Core 2 Duo Lapping

[3lfk1ng]

If your trying to get every inch of performance out of your Intel Core 2 Duo, but you want to keep your temps down, consider the following:

#1 Lapping the IHS on the Core 2 Duo
http://www.legitreviews.com/article/402/1/

#2 is for Pro's only with tons of money to spare, and works only 1 out of every 4 tries.
http://www.legitreviews.com/article/402/2/

I plan on doing #1 to mine when I get it.
#2 Isn't worth trying on a $300 processor (my opinion)

Then do the same for the Heatsink.

~3LFK1NG

 

[JCon]

I was hoping part 2 would work... Oh well!

 

[Chil]

Wow, talk about overkill.

 

[NMDante]

I use to lap my heatsinks, but found that the time I spent doing that, and the temp drop I got, was not signifigant to keep doing it. Plus, I always ended up making a mess in my work area with the all the sanding. The heatsink did look really cool, with the mirror finish, though.

The 2nd technique is just too extreme to try on a brand new CPU. I mean, even the article said that the CPU was toasted, and wouldn't post. Not really worth it, at all, imo.

As for the IHS, I never got a CPU that was convexed or concaved, so I never really lapped the IHS.

I think this is more for the advanced overclocker, since leaving any of the sandpaper residue on the CPU might cause damage, and the 2nd technique is just plain nuts.

 

[3lfk1ng]

I think the temp drop is highly worth the lapping.

If you can knock off 3 degrees by lapping the cpu....an additinal 3-5 with A5S...an additional 5-10 with a After market heatsink, and then an additional 2-3 by lapping the Aftermarket heatsink, thats a total of 13-21 degrees...

Which is fantasic for those of us with high overclocking needs.
Lower heat = Higher overclock

 

[Grimmy]

8O . o O (I wonder if it was screaming... NO!! NO!! Don't Do It!!)

 

[3lfk1ng]

LOL, it probly was felling like it was being interrogated.

 

[angry_ducky]

What a waste of a perfectly good C2D. All for the sake of keeping it a bit cooler. Option 1 seems reasonable (although I'd never attempt it), but Option 2 is just stupid

 

[turpit]

I use to lap my heatsinks, but found that the time I spent doing that, and the temp drop I got, was not signifigant to keep doing it. Plus, I always ended up making a mess in my work area with the all the sanding. The heatsink did look really cool, with the mirror finish, though.

The 2nd technique is just too extreme to try on a brand new CPU. I mean, even the article said that the CPU was toasted, and wouldn't post. Not really worth it, at all, imo.

As for the IHS, I never got a CPU that was convexed or concaved, so I never really lapped the IHS.

I think this is more for the advanced overclocker, since leaving any of the sandpaper residue on the CPU might cause damage, and the 2nd technique is just plain nuts.

I agree with you about lapping, however, posts Wusy has made (with corroborative links) indicate that there is a trend in the core 2 duo heat spreaders to be convex. Whether this is do to a poor die design (mechanical die for stamping the HS, that is ) or if it was done deliberately to insure good contact between the HS and the die (CPU die, that is) is a question which remains to be answered.

You would be someone who might be able to ask around work about this trend and the cause behind it.

Whatever the cause though, given the possibility of a convex (or concave for that matter) heat spreader, it seems as though lapping may be a good way to go with the Core 2’s

 

[NMDante]

I think the temp drop is highly worth the lapping.

If you can knock off 3 degrees by lapping the cpu....an additinal 3-5 with A5S...an additional 5-10 with a After market heatsink, and then an additional 2-3 by lapping the Aftermarket heatsink, thats a total of 13-21 degrees...

Which is fantasic for those of us with high overclocking needs.
Lower heat = Higher overclock

I honestly never seen that much temperature difference after lapping.

If you're needing that much more cooling, then I would look at exotic cooling solutions, H2O or phase cooling. Both will cool without the hours spent on lapping, and will significant lower temps of CPU for overclocking.

Of course, that's just a suggestion. I went to H2O a few years ago, but since then, I haven't really been overclocking much. Kind of ironic, that I get a setup to cool in case I overclock, only to not overclock. LOL

 

[3lfk1ng]

You forgot TEC coolers..

http://www.virtual-hideout.net/reviews/Titan_Amanda/index.shtml
http://www.pcper.com/article.php?aid=273&type=expert

This will provide sub-ambient temperatures that even watercooling cannot brag about. (and in most cases it's a lot cheaper).

http://www.newegg.com/Product/Product.asp?Item=N82E16835702002

 

[NMDante]

TEC coolers are unique, and the draw a lot of power, but I guess so does phase cooling and some H2O setups with huge pumps.

I was gonna play with one a few years ago, but I read horrible things when they were done incorrectly (but the same with H2O cooling back in the day). I guess TEC coolers have been integrated with a HSF or even made into a water block nowadays. The heat created by those things are incredible, and having a less than effective cooling setup on them will create a disaster.

I last saw one on a water cooling setup. It was pretty cool looking, but it still was a bit daunting for me to try out.

 

[trixst4r]

Method #2 was done without any knowledge on how to remove a IHS safely on a C2D.

Experiments from mainly people of XS have shown that the solder melts around 90 degrees celcius. Unfortunately, the solder holding the core onto the PCB is just as easy to melt. To use anything like a propane torch to take off the IHS is overkill. A regular lighter can do the same job and it is much safer to use since it will not increase teh temperature rapidly.

 

[derek2006]

Sanding the cpu seems like a good idea ill try that tomorrow. The only part I'm unconformable on is washing the cpu :? is this hazardous? Also does anyone know where I can get a TEC for 30-40$. I will hook it on to my tt big typhoon.

 

[NMDante]

Sanding the cpu seems like a good idea ill try that tomorrow. The only part I'm unconformable on is washing the cpu :? is this hazardous? Also does anyone know where I can get a TEC for 30-40$. I will hook it on to my tt big typhoon.

Here ya go:

Peltier for sale

Good luck installing it. I would get as much info on condensation and other things before attempting to use one. Also, I am not sure if an air cooled heatsink will keep the heat down. These bad boys generate a ton of heat on the "hot" side.

 

[derek2006]

Holy crap! Does it take 226 watts or is that how much it dissipates. And would condensation occur if the Big typhoon can't cool it down enough? My uncle gave me a pelter cooler before for like a pentium 3 or something. And the only way to get it to condensate was if you stuck the heatsink part in cold water.

 

[derek2006]

How bought this one would it have enough power to condensate? And enough power to cool the cpu? And o ya thanks for the link.

Edit: How would I have deal with condensation? What are the steps I can take against it?

 

[NMDante]

The cool side gets super cold. It starts to create condensation pretty quickly (on the cool side). I have seen a lot of TEC setups with foam/insulating foam type material around the CPU, to help reduce condensation. It's similar to those used in phase cooling.

About the 226W...I believe that is what it dissapates when it's being used. I'm not 100% about that, since I am no expert with TECs or peltiers. There is place to go, though. Peltier Info

I think they might even have a guide to setup a peltier/TEC setup there, too. Mind you, this guide is a year old, too.

Here's a quick Google link: Peltier Cooling Guides

Again, these might be dated some, or alot.

 

[hammong]

Here are some thermal and electrical specifications for the average 226 watt 50x50mm Peltier device:

Qcmax = 226W @ Th = 27C
Vmax = 15.2V @ Th = 27C
Imax = 24A @ Th = 27C
Pcmax = 364.8W @ Th = 27C


Qcmax is the maximum amount of heat energy that can be moved from the cold side to the hot side. Keep in mind that 226 watts is a best case scenario where you have no thermal losses due to the interface between processors integrated heat spreader and the much larger 50x50mm heat spreader you'll need to mate up to the peltier device itself. Allowing a 50% engineering safety factor, you can probably cool a 80-100 watt Tdp CPU with this device.

Now, at the maximum cooling effect, the total power dissapation of the Peltier device is going to be 364 watts. You need to remove 100% of that heat effectively in order to ensure that the "cold side" stays at 27C in the equation. Removing 364 watts of heat from a 50x50mm ceramic plate is difficult - you'll need a water cooling block indeed to keep your PELTIER hot side cool!


Long story short - it's not worth it for a modern CPU. The best way to implement peltier devices in this scenario are to use 4-5 of them connected to a dual-water block circulation system whereby you have a "cold side" water flow and a "hot side" water flow. The cold side circulates around a waterblock on the CPU, and the hot side circulates through a fan-cooled radiator. That way the peltier devices are not physically located inside your computer, and the duty of transferring heat and funnelling the 400+ watts of power through them can take place externally.

Greg

 

[3lfk1ng]

Don't buy just the peltier, with the Monsoon II it monitors everything and alarms in the almost not possible chance that the peltier fails. It also monitors temp to peltier to make sure no condensation is formed. The Monsoon II is the best peltier on the market for that reason.

 

[ElMoIsEviL]

Here are some thermal and electrical specifications for the average 226 watt 50x50mm Peltier device:

Qcmax = 226W @ Th = 27C
Vmax = 15.2V @ Th = 27C
Imax = 24A @ Th = 27C
Pcmax = 364.8W @ Th = 27C


Qcmax is the maximum amount of heat energy that can be moved from the cold side to the hot side. Keep in mind that 226 watts is a best case scenario where you have no thermal losses due to the interface between processors integrated heat spreader and the much larger 50x50mm heat spreader you'll need to mate up to the peltier device itself. Allowing a 50% engineering safety factor, you can probably cool a 80-100 watt Tdp CPU with this device.

Now, at the maximum cooling effect, the total power dissapation of the Peltier device is going to be 364 watts. You need to remove 100% of that heat effectively in order to ensure that the "cold side" stays at 27C in the equation. Removing 364 watts of heat from a 50x50mm ceramic plate is difficult - you'll need a water cooling block indeed to keep your PELTIER hot side cool!


Long story short - it's not worth it for a modern CPU. The best way to implement peltier devices in this scenario are to use 4-5 of them connected to a dual-water block circulation system whereby you have a "cold side" water flow and a "hot side" water flow. The cold side circulates around a waterblock on the CPU, and the hot side circulates through a fan-cooled radiator. That way the peltier devices are not physically located inside your computer, and the duty of transferring heat and funnelling the 400+ watts of power through them can take place externally.

Greg

Hehehe.. you guys are funny.

The efficiency of a Peltier can be greatly increased by simply cooling the peltier more efficiently. The trick is to create turbulence/friction within the heat exchangers.

Here's are some parts of my setup.. now I don't have any recent pics of these parts installed in the case they're in right now seeing as I haven't taken any pics yet... but the watercooling components are in the same places.

Now the next pic shows when I was putting the whole thing together. This was done back in July and this was before the drop in price on the x1900CF cards which I couldn't resist buying.

Here's another shot of the block..

This is a picture of the same setup but with my DFI/AMD X2 combo I had in there before. I really need new pics.

And the outside of the case..

Now some peeps state that it is impossible to use a 256Watt TEC to cool down an x6800 to bellow Zero temps. That's utter bullox. I idle anywhere between -25 to -20 and load around -15 to -10 at 4.45GHz (using Speedfan as Asusprobe shows 256C and CoreTemp and the Intel software both show 9C no matter what load the CPU's are on). I've done this by slightly increasing the voltage to the TEC and using two pumps and two heat exchangers. I've also increased the turbulence within the watercooling setup by using a large water drop into the secondary heat exchanger.

So TEC's are VERY much still useful. It just depends on how you set the system up to cool the TEC's. As for the case I'm using it's a highly modified Compaq Proliant 1500 series server (it used to house a Pentium 100MHz FYI).

 

[NMDante]

Nice setup, Elmo.

Pretty sweet H2O/TEC configuration. What size is that tubing? 1/2"?

Nice.

 

[wolfman140]

Hmmm...I wonder if perhaps this concave shape to the IHS is teh reason why my temps are pretty damn high even with my Zalman9500?

 

[HYST3R]

wicked set up elmo!

 

[dasickninja]

That I say is worthy of a Wusy Award for Excellence and overall cleanness. Here's your trophy :trophy:
I think I'll stick with my Monsoon II though, it works very well even with what I've put it thru.
Then again I might just go straight to a Titan and copy.. erhmm... emulate your setup.

@HYST3R
Nice to see your with the anime gumi now. Love the avatar.