The MF35-151.5 from Conrad looks ideal [...]
I agree.
However, is it truly 0.21K/W?
Great question. I typed the dimensions into this online heatsink calculator:
https://www.heatsinkcalculator.com/heat-sink-size-calculator.html
I chose 80/0.21 = 380.95 W dissipated as the Conrad heat sink is specified for 80 ºC temperature rise and 0.21 K/W. The calculator says the heat sink should be 419 mm wide and have 52 fins. In reality it's 350 mm wide with 35 fins.
I tried with the dimensions of a 0.4 K/W heat sink that I have on hand and the calculator spits out results that are ~10-15% higher than reality. So I'm thinking the Conrad specs are probably pretty close to reality.
However, where Conrad's specs do diverge from reality is on the 80 ºC temperature rise. That would mean that the heat sink reaches 105 ºC in a warm living room (25 ºC). Thats way too hot. Heat sinks get more efficient as the delta-T rises, so Conrad's specs are optimistic there. Normally heat sinks are specified for a delta-T of 70 ºC so Conrad is cooking the books a little, though not overwhelmingly so.
I recommend staying below 60-65 ºC heat sink temperature so you don't burn your fingers if you touch the heat sink. For a 65 ºC heat sink temperature at 25 ºC ambient, you're looking at a delta-T of 65-25 = 40 ºC. The heat sink will be less efficient there, so you need to multiply the 0.21 K/W specified by Conrad by a correction factor.
I've attached a table I grabbed from Aavid's website before they were bought out and the information vanished. Note that the correction factors are for 75 ºC, not 80 ºC. Close enough! The correction factor is 1.170, so your nice 0.21 K/W heat sink (specified) is actually 1.170 * 0.21 = 0.25 K/W.
Assuming you're designing for max power and run the Modulus-686 on ±36 V, you're looking at 100 W dissipated in the heat sink for an 8 Ω load driven to clipping with a music signal (14 dB crest factor). 186 W dissipated when used with a 4 Ω load driven to clipping with a music signal. So you'll need a thermal resistance of 40/100 = 0.4 K/W for the 8 Ω case and 40/186 = 0.22 K/W for the 4 Ω case.
This math assumes that the amp is driven continuously with a signal that has 14 dB CF and the peaks at clipping. Most music doesn't look like that. And few - very few - run the amp at clipping for any length of time. So I would say that your 0.25 K/W Conrad heat sink would be fine for music reproduction into 4 Ω. You could probably even get away with 0.40 K/W for that use case, but I would then definitely fit the heat sink with a thermal switch that turns the amp off once the heat sink temperature exceeds 60-65 ºC.
In case you're wondering where the 60-65 ºC comes from: I'm not aware of any standards that dictate how hot audio gear is allowed to get. But there are standards for appliances, so I use that as a guide. The controls (knobs) on the stove in your kitchen are allowed to reach 55 ºC during normal operation. Any surface that is
not a control but can be touched by the user during normal operation is allowed to reach 65 ºC. So that's what I aim for.
Those further interested in the math behind this should have a look here:
https://neurochrome.com/pages/thermal-design and here:
https://neurochrome.com/pages/power-supply-design
Then there's the Wakefield-Vette 127737 (0.19 C/W), but that has the fins mounted incorrectly if I were to use the 12" depth as the depth of the chassis. (Horizontal instead of vertical.)
Yeah. You definitely want the fins vertical. Unless you're doing forced air cooling.
Tom