What would you do if you had a power supply unit in a geometrical form of certain length, width and height (or thickness – which we call “profile”)? The unit with no conductive cooling system, but only low-power fan and ambient air to cool down. Far from having efficiency of 100 %, but with inconveniently high amount of dissipated power?
You guessed correctly – you’d grab a hammer and start flattening the unit with it in order to decrease its profile. It’s no issue that the length or width or both would be getting bigger. What’s important is that there would be increased cooling surface. At some point you’d be satisfied – and the fan and ambient air would start being more efficient, the heat would be transferred away, and unit’s reliability would rapidly increase (note: any additional decrease of temperature by 10 °C doubles MTBF!).
If you continue with the same dedication – you’d reach the ideal – the profile would become zero, and cooling surface would become infinite. Fan strength or air presence won’t matter anymore – in space or vacuum, cooling of the unit would be excellent! And thus we just came up with how to obtain a power supply unit that requires no cooling system!
This was a joke, however not completely.
If this cheered you up, then on to the essay!
Let’s start with what constitutes low-profileness (LP) or even more precisely – “planarity”.
Heat sources in the middle of the cases are represented by red circles, for example such can be transformers, inductors, semiconductors, thermistors, etc. It’s important not to let the temperature of such components rise above a certain temperature, for example +100 °C.
It’s obvious (in fig. 1) that thermal path h1 for thicker unit is much longer than for the low-profile planar unit – h2. Therefore for the low-profile unit to maintain limit temperature of +100 °C it’s allowed to have higher heatsink temperature – for example +84 °C, compared to the system on the left, which would have limit heatsink temperature of +78 °C. Then we come to a simple conclusion: let’s say ambient temperature is +60 °C, then the heatsink on the left will work with temperature gradient of +18 °C relative to ambient, while heatsink on the right will work with a gradient of +24 °C.
This means that at equivalent power loss of 100 W, left-heatsink (h1) thermal resistance must be 18/100 = 0,18 °C/W, right-heatsink (h2) thermal resistance – 24/100 = 0,24 °C/W. As a result, for a thicker power supply unit you’ll have to design a heatsink aprox. 30 % bigger that for the planar one!
Exactly because of this, the low-profile planar unit with a heatsink on the right (h2) will have significantly smaller dimensions and weight. Imagine how this may influence small-size mobile systems, aerial in particular, for example drones.
4. Very important question: what do you do with the heat which isn’t transferred into the heatsink (i.e. liquid type) but in an opposite direction? Because of such “parasitic” heat, beside liquid cooling, one will have to add additional cooling, for example forced air. Considering that “parasitic” heat, in a thick power supply design, can be 15-25 % of the total dissipated heat – the issue may become very unpleasant and lead to decreased reliability since mechanical devices (fans) must be added to the system. I had to deal with such cases when developing power supply systems for supercomputers and active-phased radar arrays. Due to use of planar low-profile power supply units it is easy enough to decrease the amount of parasitic heat to 5-7 %, in which case additional cooling options, beside fan-cooling, become available, i.e. air convection, conductive liquid, etc.
For AEPS-GROUP products we commanded – slim down fast! Pay attention to our units with LP index.
Listed benefits of planar power supply designs are not the only ones. If we meet, I’ll tell about the other.
General designer of AEPS-GROUP Alexander Goncharov.
P.S.: You can download this post in pdf