Good day everybody.
I have many heatsinks of various sizes. Some reclaimed from older equipment and many bought from reputed sellers and eBay etc.
I check the flatness of the heatsinks using a flat ruler or square prior to mounting components.
Those recovered heatsinks are truly flat and can be reused. Those purchased from online or local shops are not that flat.
I have to sand the uneven surfaces to make any use of them.
What is your experience? Do you check them also? How you make it flat?
Regards.
I have many heatsinks of various sizes. Some reclaimed from older equipment and many bought from reputed sellers and eBay etc.
I check the flatness of the heatsinks using a flat ruler or square prior to mounting components.
Those recovered heatsinks are truly flat and can be reused. Those purchased from online or local shops are not that flat.
I have to sand the uneven surfaces to make any use of them.
What is your experience? Do you check them also? How you make it flat?
Regards.
Extruded aluminium will not be perfectly flat as it deforms a little going through the die, machined heatsinks will be flat. You pay a lot for the machining though. Ideally you'd take extruded heat sink sections, machine the back flat, then anodize for superior performance thermally. That's the good stuff, but not the cheap stuff.
Most heat sinks can be made reasonably flat by sanding on an upside down 400 to 600 grit silicon carbide abrasive paper, enough for use with a small amount of heat sink gasket or paste.
Keep the paper over several layers of newspaper, you will achieve a reasonable flatness.
Anodizing later is your choice.
Machining will achieve a similar result, has a chance of causing more damage, and involves sending it out, unless you have your own machine shop.
A slight taper before machining, unless you set it up with a dial gauge, will be magnified during milling.
A 0.1 mm flatness variation per 20 mm length variation can be compensated by heat sink gasket or paste, so do not be OCD about it.
Just try it first on a not so important piece.
Also, who is going to look at your heat sink for beauty, unless it is exposed and visible to visitors?
Keep the paper over several layers of newspaper, you will achieve a reasonable flatness.
Anodizing later is your choice.
Machining will achieve a similar result, has a chance of causing more damage, and involves sending it out, unless you have your own machine shop.
A slight taper before machining, unless you set it up with a dial gauge, will be magnified during milling.
A 0.1 mm flatness variation per 20 mm length variation can be compensated by heat sink gasket or paste, so do not be OCD about it.
Just try it first on a not so important piece.
Also, who is going to look at your heat sink for beauty, unless it is exposed and visible to visitors?
Thanks for the replies.
I am using 80 > 160 > 600 grit emery sheets now. Takes too long time.
Regards.
I am using 80 > 160 > 600 grit emery sheets now. Takes too long time.
Regards.
Remove the cutting burrs with a needle file.
Practice makes perfect.
I use silicon carbide water proof paper, with a few drops of water to keep it from loading up.
And I wash the sheets after a job, and then store for future use.
Wipe clean or wash the sinks after you are done, sometimes fine powder ends up in a place you do not want it to be...
Practice makes perfect.
I use silicon carbide water proof paper, with a few drops of water to keep it from loading up.
And I wash the sheets after a job, and then store for future use.
Wipe clean or wash the sinks after you are done, sometimes fine powder ends up in a place you do not want it to be...
How many of you have checked the flatness of the components themselves?
With a surface plate and feeler gauge, or otherwise?
How many data sheets mention the dimensional flatness at all in the drawings?
With a surface plate and feeler gauge, or otherwise?
How many data sheets mention the dimensional flatness at all in the drawings?
Your best option is a milling machine. Getting a good surface finish will require that very little material is removed in the final pass. You can polish after that but there's no need to if you've done the milling right. You can shave off some cost/time by only machining the area where the devices will be mounted.
Tom
Tom
I meant that the components themselves are not flat, and are intended to be used with a gap filler like paste or sheet.
Within limits, of course.
Running a light milling cut on Aluminum is a right pain, if you use a grinder make sure a huge coolant flow is used, along with a poros wheel and a light cutting feed, it tends to clog up wheels and heat up.
Within limits, of course.
Running a light milling cut on Aluminum is a right pain, if you use a grinder make sure a huge coolant flow is used, along with a poros wheel and a light cutting feed, it tends to clog up wheels and heat up.
Depends on the aluminium - extruded sections are usually a strong Al alloy (since the main use of extrusions is structural components), and most such alloys machine very well indeed. Pure, annealed Al is another matter, its claggy as mud and not suitable for machining. Its typically seen as sheet metal rather than other shapes in my experience.
The stamping process for the copper bottoms always produces an edge feature. The bottoms always seem to be concave.
Heatsinks always seem to be concave as well. Silpads in the 2 to 3 mil range appear to be the best option for DIYers.
John
For me, If I wish to flatten a heatsink, I would need a fly cutter wide enough for the job. As my mill is only 500 watts, I would probably need to pass a 3 inch fly cutter at .5 to 1 mil (thousanths of an inch) at a time, and the cutter really needs to be balanced as I have a mini mill, 11 by 6 working area. Some sinks I've seen are over .01 inch cupped.
200 emery followed by 400 is good enough for me.
The real thing I consider best is a good spring clip. Permanent force, as I don't have to worry about grease oozing out or silpad taking a new set, the springs avoid that worry.
John
Hmm... I am no expert here but I have had good luck with either an L bracket strategically tightened to flatten the sink or simple prepping of both the sink and the part with sandpaper and a keratherm pad. I do use a heat gun to ensure relatively consistent temperature between parts and both sinks.
I purchased the hi torque mini mill 4190 from Little Machine shop. Ways are flat in X and Y to under .001 for +/- 5 inches X, +/-3 inches Y (measured), so I can't complain. Z seems good and orthogonal, haven't needed to measure vertical accuracy yet.
Oddly enough, the DRO was off in all three axis, it was a constant in the software of the android. A call to the help line solved that in about 5 minutes once I realized there was a problem.
I also purchased a lathe, the 5200 (upgrade from the 5100 which included DRO's, metal handwheels, and something else I forget..)(ah, tailstock setup) they sent me the 5100 instead. So I called, splained, and they said pack it up and send it back no charge they will replace. Shipping ain't cheap... I told them, just send the kits that upgrade my unit and I will mount them so they didn't have to pay double shipping. They did, and also sent me a set of metal change gears for threading that was not part of the deal. They didn't have to do that, but did. That tells me they are a class act. I will always recommend a vendor that is a class act. I really didn't want to clean that greasy cosmoline stuff off another unit as well, what a PITA, or even carry the unit out of my basement.
Granted, it is a low cost relatively cheaply made unit (expected at the price point), but understanding the limitations allows me the ability to produce rather accurate parts. I find now that I am unhappy when I exceed 1 mil tolerances (despite requirements of 5 to 10 mils). (not sure if that's good or bad).
I also tradeoff with chipload abilility, I can't exceed 50 mil depth of cut using a 4 flute 3/4 endmill in aluminum...but alas, I still sleep at night.
The lathe from them, same thing. Great equipment as long as you remain within the limits.
As a result of my experience with them, I go to them for all the machining needs.. 1-2-3 blocks, 4 jaw's , pin gauge sets from 11 mils to 500, boring heads, measuring tools.. I certainly believe they are worth it for what they cost. If you need accuracy under .01 mil, this is not the place. Normal humans who need half mil tolerance (I chuckle at that statement), I absolutely recommend them.
I am learning so much, so fun.
ps.. lest it be forgotten, I can easily make stuff on the mill and lathe that I purchased from little machine shop that is below .001 inch tolerance in all dimensions, using their equipment and tools. The limits I have come across are my own understandings.
hashtag "I have no association with little machine shops, I just like what I have purchased and have enjoyed their customer service.."
John
Last edited:
That's fantastic. Thanks for sharing. I now know where to go should I get the milling machine lust again. 🙂
Tom
Tom
Mark:
I get the same section (in outer size) in a weight of 1.5 kilos from recycled scrap, and 2.7 kilos from fresh alloy. Same length.
The cheap stuff is old pistons from bikes and so on, plenty of tramp elements, and tarnishes fast.
As expected, both are different in machining.
The process control during extrusion also matters a lot.
That said, the sane approach is removal of high points, on components and sink, as outlined above, and a thermally conductive gap filler.
Machining is not needed.
I get the same section (in outer size) in a weight of 1.5 kilos from recycled scrap, and 2.7 kilos from fresh alloy. Same length.
The cheap stuff is old pistons from bikes and so on, plenty of tramp elements, and tarnishes fast.
As expected, both are different in machining.
The process control during extrusion also matters a lot.
That said, the sane approach is removal of high points, on components and sink, as outlined above, and a thermally conductive gap filler.
Machining is not needed.
Unless surface is too bad or your trying break some records, like a cpu/gpu overclocking enthusiast, just use thermal compound/pad and move on.