Question Noctua NH-U14S making CPU/RAM malfunction - disables DRAM slots 2 & 4 due to weight of radiator

[giorov]

I put a build together with Gigabyte Aorus Z370 Gaming WIFI, an i7-8700k and a Noctua NH-U14S and the board won't get past the DRAM check, with the light going back and forth between CPU and RAM with a click sound -- but here is the twist:
if I take a RAM stick out of slot 2 or 4 it doesn't happen and goes to post; if the tower s set on its side though, slots 2 and 4 work... it also posts with both sticks with the tower oriented vertically, only if I hold/press the radiator up!

My conclusion is the heat sink is probably pulling the CPU out of the socket slightly at the top, so causing the CPU connection to RAM slots 2 and 4 to fail.

Has this ever happened to anyone?

I don't know if something could be RMA'd like the CPU or MoBo due to loose construction (I tightened everything as much as possible and made no difference), or if I should give up on the Noctua and try a be quiet! Black Rock Pro 4 or something. I want to keep this MoBo since it's the best for setting up a Hackintosh.

I started the RMA process with Gigabyte so I will see what their people say too.

This is so frustrating. Is it just poor socket design or is it a manufacturing defect in your opinions?

Thanks!

 

[alexoiu]

Can you post a picture of the CPU socket, to check the pins?

 

[taeioum]

Did you get a CPU fan with your 8700K? If so, try that one. I doubt the weight of your cooler is the issue. I've had a Noctua NH-D15 which is at least a third heavier than your U14S and never had an issue. I doubt the weight is causing some pins on the motherboard to be unable to connect to your CPU.

 

[giorov]

CPU came with no fan.

The system posts fine when I have the radiator and fan installed vertically and the tower laid down on its back, but as soon as I put the tower vertical again the system freezes. Funny, if I remove the fan from the radiator the problem doesn't occur when installed vertically (problem persists when radiator is installed horizontally, fan or not, even with tower on its back, although if I loosen the four bracket screws it does go to post when on its back).

Weird or what?

A Gigabyte person said they might not find any issue if they test with a lighter cooler and just said to try with other smaller coolers to see if the issue persists before trying for an RMA...

At least Gigabyte design people know I had the problem for their future reference. Now what, is the question.

At this point I'm thinking of trying for a be quiet! Dark Rock TF which connects to the back plate at four points instead of two. It is probably light enough for the issue to not present itself
He did say a possibility was to try a different processor, but I'm not going to just buy a socket 1151 to test in the board myself and send it back.

Annoying though... I wonder if maybe the socket is slightly misaligned and could unscrew it with a large star-shaped screw driver to adjust the position... Seems risky.

I sent a message to Noctua Support to see what they say before I send it back to Newegg.

 

[giorov]

 

[alexoiu]

Pins look ok. Try reinstalling the CPU and CPU cooler. See if still the same.

 

[giorov]

Yeah, I've reseated the CPU and cooler several times to no avail.

 

[alexoiu]

Then yes, RMA-ing the board might solve the issue.

 

[giorov]

Here is what Noctua said:

"The cooler can't pull out the CPU of the socket when properly installed, especially not on a LGA socket, where the CPU is held firmly in place by the retention mechanism of the socket. I suspect that either the backplate is making contact with the mainboard tray, which could lead to the symptoms you described. You could try to insulate the backside of the backplate with a some sticky tape and see if it helps.

The other possibility is that there is indeed a hairline crack at the motherboard where the little flexing of the PCB in the typical orientation of a tower chassis in combination with the weight of the cooler makes the difference."

I tried separating the SecuFirm 2 back plate with a plastic layer and nothing changed.

By the way here is what I meant by misaligned socket back plate...

Finally, here are some pockmarks I noticed to the right, bottom of the socket. do you think it could be a manufacturing defect that causes the problem to happen when the MoBo flexes?
Do you think I should try taking a look beneath the socket for cracks or adjust the socket location to be more central?

 

[giorov]

Yeah, RMAing would solve the issue but I need to convince them to try a cooler which is heavy enough to make the problem manifest itself. No easy feat, considering it doesn't even happen if I take the fan off the radiator. Maybe if I speak to a manager or something...

 

[Karadjgne]

Well here's the thing. You aren't the first to use that cooler and mobo combo, so unless there's something you did wrong in mounting the baseplate, I'm doubting that's an issue.

More likely it's a cracked trace around the socket and under one orientation the differential in weight distribution is putting a hair more pressure on that crack. Take fan out, lowers weight just enough close the crack.

I'd rma the mobo. And make sure the cooler is tight, testing possibilities by loosening the backplate again is just asking for trouble.

 

[giorov]

Yeah I've definitely tested screws all the way tight on both the four bracket screws and the two radiator mounting crews. I'm going to see if I can get them to test a cooler that is of the appropriate weight. I mean it's supposed to work so it should technically be in the scope of their testing requirements for the RMA process.