Overclock.net banner
1 - 20 of 27 Posts

RR.Bear

· Registered
R.S.M.
Joined
·
159 Posts
Discussion starter · #1 ·
Hello friends. I want to remove the cover from my 7600x. It's not difficult for me, but the problem arises how can I fix it in the socket ? I can't buy a frame from a thermal grizzly, I'll have to do everything with ordinary household tape. I don't have any problems with the mounting screws for the water block, but I'm worried that the processor cores will be lower than the socket frame, although I'm not sure about this. I already stabilized the memory at 6400 cl28-37-37-30-67-384 and I'm happy with the result. But the processor at a fixed voltage does not want to work above 5600 MHz when overclocked across all cores @ 1,25v
Thank you and have a nice day 🥰💦
 
You need a frame, or remove the mounting/loading mechanism and use the mounting pressure from the block to press your chip against the pins. Can be done but it is NOT recommended especially with AMD's multiple off center chiplet setup and without the IHS, the package loses its rigidity, meaning you put slightly too much pressure and the chip's PCB will deform. Yeah, you need a frame.
Or maybe you can find a cad drawing somewhere and let someone with a 3D printer print one for you.
 
Hello friends. I want to remove the cover from my 7600x. It's not difficult for me, but the problem arises how can I fix it in the socket ? I can't buy a frame from a thermal grizzly, I'll have to do everything with ordinary household tape. I don't have any problems with the mounting screws for the water block, but I'm worried that the processor cores will be lower than the socket frame, although I'm not sure about this. I already stabilized the memory at 6400 cl28-37-37-30-67-384 and I'm happy with the result. But the processor at a fixed voltage does not want to work above 5600 MHz when overclocked across all cores @ 1,25v
Thank you and have a nice day 🥰💦
I don't think you gonna gain anything from direct die mount. Delid, apply LM and put IHS back on.


Also you should watch this before messing with it -
 
I am not dismissing direct die generally. Intel does have bigger die, it's easier to transfer heat. Ryzens are heat dense, it's harder to get rid of that heat. I am not against direct die cooling, but it's likely not worth it. Especially if you want to run it hassle free for day to day use.
 
Discussion starter · #7 ·
I'm in process of working on my 7950x direct die and it's a pain in the ass. Definitely get the frame, it's one less variable you'll need to worry about.
the whole problem is that I physically cannot buy a thermal grizzly frame. I can buy a frame from thermalright but it is not suitable for direct contact. I will act with household tape as it was at 4790k in distant years. The processor is scalped with an iron and dental floss in 20 minutes. Solder residues are cleaned with liquid metal quickly enough

 
BZ is entry level though. I wouldn't dismiss direct die based on his results alone. Unless AMD behaves different, Intel does better direct die compared to LM and IHS.
This. Only takes a few minutes watching Der8auer delid and test showing massive temperature drops on 7900X.

OP, I must have missed it but why can you not purchase a thermal grizzly frame?
 
Mounting with tape won't be ideal, I did not say that direct die does not work. But without mounting bracket there might be some issues. With that aside Ryzen core chiplet is offset to the side, so you better have water block that accounts for this. Or else it will be harder to cool without heat spreader.
 
I would try it with liquid metal and the IHS first as you will likely get most of the benefits that way with less risk and difficulty.

Otherwise you can remove the frame and get some hardware to mount very carefully. The AHOC video above gives some clues. I think I would try to get some springs to give it constant pressure.
 
I would try it with liquid metal and the IHS first as you will likely get most of the benefits that way with less risk and difficulty.

Otherwise you can remove the frame and get some hardware to mount very carefully. The AHOC video above gives some clues. I think I would try to get some springs to give it constant pressure.
Once I finally got my block mounting sorted I went from 93c at my full out 220ppt PBO curve with LM on the heatspreader to 79c on direct die so there is a pretty significant drop. Getting the block right however was a pain in the ass. It's an EK Velocity with the AM4 mount plate. First I tried to thread M3 screws from the underside through the backplate and use the stock EK springs and thumbscrews. I think that would have worked If I used shorter screws but I was working with what I had on hand.

what I ended up doing use 4x 30mm UNC 6-32 screws with some matching 5mm nuts and threaded those all the way to the top of the screw to give it more of a thumbscrew look, put the stock EK springs on screw and a washer under that and threaded them into the AM5 backplate.

I got it relatively even, then fired up CB23 on a loop and adjusted tension until both CCDs and the IOD were showing effective cooling. I was really worried about the caps above the socket but I've got at least an mm of clearance under the bracket.

All and all, getting even tension was a massive pain in the ass but now that I'm done, I'm pleased with the results.

Automotive tire Hood Automotive design Gas Machine
 
Heating it with a 170 degree iron for a few seconds seems pretty safe.
Some people are afraid of this method, but I'm convinced it's one of the safest.
They sold a lot of 11900K,12900K,S after baking them in an electric oven at 170 degrees for 25 minutes, and they haven't had any problems since.:p
Product Sleeve Gesture Automotive design Finger
 
I will use fehonda thermal pads. They are quite dense
with pushtrough screws, or like on the chip itself ?
I'm a bit worried about oil spill by them. Its non conductive, but - its not really the place to put them

Frame looks like costs ~4000 RUB

EDIT:
Cardboard on the corners, would actually be much more reasonable🤭
 
Discussion starter · #19 ·
I've used them on a GPU and they work great. They don't leak oil. I think to combine this and cardboard as you suggested.
I did something similar on the 4790k and it worked great.
iron, dental floss thermal paste some liquid metal and cardboard. Everything you need for happiness
 
Personally, outside of all
I think delidding it
sanding CCDs down 100 microns - sanding down IHS feets 250-280 microns and putting it back on
would do you better

i am only not sure how much pressure will be there with the stock mounting system
but then also am sure that screws do bottom out
soo maaybe with one 80-100 microns washer on the screws (soft ones) ~ it should come out together nicely

EDIT:
Solder is around 270-300 microns thick
feet height is around 150ish *
and CCDs can lose at least 100 microns, if not 150
Only not sure with X3D , but non ~ can

soo you got around 250microns air gap between CCDs and inside cover ~ when IHS is flat on the CPU
I personally plan to use my phasechange pads, but not entirely sure between pure direct die or with some copper shim :)
could be LM on copper plate on phasechange pad
unsure still ~ but it needs some surface to destribute/soak heat into
 
1 - 20 of 27 Posts