I just took apart a Slot A CPU to replace the TIM between the die/cache and the IHS, and I was wondering if there was perhaps a plastic clip or something I can buy and use in lieu of the original metal retention arms.
Thanks!
it sounds like you've borked the original metal one. i think your options are rather limited. its either fix the original clips, or do something drastic.
something drastic, would be to remove the 4 center studs around teh cpu core, and replace them with screws/springs.
It is a mistake to think you can solve any major problems just with potatoes.
luckybob wrote on 2021-05-16, 05:13:it sounds like you've borked the original metal one. i think your options are rather limited. its either fix the original clips, or do something drastic.
I mean, nothing too bad this time around. The clips on my Pentium 2s, though, uh, heh, they could be better off.
luckybob wrote on 2021-05-16, 05:13:something drastic, would be to remove the 4 center studs around teh cpu core, and replace them with screws/springs.
I neither have the tools nor knowledge to do such a thing, although perhaps springs and zip-ties (?) or something to push down and retain the springs could work.
Besides finding replacements for the clips, I guess I also could get a general water block and use some string to secure it to the CPU - even less ideal. 😒
yea, i suppose if you dont have a drill and a tap/die set. it wouldn't really work.
I bet, you could ask someone with a 3d printer to make a U-shaped v-wedge to grab onto the studs.
It is a mistake to think you can solve any major problems just with potatoes.
luckybob wrote on 2021-05-16, 05:56:I bet, you could ask someone with a 3d printer to make a U-shaped v-wedge to grab onto the studs.
Another good idea.
I've been doing a bit of googling and I think I've found a solution (springs + push nuts). However, testing it will take some time, as getting the springs from China will take some time. Will update on results when the springs arrive.
i've kept thinking about it.
E-clips, springs and washers.
stack a fiber washer, metal washer, spring, washer e-clip and see if that works
It is a mistake to think you can solve any major problems just with potatoes.
luckybob wrote on 2021-05-16, 06:22:i've kept thinking about it.
E-clips, springs and washers.
stack a fiber washer, metal washer, spring, washer e-clip and see if that works
Yeah, e-clips seem basically the same as push nuts except with a cut-out on the side to make them easier to remove(?). As for metal/fiber washers, that's a good idea. Will implement.
You may also find that some clothing type snapper fasteners will clip on top of the posts there, but not sure if you need more tension than they will hold without popping off.
I had a surfeit of PII 233s back in the day, and I drilled and tapped a S370 sink and put that on with machine screws. Was able to run it at ~290 then., wooo.
Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.
luckybob wrote on 2021-05-16, 06:22:i've kept thinking about it.
E-clips, springs and washers.
stack a fiber washer, metal washer, spring, washer e-clip and see if that works
It has been a while. Mainly due to the fact that the first springs I ordered from China were no bueno, so I had to order more. Also, the push-nuts I bought wouldn't cut it--eclips were the call. The final result used 0.5x5x5mm springs (shorter/thinner-in-diameter ones could in theory, if they exist, since they'd be some weird dimension) and M2 e-clips.
One downside is that the outside metal rods aren't the same as the inner ones, which have an inner gap for e-clips to "clip" onto. With 1mm thermal pads, with only the center four rods acting as retention mechanisms, this results in the CPU card doing a bit of a bend. 😒
I can't imagine it being too serious, and after some time I'd imagine the pads would compress... I hope. If I really need to, I could see if 0.5mm pads would work, but man, it was a pain to get those tiny-ass e-clips on. Would like to not do it again.
Id remove those pads on the cache chips. But it looks great!
It is a mistake to think you can solve any major problems just with potatoes.
the_ultra_code wrote on 2021-06-29, 11:24:One downside is that the outside metal rods aren't the same as the inner ones, which have an inner gap for e-clips to "clip" onto. With 1mm thermal pads, with only the center four rods acting as retention mechanisms, this results in the CPU card doing a bit of a bend. 😒
I can't imagine it being too serious, and after some time I'd imagine the pads would compress... I hope. If I really need to, I could see if 0.5mm pads would work, but man, it was a pain to get those tiny-ass e-clips on. Would like to not do it again.
0.5mm was too thin on mine. I thought about adding in some thermal compound to bridge the gap (similar to the thermal goop that was in there originally, which didn't actually make contact on one side), but I pressed some 1mm pads in a vice before using them and that seemed to do the trick.
Congrats, that looks great! I have to do this on my four Slot A CPUs, but kind of gave up/indefinitely postponed the operation seeing how hard it is to open the cartridges.
Regarding the bending of the CPU card, aren't you going to replace the plastic shell? I believe the four 'receptacles' for the outer metal rods are shaped in a way that pushes against the PCB, thus assuring it stays flat.
luckybob wrote on 2021-06-29, 15:40:Id remove those pads on the cache chips. But it looks great!
How hot do the cache chips get anyways?
quicknick wrote on 2021-06-30, 01:15:Regarding the bending of the CPU card, aren't you going to replace the plastic shell? I believe the four 'receptacles' for the outer metal rods are shaped in a way that pushes against the PCB, thus assuring it stays flat.
Nah.
Quick follow-up: installed the CPU into my seemingly NOS Asus K7M, and it works fine as far as I can tell. Only issue is CPU temp in-BIOS is incorrectly being reported at 0C, even after some MemTest86 (the heatsink is hot, but I guess that makes sense when the system is chugging 90W from the wall). Is the CPU still not hot enough for the motherboard/CPU sensors to list an accurate value, or could I have damaged the CPU in some way for the sensors not to work? Anyone know where the sensors are anyways?
Do these chips have a on-die thermal sensor? I dont know. Because if its zero, it's usually disconnected.
It is a mistake to think you can solve any major problems just with potatoes.
There's a two pin header you connect a sensor to right next to the voltage jumpers. The sensor with leads was not (edited, sry forgot this important word! 😅) included with the mobo and would usually be taped close to the center of the heatsink to get the closest approximation of the core temp.
yea, thats how i remember things being. P3's were the same way.
i have a few processors where I placed a 10k-ohm thermistor next to the cpu core and plugged it into those sensor connectors.
It is a mistake to think you can solve any major problems just with potatoes.
https://www.omega.com/en-us/temperature-measu … /TFPS-Th-Series
they look like this. Obviously they can be found elsewhere, cheaper, from the usual skumbags.
It is a mistake to think you can solve any major problems just with potatoes.
Doornkaat wrote on 2021-07-04, 06:49:There's a two pin header you connect a sensor to right next to the voltage jumpers. The sensor with leads was not (edited, sry forgot this important word! 😅) included with the mobo and would usually be taped close to the center of the heatsink to get the closest approximation of the core temp.
luckybob wrote on 2021-07-04, 06:51:yea, thats how i remember things being. P3's were the same way.
i have a few processors where I placed a 10k-ohm thermistor next to the cpu core and plugged it into those sensor connectors.
Wow, really? That's kinda disappointing. I know my Slot 1 Pentium IIIs and even Pentium IIs have on-CPU sensors, so for AMD not to have on-CPU thermal sensors seems strange. Guess AMD was a bit behind in tech.
Maybe the AMD cpus just never needed a thermal sensor.
There's a glitch in the matrix.
A founding member of the 286 appreciation society.
Apparently 32-bit is dead and nobody likes P4s.
Of course, as always, I'm open to correction...😉
