Sign in with
Sign up | Sign in
Your question

Some notes on applying thermal paste to the Xigmatek

Last response: in Overclocking
Share
June 28, 2009 12:26:51 AM

I find myself linking another post with this in it, so rather than doing that I've made this post so it can go at the top of the thread. If you just stumbled across this and disagree or think something needs changing, feel free to chime in.

I write stuff not because I know it all, but because nobody else does.

Here are the exact steps I used to put on my Xigmatek:

I used a one-sided razor blade like this:
http://www.doityourself.com/invt/1995133

I first held the base of the cooler up to eye level and pointed it towards a light source. I then put the edge of the razor onto the surface of the base, moving it around and looking for places where the light shone in-between the razor and base.

Obviously, there are grooves between the pipes and aluminum. These need to be filled -- not because you want the contact in those areas, because the paste will be too thick there for good heat transferal. No, you want those filled so that they interfere minimally with the thermal paste you apply in the following step.

If there are OTHER inconsistencies you find with the razor method, consider lapping or, if serious enough, getting a replacement cooler. I did not feel my cooler needed any lapping. Note: A well-lapped cooler and CPU should need much less thermal paste and this method will probably need to be modified for that.

Before applying any paste, attach your hardware to the cooler (mounting brackets). I used the bracket kit, which I recommend. Push-pins suck.

Filling in the grooves:
I placed a pea sized (small pea, but ya, really) amount of AS5 onto the base of the Xigmatek. The idea is not to leave that much on the cooler, it's just to have enough to work into the grooves well. I then used the razor to work the paste around, completely filling in the grooves. This takes a little work to get just right. You want the grooves filled in, but no excess on the base anywhere once you are done. You are basically scraping the flat surfaces clean here, but leaving the grooves filled. There will be some natural "mounding" of the paste that is in the grooves... it will stick up above the level of the base just slightly. Try to minimize that, but you will always have some.

The razor works great on this step to create a flat application and scrape off the excess.

This next step is where I differ with some other folks.

Use your AS5 syringe like a pen and draw three lines down the center of each pipe, not the aluminum. Each line should be about 1/2 the total length of the contact surface. These are thin lines. (They would be even thinner lines if the CPU and cooler are lapped... perhaps MX-2 or some other ceramic paste would help here to apply a smaller amount.)

Now it's time to mount. Please note that if you got a hair stuck in the paste of some other foreign object between the two surfaces, you are probably hosed and will need to start over.

Obviously, the cooler should be placed gently and precisely so the the bolts line up with the holes. Once in place however, you should gently ROTATE the cooler back and forth. Note that the design of the CPU holder and the cooler will only let you rotate the one or two degrees. That is fine. Do this at least 5 times, then bolt it down.

Do not tighten the bolts clockwise or counter-clockwise. Use a 1-3-2-4 type crisscross pattern and make three or more passes. Never tighten one screw down all the way without first having the others beginning to tighten.

I hope this is of some use. None of these ideas are original to me, just gathered from around the internet and especially here on these forums.
a b K Overclocking
June 28, 2009 1:23:56 AM

This needs to be a new sticky!!

This cooler is one of the most popular coolers and the most popular design type now out. I wish I had this info. when I installed mine. I did it the "STD" way. I still get good temps (I believe) but the next I remove it, I'll reference this thread.

Good job on the info.!
June 28, 2009 1:50:27 AM

+1. Please sticky this. Good Job Proximon!
Related resources
a b K Overclocking
June 28, 2009 3:14:44 AM

IIRC Lupiron (What happened to him btw?)did something like this but people didn't read it lol. Good work Proxi.
June 28, 2009 3:30:18 AM

Thanks. I had made a post last year when the Xigmatek was new, because I had used one and took pictures of how it fit on my board. Later I was asked about how I applied the paste and ended up writing this, but it was far down the long thread.

I just find myself linking back to it a lot so I thought it would be nice to have it in it's own post.

I have added this link and a few others at the bottom of my guide over in the homebuilt section. It's a little "how to assemble a computer" list of links. Too bad we don't have a guide here, but then we would have to do it up right if we did it at all.
July 6, 2009 11:02:08 PM

umm.. Im building a PC soon and Im getting an AMD 955 and a Xigmatek Dark Knight...so Iv noticed some people but a litte drop of thermal paste in the middle of CPU and spread it from edge to edge as thin as possible and than mount the heatsink.. is that a good method as well? it will nice if you could include some pics.

and is the thermal paste that comes with the cooler bad and should I get AS5 instead?

and one last thing are you suppose to get rid of your old grease and apply new thermal paste like every coulpe moths to ensue good heat transfare?
a b K Overclocking
July 6, 2009 11:33:14 PM

^That old method of spreading a thin layer over CPU was good for non DHT coolers. It's not good for DHT coolers any more.

You may notice a 2-5C difference between the stock thermal paste ans AS5 esp. if they were both lapped. The HSF is pretty flat, but the CPU is not. Note: Lapping WILL void your Warranty!
July 7, 2009 6:48:15 AM

My problem with the "standard" method of applying thermal paste that way with a bare heatpipe-type cooler is that as the paste spreads out it will encounter those little valleys and air will get trapped in them. Trapped air = higher temps.

If the paste happened to spread out in a perfect circle with a perfectly regular expanding edge, then that might not happen. However, that's going to be pretty much impossible to achieve.
July 7, 2009 7:45:24 AM

oki ill follow your method proximon but could you post a pic of the thermal paste alredy applied onto the base of heatsink if you can
July 13, 2009 10:13:33 PM

Bump and STICKY! Great advice, genius method!
a c 239 K Overclocking
July 14, 2009 4:22:26 PM

If you lapp the heatsink base and the CPU heat spreader, applying thermal compound needs to be completely rethought, because the amount of thermal compound you need is almost zero.

The pea sized spread out method or the razor blade thin coat spread is applying way too much thermal compound, most thermal compounds are suggested not to use your finger because of the oils on your body.

However the finger method I've rediscovered is the best thermal compound applier of all the methods I've used in the past, use an alcohol pad to remove oils from your skin and apply a micro thin coating to both surfaces, allow no buildup what so ever, and mount your heatsink, then remove it and look at the thermal footprint.

You may be as shocked as I was to discover that the micro thin layer is almost too much, so you can adjust your thermal application to suit your actual metal to metal contact needs.

Keep in mind this is when both contact surfaces have been lapped leaving only microscopic imperfections to be filled, and I do mean microscopic.

Applying with your finger also allows you to feel the smoothness of the surfaces you're applying the compound to, just remember to clean the oils and sweat from your finger first.

Remember: The least amount of thermal compound you can apply to fill the imperfections is all you're after, for the best temperature end results.
July 14, 2009 7:52:16 PM

4Ryan6 said:

The pea sized spread out method or the razor blade thin coat spread is applying way too much thermal compound, most thermal compounds are suggested not to use your finger because of the oils on your body.



You missed the point.
-This is for HDT coolers. The grooves in between the pipes and aluminum cannot be lapped out.
-The pea sized amount is NOT the final application.
a c 239 K Overclocking
July 15, 2009 4:24:25 AM

Proximon said:
You missed the point.
-This is for HDT coolers. The grooves in between the pipes and aluminum cannot be lapped out.
-The pea sized amount is NOT the final application.


Didn't I say?

Quote:
If you lapp the heatsink base and the CPU heat spreader, applying thermal compound needs to be completely rethought, because the amount of thermal compound you need is almost zero.


I didn't miss the point I made the point, only the barest necessity of thermal compound to fill the imperfections is needed.



Are there grooves left after lapping a HDT heatsink?

Of course there is, but nothing compared to before the lapping process, but you only need enough thermal compound to fill them, so what point did I miss?

That is a Xigmatek HDT heatsink you're looking at.


July 15, 2009 10:19:58 AM

I see what you mean. No reason to waste TIM there. I think I would still do it the same way, just with smaller quantities.
a c 239 K Overclocking
July 15, 2009 2:21:41 PM

If the surfaces are unlapped your method is highly viable, I wasn't trying to discount what you've written, just add to it including the possibility that someone has lapped both the heatsink and the CPU heat spreader too.



When the two surfaces are lapped and perfectly flat the amount of thermal compound needed to fill the leftover microscopic voids is actually cut by about 90%.

This is a lapped AMD skt939 FX57.
July 15, 2009 6:46:36 PM

There we go, I made some changes that should help clarify what I mean and some cautions about the amount you leave on the surfaces when lapping. Thanks!

I'll have to run some tests soon.
a c 163 K Overclocking
September 4, 2009 7:01:17 PM

Item worth noting is that Prolimatech for example actually recommends against lapping as it decreases the performance of their milled base. It's designed to be slightly arched and made flat through the lapping process, performance actually drops.

http://www.tweaktown.com/reviews/2870/prolimatech_megah...

"The base is not exactly polished in its finish, or even level against a razor blade. Taking a quick look at Prolimatech’s site, I soon found out why. The say they have specifically milled the base in this fashion as it is part of the overall designs efficiency. They do not recommend to lap this base, as it voids your warranty for one, but also lowers the efficiency of the Megahalems."
a c 239 K Overclocking
September 4, 2009 7:19:27 PM

Wow thats an interesting discovery Jack
a b K Overclocking
September 6, 2009 7:17:13 PM

@JackNaylorPE: Yes they do have a point. HOWEVER, if one laps the CPU then they must lap the HSF as well and viceversa for best performance.
a c 239 K Overclocking
September 7, 2009 1:48:22 AM

Shadow703793 said:
@JackNaylorPE: Yes they do have a point. HOWEVER, if one laps the CPU then they must lap the HSF as well and viceversa for best performance.


Thats exactly right, and the warranty is gone for lapping the CPU anyway, so the Heatsink warranty shouldn't really matter, its a lot cheaper than the CPU.
September 8, 2009 7:16:21 AM

4Ryan6 said:
If you lapp the heatsink base and the CPU heat spreader, applying thermal compound needs to be completely rethought, because the amount of thermal compound you need is almost zero.

The pea sized spread out method or the razor blade thin coat spread is applying way too much thermal compound, most thermal compounds are suggested not to use your finger because of the oils on your body.

However the finger method I've rediscovered is the best thermal compound applier of all the methods I've used in the past, use an alcohol pad to remove oils from your skin and apply a micro thin coating to both surfaces, allow no buildup what so ever, and mount your heatsink, then remove it and look at the thermal footprint.

You may be as shocked as I was to discover that the micro thin layer is almost too much, so you can adjust your thermal application to suit your actual metal to metal contact needs.

Keep in mind this is when both contact surfaces have been lapped leaving only microscopic imperfections to be filled, and I do mean microscopic.

Applying with your finger also allows you to feel the smoothness of the surfaces you're applying the compound to, just remember to clean the oils and sweat from your finger first.

Remember: The least amount of thermal compound you can apply to fill the imperfections is all you're after, for the best temperature end results.


I wrap a Ziploc bag around my finger to smooth a tiny amount of AS5 across the heat pipes on my Xiggy. It seems to work well and temps are low, especially with an aftermarket fan.
a c 239 K Overclocking
September 8, 2009 11:12:09 AM

one-shot said:
I wrap a Ziploc bag around my finger to smooth a tiny amount of AS5 across the heat pipes on my Xiggy. It seems to work well and temps are low, especially with an aftermarket fan.


The Ziploc bag or equivalent is much better than directly on your finger like I did, AS5 is extremely stubborn to remove from your skin, even with numerous alcohol pads, traces are still left behind on the skin but Lava soap finishes the cleaning, however it would be best overall to avoid the direct skin contact.
a b K Overclocking
September 8, 2009 2:53:43 PM

I used surgical gloves available at local pharmacies like Walgreens or CVS. They are very cheap and effective.
September 8, 2009 4:28:45 PM

4Ryan6 said:
The Ziploc bag or equivalent is much better than directly on your finger like I did, AS5 is extremely stubborn to remove from your skin, even with numerous alcohol pads, traces are still left behind on the skin but Lava soap finishes the cleaning, however it would be best overall to avoid the direct skin contact.


Hehe! The lava soap only works because it removes the first layer of skin as well as the AS5.

Next time, try using a product like Goo-B-Gone instead of alcohol. Even masking tape does a pretty good job if you run the affected skin areas across the sticky side.

Direct skin contact should be avoided, but not because the AS5 is caustic, but because the oils from your skin don't mix very well with the AS5. While I wouldn't recommend eating the stuff, it's mostly silver compounds and mineral oil, both of which aren't known to cause problems to your body at the given dosage you would encounter while applying the TIM.
a b K Overclocking
December 4, 2009 10:15:03 AM

Good discussion and good information! The single edged razor blade approach is good; so is the ziploc wrapping approach. I will look at the microinch finish of the CPU and heat sink, and then decide which method to use. A sandwich bag (0.5 mil) might give me a better feel when using fingers.

What I would have liked to see from the manufacturers (CPU, heat sink) would have been the relative heat transfer coefficients (quantified). Without that, checking temps would only yield approximate (ball park) numbers.

Some day we may see thermal compounds on a small evaporative non-shedding patch which will leave a micro thin residue of the thermal compound. The amount of this layer can be controlled by the viscosity of the thermal paste as well as the carrier. An environmentally friendly product(s) can be developed for this purpose. This will surely reduce the amount of guesswork and expertise required for the proper application of thermal paste. (Computer manufacturers can use robotic spray dispensing equipment to meter out the right amount). Analogous to automotive fuel injection. Vapor deposition is also a viable alternative.

I guess, the chip guys and the CPU cooler guys are too busy with the nano chips technology, to think of ways to cost reduce and simplify computer building.
January 27, 2010 12:43:19 PM

i have the Xiggy hdt-s1283 on a amd phenom ii x3 720 (unlocked 4th core, OC'd to 3.6 at 1.45v, idle ~30C, full load 51C)

I used Dynex TIM, spread it on the Xiggy using a plastic card, filling in all the cracks between the tubes and leaving a very thin layer.

I've been told by a few people to change to MX2 or equiv and to get a different cooler (Hyper 212+) so I can turn the cooler to point the fan out towards the case exhaust fan.

I'm reluctant to spend more money on another cooler, when I think between either re-applying the Dynex differently, or maybe some MX2 and maybe spending a few dollars on a better fan might yield the same results.

Any thoughts?

How are you guys lapping these top-heavy tower HDT coolers?

I'd like to try and acheive the best cooling bang for the buck with what I have or minimal modification.
January 27, 2010 8:26:06 PM

Installing the other fan in the front of your case would be the higher priority....

Having said that, this thread is actually referenced in several current guides, so I would prefer that you not put unrelated photos in. Starting a new thread and sending PMs to the folks you would like to respond works wonders.

There were some guides to lapping heatsinks here, but perhaps they were unstickied.
March 6, 2010 10:50:40 PM

does the OP's rules still apply to ocz freeze?
March 7, 2010 6:21:13 AM

jdazer said:
does the OP's rules still apply to ocz freeze?


Well, the whole idea here is to avoid air bubbles in the grooves in HDT coolers. I think OCZ freeze is a bit thinner... but if it will stay in the grooves well enough to finish installation I would say yes. That's the key. Will it survive inverting the cooler and applying it to the CPU? You'll have to tell me.
a c 239 K Overclocking
March 7, 2010 1:45:13 PM

The best route to take is test fitting the Xiggy and inspecting the contact footprint, you should see the picket fence effect on the CPUs heat spreader, that means you've properly filled the grooves.

Note in the pictures there is no squish out around the sides, the coverage is even and very thin and that is what you want, no matter how many different suggestions to apply thermal compound you encounter, the end result needs to be whats shown here, for your best cooling results.

I do not want to think?, I may have properly applied the thermal compound, I want to know its properly applied, and that takes a little more effort on your part of the installation.



March 7, 2010 8:07:52 PM

There ya go.

Although I favor not having wall-to-wall coverage like that because then it's harder to say if you have too much or not. If the TIM spread out as far as it could but didn't quite make it to all the edges, I still have everything covered but know that it's as thin as it can be.
a b K Overclocking
March 8, 2010 2:54:51 PM

Some additional information:
During the 4th. week of Feb 2010, I upgraded my EVGA E758 motherboard to EVGA's 760 Classified board. Since I had to re-install the CPU, heatsink, etc., I decided to try out 'lapping'. The heatsink base had enough mass to prevent it from bowing under thermal stresses; therefore, lapping (or polishing as I would like to call it) was an easy decision. I lapped/polished the CM V8 base to a flatness of approx. 4 microinches in 1 1/2 inches, and obtained a surface finish of approx. 4 microinches.

Now the CPU. The heat spreader of the CPU is a shallow drawn component made in a progressive die. There is no way of predicting the mode of expansion, other than by elaborate lab experiments, or by actual 'before and after' temperature measurements. Therefore, I decided to leave the CPU alone. (It might be a good idea to polish off the high spots at the corners of the heat spreader).

To apply AS5 thermal compound to these surfaces, I wrapped my finger with plastic wrap, and applied the compound. I did need to level off the compound using a single edged razor blade. After a thin layer of compound was applied to both surfaces, I re-assembled the system and checked the temps.

I observed a consistent 5 to 8 degrees C drop in the CPU temps under identical before and after conditions. This drop in temps can be attributed to the slight concavity in the base of my V8 heatsink before I polished the base flat. CPU unchanged.

An interesting point. After changing the motherboard, I had to call Microsoft to get Windows 7 to be activated. Must be done again should I need to re-install Win 7. The activation code on the original OEM package does not work when the motherboard is changed.

Cooler Master HAF 932 - EVGA 141-BL-E760-A1 Classified - Intel i7-920 DO @ 2.66 GHz (OC to 3.0 GHz) - CM V8 - Corsair CMPSU-850HX 850W RT - EVGA 896-P3-1171-AR - GTX 275 R (GeForce) - Corsair Dominator DDR3-1600 – (6x2GB) 12GB Model CMD12GX3M6A1600C8 - Windows 7 – 64-bit - Seagate 1TB x3 (2 of these are in RAID1 for data) - ST31000528AS - Two R/W DVD - Scythe Fans Controller - Logisys Multi card reader - Dell 2208WFP 22” LCD Monitor - Bose Companion II Speakers - External backup: Seagate 1TB
a b K Overclocking
April 26, 2010 12:51:35 AM

Quote:
The Ziploc bag or equivalent is much better than directly on your finger like I did, AS5 is extremely stubborn to remove from your skin, even with numerous alcohol pads, traces are still left behind on the skin but Lava soap finishes the cleaning, however it would be best overall to avoid the direct skin contact.


This is just another reason that I prefer Tuniq TX-2 grease over AS5. It spreads like butter, wipes right off, and from what I've seen, has about a 1-2 degree performance lead over AS5. Oh, and it's cheaper, too...
This might just be my system, though... has anyone done any tests comparing these two?
!