so i started gathering up 3 bins worth of parts, so i could start assembling my 3 buck boards, and i got to thinking.... I've only tested this thing with 12-14v in, and things seemed almost too good to be true, so i decided to test with a more realistic input voltage, 44V. well, some things started showing up... The upper fet started getting warm, and both the input and output caps started warming up... so i shut everything down, reset my input voltage to 24, and things warmed up, not not nearly as much.
Upon further investigation, i could see really high spikes on the input voltage, like in the order of double the input voltage.... so, for the fet, i'm thinking i'm exceeding the source/drain voltage of the fet, which is 55v.
the other thought on the caps is ESR, i did get low esr caps, but they are rated at 35v.
SO..... in the interest of experimentation, on the input side, i strung 3 caps in series to up the voltage rating, and then paralleled a second string(using 6 caps total, and then on the output i put 2 in series, and added a second parallel string. (less total capacitance, but a high enough voltage total)
My thinking is 1) parallel strings should lower the effective ESR, and the parallel string helps offset the loss of effective capacitance from putting them in series.
well, the caps still got pretty warm, and the high side fet did too, but not as fast...
so my questions are,
- Is my thinking right on the caps, and the ESR, do i just go with more? switching at 31200Hz is pretty low in the SMPS world, so i would think that there would be more playroom...
- would the spikes on the input be able to heat the fet, without instantly killing it, and would a higher voltage rated fet solve this( and is it worth the higher RDSon???
- as the caps were heating up, i noticed that i was pulling way more input current that i was making out.... I'm wondering if all this extra input current could be attributed to the caps?