This is probably more of a physics question but it has me wondering. I bought this handy metal bracket for mounting TO-3 transistors to bigger heatsink. It's about 1/8" (0.125) thick. Would moving up to the next size give much of an advantage? Say the same thing but made from 3/16" (0.187) thick aluminum. Thanks.
I doubt it. Aluminum is pretty conductive. The vast majority of the thermal resistance will be from the heat sink that the bracket will attach to.
Tom
Tom
The limiting factor is probably the thermal resistance between the angle and the heatsink. You can calculate the temperature rise in the transistors based on the case/angle and angle/heatsink thermal resistances.
it typically looks like the L used on JLH kits with capacity multiplier, it's light but it does the job.
what worries me is the number of holes on the fixing side vs the state of flatness.
either I will leave as is but I will polish and surface the joint plane, or I will make two more holes, or I will use a solid piece that I will drill in the alignment with three holes and that I will lose in sandwich between the screws and the L to ensure the distribution of the tightening.
it's just my experience.
what worries me is the number of holes on the fixing side vs the state of flatness.
either I will leave as is but I will polish and surface the joint plane, or I will make two more holes, or I will use a solid piece that I will drill in the alignment with three holes and that I will lose in sandwich between the screws and the L to ensure the distribution of the tightening.
it's just my experience.
Thanks all. Sounds like it will be fine. I had similar thoughts of adding more holes to mount this to the larger sink. They were indeed advertised as being for a JLH kit. Picked up a pair for $15 on Aliexpress.
Put a piece of abrasive paper upside down on a piece of glass, or a few news papers. That means abrasive side up, backing side down.
Then gently rub the bracket over it, you can remove high points like that.
And when you attach it to the heat sink, check it out as well, and put a thin layer of heat sink compound in the gap to improve heat transfer.
Aluminum is like $6 a kilo here, $15 seems high for a pair of drilled pieces of extruded angle.
Then gently rub the bracket over it, you can remove high points like that.
And when you attach it to the heat sink, check it out as well, and put a thin layer of heat sink compound in the gap to improve heat transfer.
Aluminum is like $6 a kilo here, $15 seems high for a pair of drilled pieces of extruded angle.
Yes aluminum is a cheap material but now find a shop to do the machining and then another shop to anodize them. I paid $15 in total and they traveled from China right to my mailbox. I don't know how anyone can say that's not a bargain.
I have my own pillar drill, and a lathe.
Also angle grinder, pistol grinder, couple of impact drills.
Wire EDM, mills, regular EDM and so on available near by...1.5 km max.
All I need to do is supply the metal, the machines are there for me to use.
So I think differently, and use what is available at hand.
And the shipping would have cost about half the amount paid.
So, convenient, and reasonable for you.
Just check for flatness, and use thermal grease, as above.
Also angle grinder, pistol grinder, couple of impact drills.
Wire EDM, mills, regular EDM and so on available near by...1.5 km max.
All I need to do is supply the metal, the machines are there for me to use.
So I think differently, and use what is available at hand.
And the shipping would have cost about half the amount paid.
So, convenient, and reasonable for you.
Just check for flatness, and use thermal grease, as above.
You can do the back-of-an-envelope maths. Aluminium alloys can be around 100W/m/K, although pure Al is much better. Given a path length of say 4cm, 3mm thick by 15cm wide gives a conductivity of 100 * 0.003 * 0.15 / 0.04 = 1.1 W/K, so 50W dissipation is 44C drop or so. Compare to the conductivity of a typical TO3 transistor and you'll see this thin sheet is probably comparable to the thermal resistance of a pair of TO3's, and if you know the thermal resistance of your heatsink you can figure out if its worth improving on.
So going from 3.2 mm to 4.75 mm thickness:
3.2 mm: 100*0.0032*0.15/0.04 = 1.2 W/K, so 42 ºC temperature difference for 50 W dissipation
4.75 mm: 100*0.00475*0.15/0.04 = 1.8 W/K, so 28 ºC temperature difference for 50 W dissipation
That's actually a bigger difference than I would have thought. But probably not enough of a difference to wreck the design.
And just for fun let's try 1/4" and 10 mm:
6.3 mm: 100*0.0063*0.15/0.04 = 2.4 W/K, so 21 ºC temperature difference for 50 W dissipation
10 mm: 100*0.01*0.15/0.04 = 3.8 W/K, so 13 ºC temperature difference for 50 W dissipation
Though, I suppose we need to include the spreading resistance for the thicker pieces.
Tom
3.2 mm: 100*0.0032*0.15/0.04 = 1.2 W/K, so 42 ºC temperature difference for 50 W dissipation
4.75 mm: 100*0.00475*0.15/0.04 = 1.8 W/K, so 28 ºC temperature difference for 50 W dissipation
That's actually a bigger difference than I would have thought. But probably not enough of a difference to wreck the design.
And just for fun let's try 1/4" and 10 mm:
6.3 mm: 100*0.0063*0.15/0.04 = 2.4 W/K, so 21 ºC temperature difference for 50 W dissipation
10 mm: 100*0.01*0.15/0.04 = 3.8 W/K, so 13 ºC temperature difference for 50 W dissipation
Though, I suppose we need to include the spreading resistance for the thicker pieces.
Tom
No direct relationship between those prices.
One is a commodity/raw material and can be sourced from 1000 suppliers; the other is a processed product, include extruder machine, dies, shop space/rental/tax, salaries, aluminum bar cutter, puncher and dies, anodizing, packaging, plus manufacturer and reseller profit in the mix.
Here in Argentina cheapest raw aluminum available is recycled ingots cast out of old engine blocks, actually rasonably good quality, and costs some U$2.40 a kilo or some U$2000 a Ton
But again, there is a long and (relatively) expensive path from raw ingots to finished ready to sell product.
Besides that, those are not "generic" brackets but made for a specific project, a JLH kit.
That alone carries a price differential.
@mark - I think your estimate is pessimistic. Those TO-3 base-emitter holes are about 11mm apart and the back of the heat bracket is about another 11mm. The centre of the transistor can would therefore be approx. 16 mm from the mounting plane. If each transistor section is treated as a uniform heat conductor about 50mm wide that makes the thermal resistance about 0.5K/W per transistor, assuming 200W.m/K for the aluminium. Though what the actual alloy resistance is would need to be measured I guess. The transistor insulator is likely to add another 0.4K/W. That's creeping up. Depending on the transistors the junction thermal resistance could be between 0.8 and 1.5K/W so taking a middle figure it's getting to be about 2K/W per transistor. Or, if nominal operating voltage and current, 16W per device , giving junction temp rise of 32C above the main heatsink. The main heatsink has to dissipate heat from both transistors and the capacitance multiplier.
Elsewhere, in the JLH thread, the main heatsink thermal resistance is recommended to be 0.5K/W (both transistors) and if your bracket is much more than this it would not be suitable.
I've typically used L brackets for power transistor mountings and three M5 holes seems adequate to mount on a flat based heatsink. With silicone compound to maximise the thermal contact.
Elsewhere, in the JLH thread, the main heatsink thermal resistance is recommended to be 0.5K/W (both transistors) and if your bracket is much more than this it would not be suitable.
I've typically used L brackets for power transistor mountings and three M5 holes seems adequate to mount on a flat based heatsink. With silicone compound to maximise the thermal contact.
Here in Baroda, and in fact most of India, the extrusion plants sell many types of section, in 12 or 20 foot lengths.
15 foot pieces are unusual.
The traders / dealers do the chopping if needed, or I get it done by a fabricator with a chop (abrasive) saw.
The drilling, and so on are easy.
Anodizing, again we have many shops, going there is the hassle.
Though in most cases, if the aluminum is good quality, will not corrode here in our local climate, so anodizing is a decorative option.
Generic net image, a typical shop storage. No ties to seller.
Correction: The price seems to have reduced to the $3.5 to 4 per kilo range for sections, and may be less in truck load quantities.
Also, if it is a standard extruded section, all the bracket maker is doing is cut to length, and make holes, with a press if the volumes are large. Then anodize.
Not complex, I think.
15 foot pieces are unusual.
The traders / dealers do the chopping if needed, or I get it done by a fabricator with a chop (abrasive) saw.
The drilling, and so on are easy.
Anodizing, again we have many shops, going there is the hassle.
Though in most cases, if the aluminum is good quality, will not corrode here in our local climate, so anodizing is a decorative option.
Generic net image, a typical shop storage. No ties to seller.
Correction: The price seems to have reduced to the $3.5 to 4 per kilo range for sections, and may be less in truck load quantities.
Also, if it is a standard extruded section, all the bracket maker is doing is cut to length, and make holes, with a press if the volumes are large. Then anodize.
Not complex, I think.
Last edited:
For the record I’m not building a JLH but something similar. Dissipation is around 25w per channel. https://www.diyaudio.com/community/threads/my-germanium-based-amplifier.388996/
thickness works more as a heat spreader which will spread heat out from the hot spot created by any transistor mounting area.
thickness is helpful as a heat spreader, so its more about the thickness and width it can spread too.
as with any heatsink spreading the heat outward allows more cooling fins to be used.
thickness is helpful as a heat spreader, so its more about the thickness and width it can spread too.
as with any heatsink spreading the heat outward allows more cooling fins to be used.
there you go heat spreader. but to both sides with fins.
But the angled method to heatsink has been done many times as well.
heatsink is a heatsink, every one has thermal rating.
but that style heatsink in #16 post. I have seen many times with voltage
regulators. very common in vending machines or arcade games.
its the fun of TO-3 packages , love it
But the angled method to heatsink has been done many times as well.
heatsink is a heatsink, every one has thermal rating.
but that style heatsink in #16 post. I have seen many times with voltage
regulators. very common in vending machines or arcade games.
its the fun of TO-3 packages , love it
Last edited:
Mass absorbs heat energy until it can't anymore. Surface area dissipates heat. (The mass also helps the semiconductor device from thermal oscillation.)
Those heatsinks were good for breadboarding and testing but not large enough. The boards are going to be mounted in a chassis from the diyaudio store. https://diyaudiostore.com/collectio...oducts/mini-disspiante-2u?variant=12171525316