Interesting problem, at least to me, and might help someone in the future with these. Fired up the 3rd board yesterday, and all looked well (LEDs, etc.). However, there were problems:
Problem 1: Went to check DC offset, and the R channel (both +/-) was showing 4V offset. Not good. After looking the board over, I suspected one of the small transistors (MMBTA06/56), as there were a couple that were a bit crooked and I couldn't tell if all pins were soldered or not. Q3_2 looked suspect, so removed with hot air, which also took the 3.01K resistor right by it. Cleaned and fluxed the pads and got it put back together. I also touched up the solder connections on the others on both channels in this section of the board. Fired it back up, same problem. I then looked at Q1_2 and Q2_2. Straightened these and touched up the solder joints. Success! But I wasn't quite done...
Problem 2: Adjusted DC offset (no pot in yet, so will recheck once that is in). I then started going down the 20ohm resistors checking the biasing. All looked good until I got to the L+ channel. These alternated from 290/220mV (they should be balanced between the NPN and PNP sides). The 20 ohm for the closest to the center NPN showed no drop across it. I retouched the solder joints for this transistor, but same problem. Removed and replaced the transistor. Success. Now bias was the same as the other channels and consistent across all resistors.
Ohming that transistor out, it appears to be good. I couldn't see any evidence of solder on the bottom of the tab (collector). I know when I checked voltage on it, it showed V+, but I was probably pushing it down with the probe. The collectors aren't connected on the middle small pin on these from what I can tell in the gerbers. I had used hot air to install these originally, but had hit the tabs with an iron with a wider chisel tip.
These are considerably more difficult to troubleshoot/rework than the full size dynalos are, so it pays to take your time building them. Before I checked the end transistors in #1, I wasn't relishing having to do more extensive troubleshooting/rework on this. The other two boards came right up with no problems. And I'm still not fond of SMT pots
Also, since I don't think this is documented anywhere, with the board facing you (pot, outputs, etc.), looking at the channels as quadrants, they are laid out as follows:
L- L+
R- R+