Cpu and heatsink lapping.

[EcoGeek]

In short I'm wondering how effective lapping cpu's or heatsinks can be.

I'm just considering the process of actually a lapping a cpu and thinking that ok, I've seen it done on some forum or another a couple of times and it saved someone 2-5c after both cpu and heatsink is lapped or whatever, and all looks good etc.

But lets say I decide to lap a cpu for the first time (whereas someone posting on a forum may already have tried a couple of times), get what I need and go ahead and do it. Can I really guarentee that by lapping a cpu or a heatsink that i'll end up with an even surface?

We know machine cutting isn't perfectly accurate but assuming I was careful in doing this and measured the lapped surface using the chart regularly whilst working on it etc, how accurate is that likely to be?

Surely I could be 'just out' somewhere like a machine would be and end up with a just as inaccurately 'cut' surface?

Also, the majority of heatsinks out there now have polished surfaces. But I'm dubious as to if that means it's actually a good and accurate cut. Sure it shines, but does that prove it's an accurate cut and would it be as accurate as a properly lapped heatsink? In other words, should a 'polished base' heatsink be lapped anyway if one intends to lap?

Thanks for reading

 

[Mr.Scott]

Finish means nothing if it's not flat.
Lap on a piece of glass. Take your time. Use multiple grits of paper. Finish with 2000 grit wet and dry. If done properly you'll see a few degree decrease in temp. NEVER lap direct touch heatpipe coolers.

 

[EcoGeek]

Finish means nothing if it's not flat.
Lap on a piece of glass. Take your time. Use multiple grits of paper. Finish with 2000 grit wet and dry. If done properly you'll see a few degree decrease in temp. NEVER lap direct touch heatpipe coolers.

Thanks for the advise.

Is few degrees gain for lapping one or both?

What are direct touch heatpipe coolers? Not sure if I've seen them.

I use a Venomous X with a 'polished base'.. I guess lapping this would be fine, or would I "undo" a really fine/flat cut?

 

[Mr.Scott]

Is few degrees gain for lapping one or both?

What are direct touch heatpipe coolers? Not sure if I've seen them.

I use a Venomous X with a 'polished base'.. I guess lapping this would be fine, or would I "undo" a really fine/flat cut?

Results depend on how bad the surface was to begin with. I've personally had from a 0 decrease to over a 10c decrease. Your Ven X is ok to lap if you like.
Example of a direct touch heatpipe cooler would be like this.

 

[EcoGeek]

^ Interesting looking cooler. How are they?

 

[Mr.Scott]

Many different manufacturers use this type of technology now. It's pretty much the industry standard on performance coolers today. As with any other product, results vary pending the product.

 

[EcoGeek]

Ahh it's been a while since my last build, so I'm not up to speed on it.

Any opinions on GPU lapping?

 

[Mr.Scott]

Ahh it's been a while since my last build, so I'm not up to speed on it.

Any opinions on GPU lapping?

Yes, not worth the agony. Too much can go wrong.

 

[Tech Tweaker]

In short I'm wondering how effective lapping cpu's or heatsinks can be.

I'm just considering the process of actually a lapping a cpu and thinking that ok, I've seen it done on some forum or another a couple of times and it saved someone 2-5c after both cpu and heatsink is lapped or whatever, and all looks good etc.

But lets say I decide to lap a cpu for the first time (whereas someone posting on a forum may already have tried a couple of times), get what I need and go ahead and do it. Can I really guarantee that by lapping a cpu or a heatsink that i'll end up with an even surface?

We know machine cutting isn't perfectly accurate but assuming I was careful in doing this and measured the lapped surface using the chart regularly whilst working on it etc, how accurate is that likely to be?

Surely I could be 'just out' somewhere like a machine would be and end up with a just as inaccurately 'cut' surface?

Also, the majority of heatsinks out there now have polished surfaces. But I'm dubious as to if that means it's actually a good and accurate cut. Sure it shines, but does that prove it's an accurate cut and would it be as accurate as a properly lapped heatsink? In other words, should a 'polished base' heatsink be lapped anyway if one intends to lap?

Thanks for reading

I'd say it depends on the heatsink whether it "should" need to be lapped. Some manufacturers are known for making uneven "not flat" heatsinks that can benefit from lapping, but there are some others that are known for being very flat already and would have little to no benefit gained from lapping them.

When I lapped my first heatsink it took me about three hours, and it was hardly worth the effort. My idle temps dropped by 3°C, but my load temps were exactly the same as before. I've lapped probably three or four heatsinks now, usually with minimal improvements to each one. I might gain 2-3°C, but that's about it. Frankly I don't have the patience for a really first-rate pro lapping job, since I hear from folks that have done this a lot that a really proper job takes 4-5 hours.

What I do is get a flat piece of plexi-glass, tape it down to a table and then tape my piece of sandpaper to that and rub the heatsink against the sandpaper while making sure to keep the base of the heatsink flat against the sandpaper. Oh, and I wet-sand mine (get a bowl of water and occasionally put a few drops on the sandpaper), you also have to wipe down the sandpaper every now and then or it will "load up" with the bits of metal off the heatsink and the particles that have come off from the sandpaper. If you don't wipe off the sandpaper the lapping will do nothing with loaded up sandpaper as the heatsink just glides across the bits that are stuck between the sandpaper and heatsink (Think of it like hydroplaning in a car, the tires don't touch the road and just skid on the water. So here the heatsink rides on the bits left behind and doesn't make full contact with the sandpaper.). My grit progression is usually something like 220-320-400-600-800-1200-1500, or 220-400-600-800-1200. I've tried grits around 1800-2000 before, but I didn't really see any noticeable improvements past 1500 grit.

Yes, not worth the agony. Too much can go wrong.

Probably too difficult to be worth the attempt too.

At least with lapping a heatsink the surface you wish to lap is easily accessible. With a GPU you have to worry about capacitors and other onboard components getting in the way.

 

[EcoGeek]

Yes, not worth the agony. Too much can go wrong.

Ah, I expected it could be dodgy. I'll do just the heatsinks for my graphics in that case.

Any recommendations for a lapping guide?

 

[EcoGeek]

With a GPU you have to worry about capacitors and other onboard components getting in the way.

They're also not thick chips to start with, I guess that could be a problem

 

[EcoGeek]

I'd say it depends on the heatsink whether it "should" need to be lapped. Some manufacturers are known for making uneven "not flat" heatsinks that can benefit from lapping, but there are some others that are known for being very flat already and would have little to no benefit gained from lapping them.

[etc]

Nice update. That's some good info, thanks . Ideally if I were to do this for myself I'd want to get the most out of it and don't mind doing a little work some sunday afternoon or something, if there's something to be gained from spending more time on it.

 

[EcoGeek]

So, what grade of paper should I use and how many different grades should I give it for the best result? Does it vary based on wether or not you're lapping a cpu or a heatsink?

 

[Conumdrum]

Tech Tweaker said in his earlier post:


What I do is get a flat piece of plexi-glass, tape it down to a table and then tape my piece of sandpaper to that and rub the heatsink against the sandpaper while making sure to keep the base of the heatsink flat against the sandpaper. Oh, and I wet-sand mine (get a bowl of water and occasionally put a few drops on the sandpaper), you also have to wipe down the sandpaper every now and then or it will "load up" with the bits of metal off the heatsink and the particles that have come off from the sandpaper. If you don't wipe off the sandpaper the lapping will do nothing with loaded up sandpaper as the heatsink just glides across the bits that are stuck between the sandpaper and heatsink (Think of it like hydroplaning in a car, the tires don't touch the road and just skid on the water. So here the heatsink rides on the bits left behind and doesn't make full contact with the sandpaper.). My grit progression is usually something like 220-320-400-600-800-1200-1500, or 220-400-600-800-1200. I've tried grits around 1800-2000 before, but I didn't really see any noticeable improvements past 1500 grit.

 

[EcoGeek]

My mistake lol. It's late here