The approach proposed by @K.C. Lee seems to be working: using the Forth snippet from the last log entry and a quick&dirty RC filter I can control the output voltage of the DC/DC converter even if the current filter output leaves a lot to wish for.
With an unregulated 12V power supply from the AR-box *), 0% PWM results in 14V output voltage. The output voltage drops to about 2V at 50% PWM. The RPM of a connected DC motor can be controlled in small increments by changing the PWM value (I know, for a motor a constant voltage supply and a PWM output would have been just as good - the motor was just a load for a test).
All in all it looks promising, but I first need to build a set-up for testing. For this I need a switchable load to do basic dynamic testing, a more reliable primary power supply, and good wiring.
*) thanks to @esot.ericor this "croponym"