Should I lap copper-plated stock chipset heatsinks?

[locustella]

I recently put together a system (my first) incorporating:

intel q6600 g0
asus p5k premium
thermalright ultra 120 extreme

I've had it prime stable at 490 x 7 MHz which I'm happy with, really, but I'm a bit obsessive-compulsive about even numbers and decided to see if replacing the TIM on the stock chipset heatsinks (combined with some airflow alterations and lapping the cpu) would get me to 500 MHz on the front side bus. Some people seem to be of the opinion that replacing stock TIM is always a good idea on Asus boards.

On the p5k premium the southbridge, northbridge and one set of pwm chips share a heatpipe. There's a separate removable heatsink on another set of pwms. Here's what the northbridge and southbridge looked like when I got the heatpipe off.

northbridge:

northbridge heatsink:

southbridge:
No picture of that.

southbridge heatsink:

The pwm heatsinks were clean since they had pads (or tape, unsure of the terminology here) that had not melted. This is the one on the heatpipe with a pad that's thicker than the other one:

Question 1) The application of the stock TIM on the northbridge and southbridge does cover the whole contact area, but it looks a bit thick, doesn't it? Do you think I will see some performance gain from replacing this with AS5?

Having already lapped my cpu heatsink I thought I should lap the chipset heatsinks too. This is what they looked like when I finally got the TIM off (I had to use acetone).

northbridge heatsink:

southbridge heatsink:

pwm (the heatpipe one):

Question 2) Do you think the northbridge, southbridge and pwm heatsinks will benefit from lapping?

However, when I started lapping the separate pwm heatsink it turned out to only have a copper coating:

So, as far as I can tell, it's copper plated aluminum. I should have guessed from the weight, of course ... This is also the case for the whole heatpipe section.

Question 3) Should I lap these and wear the copper layer away or are they better left alone? I'm no expert in this area but I wonder what use a thin copper layer is.

I'm also wondering what to do about TIM on the pwm heatsinks. Since they cover 16 separate chips that may or may not be completely even in relation to each other I'm concerned that all may not contact the heatsink if I replace the pads with Arctic silver ceramique (which I've bought for this purpose). Especially for the one that's part of the heatpipe since it doesn't move as freely. However, when I put the whole heatpipe assembly in place "dry", it doesn't wiggle or rock hardly at all.

Question 4) Do you think I should replace the pwm heat pads with Arctic silver ceramique?

I read in this thread about the S775 board Abit IN9 32X-MAX:

"One thing I would mention though is PWM temps. Mine were high and I did add an 80mm fan and I also folded the heat tape over the PWM chips to make it thicker and that knocked a good 20-30c off my PWM chips. I would recommend you do that as soon as you get the mobo since the tape gets a little gooey after awhile, plus on a bare board it only takes like 15 seconds to remove the heat pipe assembly. I also replaced the heat pads on the NB and SB with AS5."

Is folding the heat pads/tape an option? I thought the object of any TIM, compound/pad/tape, was to give the heat a short and easy way to the heatsink?

Thanks for any input!

 

[a573573]

I too am wondering about this tape on mosfets/pwm question. I just replaced my DS4 heatpipe arrangement with TX-2 and it had the tape on the SB and the Pwm/Mosfets I did add a little more than a thin layer on the mosfets and it seems to be tight enough for proper transfer, but I have no temp sensor or gun to see just how hot they are getting.

I was thinking already about changing the pushpins/springs in for real nuts and bots, just havent had a chance to do that yet. I know then for sure I will get good transfer, but if I dont need to do all that it would be great.

So I am waiting for some good responses as well about the tape and such

 

[BillParrish]

Nice to see someone paying attention to the details.

Since you have it off I would touch up the NB and SB sections of the heatpipe but not worry a lot about it, maybe stop at 400 grit once they are flat, the screws nuts and bolts will probably be the best as the extra clamping force will help ensure the pipe assembly sits flat on the chips. Be careful, you don't want to crack a core. #4-40 nylon hardware from the specialty drawer at Lowe's or a good hardware store works great, consider double nutting (use two nuts per screw) with split lock washers to help keep them from loosening up due to thermal cycling. Spring mounting them is even better but finding small strong springs is a pain.

The Mosfets are a whole different ball of wax as you have found out. You will just have to see how much gap is between the mosfets and the pipe. Do a trial fit with no paste and see how much board flex (you will need to have your final CPU cooler installed) and the board in the case which makes it very difficult to see the gap. (I pull off the I/O plate and use a small flashlight as a backlight and see how much shines through the gap, hopefully none right where each mosfet sits. ). Many times the board is so warped you have to really do some special things like fabricate an insulated backplate to fit under the Mosfet area to help flatten it out so the gap is reduced small enough to use thermal paste. Take care of those original pads, many times you have to use them as the gap cannot easily be made small enough and they will have to be reused. The mosfets are high temp rated devices, much higher than typical IC's so as long as you are getting good pad contact they should be good and getting airflow on the heatpipe will also help a ton. I would not Lapp the mosfet area its just not worth it IMO but it cant hurt. Again good solid contact between the mosfet and the pipe with improved hardware and increased clamping force will probably be where the main benefit is. In theory the copper plating would help a bit with conduction heat, much like copper clad cookware, it helps spread the heat conduction evenly, in this case I very much doubt it will matter at all, its too thin of a coat and too crappy. I would be more worried about the base metal corroding or oxidizing preventing good heat transfer and with the paste or pad keeping oxygen out that should not be a concern, at least I have never seen it or a post about it.


A small fan on the NB part of the heapipe does wonders for keeping the entire assembly cool.

Thats about all I can think of atm about it based on my heatpipe adventures.

 

[locustella]

Thanks Bill! Lots of good stuff there. I think some of the things you mention are a bit beyond what I'm willing to put into it. I've hade it in pieces for over a week now, just waiting for replies on the forums and am itching to actually start using it ... ;-)

I think I will leave the heat pads and push pins on the pwm heatsinks. I hadn't thought about the board bending and such. I'm afraid it would take me ages to sort that out.

For the north and southbridges I will lap and use AS5. There are two things that have come up with regards to the northbridge, though.

1) The northbridge itself looks kind of fragile, the material is so shiny, see the picture below. Should I be worried about cracking or otherwise damaging this component if I use nuts and bolts and a thin layer of AS5?

2) The other thing is the black cushioning stuff around the edges of the northbridge heatsink. Here's that picture again:

Is that there to prevent heat escaping from the heatsink through airflow or to keep the heatsink from tilting to get better contact? I'll have to remove that if I'm to lap the heatsink flat. I could probably make the heastink smooth using just a finger to sandpaper it but I don't want it smooth, I want it flat, right?

I've tried cutting beneath it slightly with a razor and it's not going to come off easily. I don't think I'll be able to put it back. If it's needed to stabilize the heatsink, maybe I should just leave it be?

3) The last thing I'm thinking about is electric isolation if I use metal nuts and bolts. I was thinking I would cut little pieces of bicycle tubing to put between the nuts and bolts and the motherboard. Good idea?

 

[Forceman]

It looks like the real problem isn't that the heatsinks aren't flat, it's that they aren't making good contact because they aren't even with the motherboard. I mean they are tilted at an angle instead of lying flat like they are supposed to be. Since they are connected by the heatpipe, if that isn't quite in the right shape it will prevent the heatsink from making good contact. Can you just reseat the heatsink/heatpipes so they make better contact? That black material is there to stabilize the heatsink and keep it more or less flat on the core - I wouldn't want to remove it unless you can put it back on, otherwise you might make your situation worse.

 

[bardos]

In a nutshell, copper works out to be the best cost effective material for heat conduction available to us computer users at the moment. meaning it removes the heat from a cpu faster than aluminum does. on the other hand, aluminum is cheaper.

i think it was about the time of the s370 cpu's when manufacturers began turning to "hybrid" hsf's: thick copper bases with the rest of the heatsink being aluminum.

when i find a copper plated heatsink my theory is that the copper "dusting" of what is basically an aluminium HS is just for looks and has no real world effect on the actual heat absorption. ergo, lapping the hs, if skewed, would have beneficial effects. just my opinion.

thermal conductivity chart i found on the net:

Diamond --- 2300 W/mK
Pyrolytic Graphite --- 1950 W/mK
Silver --- 429 W/mK
Pure Copper --- 401 W/mK
Pure Aluminum --- 237 W/mK

waiting for a solid diamond hs...

 

[BillParrish]

The foam acts as a cushion more than anything else and a not too good one at that. Usually it is a little denser than what it looks like in your picture, and it might also be there to prevent factory applied thermal compound from oozing out onto the mb which would require manual cleanup.

I have removed the foam from around several heatpipes and most commonly I find it on video card heatsinks, never had any problem from taking it off.

Go with the hardware, even reqular metal screws, the little red washers you probally have in a drawer that come with the case is perfect to put under the screw heads on the backside of the board or any non-metallic washer will do.

As far as tightening them, smartly snug is good enough. I alternate back an forth, running the nts down until they start to compress the split washer. It then takes just about one turn on 40 thread per inch hardware to fully compress the split washer so I give each nut one turn, this typically snugs down the pipe. With all nuts at this stage I use just thumb and index finger on the shaft of the nutdriver or screwdriver if tighening by holding the nut and using a screwdriver on the backside, and "feel" the snug-ness. I try to guage that all the fasteners are about the same tightness and if I "feel" one ( I even close my eyes during this so I can concentrate on the "feel") that seems not as tight as the others I give it a little more, again at this stage I just want them all snugged down and the lockwashers fully compressed, no more. Once I am happy all the hardware is snug, the pipe has not shifted etc. I then, again with the "two finger" method give each screw about a 1/2 turn more as a final "tighten". I again go to all hardware and "feel" the tightness and if all good I quit. Using just the two fingers on the shaft of the tool allow a very fine "feel" for the tightness and you would have difficulity genetating enough torque to damage anything unless you really tried.

On the manuf line we speced 7-9 inch pounds of tighening torque for #4-40 metallic hardware with splitwashers when used to mount devices to heatsinks if that helps, but accurate torque screwdrivers are expensive.

I know what you mean about wanting to put it together, but if its not too late I highly recommend the replacement hardware, it will do more than any lapping, use AS where you can and dont worry about the rest. If you press down on those pushpins you will easily feel how loosly they hold on the pipe. Low clamping force. I think that is why they use so much goop, it almost glues the pipe to the chips as those push pins are more for keeping the pipe from falling off than for making the pipe contact the chip well. Really dump those pins, 4-40 nuts, split lockwasher and 3/4 (1/2 may do it) inch screws and some nylon or fiber +4 washers and gently snug her down. The risk is minimal. I can tell by the quality of your pictures you have the skill and patience to do the job correctly, its just a matter of "feel" and just a little care and having your brain switched on when you do it is all it takes to get excellent results. To this day I firmly beleive I am able to get such high stable FSB speed is because I spent most of a Friday nite and sat morning working on the same thing and I didnt even try to lapp mine (had to fabricate a stiffener for the board for the mosfets and was probally wasting my time but got it so I could use paste.)
Stock the heatsink was warm, after the AS and hardwware it was damm HOTTTT !!!. This is what lead me to my fetish about putting a MCH/NB (or somewhere) on the heatpipe because as soon as I did the thing whet immedately cool and suddenly I could do FSB of 425+ which had been a wall in casual first day scewing around testing out of the case.