Painting Heatsinks

R

richie

New member
Hello,

Having searched for this on the internet, so far I have only found pretty uninformed, extreme reactions, and nothing that gives me any confidence in what the poster has put.

I've built a pretty nice computer so far, in a white, black and pastel green theme in a white Corsair 680T, but my Gigabyte Z77-ud5 board has blue anodized heat sinks on the PCH and VRM that is really spoiling the look. Apparently it is not possible to anodized to white so... spray paint.

Is this just a terrible idea, of course it will affect the thermal properties of the heat sink, but by an amount that I care about? The system is water cooled, so heat from GPU and CPU is being dumped outside the case. (I've come to the air section as the board is air cooled, hope this is the right place).

Has anyone tired spraying heat sinks and have any advice on how much they are affected thermally?.

Of course it will be the thinnest coat I can get away with, and I was thinking car spray paint (already have a can from doing a few other non-heat related parts).

Any advice would be great.

cheers
 
With heatsinks like that it will insulate them to a degree, and will indeed increase the chipset and mosfet temps. By how much will only ever be guess work.
 
If green and black are your other colours why not annodize to one or other of those and not get involved with the whole paint issue. As TTL says it will be difficult for anyone to give you any quantifiable answers.
 
There is no reason that you can't get them sand blasted and re-anodized. Who said you cant do that?
 
You may be able to get the answers you want posting this question in the modding forum, there are a bunch of people who are very knowledgeable and who may at very least be able to help you chose paint.

Also check out this thread
http://forum.overclock3d.net/showthread.php?t=44299
G-Dubs painted his heatsinks and goes into a bit of detail about how he did it and I'm sure he would be happy to answer some questions for you. As far as I can see you biggest problem is getting the right paints as you want to go for a white paint you want something that will be good with heat and most importantly keep its colour without tarnishing under those condition's this will be the tricky part and I'm not even sure it can be done.
You might want to check and make sure you heatinks aren't getting to hot and if they are go with a black for the heatsinks and balance it out with white on something that won't be getting hot. If they are running cool then you may be able to get away with it although you may want to make sure you keep good airflow to make sure they stay that way. This is just my speculation as I have no experience painting heatsinks and not much experience painting in general, so I would take some of the things I have mentioned and ask some questions over at the modding forum.
 
PEXON said:
There is no reason that you can't get them sand blasted and re-anodized. Who said you cant do that?
Click to expand...

From what I've been reading about the anodizing process, you can not anodize to white as the molecule for white dye is larger than the holes that are formed in the alu, so it will not take.
 
[FONT=Liberation Sans, sans-serif]Hi,[/FONT]


[FONT=Liberation Sans, sans-serif]Well firstly thanks for the posts about this, it is appreciated, however I was really hoping for some idea of how much heatsinks are affected by painting them in general. So that left only one thing to do... an experiment![/FONT]

[FONT=Liberation Sans, sans-serif]Ok, for people who don't want to read all of this:[/FONT]
[FONT=Liberation Sans, sans-serif]
[/FONT]
[FONT=Liberation Sans, sans-serif]Conclusion Summary[/FONT]
[FONT=Liberation Sans, sans-serif]My conclusion is that the paint has little to no affect on the thermal properties of heatsinks and I will be trying this on my Gigabyte z77-ud5. There is still some uncertainty in the results but you'll just have to read all of this to find out why. [/FONT]


[FONT=Liberation Sans, sans-serif]Don't Blame me![/FONT]
[FONT=Liberation Sans, sans-serif]I suppose I should state that the following data will not be 100% the same for every heatsink application and this post is meant as for info only, if you paint your heatsink and get bad results, don't blame me - you chose to do it![/FONT]


[FONT=Liberation Sans, sans-serif]The Experiment[/FONT]
[FONT=Liberation Sans, sans-serif]I tend to keep old bits of hardware when I get it, because you never know, but recently mainly because of the build of my new computer, I got rid of a load of old bits, how I regretted it! I put together a really crappy set up to test the difference in the heatsinks sprayed and not sprayed. Neither of the heatsinks were anodised just bare aluminium.[/FONT]

scaled.php


[FONT=Liberation Sans, sans-serif]The cable management is awesome I know! Though it's hard to tell, the heatsinks have been sprayed white in this picture.[/FONT]


[FONT=Liberation Sans, sans-serif]This is a 2.66GHz [/FONT][FONT=Liberation Sans, sans-serif]Celeron and an unknown motherboard. While there is a HDD and DVD in the picture it turned out the HDD did not work anymore and was the only IDE drive I could find laying around. But I wasn't about to give up so ran it without a HDD or DVD.[/FONT]


[FONT=Liberation Sans, sans-serif]The System was booted from a USB stick with a live linux distro (openSUSE). lm_sensors was used to monitor the CPU temp and the temp of the chip being cooled by the passive heatsink in the picture above.[/FONT]


[FONT=Liberation Sans, sans-serif]The system was left idle for 15 minutes to stabilise the temperatures, then:[/FONT]


[FONT=Liberation Sans, sans-serif]For the next 10 minutes the room temperature was noted, along with the temperatures reported by lm_sensors each minute.[/FONT]


[FONT=Liberation Sans, sans-serif]The fans were not speed controlled and were just on.[/FONT]


[FONT=Liberation Sans, sans-serif]These are the results:[/FONT]


[FONT=Liberation Sans, sans-serif]Not painted, system at idle:[/FONT]
nonpaintedidlechart.jpg

scaled.php


[FONT=Liberation Sans, sans-serif]Next, the system was put under load using stress http://weather.ou.edu/~apw/projects/stress/[/FONT]
[FONT=Liberation Sans, sans-serif]
[/FONT]
[FONT=Liberation Sans, sans-serif]The system was again left for 15 minutes under load, and the temperatures noted as before.[/FONT]

[FONT=Liberation Sans, sans-serif]Not painted, system stressed:[/FONT]
nonpaintedstressedchart.jpg

scaled.php


[FONT=Liberation Sans, sans-serif]Right, we now have base levels for the temperatues of the system. Next the system was taken apart and the heatsinks spray painted. The paint used was Halfords own brand Renault Glacier White (of course the contact area of the heatsinks was masked off and so not painted). Two liberal coats were applied.[/FONT]

scaled.php


[FONT=Liberation Sans, sans-serif]The system was reassembled, and the above tests were repeated recording the data for the temperatures at idle and stressed.[/FONT]

[FONT=Liberation Sans, sans-serif]Painted, system at idle:[/FONT]
scaled.php

http:


[FONT=Liberation Sans, sans-serif]Painted, system stressed:[/FONT]

paintedstressedchart.jpg
scaled.php


[FONT=Liberation Sans, sans-serif]Not everything went smoothly[/FONT]


[FONT=Liberation Sans, sans-serif]I'm not sure if it was when removing the heatsink or when putting it back on, but a plastic clip of the fan assembly that provides the pressure to the heatsink snapped off. I say it's due to the plastic becoming brittle with age, though Peanut (I want to say an aquaintance, but will go with friend has he helped me in this experiment so much, with only a small degree of moaning) would say it was because I was too heavy handed with it. I also blame Intel on the remarkably bad design either way.[/FONT]


[FONT=Liberation Sans, sans-serif] This meant that the test was compromised at this point. First I tried zip tying the assembly down, and checked the tempatures in bios, the CPU was high... very high, and nearly ended the experiment. As we had gone so far, I still wanted some data I could use. So the idle test was carried out. This showed a massive increase in the temperature of the CPU but not the other temperature. This seemed wrong as the paint should have affected them both by about the same amount. So looking at the CPU heatsink it became apparent that the zip ties were not working, so we tried putting manual pressure onto it, and this resolved the problem. However, this meant the side panel of the case had to be off to measure the painted set of results which may have increased the air flow, providing better cooling. As the pressure on the heatsink may also now be different, this too may have affected the results.[/FONT]


[FONT=Liberation Sans, sans-serif]A further area that may have affected the results was the thermal compound. This occurred to us a little too late. The first set of results had the original compound on, but when painted had new paste added, probably a different brand, which may have different thermal properites. In hindsight, we should have changed the compound for the first test. That said the passive heat sink (to my suprise) didn't have any thermal paste at all, so the results for the “other” temp would not have been affected by this.[/FONT]


[FONT=Liberation Sans, sans-serif]A nice graph of the results:[/FONT]

http:
scaled.php


[FONT=Liberation Sans, sans-serif]From this graph it would seem that a painted heatsink actually cools better than the non painted ones. Except for the other temperature under load.[/FONT]


[FONT=Liberation Sans, sans-serif]I would have to speculate that the drop in temperatures was due to the above errors in the experiment, as it was expected the paint would reduce performance and not improve it. That said when all is taken into account my conclusion has to be, the paint makes very little, if any difference to the performance of the heatsinks.[/FONT]


[FONT=Liberation Sans, sans-serif]If I had more old hardware I would have repeated the experiment when the clip broke, to eliminate the errors, but unfortunately I didn't[/FONT]


[FONT=Liberation Sans, sans-serif]So over all I have decided based on these results, to try spray painting the heatsinks on my z77-ud5 board. I will try to get similar data when I do paint them (hopefully next few days as I have this week off work) and will post back how it went.[/FONT]


[FONT=Liberation Sans, sans-serif]I hope someone finds this useful, and if any one has any comments about how this experiment could have been improved or anything about anything else, please feel free to post them, and I'll try to take them into account for the z77.[/FONT]


[FONT=Liberation Sans, sans-serif]Thanks for reading and thanks to Peanut for the help![/FONT]
 
Last edited:
Really interesting experiment. I would have expected a much more pronounced effect. It will be interesting to see the comparison on your new board also.
 
[FONT=Liberation Sans, sans-serif]Hello again,[/FONT]


[FONT=Liberation Sans, sans-serif]The heatsinks have now been sprayed, and two important things to mention early on is:[/FONT]



      1. [FONT=Liberation Sans, sans-serif]My motherboard still works and temperatures are just fine – Phew![/FONT]
      2. [FONT=Liberation Sans, sans-serif]I really am very pleased with how it's turned out aesthetically.[/FONT]


[FONT=Liberation Sans, sans-serif]I have also recorded the data as before (or at least similarly to before: there are differences). [/FONT]


[FONT=Liberation Sans, sans-serif]Let's start with a picture of the system I'm actually using, and more importantly the motherboard that's being used.[/FONT]


scaled.php


[FONT=Liberation Sans, sans-serif]This is a Gigabyte Z77-ud5h motherboard (in case any missed it in earlier posts). I also want to point out, the green coolant isn't the right shade of green I wanted, and will be changed at some point. Also this isn't a finished system, while it's fully functional, the black corsair cables are being replaced with white braided ones, I've just not finished making them all yet (not looking forward to the 24-pin), but I have done the 8-pin and a couple of others.[/FONT]


[FONT=Liberation Sans, sans-serif]More important to mention is the air flow in this case, there are 4×180mm fans on the front rads blowing into the case, the top rad has 2×120mm fans blowing out of the top of the case, and the main point to make is the rear fan is blowing into the case, it's not an exhaust like normal. Will go into why another time in another post, but just thought it important to mention.[/FONT]


[FONT=Liberation Sans, sans-serif]As can be seen in the picture, the heatsinks are anodised blue on this board unlike before. They will be sprayed over without attempting to remove the anodising.[/FONT]


[FONT=Liberation Sans, sans-serif]To monitor the experiment, as this was a fully working system, I went with CPUID Hardware Monitor and Prime95 for the stress test, all done on Windows 7.[/FONT]


[FONT=Liberation Sans, sans-serif]I have not recorded the CPU temperature as before, as it is a water cooled CPU and the water block isn't being painted. HWmonitor reports 5 temperatures from the motherboard:[/FONT]


[FONT=Liberation Sans, sans-serif]TZ00[/FONT]
[FONT=Liberation Sans, sans-serif]TZ01[/FONT]
[FONT=Liberation Sans, sans-serif]TEMPIN0[/FONT]
[FONT=Liberation Sans, sans-serif]TEMPIN1[/FONT]
[FONT=Liberation Sans, sans-serif]TEMPIN2[/FONT]


[FONT=Liberation Sans, sans-serif]As the bios on this board only gives 3 temperatures, and TZ00 and TZ01 didn't change thoughout any of the tests (nor have I ever seen them change) I have eliminated them from the results. [/FONT]


[FONT=Liberation Sans, sans-serif]Throughout the rest of this TEMPIN0, TEMPIN1 and TEMPIN2 will be referred to as T1, T2 and T3 respectively. [/FONT]


[FONT=Liberation Sans, sans-serif]The rest of the experiment was carried out just as before, 15 minutes to stabilise the temps and then 10 minutes of results taken each minute.[/FONT]


[FONT=Liberation Sans, sans-serif]Not painted, system at idle:[/FONT]
nonpaintedidle.jpg


[FONT=Liberation Sans, sans-serif]Right, the first thing to note is T3. This is a sub-ambient temperature so is making the delta temperature a negative number. This is clearly not reporting correctly (From a recent video I have seen, I think someone else might have had this happen too). However, This isn't really much of a problem as it's not so much the actual temperature we're looking at rather the change in temperatures (delta temps) comparing the non-painted results to the painted ones.[/FONT]


[FONT=Liberation Sans, sans-serif]Not painted, system stressed:[/FONT]
nonpaintedstressed.jpg


[FONT=Liberation Sans, sans-serif]Next the heatsinks were removed and spray painted with 2 coats using the same paint as before.[/FONT]


[FONT=Liberation Sans, sans-serif]For anyone who is thinking about painting the heatsinks on their Z77-ud5h, and comes across this, a few tips/statements (sorry if any of this is stating the obvious). [/FONT]



  • [FONT=Liberation Sans, sans-serif]The heatsinks come off and reattach really easily, they just screw on with spring loaded screws. All 3 and the heat pipe come off as one unit.[/FONT]
  • [FONT=Liberation Sans, sans-serif]The heatsink above the CPU and to the left of it use heat pads, but the bottom one does use thermal compound.[/FONT]
  • [FONT=Liberation Sans, sans-serif]When masking off the contact area of the two heatsinks using the thermal pads, a grease is left behind that stops masking tape from sticking, this needs to be cleaned off.[/FONT]
  • [FONT=Liberation Sans, sans-serif]The heat pipe can't be removed (or at least I didn't see any way) so will have to be masked off if you don't want it painted.[/FONT]
systemafter1.jpg


[FONT=Liberation Sans, sans-serif]As I said earlier, I'm very happy with the look of this (shame about the blue connector in the top left). The only other things that were changed were new coolant as the system had to be drained anyway; and the addition of the back plate on the graphics card. Neither of these would make any difference to our results. However, the dust filters were cleaned.[/FONT]


[FONT=Liberation Sans, sans-serif]As a side note, the backplate was always meant to be there, but was out of stock when I ordered the rest of the hardware and was holding up delivery, so it was dropped from the order, that's why it went in at this point. I must confess to stealing Tiny Tom's idea of spraying it white, that may have helped lead to all of this).[/FONT]
[FONT=Liberation Sans, sans-serif][/FONT]
[FONT=Liberation Sans, sans-serif]Painted, system at idle:[/FONT]
paintedidle.jpg


[FONT=Liberation Sans, sans-serif]Painted, system stressed:[/FONT]
paintedstressed.jpg


[FONT=Liberation Sans, sans-serif]Not a lot to say about the tables of data, but they do seem to show the same trends as before, so here's the graph...[/FONT]

graphad.jpg


[FONT=Liberation Sans, sans-serif]Those negative delta T's have made it a bit harder to follow, but it does seem to show the temperatures are reduced when painted just as before.[/FONT]


[FONT=Liberation Sans, sans-serif]Since doing this, the system is running completely normally and I have seen no ill affects at all. No heat build up over time, temperatures drop just as quickly after the system has been unloaded.[/FONT]


[FONT=Liberation Sans, sans-serif]Although I have to say, I find it hard to believe the paint is cooling better than not painted. Over a couple of beers Peanut and I discussed the results and I don't think either of us are convinced it's cooling better painted than not, it's just too counter-intuitive. Maybe the reduced temperatures are due to the air filters being cleaned out (I know for experimental purposes I should not have done this, but I use this computer, and so maintenance was done when it was convenient to do so). [/FONT]


[FONT=Liberation Sans, sans-serif] However, I can't help but feel that I'm now trying to finding reasons to justify my original hypothesis that painting the heatsinks would increase the temperatures, and it was by how much that was in question. Maybe I should believe the two sets of results that pretty much show the same thing, cooler with paint...[/FONT]


[FONT=Liberation Sans, sans-serif]One last thought is, if paint aids cooling why are heatsinks anodised rather then painted? Well, the only explanation I can think of, is that anodising is much more scratch resistant than painting, and the difference in temperature is minimal, so anodising is better, but not for heat reasons.[/FONT]


[FONT=Liberation Sans, sans-serif]I would love to know your thoughts on these results, and if any one else fancies contributing data to look at regarding this, I'd love to see it![/FONT]


[FONT=Liberation Sans, sans-serif]Thanks.[/FONT]
 
Nice write up dude.

To everyone looking, the top rad is not painted white, as it looks in those pics :D

The other thought on why painted was cooler than unpainted would be the thermal interface properties of the different mediums, that paint transfers heat to air better than aluminium and sufficiently well with aluminium.

One reason for manufacturers anodizing rather than painting would be longevity as the paint will discolour before the anodize.

Oh but no, I still refuse to believe the data, it must be wrong somehow, TTL hacked our HWmon too or something.
 
Very interesting, confusing....but interesting :p White will reduce radiative absorption but then I wouldnt have thought there would be a large enough flux incident on the heatsink from any particular source for that to make a difference...Plus it will reduce radiation from the heatsink its self, which again is probably not a huge factor either. So IF the results are a true indication of paint improving performance it would have to be improving the conduction properties somehow, which also sounds unlikely...I would assume for now that it is some error that makes it look better and that it is infact the same and the paint had a negligible effect. Still that in its self is good to know!
 
I would assume for now that it is some error that makes it look better and that it is infact the same and the paint had a negligible effect.
Click to expand...

I completely agree with that statement, it still just seems wrong that it would reduce the temps.
 
OK.. not sure what was going on.. but the image didn't display form me, but now seems to be working... Sorry for the cock up.

Bit more research into this and a leslie Cube seems to be the correct experiment for this sort of thing. I've not found out too much about this but found this video http://www.youtube.com/watch?v=WEPOhKWRyKU

It seems while black is the best absorber of radiative heat, white is just as good as an emitter, so this should mean being black or white will make no/little difference, also not something I would expect.
http://www.youtube.com/watch?v=WEPOhKWRyKU
 
I Find it quite easy to believe that it would improve thermal performance actually. To put it in a way everyone can understand: When you get out of the shower and still have water on you then you feel cold if you catch a bit of a breeze right? (:o) This is because the water wicks the heat away from your skin and passes it into the air more efficiently than from skin to air.
I can see this principle working on the coating on the heatsinks :D
Also @richie, there are desert tribes that wear black robes :p
 
mmm 2cents
applying a seal coating on a bare metal should impede thermal transfer, unless the
coating has a better coefficient of thermal transfer than the base metal. petroleum
base enamels and epoxies are horrible heat transfer products, due to their thickness
in coatings (excluding a base primer which will decrease transfer further). at best
a ceramic or metalic coating of higher thermal transfer can be applied, but at
a severe cost negating any recovery of delta vs investment.
a 2°-5° can be considered error without a controlled environment.
in racing, we worked very hard to make the engines reliable and durable when
a heat releated question came up. we tried mega-dollar paints, expoxies, polishes,
etc.. findings went
1. bare metal (rough cast deburr rough edges) or
2. bare metal with light color coat (prevented corrosion) color was no difference.

seal in the heat for looks... my 2cents

airdeano
 
You must log in or register to reply here.

Similar threads

P
Do some people never learn from the past?
Replies
2
Views
754
Pugsport
P
AlienALX
PUBG Rig V2.
Replies
1
Views
2K
AlienALX
AlienALX
AlienALX
Modular.
4 5 6
Replies
103
Views
15K
AlienALX
AlienALX
Dicehunter
Painting Metal ?
Replies
4
Views
3K
Pugsport
P
AlienALX
Wub Wub, time for something different.
4 5 6
Replies
111
Views
13K
AlienALX
AlienALX
Back
Top