lapping did jack all for me

T

THRESHIN

2[H]4U
Joined
Sep 29, 2002
Messages
3,622
i spent a couple hours last night lapping my new AC freezer 64 (it was cheap, OK?!?) to get all the nasty machining marks off it and even lapped the top of the IHS of my opty 165. what an utter waste of time! i dropped maybe 1-2ºC tops, probably less! did i do something wrong? lol.
 
nope, thats about all its good for my man.
waste of time IMO.
 
It really depends on how well it was lapped and how bad it was before it was lapped.

I used an electric orbital sander to lap both my heatsink and CPU. Dropped about 3-5c under load. Hand lapping takes way too long and isn't worth the extra effort to get the same exact results.

The temps were still too high and not even enough on my Opty 175 so I then removed the IHS.

Dropped another 15c and made the core temps a lot more even.
 
Unless your cooler is significantly f'ed up your not going to see anything, thermal paste will fill the gaps for the most part
 
Hand lapping can produce invisible concave surfaces because of the uneven moment of your hands.
 
THRESHIN said:
i spent a couple hours last night lapping my new AC freezer 64 (it was cheap, OK?!?) to get all the nasty machining marks off it and even lapped the top of the IHS of my opty 165. what an utter waste of time! i dropped maybe 1-2ºC tops, probably less! did i do something wrong? lol.
Click to expand...

It depends on how rough the surfaces are to begin with... if both are relatively flat, you won't see a whole lot of delta temp. I dunno about opty chips or your ACF64, but I knocked 7-10 °C off my temps by doing the IHS on my q6600 and the base of my ultra-120 ex. That said, both components are known to be irregular. In fact, C2D/C2Q chips are notoriously concave. The official word from TR is that the U-120/U-120X are both "designed" to be convex as a result.

Anyway, how did you measure the flatness of both surfaces? Also, did you apply the right amount of TIM and what are you using? If you're using AS5, beware that there is a break-in period of 100-200 hours. I saw a 4-5 °C drop after the first 24 h and then an additional 2-3 °C after about 300 h.
 
graysky said:
It depends on how rough the surfaces are to begin with... if both are relatively flat, you won't see a whole lot of delta temp. I dunno about opty chips or your ACF64, but I knocked 7-10 °C off my temps by doing the IHS on my q6600 and the base of my ultra-120 ex. That said, both components are known to be irregular. In fact, C2D/C2Q chips are notoriously concave. The official word from TR is that the U-120/U-120X are both "designed" to be convex as a result.

Anyway, how did you measure the flatness of both surfaces? Also, did you apply the right amount of TIM and what are you using? If you're using AS5, beware that there is a break-in period of 100-200 hours. I saw a 4-5 °C drop after the first 24 h and then an additional 2-3 °C after about 300 h.
Click to expand...

Very true. I lapped both my Q6600 and TR Ultra-120 E. The Ultra-120 was actually fairly flat, but the Q6600 had two concaved spots. I ended up dropping 7C load / 10C idle. I stopped at 1000grit and did not use any sort of polishing compound.
 
well i just used some very very high grit automotive finishing sandpaper, water with a touch of dishsoap for lubrication, and did the lapping in circular motions using a machined flat surface - an old project from machine shop at college actually.

i also lapped the cpu and found it was slightly concave. and yeah i measured the temps right after installing it, used AC MX-2 tim...meh. supposed to be about the same as AS5. i havn't removed the IHS yet because i don't want to modify anything on the heatsink to keep good contact - probably going water in the spring and since its going to be bolted down, thats a non issue.
 
I used an electric orbital sander to lap both my heatsink and CPU.
Click to expand...

:eek:

Dude, you "sanded" them you certainly did not lapp them in the conventional sense. But I bet it was quick and if they were off a lot you probally came out better than I did. /boggle

anyhow, its all crapshoot, as mentioned before just depends on how bad things were to start. Ripped my machine apart to lap the CPU on lapidary (jewelers) equipment as I had done the water block before installation but did not do my CPU. It was obvious right off the IHS was not close to flat. Long story short, didn't make a single deg of difference that I could measure. But here is the key thing, I feel much better now. I no longer lay awake at night wondering about it. A calm has descended over my life now that I know everything is as good as I could make it. :p God loves me again.
 
anthongy817 said:
Very true. I lapped both my Q6600 and TR Ultra-120 E. The Ultra-120 was actually fairly flat, but the Q6600 had two concaved spots. I ended up dropping 7C load / 10C idle. I stopped at 1000grit and did not use any sort of polishing compound.
Click to expand...

Glad to hear it worked for you; those are some excellent temp reductions for the small amount of money you spent on the sandpaper, eh?
 
Get both surfaces perfectly flat then use no TIM and I guarantee your temps will drop :)
 
I would say not at all a joke. If your surfaces are flat and even enough so that they are flush, metal on metal has a much better heat transfer than thermal material. This is something that should only be done with professional/industrial equipment though.

One of the most important things TIM does is keep the air out. With truly flat surface mating there would be a minimum amount of air and better heat conduction.
 
Right, but even lapped surfaces will have major unevenness when magnified, although they look quite flat to the naked eye. The TIM bridges these microscopic gaps, it is certainly needed even when working with lapped surfaces.
 
tcraig_4096 said:
I would say not at all a joke. If your surfaces are flat and even enough so that they are flush, metal on metal has a much better heat transfer than thermal material. This is something that should only be done with professional/industrial equipment though.

One of the most important things TIM does is keep the air out. With truly flat surface mating there would be a minimum amount of air and better heat conduction.
Click to expand...

To run a TEC without TIM, a mounting force of 250lbs is recommended... Good luck ;)
 
i put a scratch in my E2180 when I was trying to apply thermal paste. Would that screw my cooling over? would lapping fix this?
 
tcraig_4096 said:
I would say not at all a joke. If your surfaces are flat and even enough so that they are flush, metal on metal has a much better heat transfer than thermal material. This is something that should only be done with professional/industrial equipment though.

One of the most important things TIM does is keep the air out. With truly flat surface mating there would be a minimum amount of air and better heat conduction.
Click to expand...

As mentioned this is not for normal people, I use a gem fauceting machine to achieve extremely flat and .05um surface roughness but still use a hazing of AS5 on both surfaces. LOL and as mentioned above sometimes I see an improvement and sometimes I don't. /shrug.

BTW to truly "Lapp" the IHS and the Heatsink/block should be lapped together, with a super fine abrasive (commercially crushed walnut shells available as an polishing powder, believe it or not, is excellent) between the two parts and rubbed against each other. This removes machine and setup tolerance errors. If done properly the two part will fit together so well, again as mentioned above because of air pressure, that they will "stick" of their own accord and considerable force required to separate the parts.

All very interesting, but just get it as flat as you can and don't worry about it too much.

And of course there is another completely different school of ideas on (not about flatness that is pretty much agreed on) but surface roughness. IBM has released research/papers that a precision, on purpose, "pattern" of roughness can improve heat dissipation.

so its all good.
 
You must log in or register to reply here.
Back
Top