Group Actions
Engagement
  • Author Author: michaelkellett
  • Date Created: Date Created
  • Views 948 views
  • Likes 10 likes
  • Comments 2 comments
Related
Recommended

Heatsinks

michaelkellett

Recently /members/andrewj posted about some issues with heatsinks. It so happened that he was thinking of using heatsinks similar to one I'd already chosen for a resistor based adjustable load.

I'd run into some issues using active loads with a power supply design and I need something I can trust, so the plan is two heatsinks (Farnell 4621920) rated at 1.25 K/W each with 5 x 50W 100R resistors. There will be a way of selecting which resistors are paralleled statically. It will also be possible to connect selected resistors acroos the load but with a big MOSFET in series so I can check out the response of my power supply to sudden steps in load current.

I'm not too bothered about the arrangement sinking the maximum resistor rating of 500W (250W per heatsink) but I would like it to manage at least 250W continuously.

I think I'll be able to achieve this by blowing the air across the heatsinks.

So I've drilled one of them an fitted the resistors for a quick bench test of the arrangement, and hope to answer the question, how much power can a 150mm x 100mm x 27mm  heatsink cope with if blown good and hard.

I started off with no fan, 55V across the paralled 5 resistors which gave me 2.473A and 150.8W - this power was chosen because the power supply I used can only manage 180W and 150 W is Andrew's target for his project.

The starting temperature of the heatsink was 25.7C and quickly rose (9 minutes) to 95.6.

Then I turned the fan on and the temperature dropped to 49.7 in about 3 minutes and eventually settled at about 44C.

The heatsink with air blowing is managing a 20K riise for 150W which is a thermal resistance of approx. 0.13 K/W, to be compared with the claimed convection rating of 1.25 K/W.

The fan drew 0.7A at 12V (Farnell 1735922). It's quite noisy so in a real application it would be worth controlling its speed.

I checked the heatsink temperature with an IR thermometer which agreed closely (within 2K) with the thermocouple embedded in the heatsink in a 2mm hole.

image

image

image

image

I feel quite good about this - I've confirmed my rule of thumb that a feasible, if slightly extreme fan assitance, can achieve a 10:1 gain in therml performance over the convection rating for an extruded heatsink.

The 150mm x 100mm heatsink will be OK with my resistors at their full rating (the resulting 250W wil l result in a maximum heatsink temperature of about 72.C which is fine because the resistors can cope with 150C case temperature.

  • in reply to Andrew J

    Fan was blowing, I find that works best for extreme heatsinklng.

    The toolmakers vices are a recent thing - the metalworking stuff supplier I use has them on offer at intervals for half the normal price. You can get similar things on Ebay ( https://www.ebay.co.uk/itm/370536458297?epid=1249319391&hash=item5645b1a839:g:kN0AAOSwfgZhcq5c) but a bit pricey.

    I paid £39.35 for the little ones (50mm wide)  and £47.48 for the big one (63mm wide) from mscdirect.co.uk  - that's an offer price - standard price is twice as much.

    The great thing about them is the very stiff construction and truly // jaws. They make a much better job of gripping pcbs ( or fans by the edges ) than any other little bench vices I've tried.

    MK

  • That is a really interesting and useful result to see, thanks for doing it Michael.  Empirical testing seems to be the only answer as the reality seems that whatever a manufacturer claims for thermal resistivity or conductivity has to be treated with some circumspect.  For example, some of the figures that Bergquist quote for its Sil-Pads is based on 200PSI pressure applied to the pad.  The caveat always given is do your own testing in your own setup!  Fair enough I suppose, but it would be useful if the testing was a bit more realistic and I certainly wish manufacturers of these things would standardise on a specification unit!  

    It's a beast of a fan, mind, at nearly 8.5W: it was blowing, rather than sucking, across the heatsink?  This does show that with some monitoring and automatic control, a cheaper solution can be achieved than, say, lumping out £70 for a tube heatsink which actually does no better.  I think in my case, I'm right in revising my approach downwards to a more manageable power level but it will be interesting to see if I can set up a controlled test to drive it to 150W.  Like you I do have a supply that can do 180W but I don't really need a DC load rated to these sorts of levels.

    PS: I like those vice.