ATX 3.0, Power Cables 12VHPWR, (3.0 GN report)

[Kisai]

48 minutes ago, ravenshrike said:

Aaaand GN comes through once again.

 

 

 

 

So pretty much half the problem is user-error. see

It seems that the theories about the actual connector not making enough contact/contaminations from re-insertions are the direct reason, but the meltings are caused by not getting the cable to make full connection, which is again because the connector isn't deep enough. So the under-engineering of the connector itself is still correct.

 

A bigger/deeper connector would result in user-error being less possible. But it seems like it's very easy to hit this failure condition because the connector softens when it warms up when it's not fully inserted. Which again tells us the connector isn't really deep enough.

 

The conclusion I would make is to avoid using the adapters in the first place, but to also to visually inspect that the connector has no gaps before turning it on.

[ravenshrike]

17 minutes ago, Kisai said:

The conclusion I would make is to avoid using the adapters

The problem isn't confined to the adapters. The stock connectors use the exact same end connector.

 

Addendum - Also, a big issue is going to be prebuilts, because as the GN series has shown, they can get pretty sloppy in assembly.

[Kisai]

1 hour ago, ravenshrike said:

The problem isn't confined to the adapters. The stock connectors use the exact same end connector.

 

Addendum - Also, a big issue is going to be prebuilts, because as the GN series has shown, they can get pretty sloppy in assembly.

You focused on the wrong part of that sentence.

 

Quote

visually inspect that the connector has no gaps before turning it on.

The contaminations are going to happen and that's unavoidable, and the only thing that can be chalked up to manufacturing problems. The fact that the contaminations can heat up and cause a runway heating means these connectors are not designed with enough surface area to avoid it.

 

[Quackers101]

Wonder how PCI-SIG is going to update around this. As thought, it has some issues and some still unconfirmed ones.

connecting and user issue, not either one, but both. at least its good to know it should be safe-ish from it being "fully connected" and if we know that it would stay fully connected in every situation. but also brings into question some of the failed cables too, not properly connected wilfully or not, and if some had other issues.

 

Also to be noted, how igor mentioned about the socket position which can help it being a further problem. how he cut himself on the sharp edges from the position of the connector socket and maybe the needed push to get the connector fully or "rightly" inserted. one of my issues with motherboard sockets that often is a pain to get into and can ruin motherboards...

[ravenshrike]

6 hours ago, that_dude said:

Not at all. As company implement SOP and enforce them.

This is a spherical horse solution. Companies with lackluster assembly techniques are not going to suddenly get better merely because a new video card is on the market. 

[Poinkachu]

What's funny to me is that the one that went into the rabbit hole for clarification is a youtube techtubers instead of the big companies.


I hope GN gets lots of orders on their merch to cover the cost of their test.

[starsmine]

7 minutes ago, Poinkachu said:

What's funny to me is that the one that went into the rabbit hole for clarification is a youtube techtubers instead of the big companies.

that isnt true in the slightest.

the "big companies" ie, the conglomeration, PCI-SIG, and Nvidia are both investigating, just because they have not published anything is not proof that they are not going into the rabbit hole.

[Kisai]

55 minutes ago, starsmine said:

that isnt true in the slightest.

the "big companies" ie, the conglomeration, PCI-SIG, and Nvidia are both investigating, just because they have not published anything is not proof that they are not going into the rabbit hole.

Nah, PCI-SIG and Nvidia don't have the means of analyzing failures that are apparently "random chance" and infrequent.

 

"Usually", and I say this while making airquotes, when stuff fails, it gets binned (as in garbaged,) not investigated unless there is a pattern to the failures. Usually the first few times you see something, it's "random chance", not a pattern.

 

How many failures, and who is to blame, well consider that this issue was identified BEFORE any of the cards were even sold...

https://wccftech.com/atx-3-0-12vhpwr-gen-5-connector-major-safety-risk-using-adapter-confirms-pci-sig/ , that is dated Sep 10th.

Quote

Dear PCI-SIG Member,

 

Please be advised that PCI-SIG has become aware that some implementations of the 12VHPWR connectors and assemblies have demonstrated thermal variance, which could result in safety issues under certain conditions. Although PCI-SIG specifications provide necessary information for interoperability, they do not attempt to encompass all aspects of proper design, relying on numerous industry best-known methods and standard design practices. As the PCI-SIG workgroups include many knowledgeable experts in the field of connector and system design, they will be looking at the information available about this industry issue and assisting in any resolution to whatever extent is appropriate.

 

As more details emerge, PCI-SIG may provide further updates. In the meantime, we recommend members work closely with their connector vendors and exercise due diligence in using high-power connections, particularly where safety concerns may exist.

 

Thank You,

 

When you work with tech stuff, especially when you work in a customer-support/technical-support environment, you are in the position to see statistical anomalies. When I worked for the wireless company, I saw a pattern of phone failures for korean flipphones, motorola flipphones that just did not exist for the candybar phones. If I told you point blank to never buy Samsung or LG based on that, you'd probably ask "well how often is that failure?", from my POV, it was 100% for LG, nearly 100% for Samsung, and a not insignificant amount for Motorola, but nearly 0% for Nokia. Now what does this really say? What if I told you that the LG and Samsung phones were typically "free phones" that were chosen over the Nokia? That explains the frequency, the company simply "sold" more of the free phones, but if that's the case, why is the nokia warranty returns so low? Could it be simply that the LG and Samsung phones shared a design defect (eg bad 850Mhz support)? Was it more likely that the radio network **&* wireless used had bad interoperability with the Samsung and LG devices? Were people sitting on the phones and breaking the antennas? That is information I don't have unless the customer said so. When a phone is refurbished, that's when you find out that information, and if the refurbs also come back, then the problem is still that model phone isn't it?

 

We will probably find out by January 30th (eg a month after Christmas) how frequent actual melting connector problems exist, because by then all the prebuilts people got for Christmas will have entered the retail channel. That also means that Dell, HP, Lenovo, et al have had time to come up with proper solutions, and System Builders like Origin PC, would have had time to ensure that they address this so that they aren't sending PC's back and forth under warranty. After all, if the user doesn't open the chassis, they won't know if the GPU cable has come loose in shipping.