I bought one of these as a fairly random impulse, not necessarily greatest or cheapest:
It's got the d2pak or something similar FET and diode. Those are 470uF electrolytic input and output caps, however the output is shorted with some sizable SMD cap, most likely ceramic. That seems to be a common practice power through output transients while maintaining low ESR of the ceramic.
The input cap is shorted by an "M4" diode (you can see it on the bottom of the board). That's interesting. It seems to be rated for 1A and basically shorts reverse polarity input to ground. However I don't how that wouldn't fry in a couple of ms, so I'm not sure it really serves as reverse polarity protection.
The whole thing is 26mm by 60mm including pots, terminals, and the giant caps, however it's a two sided board. Of course there's no mcu. Or I don't think so. There is an IC, that I'm fairly certain is a buck controller. I've tested it so far on a wire short, and it was horribly inneficient, (like 10% efficient) but that was waay into DCM and way below the diode Vf. The regulation control got very wonky down there too. I also tested it at 2A output to an XM-L2. It was 87% efficient, for input anywhere from about 4.5V up to 17V, and seemed to control very reasonably. The input power really didn't change. So it seems to work ok for that.
For the wire short I pushed it up to 10A for awhile and it handled it pretty well inspite of that huge inefficiency. I didn't have a chance to see how long it could keep it up or really measure how hot it got het. Of course that wasn't much actual power, I think 7W total power as recall, about 0.7 output. I don't have any good way to test it at high power at the moment. On the other hand most of the losses depend more on current than power.
It would kind of be interesting to see what other cheap similar sized bucks are capable of. Of course as for cheap, they have the advantage of mass production. I'm kind attracted to some of the simple analog controls and self oscillating designs just for the simplicity of it, but it's not clear how much cheaper any of it really is. You still need a good inductor, good fet, good diode, and for anything this big, of course the big board. There are some voltage regulated designs that drop the sense resistors.
It's hard to see how to make anything cheap on a one off DIY scale though. Of course the board is nearly half the price for one-off. It can be worth thinking about what it costs to make 100 though and maybe in some design MTN would do it eventually so it's not only one-off costs that matter. Boards get 3x cheaper automatically when you build more than three at a time, and many of the components get cheaper if you build 10. It might be possible to build 10 for the price of three. This is not necessarily meant to be correlated to my thoughts about simpler designs.