[nick10840]

I am thinking about lapping my h70... actually I'm more than just thinking about it. I already bought 600, 800, 1000, and 1500 grit sandpaper and I've watched many videos and read multiple guides on how to lap CPUs. I couldn't really find any guides/tutorials for lapping a h50 or h70 (I assume it's pretty similar).

My only question is how many reps per rotation and how many rounds per grit should I do? The video that everyone links for CPU lapping says 30 reps per rotation and 3 rounds per grit. Would this apply to a h70 as well?

The reason why I ask, it seems to me that my h70 might need less reps/rounds since I'll already be starting on the copper. So if you had good results with lapping your h50/h70 and you remember how many reps per rotation and how many rounds per grit you did I would greatly appreciate you sharing that information with me. Thanks!

BTW: I'm doing this because I have quite a few scratches on my H70 and I think it will help.

 

[Rick Arter]

I have lapped a H50 in the past and personally do 20-25 per side then turn and do it equally until it looks uniform. Then move onto the next grit up it takes a bit but after a while you will get the hang of it. Since you are already starting with copper base I say start with 20 per side then rotate and check after you have gone around all sides once. If not looking the way you want and not fully flat yet keep lapping. Main thing it to use the weight of the part and not push down as you might cause a none true surface for the block to mount to CPU surface. Hope that made sense I am not the best with explanations sometimes sorry.

 

[Caleal]

I lapped my H70, but it had 0 impact on performance. Mine wasn't scratched to start with though.

Lapping my CPU made a pretty significant difference though.

 

[nick10840]

Well I did it and it passed the razor test and all but I only saw about maybe 2 C drop. Maybe it has to do with the "break in" period of AC5 but I don't know if that is really real. If such a thing exists then that is the only difference because I tested my non lapped temps about 2 hours before testing the lapped temps. My air conditioner was on and the ambients were the same. I guess it would have been more scientific of me to apply new TIM before testing my non lapped temps but oh well.

Either way at least I didn't ruin the copper plate... and it's shiny lol... I wanted to keep it light. So I ended up doing 15 per rotation on the first two rounds and then I only did 10 per rotation on the third round because I didn't feel like I was getting any flatter. So I just did the same thing on all the following grits (600, 800, 1000, then 1500). Sorry no pictures I have a cheap old phone.

Thanks for the responses!

 

[Caleal]

Yeah, my H70 was pretty darn flat, all lapping it really did was give it a smoother finish.
My CPU was somewhat less than flat though.

 

[staryoshi]

I would have opted for a different thermal paste... AS5 had it's day in the sun but new technology has surpassed it. 2C is an improvement at least