W GPU when you add the 75W from the slot. The only reason the 8-pin connector for PCIe is limited to 150W is because it's an evolution of the 6-pin connector in which only two pairs carried power because a pair is wasted for sense pinouts.

All of the server GPUs use EPS already; The sense pinout isn't necessary for extremely hungry datacentre GPUs.

This dumbshit HPWR connector with it's tiny fragile pins and create a need for adapters with just about any PSU was an Nvidia douchebag move that has been wholly unwelcome in what had previously been a robust, problem-free, compatible ecosystem of Minifit Jr PSU connectors. Nobody likes the ugly adapters, and the whole thing was a dumb concept to save a tiny amount of PCB space in Nvidia's FE cards.

 

[enb141]

If anything is going to replace the 12+4 pin HPWR connector, it'll probably be multiple 8-pin EPS (ATX12V) connector used by motherboards for CPUs. The connector and wires are rated for 336W (7A per wire pair) so a couple of those will do any current GPU without difficulty, using just one will power a 400W GPU when you add the 75W from the slot. The only reason the 8-pin connector for PCIe is limited to 150W is because it's an evolution of the 6-pin connector in which only two pairs carried power because a pair is wasted for sense pinouts.

All of the server GPUs use EPS already; The sense pinout isn't necessary for extremely hungry datacentre GPUs.

This dumbshit HPWR connector with it's tiny fragile pins and create a need for adapters with just about any PSU was an Nvidia douchebag move that has been wholly unwelcome in what had previously been a robust, problem-free, compatible ecosystem of Minifit Jr PSU connectors. Nobody likes the ugly adapters, and the whole thing was a dumb concept to save a tiny amount of PCB space in Nvidia's FE cards.

That had to be the correct way to replace the 8 PIN cable but they somehow decided to think "different" so hopefully they will move to EPS but I doubt, anyways, in the not so distant future video cards will need 2 or even 3 12VHPWR connectors anyways.

 

[maxfly]

Then why they decided to create a new connector?

My guess is that this connector will be like the AGP, a short lived, replaced by PCIe.

Again, this being a revision...NOT a new connector. They are simply going to modify some of the pins to allow for better contact.

If your honestly concerned with this revision affecting your PSU, I encourage you to read one of the readily available articles.

 

[trsttte]

This dumbshit HPWR connector with it's tiny fragile pins and create a need for adapters with just about any PSU was an Nvidia douchebag move that has been wholly unwelcome in what had previously been a robust, problem-free, compatible ecosystem of Minifit Jr PSU connectors. Nobody likes the ugly adapters, and the whole thing was a dumb concept to save a tiny amount of PCB space in Nvidia's FE cards.

I get that back when pcie power connectors first became a thing they didn't know how power specs would evolve and all that, but even on the consumer space the sense pins make little sense - especially when they are just passive jumpers to 12v or gnd on the connectors, no active signalling is involved with either the old or new 12vhpwr specs.

Pretty much everyone has been using 8x pin connectors on the psu side to carry up to 300w for years, refusing to do so on the gpu side is just silly.

 

[Chrispy_]

Pretty much everyone has been using 8x pin connectors on the psu side to carry up to 300w for years, refusing to do so on the gpu side is just silly.

Silly indeed. Both Nvidia and AMD have had >300W graphics cards since March 2011.

Why on earth did it take Nvidia 11 years to do something about it when it wasn't an issue beforehand, and why was the inferior HPWR their solution? It's more fragile, the pins can handle less current and apparently it doesn't like being bent but Nvidia put it on the outside edge RIGHT UP AGAINST THE SIDE PANEL of their largest GPUs ever realeased to market...

Talk about shooting yourself in the foot, I guess there's a lot of hope and prayers behind "it just works" these days...

 

[enb141]

Again, this being a revision...NOT a new connector. They are simply going to modify some of the pins to allow for better contact.

If your honestly concerned with this revision affecting your PSU, I encourage you to read one of the readily available articles.

No, I mean that if this connector becomes a fiasco, then they (intel / nvidia) will come with something different, the same that happened to the defunct AGP.

 

[TheoneandonlyMrK]

No, I mean that if this connector becomes a fiasco, then they (intel / nvidia) will come with something different, the same that happened to the defunct AGP.

Intel recently (after this pr but not today)released a press release stating the dimple crimped version should not be used and also evolved the latching mechanism to engage better.

Why.

Because generation one(or possibly ++++)of the design was shit.

So different has arrived already That's how great those adapter's are.

I expect several house fires in a couple of years once the second hand market sees mostly 4### series because degrading connections Degrade, well made ones don't.

 

[enb141]

Intel recently (after this pr but not today)released a press release stating the dimple crimped version should not be used and also evolved the latching mechanism to engage better.

Why.

Because generation one(or possibly ++++)of the design was shit.

So different has arrived already That's how great those adapter's are.

I expect several house fires in a couple of years once the second hand market sees mostly 4### series because degrading connections Degrade, well made ones don't.

In a few years we'll know if this connector becomes a fiasco (very likely) or just get improved (I doubt).

 

[arni-gx]

so, MSI a1000w, corsair rmx shift 1000w, also all PSU with full atx 3.0 + pcie 5.0 that has been released, early this year, alll became obselete .... ?? just because all of them has only 3 dimple, not 4 spring connector of 12pin... ??

 

[TristanX76]

3000W max. Problem solved.

Oh, its a mass production part too. Cheap.

 

[RegaeRevaeb]

View attachment 292639

3000W max. Problem solved.

Oh, its a mass production part too. Cheap.

Goes well with four-slot designs as well from the looks of it.

 

[LabRat 891]

View attachment 292639

3000W max. Problem solved.

Oh, its a mass production part too. Cheap.

It's a travesty that 'code' doesn't expect all outlets be 'split-phase' in the US/Can.

All NA homes are served by 120v/120v 'split-phase', and it's arguably 'safer' than WW 208-250VAC -Only the most un-lucky and clutzy could ever receive more than 169.71V of potential at any given moment.

Also, all 125V-NEMA outlets can be 'split', such that 1 'side of the phase' goes to one outlet, and 'the other side of the phase' goes to the other.
It used to be somewhat common for Kitchens to have outlets wired like this, to separate Appliance loads. It's uncommon now, as Neutral-sharing can be dangerous in some failure-modes, and nobody's (usually) willing to spend $ on another run of White Insulated Copper for an additional dedicated Neutral.

15A @ ~240VAC60hz is 3600W, and 2880W+ continuous. A dual NEMA5-15P <-> to IEC cable would be all that'd be necessary to get 2.8-3.6KW safely.



In my case, I retrofitted the wall-heater circuit into a locking 250V NEMA receptacle, and connected that to an APC PDU for 240VAC to my 'office' equipment.

 

[trsttte]

it's arguably 'safer' than WW 208-250VAC -Only the most un-lucky and clutzy could ever receive more than 169.71V of potential at any given moment.

Safer in what way? IMO NA electrical systems are simply a mess, the voltage being halved doesn't make it safer in any meaningfull way while making it more dangerous by using double the current for anything and also wasting more power doing so.