Can't really comment until we see the LM1815 input (pin 3) and output (pin 12) waveforms together, and on a shorter time base like 50us/div.
Hopefully I can get that setup by the end of next week, then we can see what happens.
I think should be better use battery against to the switched supply...
The output on these looked decent on the scope, but I'll see if I can track down a spare car battery.