Making heat sinks.

[jcdammeyer]

For this one project I need 6 more heatsinks plus a few spares. Total is 8 units. But even just ordering a minimum of 10 the price is $45 each. And from an online CNC shop it's $55 but that's in US dollars. They were cut on a HAAS but that machine shop owner has since retired and his main employee didn't want to buy the business.
So I ordered what I thought were 3x3Dx50L cutters but today received 3x4Dx50L. The 4mm shank diameter is too large to fit between the 3.5mm fin gaps. So I compromised.
Redid the drawing to allow the first 6.35mm depth to be done as a 4.2mm width and then the subsequent depth to be done at 3.5 width. More or less just like the ones done by the HAAS although I think he has a 3mm mill and just widened it a tad for shank clearance.

There will still be some clearance with the new version. I may make the lower gap a bit wider to create a smoother transition.

Probably try milling some 6061 this week. Pictures of broken cutters to follow.

 

[slow-poke]

Nice looking heat sink.
Dual power supply or amplifier?

DFM, DFT. Custom heat sinks are expensive. <period!

Obviously better to design using something OTS vs. need to figure out that heat sink when the board gets back. I have spent decades fixing boards that "worked" but were not economically feasible to manufacture or test. That being said sometimes custom is the only way to go, and with large enough volume price can be okay.

I crammed > 3kW of resistors in a small 213x103mm HP chassis and only needed one small custom heat sink ( that I made in my mill.) A bit like a hair dryer coming out the back. It was a proof of concept prototype.

 

[Susquatch]

For this one project I need 6 more heatsinks plus a few spares.

Who did the initial heat transfer calculations on the heat sink?

Many times, they are simply rated by measuring their performance. Good enough is good enough.

I'd bet you could get away with doing something similar. Make the heat sink base as one part and buy or make two simple heatsinks to attach to the base using screws & heatsink transfer grease. If you make them, there is no real need for the 3mm spacing. In fact, wider spacing might be much better. It's more important to have airflow in there than it is to have small spacing. Basically, it boils down to surface area and heat removal capacity (air flow). So make them longer, wider, or whatever fits.

Basic heat transfer calculations are linear and the ratings are easy to establish. Just measure the power produced by the device and the power device temperature and then modify the heatsink and retest till you have something you can either buy or make.

 

[kstrauss]

I have had more success at heatsink fabrication using a slitting saw than trying to cut deep slots with an endmill.

 

[Susquatch]

I have had more success at heatsink fabrication using a slitting saw than trying to cut deep slots with an endmill.

Yes!

In fact, I have, and you can buy, thicker saws. 3mm should be no problem.

 

[YotaBota]

Might be viable to use a chop saw to remove the majority of the material and then mill it to final size.

 

[jcdammeyer]

I have had more success at heatsink fabrication using a slitting saw than trying to cut deep slots with an endmill.

That was the way I originally expected to do this.

 

[jcdammeyer]

Nice looking heat sink.
Dual power supply or amplifier?

DFM, DFT. Custom heat sinks are expensive. <period!

Obviously better to design using something OTS vs. need to figure out that heat sink when the board gets back. I have spent decades fixing boards that "worked" but were not economically feasible to manufacture or test. That being said sometimes custom is the only way to go, and with large enough volume price can be okay.

I crammed > 3kW of resistors in a small 213x103mm HP chassis and only needed one small custom heat sink ( that I made in my mill.) A bit like a hair dryer coming out the back. It was a proof of concept prototype.

Dual output DC to DC power supply 250W - 12.8V 10A, 24V 5A. There is a fan. Under full load and 21C ambient the DS1822 that supplies serial number and measures temperature reports 38C under full power. The unit is designed to operate in a closed cabinet that has fans exchanging air in desert heat. Ambient can reach 60C inside the cabinet I'm told. I've never had logged data returned so I have no idea if it actually has run in that environment.

The supply also measure 220VAC AC voltage from two difference sources: Genset and Shore Power. Also DC from Solar and Alternator. Can control two relays and reads two dry contact switches. Talks to the world via CANopen.

To deal with the slightly wider spacing between fins I did have to make the heatsink 4mm wider. It still fits just fine.

 

[jcdammeyer]

Oh and the Inductor the 127-2R2 gets hotter than the heat sink as measured with my IR reflective temperature sensor. This is without fan cooling.

 

[jcdammeyer]

This is how it's machined in the HAAS before it's flipped over and the bottom is done. There are recesses for parts and raised areas to position it so the SILPAD material gets compressed just enough for good heat transfer. I watched the HAAS profile around the top edges of the fins to round them so it wouldn't have to be done by hand. Therein lies the problem with the slitting saw. Sharp edges.

And of course everything is tested out first with the 3D printer. Lots of clearance on the new model for the 3x4Dx50L carbide mill.

I did check out what I could do with the slitting saw on a bit of aluminum sprue.

I actually cut the first one from an old PC CPU heatsink to test heat transfer etc. This was done on the JGRO CNC router (MACH3).

 

[kstrauss]

This is how it's machined in the HAAS before it's flipped over and the bottom is done. There are recesses for parts and raised areas to position it so the SILPAD material gets compressed just enough for good heat transfer. I watched the HAAS profile around the top edges of the fins to round them so it wouldn't have to be done by hand. Therein lies the problem with the slitting saw. Sharp edges.
View attachment 35727

And of course everything is tested out first with the 3D printer. Lots of clearance on the new model for the 3x4Dx50L carbide mill.View attachment 35728

I did check out what I could do with the slitting saw on a bit of aluminum sprue.
View attachment 35729

I actually cut the first one from an old PC CPU heatsink to test heat transfer etc. This was done on the JGRO CNC router (MACH3).

View attachment 35730

View attachment 35731

"Therein lies the problem with the slitting saw. Sharp edges." Why? Yes, it requires a tool change but I mostly clean any sharp edges with a 90-degree chamfer cutter. For something like your heatsink you could use a double-angle cutter (like a thread mill) to eliminate the need to rotate the work (or use a right angle attachment to hold the slitting saw).

 

[jcdammeyer]

Most of the work is with it horizontal so it's actually easier to use and and mill if one is making 10 or more.
If I were to build a single custom heat sink for say a motor driver I wouldn't hesitate to use the slitting saw.

 

[slow-poke]

Dual output DC to DC power supply 250W - 12.8V 10A, 24V 5A. There is a fan. Under full load and 21C ambient the DS1822 that supplies serial number and measures temperature reports 38C under full power. The unit is designed to operate in a closed cabinet that has fans exchanging air in desert heat. Ambient can reach 60C inside the cabinet I'm told.

Modern switchers are such a pleasure to use compared to what was available 20-30 years back. Especially the higher frequency ones. What part is under the heat sink?
switching frequency? Guessing 750kHz?

Not trivial power, hopefully you breeze through EMC testing.

 

[jcdammeyer]

Modern switchers are such a pleasure to use compared to what was available 20-30 years back. Especially the higher frequency ones. What part is under the heat sink?
switching frequency? Guessing 750kHz?

Not trivial power, hopefully you breeze through EMC testing.

We had to pass all sorts of tests including really high voltage transients when the systems were part of a Radiation Portal Monitor project. They are also used in military operations where emissions would be an issue.
The LTC3892 is on the bottom of the board. The 4 FETs are on the top of the board under the heat sink. It's a 6 layer board. The metal bottom of the IC is connected through to the top of the board which in turn with heat sink grease is connected to the heatsink.

I measure both the Input current and voltage from two different sources and the output voltage and current for both outputs. The unit has the ability to be remotely powered off. It allows the rest of the system to cleanly power down. It then shuts off the BMS which removes 24V from the rest of the system. Then shuts down the 24V output and 12V instrumentation bus.

At this point you have to manually switch the power switch OFF and then ON again to restart. Generally this only happens when the batteries have reached 5% SOC and the BMS would shut it down without warning at 3%. So we do it cleanly. Then an operator has to go to the site and figure out why perhaps the generator didn't start or what else when wrong that the batteries would reach that low an SOC. Usually it's go put fuel in the generator.