Heatsinks for DIY LED lamps
- Thread starter SupraSPL
- Start date
nevergoodenuf
Well-Known Member
A jigsaw with the right blade would work great. Even a hacksaw, it's only aluminum.
testiclees
Well-Known Member
thanks guys. I got a fresh, quality blade on my saw. im gonna try that first (observing good safety practices).
HydroDC
Well-Known Member
It's gonna be LOUD so you'll find out if you've really got a pair. But you'll be fine with your new sharp carbide blade - slow and steady wins the race.
testiclees
Well-Known Member
C'mon bro my name is Testiclees!
littlejacob
Well-Known Member
Hello weeders
I plan to order vero18 3000k 80CRI for a blooming box.What is the best? 5 on a hlg-120-1050B or 6 on a 185-1050B and passive cooling is going to be too expensive for me active one would be better, so do you know Akasa AK-865 or Asetek vapo chill or Evercool PTO3-9232CP? Anyone know how many RPM or CFM a fan should run to be good and the size of the heat sink you need for 30, 35w of cob and if 18 cob are ok for a 105x105x135cm box (3.5x3.5x4.5 feet i use to run a 600hps in it).
Thanks from across the ocean★
I plan to order vero18 3000k 80CRI for a blooming box.What is the best? 5 on a hlg-120-1050B or 6 on a 185-1050B and passive cooling is going to be too expensive for me active one would be better, so do you know Akasa AK-865 or Asetek vapo chill or Evercool PTO3-9232CP? Anyone know how many RPM or CFM a fan should run to be good and the size of the heat sink you need for 30, 35w of cob and if 18 cob are ok for a 105x105x135cm box (3.5x3.5x4.5 feet i use to run a 600hps in it).
Thanks from across the ocean★
CellarDweller
Well-Known Member
That's really rather sexy sir.
harleydavid
Member
I unscrewed the fans and removed that little bracket, as for flatness i spend a few min on a mirror with P800 grit [tip from supra if i remember], although the face looks like it's been machined anyway,
overall happy with them, no more overghanging cobs and i expect them to last a lifetime for that price
sanjuan
Well-Known Member
The flatness of the 3.950in HeatsinkUSA product was disappointing. I'm not surprised, given the depth of the fins but a fan will not lie flat on top and the bottom surface is cupped at least 0.01inch. I've spent way too much time draw filing and sanding--should have requested a machining quote for fly cutting or surface grinding.
I have CXB3590s to mount and the surface isn't just cupped, I can feel waviness. The extrusion is flat along its length, at least. I was hoping to come up with an easy DIY COB bar but this shoots it right down from the get-go.
I have CXB3590s to mount and the surface isn't just cupped, I can feel waviness. The extrusion is flat along its length, at least. I was hoping to come up with an easy DIY COB bar but this shoots it right down from the get-go.
Last edited:
Getgrowingson
Well-Known Member
What would be the smallest/ cheapest heatsink type at 44" from husa I could use with a 200w (4x3590@50w each)? I wanted to use the 5.88 inch heatsink but it's going to cost me over 350$ for 3 including shipping. Starting to mount up costs on this light I want to build and it's getting frustrating to say the least.
SupraSPL
Well-Known Member
@sanjuan I used to expend untold amounts of energy flattening heatsinks but when I got the tools to test the actual difference in temp droop, to my surprise there was no measurable difference between a carefully flattened and polished heatsink VS a raw extruded aluminum surface. Very unintuitive but due to the high thermal conductivity of modern nano-particle thermal paste, the relatively thin layer and the relatively low heat density in the thermal path, there is nothing to be gained. I am known as efficiency nazi, you can bet that if there was something there I would be all over it. So take heart and go ahead with your COB bar 
I run mine soft ~ 25W each and temp droop is consistently less than 1% and more likely less than, .5% although it is hard to measure with that level of precision.
I run mine soft ~ 25W each and temp droop is consistently less than 1% and more likely less than, .5% although it is hard to measure with that level of precision.
sanjuan
Well-Known Member
My design is a pair of 3000K driven at 75W each, under 100mm diameter glass lenses. (I think somebody here said lenses shouldn't be used at over 75W?)@sanjuan I used to expend untold amounts of energy flattening heatsinks but when I got the tools to test the actual difference in temp droop, to my surprise there was no measurable difference between a carefully flattened and polished heatsink VS a raw extruded aluminum surface. Very unintuitive but due to the high thermal conductivity of modern nano-particle thermal paste, the relatively thin layer and the relatively low heat density in the thermal path, there is nothing to be gained. I am known as efficiency nazi, you can bet that if there was something there I would be all over it. So take heart and go ahead with your COB bar
The flatness of the heatsink is commercially acceptable, I guess, but I'm coming from a PC modding background.
Edit: I plan to epoxy K type thermocouples to the COBs (under the lenses) on a test unit.
My heatsink is similar to the one pictured in post #832 but not as bad as that.
I wrote the wrong width in the post above, it is actually the 3.945 extrusion.
Last edited:
HydroDC
Well-Known Member
Supra certainly knows a lot more about this than I do, but I wouldn't be putting my COB on the one pictured in Slitlos' picture. As I recall, Supra was spending hours getting to a mirror like finished but, through testing, determined that that level of effort produced negligible benefit. I think there's a world of difference between going for a mirror finish vs. trying to get flat enough to insure good heat transfer.The flatness of the heatsink is commercially acceptable, I guess, but I'm coming from a PC modding background.
To get enough TIM to fill the gap that Slitlos had would seem to negate much of the effort DIYers are making to control the heat from high powered COBs. You say that your gap isn't as bad as that, but I'd sure be getting out the belt sander.
In the interest of full disclosure, I did stick some 240 and 360 grit paper on my jointer bed to flatten the central area of the Alpine 64 Plus coolers I used in my build, so maybe I'm way too far over on the OCD scale for valid input.
I hope Supra comments further on this issue.
sanjuan
Well-Known Member
I ordered a proper set of Nicholson 12 inch flat files: (double cut) bastard, second cut, and smooth. I've been struggling with a 10" smooth that constantly clogs and 60 grit emery cloth from China that immediately wears out. A mirror finish has never been my goal for a heatsink. I've bought a half dozen of these 16" bars.
My friend has my Makita jitterbug palm sander or else I'd probably use that to knock down the "shoulders" and then smooth file the COB-heatsink contact area. Using just files should result in a flatter surface.
My friend has my Makita jitterbug palm sander or else I'd probably use that to knock down the "shoulders" and then smooth file the COB-heatsink contact area. Using just files should result in a flatter surface.
Last edited:
SupraSPL
Well-Known Member
I address my own slight OCD by sanding heatsink bars quickly with 200 grit just to make them pretty, but if you are trying to get them truly flat you have to start with 60-80 grit, horizontally and vertically and then progress to finer grits from there. But the same thing holds true, it makes absolutely no difference, tested many times at high dissipation wattage, on CPU coolers and extruded heatsink bars with a variety of TIM. Very unexpected but very welcome data.
So apparently the thermal interface is not a bottleneck in our designs. What DOES make a difference is the heatsink itself and the drive current. The best CPU cooler I have come across, copper core blow-through style like the Rosewill RCX-Z300 reduces temp droop by half compared to something like the Alpine 11. But even those are insignificant gains in output compared to the gains from driving the COBs at a lower current (current droop). And if you are running soft your temp droop can easily be less than 1%, so there is even less reason to spend anytime on the heatsink surface in that case.
When it comes to the ideal heatsink for bar style, look for a thick base, thick fins and use sufficient surface area. If passive cooling, you want wide fin spacing and short fins, so the air between the fins can convect on its own, but still with sufficient surface area. Even better if they benefit from slight airflow from circulation fans.
Last edited:
Greengenes707
Well-Known Member
Are your findings all with heatsink temps?
What about using a thermocouple on the Tc measuring spot to know exactly?
You use kapton tape most of the time right?...the measuring point is accessible for cxa/b's that way unlike with certain holders.
sanjuan
Well-Known Member
Hmmm . . . maybe I'll run three 72V COBs on my bar @ 700mA instead of two @ 1050mA.
I just received some 3.500 wide heatsinkUSA bars and I checked both types with feeler gages and a machinist parallel. I'm seeing 0.009 inch dish in the 3.500 extrusion and 0.0105 for the 3.945.
I just received some 3.500 wide heatsinkUSA bars and I checked both types with feeler gages and a machinist parallel. I'm seeing 0.009 inch dish in the 3.500 extrusion and 0.0105 for the 3.945.
HydroDC
Well-Known Member
So let's say the dip in the middle of the heatsink in Slitlos's picture is .75mm - right in the middle where a COB would be mounted. And let's also say that a properly applied grease layer is .25mm. So right in the hottest part of the COB, there's a 1mm thick layer of TIM. That much is OK? Has this horse already been beaten too much?