Filtering the PWM for the contrast didn't work with a 4k7 resistor and a 10uF capacitor. The contrast pin appears to supply a current as even switching the output to zero resulted in 1.5V on the contrast pin. To fix it, the two components in the top right are the resistor and capacitor, so I soldered a 1K 1206 resistor diagonally between the bottom of the resistor and the top of the capacitor and all works well.
As I mentioned in post #12, I used a 1K and 10uF for PWM, which has always worked fine - however, nothing 'clever' about it, I just copied the values from an example I'd seem somewhere on-line.
Interesting to hear that using a higher value resistor messes things up.
Nice looking board though.