Whenever I design a new circuit, I order set of prototype boards (those 10 boards for 10 bucks or so services are really useful) and hand-assemble the first one testing it as I go. This means that I usually start from "outside" - power converters and thus are first parts so I can test them without risk of blowing up PITA-to-solder MCU due to bad output voltage, then proceeding inwards adding components. This is when I am building full board and not just testing specific parts of it separately, like building just MCU and parts it requires to see that JTAG works and MCU runs.
This is the main reason I don't use BGA packages - I don't have suitable tools to solder them, so I would have to order manufactured boards which can cost easily thousand or so for even small batch.
I just recently made a small new board which works as "glue" part for larger interoperating system, containing just power converter and needed connections to both directions. So as usual I built it up and tested it quickly. First attempt is success - I get correct output voltage from device so I am happy. But then it was time to go home for the day.
On morning I arrive and fire up my BK Precision 8540 electric load. As a side note, this kind of load is incredibly useful if you even semi-routinely design power supply parts. No more fiddling with resistors or lamps as load, I can actually properly test the damn thing so very easily on various voltages and loads!
For this design I used a buck converter chip I haven't used before as I needed more output current as usual, meaning this is "all new design" (well, as new as basic buck converter can be) so I start easy. 50 milliamp load? Ok. 100mA? Still ok. Up to 300mA ... and regulation just falls completely apart. And this is design that is supposed to handle up to 2A loads. Fsck.
I start probing the circuit, trying to figure out what is wrong. Bad inductor? It was quite small one, just 1.5uH, which I though to be too small for rated current, so switch 10uH one in. Issue continues. Diode? Feedback? All good. Damn it.
Finally - after hour or so - I remember to the check power supply. Which is set to 13v, but limited to just 100mA or so. Just like I always do on first try - if I expect new board to consume specific amount of current, I set the first try limit to just above that to make sure nothing blows up if there is for example short somewhere. I happily managed to forget this limit over night.
Now up goes the current limit and what do you know, now the boards works just fine up to 2.5 amps output current, well past rated maximum output (I upped inductor to 3.3uH from original, though, as tests told me that it gives better overall efficiency than originally calculated one. Might try 4.7uH if I had one to test it with.)
That is, until converter chip gets to 130+ degrees C and starts cutting power output. This was not unexpected either, since this chip has thermal pad on bottom which is PITA to solder (read: it very likely isn't soldered at all) so thermals aren't as good as they would be on properly manufactured board. Unfortunately that isn't something I can check until I get first batch of said properly manufactured boards, so that will be the next step.
