# Cutting extruded heatsink

I have an heatsink extruded, with this data:

It is extruded and is quite long (10cm) with 4.9°K/W. I had a similar heatsink but cutted to be square rather than rectangular, therefore with a side of 1.5cm and a thermal resistance of 25°C/W.

With these considerations, I thought that cutting it in half and putting the two pieces in a parallel geometry could be the same, if the hotspot which is a square of around 4cm is still covered. I'd like to know if it can work.

• What "hotspot" are you talking about specifically? Most devices requiring a heat-sink are one-sided devices. Feb 3, 2016 at 21:36
• It could work, if I understand your question correctly. Having said that, a sketch of your proposed setup would make the question clearer. (Horizontal or vertical? Forced convection, or natural?) Feb 3, 2016 at 21:53
• I use natural convection, to cool a flat surface where the hot components are on the other side. The use is with the heatsink horizontal with wings towards up, but could be also vertical. Anyway I think I will not use this solution, since I risk to not prove the correctness of my design due to bad/not precise implementation of the cutted extruded hsink (the main purspose is a challenge to myself to prove if thermal calculus are right). Feb 6, 2016 at 11:26
• @thexeno The link to the heatsink on Farnell's site is dead, so the question had lost a lot of context. Could you fix it? Oct 8, 2016 at 23:05
• Actually the link is still working... but I changed it anyway, so is available also when Farnell is down. Oct 9, 2016 at 9:20

I would not want to try it. Although the idea is reasonable from a thermal point of view, mechanically it would be very chancy. The problem is getting good contact over your 4 cm hot spot. With a single heat sink, putting even clamping pressure on the four corners will make a reasonable contact with the hotspot, assuming the upper surface of the hotspot is parallel with the board surface.

With two pieces of heatsink, the hotspot is off-center for each piece. Trying to get good clamping pressure without tilting the heatsinks and ruining the hotspot/heatsink interface would be fairly tricky.

• I need to cool a flat aluminum surface which mounts hot components (LEDs), with naural convection and I have already estimated all the thermal data. I am the component selection phase. Anyway I agree with you, I thought the same: with two of them the center of the heat source is not the center of every (two) heatsinks... this will reduce the efficiency and I cannot prove the correctness of my calculus. Feb 6, 2016 at 11:20
• If the aluminum surface is the same size as the combined footprint of the two heatsinks, I withdraw my objection. You can simply screw the two heatsinks to surface (with a very thin layer of thermal grease between them for good measure) and be on your merry way. For that matter, if your new form factor of the heatsinks is closer to square than the uncut heatsink, you will get slightly better performance. The reason is that the worst-case thermal path in the heatsink will be shorter, and its thermal impedance less. Feb 6, 2016 at 15:44

If you cut it in half and place the two cut pieces side by side, it will be a similar heat conduction pathway to the air, but not the same. You can think of the heatsink like a giant resistor, but the resistance changes with surface area to the air. By placing the heatsinks side by side you are reducing the surface area of the heatsink to air by what I would estimate to be 4-10%. If you left a gap between the two new heatsinks it would be the similar to having one piece.

• I would guess that you've made an assumption that the hotspot is horizontal and convection is natural. This assumption is probably not incorrect, but we don't know for sure. Feb 3, 2016 at 21:55
• Your right. That's why I put 'similar to' instead of 'the exact same as'. We don't know if this is in a still air environment or if there is a fan. We don't know what the ambient temp is or how much cooling is needed. Feb 4, 2016 at 0:12