1) All prototype boards use thru-hole components when possible.
Ever went on your third date with a human and they start talking about marriage? HANG ON!! SMD parts, generally speaking, are marriage. That’s what you do when you are gearing up for use of a pick and place machine. (Note: There are some circuits in which parasitic inductance and such is a factor and smaller SMD parts are required inherently.) Obviously, some chips aren’t made in anything but SMD. Those are okay. If you’ve already used a chip before, that’s moving in the direction of SMD being acceptable.
When I was working on my big, senior capstone project back in school (Boost Converter in the DQ Reference Frame), I stuff a board almost as densely as I could with 1206 parts. I wanted the lower noise of SMD. It turns out that I still had noise due to other errors. What really sucked wasn’t the noise. We had a differential setup that killed most of that. What really really sucked was how damn hard it was to make changes to my screw-ups. I put many test points on that board. I needed many, many, many more. Thru-hole components come with test points on every component (as long as you don’t stuff the electrolytic capacitors in too far). The requirements for the board changed slightly and making slight changes was a real chore with SMD. I’m not great with a heat gun, apparently, so swapping components often required fixing all the new problems I caused when I blew my components off the board. (I’m only slightly exaggerating. There’s tact required in that one.) When I made a bodge, I always hated soldering a wire to an SMD pad. It was just begging to get torn off. Thru-hole wouldn’t have these problems.
All of the above are really only an issue during the dating phase of the circuit. Once you are engaged, your fellow human knows now you like to stack your underwear and doesn’t have to worry about offending your religious sensibilities. Once you and the circuit are intimate, hammering down with SMD is fine and slightly better on some accounts.
