Lapping a heatsink:Whos done it? Was it worth it?

[RSV_Ecosse]

I'm at a stage in my PC building, where I'm thinking of ditching air cooling and pushing the boat out a bit, going either down the road of watercooling, or the "whole hog" and treating myself to a VapoChill.

However, before I do that, I'd like to try lapping my HS, which is a Thermalright SLK900A.

Anyone done this recently? Did it drop your temps a little, or were you disappointed by the results?

I notice the Kustom guys sell a kit for £8.00, which I think I'm going to order, and try out.

[Slamdog]

i dont think that you need to lap an slk.. they are made pretty well anyway....

if you were talking about coolermaster hsf or akasa ones then yep, probably worth it...

[Sticky]

Yup, it's a winner matey. Definately.

Back on air cooling I lapped all my heatsinks, both Intel stock and Thermaltake Spark H/S. There was a good chunk of a drop in temps, around 5 degrees if I recall (Intel P4 Northwood core).

I used Artic Alumina (not electrically conductive) for the thermal interface, and it dropped temps a further 2 degrees after a few days "burn in" :p

I still wasn't satisfied, and basically because I can't leave well enough alone, I lapped the P4 heatspreader on the chip itself. Not as scary as it sounds, but only do if you are prepared for the consequences, i.e. shafted CPU. Worked a treat, more temp drops (a further 3 or 4 degrees I think it was).

I stopped short of popping off the heat spreader and gently lapping the core itself (I reckoned I would be asking for trouble at that point). Although modern H/S have a good finish, I would never do an install without lapping now, if nothing else it gives you a comforting feeling that ye've done yer best.

I'm on a Prescott core now with water cooling, but same goes, lap the water block and take care of the interface with Artic Silver (careful with excessive use!). Contrary to what a lot of people will have you believe, Prescott cored P4s are excellent chips. With a stock cooler, ye'd be as well with a sock between it and the CPU!

Go for it I say. Start with the Heatsink itself, and give it a right good going over, if it takes you less than an hour, then there's more work to be done ;)

[RSV_Ecosse]

I think I'll give it a go.

My SLK's base isn't as good as I would have expected, in fact, its not as good as the Thermaltake Volcano 11's base that I replaced with it, so I'm hoping it will help a little. I'm idling at 45-47 deg on an Athlon 3000, with a Tt Smartfan II, and under load it can reach as much as 50-52. Case cooling in general, is as good as I can really get it, 2 80mm Akasa's as intake and two as exhaust, so I'm looking at other ways to lower the temps.

[Big Adam]

Sticky,

If lapping dropped your temps by 10 degrees, you're either talking in Farenheit or your original surfaces must have had a profile akin to the Himalayas!!!

:eek:

[Sticky]

.... a profile akin to the Himalayas!!!

You should see what Brad Pitt managed with a Celeron 300A, he had seven years in Tibet........



If lapping dropped your temps by 10 degrees, you're either talking in Farenheit or .....

Anyway, hmmm.

Not quite sure how to take that. Questioning my integrity, a genuine query, or a mark of respect to my ability?

Sure, adding the numbers would seem to indicate quite a temperature drop....... and it was. What you haven't factored in was that this was on a retail Northwood 1.6A overclocked to hell (Kustom guys can confirm this), initially hiked up with std heatsink and the horrendous Intel TIM. Neither was up to the job, and it held back my progress in cranking the CPU.

Next up we slapped on a finely lapped Thermaltake Spark copper H/S with modified fan assembly. That did indeed drop the temps by 5 degrees C. And in the period of the next few days, the idle temps dropped by another 2 degrees, just as I had been assured they would. Top stuff at this point, was looking good.

Not entirely satisfied, I asked myself what was the point of lapping one side of the interface, and not the other? The P4 heatspreader has raised lettering on the top, clearly the Alumina would be functioning like tile grout.

I wanted a surface finish that could rival slip gauge adhesion! So, off with the Thermaltake and we lapped the P4 heatspreader. Don't get me wrong, it took some time, and I did think I had drowned the poor Northwood in oil and ink. But he soldiered through, and made it...... in style. The surface finish was just as I had hoped for, and I could use a lot less Alumina between the two interfaces. A further 3 degree C drop was observed.

The 1.6A was retired earlier this year after some sterling work, and placed in a 2nd machine ............ at stock speed :eek:

So, as I have detailed, it can be done, and all with reference to the fine work by Anders Celcius (1701-1744) ;)

[reaper]

I have to back my good mate Sticky up on this one! :)

Following his "wet & dry" antics I was intrigued by his reduction in temps and vowed to give lapping a go myself. At my first attempt I was loathe to risk my chip itself (P4 2.4B) so stuck to lapping the stock Intel heatsink itself. After a fair bit of perseverence I attained quite a reasonable finish. The heatsink was refitted with Arctic Silver, and after settling in for a few days, I observed a reduction in temp of almost 5 degrees! :p

After some further "persuasion" from Sticky, I later took the CPU itself out and gave it a very mild lap as well. As he has mentioned, and you probably all know anyway, there is writing etched on the surface of the CPU itself. All I did was basically flatten this off a little to inprove the mating surface. I have to admit that during this process there were all sorts of "What on earth am I doing sandpapering my £140 chip??" thoughts running through my head. :rolleyes:

Post the mild lap of the CPU itself, I observed a further 1-2 degree drop in temp.

All in all, I have to say I believe there are worthwhile gains to be had from lapping. However, it should only really be attempted if : -

a) You can afford to put right any unfortunate outcome.

and

b) You have big enough nads to try it in the first place!! :p

[Big Adam]

Next up we slapped on a finely lapped Thermaltake Spark copper H/S with modified fan assembly. That did indeed drop the temps by 5 degrees C.
Which only reinforces my point. Your biggest temp drop was quite probably due to the change in HSF.

Don't get me wrong, lapping does have it's benefits and I'm not denying that lapping both mating surfaces dropped temps significantly......

Just that your original post implied the process alone resulted in a 10 deg drop...which is misleading. :)

[daniel_owen_uk]

I lapped my waterblocks, although not really same thing.

Can't beat a bit of autosol to finish with either. What grit does the kustom kit go to?

[reaper]

Your biggest temp drop was quite probably due to the change in HSF.


All I can say is that I obtained at least a 6 degree drop in temp by lapping the stock Intel cooler and the surface of the CPU.

In my mind that is a worthwhile improvement for minimal cost.

Oh, and it gives you an enormous sense of well-being! :p

[Sticky]

Which only reinforces my point ....

By implication, you are referring to your first post....... and erm, where was the point that ye were reinforcing :confused:

What ye have reinforced was that yer original post was indeed questioning the validity of the temperature drop. I post to help others where I can, and usually don't where I can't. I didn't post the above so someone could basically post "yer ar$e" immediately after it (again, implication?). Sure, it's a forum, so feel free to do so, but please do not say my post was misleading, when it, erm, clearly wasn't :)

a) 5 degrees from intro of the lapped HSF
b) 2 degrees after settling in with the Arctic Alumina
c) 3 degrees with lapped heatspreader

Read into that how ye will, but the point I was trying to make was that I personally don't see the point in spending a large amount of time, money, and effort when installing a new cooling solution to only see 60/70% of the potential gain, especially under heavy overclocking. The importance of a good HSF is clear, but why not take the time to prepare both mating surfaces prior to installation? And that was the reason for my response to RSV_Ecosse.

Sure, not everyone has the patience, manual dexterity, or "nad size" (reaper yer a shocker) to lap both surfaces. It's messy, potentially costly, mistakes can easily be made, and more importantly it delay's the commissioning of the rig. However, it does not make sense to ignore the interfaces. The manufacturers simply do not have the time to provide the surface finish that can be achieved after a bit of hard work by the end user. The surfaces on the HSF and CPU heatspreaders are neither smooth nor flat. Again, one might as well fit a sock between the interfacing surfaces if one was to ignore that :p


I lapped my waterblocks, although not really same thing....

I totally support that, that is exactly what I am getting at. The quality of the surface finish on the machined waterblocks can vary, so taking a little time to prepare the surfaces prior to installation makes perfect sense.

RSV_Ecosse stated in the first post that he may try water cooling next. And I'd offer the exact same advice at that point too.


Anyway, Big Adam............. what temperature drops did you see when you lapped your CPU heatspreader/core and what methods did you use............. just for comparison, like.

[RSV_Ecosse]

All interesting stuff fellas.

I'm certainly going to try it out, at least on the SLK, I've not got big enough "nads" to try lapping the Athlon though. :eek:

[Big Adam]

Sticky,
Please understand that I am not trying to argue the toss. The fact that lapping improves the mating interface between CPU and HSF thus improving heat transfer is a given!

I'm not even arguing against the fact your lapping skills may be be worthy of a medal from the Queen.

All I was merely pointing out is that under normal circumstances, lapping will NOT result in a 10C temp drop.

As your second post plus the quote below clearly confirms, your initial 5C drop was due to upgrading from a stock Intel HSF to a lapped TT Spark HSF.

a) 5 degrees from intro of the lapped HSF
b) 2 degrees after settling in with the Arctic Alumina
c) 3 degrees with lapped heatspreader

Now, I'm not saying that 10C is not achievable if comparing like-for-like HSF units. Just that anyone considering lapping their kit for the first time should not be expecting to achieve that kind of result. 4-5C is a good target.

I'm not trying to inflame things, just trying to present a balanced and realistic viewpoint. Feel free to respond but I have said all I need to say. :)

[Sticky]

Sticky,
Please understand that I am not trying to argue the toss.

Fair play :)




I'm not even arguing against the fact your lapping skills may be be worthy of a medal from the Queen.


Knight Grand Cross in The Most Excellent Order of the British Empire for m@d l@pP1ng sk1llz ...... no? :p



I'm not trying to inflame things, just trying to present a balanced and realistic viewpoint. Feel free to respond but I have said all I need to say. :)


Aww :(

[daniel_owen_uk]

I totally support that, that is exactly what I am getting at. The quality of the surface finish on the machined waterblocks can vary, so taking a little time to prepare the surfaces prior to installation makes perfect sense.

Not just that but if you look at the way the slk is made its not actually flat copper at the base, its fins pushed together, so you are always gonna get gaps between the fins no matter what.

Waterblocks are solid copper on gaps for fins on base, so lapping makes sense.

[pdf27]

Just a warning: not all waterblocks will be improved by hand lapping. Some (Little River ones spring to mind) are machined to such a high level of flatness anyway that hand lapping will only make things worse. Others are very crudely machined - so check what the base is like before lapping.

[jnm21]

I found the SP97 to be quite poor with regards finish, so I did lap it.

[Sticky]

Waterblocks are solid copper on gaps for fins on base, so lapping makes sense.

Not sure what you mean? I know the SLK900A construction, you saying you can't improve the factory finish on a SLK900A H/S :eek:

And what waterblock do you mean has fins on the base? Ye've lost me a bit :confused:

Both the SLK900A and solid copper waterblocks will respond very well to effective lapping.



Just a warning: not all waterblocks will be improved by hand lapping. Some (Little River ones spring to mind) are machined to such a high level of flatness anyway that hand lapping will only make things worse. Others are very crudely machined - so check what the base is like before lapping.

Hmm, hang on here.

a) I could be wrong, but aren't Stew Forster's blocks already hand lapped?
b) Given the above, if the surface finish (note: not the shine!) is genuinely good, then unless confident of ability, then you wouldn't want to lap it.
c) The actual result is very subjective, different people will attain different results depending on effort, skill, materials used, experience etc.

Effective lapping takes hours. It is not a simple case of making it shine. The geometry of the surfaces should be addressed for optimum heat transfer.

[RSV_Ecosse]

And after all this discussion, I go and attempt to order one from the main site and they're out of stock!!!

Finding it difficult to track one down anywhere tbh, I suppose Halfords is an option though. :cool:

[pdf27]

a) I could be wrong, but aren't Stew Forster's blocks already hand lapped?
I think so, but they are deliberately not shiny. Main point was that lapping everything isn't a good idea - you can't just assume you will be able to lap better than what you already have.

[jnm21]

Didn't like to say, but if it's outta stock, how about B&Q - various sand papers then wet'n'dry paper - dry first then (a little) wet to finish.

[Sticky]

.... you can't just assume you will be able to lap better than what you already have.

If the surface finish on the component is poor, of course you can :)

Lapping a heatsink, waterblock, heatspreader, or even the top surface of the core (steady) is merely one of the many methods that can be used to get more out of the components you install. There is nothing ground breaking being discussed here, it's basic engineering principles we are talking.

Most manufacturers simply can't take two hours or more to hand lap each component. An end user can afford that time if so desired, indeed if at all required.

Fact is, if the thermal interfaces do have surface imperfections, rectify them by the means already discussed. Get busy with the tools in the workshop and do justice to the rig that is being built.

The alternative is to be content with stock components at standard speeds. Now, there's nowt wrong wi that............ it all just sounds a bit dull to me :(

Edit: typo

[pdf27]

Most manufacturers simply can't take two hours or more to hand lap each component. An end user can afford that time if so desired, indeed if at all required.
True, but depending on the manufacturing method used the surface may well be flatter than you can get by lapping anyway. One place I went for interview a tolerance of +/- 1 micron was considered unacceptable - and for a person to get a block both flat AND smooth to that sort of tolerance is IMHO unlikely. It's probably possible with practice if you can measure flatness, but most people don't have a chance of measuring to that accuracy.

[Sticky]

.... the surface may well be flatter than you can get by lapping anyway.

Woah there! Have you also considered that it may well not be? :)


.... for a person to get a block both flat AND smooth to that sort of tolerance is IMHO unlikely.

It's also entirely possible that there's no chance a stock component will be at that level, and i've yet to see a heatsink that couldn't benefit from a bit of "persuasion".


.... It's probably possible with practice if you can measure flatness....

And while I am loving the debate, you seem to want to contend the ability of the typical Kustomer being able to improve upon the surface finish of a CPU heatsink/waterblock. I may be wrong, but your entire tone seems to indicate that the average Kustomer may well be what some term, well, "haunless".

Now, thats plain insulting. As you know, there's a lot of damn talented people on these boards :confused:

So, like before, I put RSV_Ecosse's question regarding lapping to you; "Did it drop your temps a little, or were you disappointed by the results?" Can you advise us whether your lapping did indeed drop your temps? Did you lap the heatspreader too (if fitted)? Did you perhaps prise off the 'spreader and give the core top a light going over? I'd be keen to hear how that went, after all, we're all here to pick up tips and experiences where possible :)

[pdf27]

Woah there! Have you also considered that it may well not be?
Yep - notice my original post only said that hand lapping may not improve the surface, and not to automatically lap everything.

And while I am loving the debate, you seem to want to contend the ability of the typical Kustomer being able to improve upon the surface finish of a CPU heatsink/waterblock. I may be wrong, but your entire tone seems to indicate that the average Kustomer may well be what some term, well, "haunless".
Now, thats plain insulting. As you know, there's a lot of damn talented people on these boards
It wasn't intended that way. There are a lot of very clued up people about, but there are also a lot of people who do things because they have been told to rather than actually unterstand what they're doing (remember many people read these boards who don't actually post). The comment was intended to get people like that to stop and think before they actually do anything.
Note that there are also people like me about who while they are really happy with the theory of why/how the things work, are totally cack-handed when it comes to actually modifying them...

[Sticky]

It wasn't intended that way.

Over sensitive moi :)

You do have a very valid point though, right now I am picturing some lurker thinking, "lapping - i'll have some of that"............. only to find that P4's can make great shiny keyrings.

[reaper]

I have a smashing shiny P4 2.4B keyring courtesy of a friend who killed it off after about 18 months worth of running it at 2.7GHz. :p

Cue me sanding the pins off it and polishing it up!