CPU/CPU Heatsink Lapping Experiances

[Slick]

What experiance do you have with lapping your CPU or CPU heatsink?

 

Have you done it before? Know someone who has?

 

Did you see improvements? If so.. What cooler or CPU did/do you have?

 

Did you NOT see improvements? If so.. What cooler or CPU did/do you have?

[ThinkWithPortals]

2 minutes ago, Slick said:

What experiance do you have with lapping your CPU or CPU heatsink?

 

Have you done it before? Know someone who has?

 

Did you see improvements? If so.. What cooler or CPU did/do you have?

 

Did you NOT see improvements? If so.. What cooler or CPU did/do you have?

Oh. Hello Luke. Don't see you posting around here often.

 

EDIT: Yay, 3000 posts on this site!

[MegaDave91]

Experience

[Imakuni]

4 minutes ago, Slick said:

Know someone who has?

Did you ever do it?

[ChrisCross]

i think its like with cars. in the past you could modify them really easy. but know the smartest thing to do is just leave it as the engineers designed it.

 

hopefully you will find something that interesting in the workshop soon. maybe about ram speed? we havent got one of those for a while.

[done12many2]

3 minutes ago, Slick said:

What experiance do you have with lapping your CPU or CPU heatsink?

 

Have you done it before? Know someone who has?

 

Did you see improvements? If so.. What cooler or CPU did/do you have?

 

Did you NOT see improvements? If so.. What cooler or CPU did/do you have?

I did it years ago with a Pentium 4.  In my experience the temp changes were so small that they fell within margin of error.  I can't remember the CPU heatsink used.  The results were so insignificant that I never bothered with it again. 

[mikat]

what is lapping?

edit: it's sanding your heatsink to make it transfer heat better

[FilipSebik]

PC Intel P4 with intel stock heatsink

Non lapped

64`C Load

40`C Idle

Lapped Heatsink and IHS

63`C Load

38`C Idle

 

The thermal paste was Arctic MX-2 on both

[Gofspar]

Lapped and delidded my old 3770k dropped 28c load at 4.9GHz at 1.24v

 

Don't have the balls to lap a binned 5820k. 

[Anaerin]

In the good old days of Intel Pentium IIIs and AMD Athlons (Before Phenom), lapping the CPU die and heatsink contact patch made a LOT of difference. Thinning the die made for less insulating silicon and greater heat transfer, and the flat and polished contact area meant there was more direct contact and less "Through heat paste" contact which, while good, wasn't as good as a direct transfer. This is why the old methods of heat paste application (razor thin amounts in specific areas) were used - Your heatsink would ideally touch the die directly, with as little paste in the way as possible, and the paste was only there to fill in the slight imperfections in the die and HS base.

 

These days, with Integrated heat spreaders, there is no appreciable difference. If you are brave enough to de-lid your CPU you might get some bonuses, but that's not what's being asked or tested here. Theoretically the IHS should spread the thermal load out (hence the name) meaning contact area is not nearly as critical (Compare the size of contact area of an IHS with the size of the die underneath). Losing 1 or 2 degrees per square centimeter does next to nothing when you have 5 times the area, and the differences get lost in the margins of error. And generally speaking the "Weak point" in the transfer chain is the IHS material - It doesn't matter how good your contact is, how well performing your paste is, or how perfect your interface is, your thermal transfer performance is limited by that of the IHS. So even if your thermal paste transfers 10 degrees per cubic milimeter, if the IHS material only transfers 2, you're going to see no benefit.

 

Also, with Intel especially, you can't get the same mounting pressures that you used to, as pin-based sockets have the plastic of the retention mechanism against which to brace and spread the load of down-pressure from the HS, whereas the LGA style puts all that load on the pins themselves.

 

Of course, I am more than willing to be wrong about this, and if you can go ahead and prove this conjecture wrong, please do so.

[helping]

If you lap both CPU and Heatsink, the benefit becomes pretty apparent as you increase TDP (assuming imperfections)

Cooler was a Xigamatek Loki, pretty underwhelming, especially for a BD chip.

 

This was a long time ago, and I only did it because I was curious and was going to sell it anyway. Most improvement was likely from CPU lapping.

 

8150 was lapped down to mostly copper finish with varied grit from 400 to 2500 on glass. CPU IHS was noticeably concave, you can tell easily from where the copper begins to show during the lapping process.

 

Because this was so long ago I don't remember exact details and I didn't do any hardcore data collection, but IIRC temperature dropped approximately 5c down to 48~50c under load across cores. Ambient temp would have varied 23~26c, but should be mostly consistent. Slight undervolt and overclock to 4Ghz.

 

Even 2-3c difference is a big deal for an AMD chip.

 

 

[Wolfy]

This video piqued my interest, as I just recently lapped my NH-D14. 

I got a temperature decrease of 7 degrees Celsius.

 

 I think the problem in this video is that Luke stopped at 2000 grit sandpaper. I used a 4000 and an 8000 grit whetstone (new, to ensure flatness) and finished it off with a piece of tin that was machined flat. The surface became clearer than any mirror I've seen. I also lapped the cpu (don't worry it's not my daily use cpu) up to 4000 grit.

[Euphoria]

My Experience With Lapping an Ivy Bridge

 

Last year I decided to revamp my rig, building a new custom case for it and generally tidy it up with some nice new cables etc... I was considering De-lidding it (3770k) as we all know, Ivy Bridge is pretty hot, and my chip even on an overkill custom loop was still pretty hot which was hold back my OC.

But before de-lidding while the chip was out the mobo, I tested it on an off-cut piece of toughened glass I have in the garage, to my surprise, it was ridiculously bevelled, I mean when laid on the glass it was like a spoon! It had a massive bevel in the center. So I tested the waterblock too, and again, this was pretty bevelled, not quite as bad but still enough to annoy the nerd in me. Especially as the bevels were the same direction on both, curving down from the edges to the center, resulting in when mounted, there would be a decent sized gap in-between the center. Enough is enough... *Reaches for sand paper*

 

I went with: 400 grit wet, then onto 600 wet, 800 wet, 1200 wet, 1500 wet, 1800 wet, 2400 dry and finally 3000 grit dry.

The experience of the lapping itself was actually not that tedious like Luke said in the video, but then anything nerdy gives me "The Tingle". The effort involved for me at least was actually a lot! The waterblock wasn't too bad, 10 laps in each direction, with each grit paper was enough to get a nice smooth flat finish. Whereas the IHS, jebus H. christ, that took hours! With the first few grit papers I spent what seemed like a good hour plus lapping as the paper was touching the 4 corners of the IHS, but wasn't even close to touching the center until 30-40 minutes of continuous lapping, only stopping briefly for a few seconds to give my hand a rest.

Finally, once the center was contacting the paper I started to progress up through the grits, spending anywhere from 20-50 laps with each paper in each direction.

 

The final result was pretty nice I thought... Screenshot:

Spoiler

The IHS now has a smooth, shiny and much more flat finish. When laid on the glass with a little moisture between them now, there is a good resistance when lifting it off the glass, meaning the surfaces are flat enough to cause the suction effect, (I think this is known as wringing).

Now with the CPU back in, the TIM spreads much more evenly resulting in me applying less. As for Temps... well I know some may say it's within the margin of error, but as my man cave is very well temp controlled I would disagree and say it did indeed make a difference. It seemed to be around 3 degrees Celsius reduction, which I'm happy with. When benching or stressing an OC my temps are extremely consistent due to the room being temp controlled, so it was a noticeable drop.

 

De-lidding - (For those interested): I did then go on to De-lid it. I used the Bench Vice, block of wood and Hammer method to de-lid this chip, as I f***ing hate the razor blade method. I've used that method before and I find the razor is extremely tedious, dangerous to the CPU more so than the vice and hammer, and bloody dangerous to My Fingers! I would HIGHLY recommend the Hammer and Vice method, it worked beautifully and took seconds.

Temps from De-lidding obviously were a Lot more noticeable. The amount of cheap, nasty and dried crusty TIM on the die was painful to see. I replaced it with an extremely wafer thin layer of liquid Gallium, then replaced the IHS but I did not stick it back down to the PCB, so I can access it again if needed.

Both the De-lidding and lapping resulted in a drop of roughly 17 degrees Celsius combined, which gave me a big Nerd Tingle And an even bigger nerd tingle when it was able to finally hit the magical 5GHz!! Yeah Baby

 

I digress. In conclusion, would I recommend lapping... well... Yes... and No. I would first check to see what your chip and/or heatsink/waterblock is like first. If when laid on glass with a tiny amount of moisture they stick quite tightly together, then you probably have an ok surface and lapping may not net you a noticeable result. But, if like me, you have a CPU more curved than a banana... Yes, yes I would recommend lapping (and De-lidding if you want to take the risk).

[SgtSnyder]

I agree with Wolfy that he needed to use finer grit. I went to 2000 and then used polishing compound to finish it off and also got a mirror like finish. Did the same on the IHS. But the thing that made the biggest difference for me was increasing the mounting pressure. I don't know about the Hyper 212, but I was using the Hyper T4 and it uses an AMD style mount, so when you remove all that material it still holds the cooler at the same distance instead of being able to bring them closer together. I just took some paper and wedged it into where the clip presses into the cooler and that made a substantial difference. Between the two changes I got about a 7C drop in temperature. Not sure how you'd do it on a 212 because I'm not too familiar with the mounting. 

And I'm sure you this, but remember to clean it with alcohol when you're done! The amount of residue left behind is incredible. It would be awesome if Luke tried doing this again using finer sand paper and increasing the mounting pressure and making another video if it ends up making an improvement! 

[Jumper118]

using sandpaper to make something smoother isnt lapping.

 

[ShaunC]

Back in 2004-2005 i lapped a heatsink for my PC.  It had an AMD athlon xp 3200+.  The processor was running hot all the time.  When i got it with the stock heatsink it was idling at 70*c and i had to open the side and put a fan on it to make it so my PC wouldn't reboot. I picked up an all copper heatsink and gave that a try.  It was still heating up to the point of rebooting under load.  So i lapped it.  I picked up a lapping kit that came with a small piece of glass and all the sand paper grits.  After lapping, it would hit 70*c under load, but wouldn't make the system reboot.  I think at idle it dropped around 8*c and worked really well.  I cannot remember which heat sink it was, but it was all copper, no heat pipes, just fins and a fan on top.  I remember the surface of the heatsink where it would make contact with the processor/thermal paste was matte before the lapping, and very shiny afterwards.  Also, the instructions were different than what you did in the video.  It said to make small circular motions for a certain amount of time, and then change directions for a certain amount of time.

[Fyrelith]

Hey Luke, great video but you didn't finish... well technically.

I commend you on a great first try, you really took your time which is key. Usually people trying this for the first time will end up applying too much pressure and wearing through the CPU die / heat sink.

I've lapped 5 CPU/Heat sink combo's in the past, not a lot of experience really, but I found applying a small amount of pressure to the piece being lapped can be beneficial. You have to make sure the sandpaper is taped down "tightly", preventing buckling of the sandpaper in front of the surface being pushed. If you push too hard though, you run the risk of sanding down the edges of your perfectly flat surface (ie. buckling of the paper).

Another trick is in the sanding pattern/technique, straight lines & figure 8's. Figure 8's are definitely more prone to wearing the corners down when applying pressure, it's just about balancing the pressure for the best trade off.

Here's the... 3rd I think, lap job (sounds kinky ) I did on an Intel E8500 C2D...

Spoiler

 

I usually found anywhere between 3-6°C reduction in cooling from the whole process, which seems like little payoff for the time/labor. I enjoy doing this sort of hands on work anyway, so these projects never really annoy me, but I can see how it would get frustrating.
Here's a link to one of the first lap jobs (heh) I did on an  AMD Athlon 1800+ / Typhoon heat sink.

 

Hope you enjoy pics, and LUKE!... give it another try, get those surfaces shining!

 

 

Cheers, Jay

 

(ps. a little incentive if you will, when both the processor lid and heat sink look like mirrors... you don't need any thermal paste. (from personal experience anyway, sure someone will prove me wrong )

[ssnakee]

this video triggered me so hard that i even created an account to rant LOL

tired of things beeing WHATEVER? look at the physics first dude

the argument of the video is: does sanding the flat surface of the heatsink affects the performance of the heat transfer?

Well lets take a look in to the problem

heat transfer is constrained by 3 things surface area,material of the things transfering heat and power(in our case intel or amd gives us a easy way to figure this out as TDP if you want to go further you need to look into the voltage that the processor takes and the amperage)

 

Now lets analyze the problem in easy parts:

 

Does sanding the flat surface change the thermal conductivity of copper? NO, sure it makes it more shiny and maybe you could make it more flat but it does not change shit about the properties of the material 

 

Does sanding change the surface area in contact with the CPU? Maybe, if you had a shitty manufactured heatsink it could change the surface area but a better question is:

Does the change on the surface area could be meaningful to the performance? It depends on how much the change in contact area is if is >1mm it will do shit and unless u had a heatsink made with stone tools and glue that change will be almost nothing

 

does sanding change the power? lol NO

 

by just taking the physics in consideration first you realize that sanding does nothing

 

TL;DR if you want that things stop being whatever check the physics first, either change the materials of the heatpipes or increase the surface area by carving a grid on the heatsing and then embed them on to the cpu or i dont know but HOLY SHIT DONT COMPLAIN THAT THINGS ARE WHATEVER IF YOU JUST MAKE THE THING THAT WAS FLAT EVEN FLATTER

 

\rant

[Gerr]

I have read several comments that spoke against "sanding flat surfaces".  If your CPU & HSF are already perfectly flat, then no, lapping won't do anything.  The point of lapping is that most CPU's and/or HSF's are NOT perfectly flat and the process of lapping is to get them flatter than they currently are, which in turn will help with heat transfer from the CPU to the HSF and thus reduce temps.

 

One possible improvement...once you have lapped your HSF to as flat as you can get it, you should sand the edges down a hair more to make the HSF slightly convex as it seems that a slight convex HSF works best as it places the most pressure at the center of the CPU, where most of the heat is.  Remember that it's very unlikely that anyone could obtain perfect flatness when lapping by hand, so cheating towards convex a bit makes sure you don't accidentally still have a concave shape.

[Gerr]

I have read several comments that spoke against "sanding flat surfaces".  If your CPU & HSF are already perfectly flat, then no, lapping won't do anything.  The point of lapping is that most CPU's and/or HSF's are NOT perfectly flat and the process of lapping is to get them flatter than they currently are, which in turn will help with heat transfer from the CPU to the HSF and thus reduce temps.

 

One possible improvement...once you have lapped your HSF to as flat as you can get it, you should sand the edges down a hair more to make the HSF slightly convex as it seems that a slight convex HSF works best as it places the most pressure at the center of the CPU, where most of the heat is.  Remember that it's very unlikely that anyone could obtain perfect flatness when lapping by hand, so cheating towards convex a bit makes sure you don't accidentally still have a concave shape.

[Prysin]

On 28/04/2016 at 2:09 PM, ChrisCross said:

i think its like with cars. in the past you could modify them really easy. but know the smartest thing to do is just leave it as the engineers designed it.

 

hopefully you will find something that interesting in the workshop soon. maybe about ram speed? we havent got one of those for a while.

oh please, i hope they dont do another bullshit video on RAM. They have done enough damage as it is.

[ChrisCross]

37 minutes ago, Prysin said:

oh please, i hope they dont do another bullshit video on RAM. They have done enough damage as it is.

maybe just something to reply on people who think they need a high-end 3,4ghz ram kit for gaming. 

[TheGamingBarrel]

1 hour ago, ChrisCross said:

maybe just something to reply on people who think they need a high-end 3,4ghz ram kit for gaming. 

There is a notable difference in mobile devices and apus. 

[ChrisCross]

11 hours ago, TheGamingBarrel said:

There is a notable difference in mobile devices and apus. 

i have never saw a mobile device with such high-end ram.

your right with the apus. but i was more talking about the people who build a 1k gaming pc and waste their money on super high-end ram. 

[TheGamingBarrel]

1 minute ago, ChrisCross said:

i have never saw a mobile device with such high-end ram.

your right with the apus. but i was more talking about the people who build a 1k gaming pc and waste their money on super high-end ram. 

On DDR3, 1066 To 1600 MHz CL11 Gives A near 30% Improvement In heaven.