Heatsink size for Fan-less SMPS design is very big. Use multiple paralleled SMPS's instead?

[zenerbjt]

Dear Engineers,

We are using a Vicor DCM3623T75H26C2T00 DCDC module for a spec of

Vin=42-59VDC, Vout = 24V 5.8A.

The Module sits on a PCB, which is itself in a large-ish, sealed enclosure. The module will dissipate 17.2W of power at full load. The DCDC module’s internal temperature must stay below 110degC.

No cooling fans are allowed, just a heatsink.

DCM3623T75H26C2T00 DCDC module datasheet:
https://www.vicorpower.com/documents/datasheets/DCM3623x75H26C2yzz_ds.pdf



The DCDC module’s Rth(junction-case) is 1.93degC/W.
With the module’s power dissipation of 17.2W, the ambient temperature inside the enclosure will be 67 degC, we know this from temperature tests.

The heatsink used is the “180AB” by abl-heatsinks (datasheet below). Dimension is 215mm x 170mm x 7mm (height). Heatsink Rth = 0.3degC/W.

Anyway, DCDC Module internal “junction” temperature will be 67degC + 1.93*17.2W + 0.3* 17.2W = 105degC.

Since we are not using a fan, we do need this very large heatsink. This is unfortunate. Would you agree with these calculations?

180AB heatsink datasheet:
https://www.abl-heatsinks.co.uk/index.php?page=extrudedproduct&product=170


We are starting to think that since we are doing it fan-less, and the heatsink size needed is so big, we would actually be better off doing this with four paralleled Flyback converters instead(..with one single feedback error amplifier so they all share output current equally). Then the heat is spread between the four flyback converters, and the overall solution size will be less. What are your thoughts?

 

[Diver300]

The heatsink is much thicker than 7 mm. I think that is is 70 mm ish thick. The dimension number on the datasheet has been cut off after the 7 of some number between 70 and 79. The solid bit at the bottom is 17 mm thick. It needs to be mounted with the fins vertical

Altering the number of converters won't alter how much heat needs to be dissipated in total. What effects that is the efficiency that can be achieved. Your figure indicates an efficiency of just under 90%. If you can get a better efficiency for the four buck converters, there will be less heat to be dissipated.

The temperature of 67 °C seems to imply that a lot of the problem is getting heat from the enclosure, so any change in efficiency will alter that quite a lot.

I can't see where you get the figure of 1.93 °C/W from. It's not mentioned in the datasheet. Anyhow, one of the problems you have is that the temperature of the case, plus the temperature difference between the junction and the heatsink has left very little temperature difference for the heatsink, and resulted in a very large heatsink.

You are getting 5.16 °C due to the heatsink. You have a margin on 5 °C so you could possibly double the heatsink loss which would make the heatsink a lot smaller.

The other consideration is how much temperature is lost between the heat source and the surface of the heatsink that actually transfers heat to air. The heatsink is specified with a distributed heat source. The converter is much smaller than the heatsink, so there will be a larger temperature difference with a smaller heat source. That problem would be reduced with multiple converters.

There is also the fact that the longer the fins are, the more the hot air from the bottom of the fins slows the loss of heat from the top of the fins. That is why that heatsink performance is 0.3 °C/W for 170 mm of heatsink, but for twice the temperature drop, only about 1/3rd of the size of heatsink is needed. Splitting the heatsink into two, if you can get them both vertical, and the heat from one not rising onto the other one, you will get better performance.

 

[zenerbjt]

I can't see where you get the figure of 1.93 °C/W from.

Thanks yes,...sorry, its 1.5degC/W as on page 15...

https://www.vicorpower.com/documents/datasheets/DCM3623x75H26C2yzz_ds.pdf