Modular PSU connection possibly solves boot issues

[Splinterdog]

I was handed a nice Ryzen PC, built by a first timer 12 year old lad and considering it was his first, he did a very good job, and I don't mean that in a patronising way. He really did, with very few errors and none that would have killed the PC.
It wouldn't boot and there was no video signal, so I stripped it down and luckily had a new AM4 mobo to test every component. The odd thing was that nothing failed on that bench set up and I even tried it with my RX 580 and it worked.
Put everything back into his Rog Strix B450F Gaming II and the Evga RTX 3060 again failed to boot the system, giving one long and three short beeps - VGA problem, apparently.
But then on the off chance, I looked closely at the modular Corsair RM750X, then switched the PCIe and 4+4 PSU connectors over, placing them in their respective connectors as indicated by the back of the PSU and the system booted.
I had repeated this exercise over and over and everything pointed to the GPU, which turned out to be right in the end, but it's still a head scratcher.
Do these voltages vary?


PS
Very impressed with the RTX 3060 while testing it on Tomb Raider DOX, which is quite a demanding game.

 

[Bill_Bright]

switched the PCIe and 4+4 PSU connectors over, placing them in their respective connectors as indicated by the back of the PSU and the system booted.

This has caused problems for many and why it is so important to check the labeling during assembly.

Personally, I think the ATX standard is a godsend and I am thankful Intel created it all those years ago. Without it, there would be no build-it-yourself PC industry and everything would be proprietary just as it is with laptops.

And while Intel really has done a great job at keeping the ATX standard current and "neutral" (not biased towards any one company or brand), I feel they really messed up with those two connectors specifically. They are too much alike and therefore, too easy to mistakenly mix up. The connectors should be physically different enough (shape, size and how they are keyed) so that it is impossible to insert one plug in the wrong socket without standing on it, or hitting it with a 10lb sledge hammer.

Part of the problem is there are no standards at all for the PSU-end of modular cables. This means cables that come with different PSUs cannot be mixed or matched. Even within the same brand, in particular, where different OEMs are used, you cannot mix and match cables. This means users have to be careful how they keep separate, label, and store their unused cables - forever! This can be a logistics nightmare if you have dozens of computers you are responsible for. This is not a problem with hard-wired PSUs.

The solution to the ATX issue is there needs to be a consortium of manufacturers (not just one company) who come to a consensus when it comes to physical characteristics for cable connections. The problem is, that will never happen. It is like the constant battle between Apple and Android, or the two sides in politics these days - each side demanding everything be their way or no way - unwilling to meet in the middle. And who suffers the most? The consumer/citizens.

 

[Splinterdog]

This has caused problems for many and why it is so important to check the labeling during assembly.

Personally, I think the ATX standard is a godsend and I am thankful Intel created it all those years ago. Without it, there would be no build-it-yourself PC industry and everything would be proprietary just as it is with laptops.

And while Intel really has done a great job at keeping the ATX standard current and "neutral" (not biased towards any one company or brand), I feel they really messed up with those two connectors specifically. They are too much alike and therefore, too easy to mistakenly mix up. The connectors should be physically different enough (shape, size and how they are keyed) so that it is impossible to insert one plug in the wrong socket without standing on it, or hitting it with a 10lb sledge hammer.

Part of the problem is there are no standards at all for the PSU-end of modular cables. This means cables that come with different PSUs cannot be mixed or matched. Even within the same brand, in particular, where different OEMs are used, you cannot mix and match cables. This means users have to be careful how they keep separate, label, and store their unused cables - forever! This can be a logistics nightmare if you have dozens of computers you are responsible for. This is not a problem with hard-wired PSUs.

The solution to the ATX issue is there needs to be a consortium of manufacturers (not just one company) who come to a consensus when it comes to physical characteristics for cable connections. The problem is, that will never happen. It is like the constant battle between Apple and Android, or the two sides in politics these days - each side demanding everything be their way or no way - unwilling to meet in the middle. And who suffers the most? The consumer/citizens.

Thanks for your reply. Are you suggesting that there's a power difference between the PCIe 6+2 connection and the 4+4 CPU connection?

 

[JL87]

Not so much as a power difference as it’s the possibility that they are wired differently.
I’ve run into similar issues before when someone replaced a bad power supply and tried to reuse the cables already in the computer (different manufacturer). Once I took a look at it and asked the question, I used the cables from the new PSU and everything worked fine.

 

[Bill_Bright]

Power difference? Not sure what you mean there. Power (as in available wattage) could indeed be different depending on the capability of the PSU on that specific rail.

But that was NOT my point.

All I am "suggesting" is the ATX Form Factor standard should, like it does with other connectors, dictate specific and unique physical characteristics for every connector to ensure with 100% certainty that it is impossible to connect the wrong plug to a socket.

 

[Splinterdog]

Power difference? Not sure what you mean there. Power (as in available wattage) could indeed be different depending on the capability of the PSU on that specific rail.

But that was NOT my point.

All I am "suggesting" is the ATX Form Factor standard should, like it does with other connectors, dictate specific and unique physical characteristics for every connector to ensure with 100% certainty that it is impossible to connect the wrong plug to a socket.

I get that and what I have noticed is that all the PSU manuals don't cover the female connections on the PSUs themselves. I don't often read PSU manuals, but on this occasion I did, but there's no reference to this aspect at all.
There are Corsair guides, however:

PSU CABLE COMPATIBILITY

Corsair

www.corsair.com

No mention of the PSU connector slots here:

https://www.corsair.com/corsairmedia/sys_master/productcontent/RMx_2018_Manual.pdf

 

[Bill_Bright]

Yeah, you point out the problems very well. Many manuals are lacking information or are incomplete.

Why is there even a PSU cable compatibility chart? Because there is no industry standard being used even within the same brand.

Connecting a PSU to the components inside the case should be totally intuitive with the manual simply being a formality and list of standard safety precautions. The manual (or Quick Installation Guide) should at least be fully illustrated with pictures of the connector side of the PSU showing which cable goes where - on both ends.

I think the problem is the first modular PSUs hit the market, more or less as a surprise. And the other makers quickly came out with their own versions. This happened so quickly and before the ATX Standard could address this, that each maker had their own standards. Then it was too late to get them all to agree on one - especially after patents were given.

 

[Splinterdog]

Now that it's up and running, there's a lot of RGB going on

 

[mstenholm]

Now that it's up and running, there's a lot of RGB going on
View attachment 226942

I'm sure somebody can help you turn it off

 

[Splinterdog]

I'm sure somebody can help you turn it off

It belongs to a customer, a kid who built it himself.

 

[mstenholm]

It belongs to a customer, a kid who built it himself.

I know, I read all the posts. It was a joke. Sigh.

 

[looniam]

I get that and what I have noticed is that all the PSU manuals don't cover the female connections on the PSUs themselves. I don't often read PSU manuals, but on this occasion I did, but there's no reference to this aspect at all.
There are Corsair guides, however:

PSU CABLE COMPATIBILITY

Corsair

www.corsair.com

No mention of the PSU connector slots here:

https://www.corsair.com/corsairmedia/sys_master/productcontent/RMx_2018_Manual.pdf

try here

Corsair® PSU Type 3 Cables Pinout

Type 3 cables are compatible with the following Corsair® PSUs: AXi (Except AX1600i) HXi HX (GOLD) RM (previous to 2019 version) Limited compatibility: Only compatible with 24pin ATX cable: AX1200 Only compatible with 8pin PCI-E, 4pin Power and 5pin SATA Power cables: CS-M CX-M TX-M...

pc-mods.com

you'll see the EPS is 4 hot wires and the gpu is 3:

taking a guess - i know a lot of 8pin cpu connections work with just a 4pin, but they need to be the right four pins and that open connection may have been the issue.
edit:
sorry for the late edit but i forgot you said it was a gpu issue - well for whatever the reason the card didn't recognised the connection. sensor(s) can/are jumpered to ground and w/ground wires being the same . . it just didn't like it.

 

[Splinterdog]

try here

Corsair® PSU Type 3 Cables Pinout

Type 3 cables are compatible with the following Corsair® PSUs: AXi (Except AX1600i) HXi HX (GOLD) RM (previous to 2019 version) Limited compatibility: Only compatible with 24pin ATX cable: AX1200 Only compatible with 8pin PCI-E, 4pin Power and 5pin SATA Power cables: CS-M CX-M TX-M...

pc-mods.com

you'll see the EPS is 4 hot wires and the gpu is 3:
View attachment 226949
taking a guess - i know a lot of 8pin cpu connections work with just a 4pin, but they need to be the right four pins and that open connection may have been the issue.
edit:
sorry for the late edit but i forgot you said it was a gpu issue - well for whatever the reason the card didn't recognised the connection. sensor(s) can/are jumpered to ground and w/ground wires being the same . . it just didn't like it.

That's an excellent observation and bears out what @Bill_Bright was saying earlier. In defence of the kid who built the PC, it's a very easy mistake to make. Heck, I'm pretty sure I've done it myself.

 

[Outback Bronze]

Pretty sure my Corsair PSU's don't matter which 8 pin connection they go into for CPU or gpu. I've never had a problem so this issue is new to me. I didn't think it made a difference.

Models I have been using are AX's, HX's, RM's and even a 450SF all no issue's whatever 8 pin port on the back of the psu, so strange yeah.

 

[Bill_Bright]

Pretty sure... ...don't matter which 8 pin connection they go into for CPU or gpu.

When it comes to potentially destructive voltages destroying expensive components, I don't have confidence with "pretty sure" doesn't matter!

Like you, I personally have not experienced problems in this area. But that mostly is due to dumb luck and good timing because I personally know other folks who did encounter these issues first. So this time, I managed to learn from the mistakes of those before me.

 

[RJARRRPCGP]

giving one long and three short beeps

For me, it 99 percent means I have to reseat the video card for the gazillionth time! The number one rule for me: Never bump the video card horizontally, or else you probably need to pull out the video card, reinsert it and make sure there's that satisfying click sound. (of the slot latch or slot clip)

Seemingly especially for AGP PCs, even moving the PC, can result in the above BS!

 

[Splinterdog]

For me, it 99 percent means I have to reseat the video card for the gazillionth time! The number one rule for me: Never bump the video card horizontally, or else you probably need to pull out the video card, reinsert it and make sure there's that satisfying click sound. (of the slot latch or slot clip)

Seemingly especially for AGP PCs, even moving the PC, can result in the above BS!

Thanks, but that wasn't the cause. I reseated it more than a dozen times.