Hi,
I've a very small circuit where the LMR36015 (1MHz) is feeding 4.85v into a load. Input current is fixed at 0.5A with a theoretical 20V point, so 10W, but in reality it'll be more like 12-13V. With no load the input voltage can rise to 48V where it is capped by Zener.
I'm using the 1MHz chip and a CoilCraft 4030 10uH inductor (I can't go a bigger foot-print than this). Input capacitance is a close 4.7uH X7R and a 1000uH electrolytic a little upstream. Output is a 22uH X7R and a 220uH tantalium (1 ohm). Ripple is good, not as good as 3x 22uH but board space is very tight and there's no room for more. It's acceptable.
When I ran a test I managed to get 1.82W at the output while I know 2.03W is possible. While I'm not providing the detail for this test here it's another buck sharing the same input and load, demonstrating that the other device is being more efficient.
To find out where any losses are I did an infra-red check. The chip isn't hot by any means, less than 28c with a 200mA load (while the green area is 21c), but shows if I'm to optimise then I need to consider it first.
At 1MHz a 4-30v buck would typically use a 4.7uH inductor. As I'm not drawing more than 0.5A current above an input of 20V, I'm wondering if I'm also better going for this rather than the 10uH recommended by the datasheet. Or, maybe I'd be better sticking with the 10uH and going for the 400KHz chip?
Any other thoughts warmly appreciated!
Andrew