I don't see any problem to use an external power supply by putting the 0v lines in common. As you say the opto-isolatiors can do this just as well.
However, I would suggest giving it a try with the on-board HV generator first, and possibly with a heatsink on the MOSFET. The IRF 840 is super robust but does not have a great Rds, so might get a bit hot if you drive it very hard. There are other lower resistance versions out there, but I tried to use jelly-bean components where possible to make it easier for people who just get the board and source the components themselves.
I am building rev5 wifi clock with 6 IN-18 tubes. Yesterday I istalled 4 neon indicators and I noticed problem. On full brightness (LDR 999) when neons reach full brightness all 6 tubes dimms for little while.
Is this too much load for HV power supply? For neons I have 50k resistors, as reccomended in manual. Now I tried 100k and it is better, but dimming is still here.
What is your sugestion? Try even higher resistors for neons or change PWM or HV voltage?
I have good 9V 2A power supply, I tried reset and calibration - didn't solve my problem.
What is your sugestion for resistor values?
Here is short video, but dimming is barely visible. In person it is much more visible.
I think heatsink is not needed. On full brightness with 4 neons is temperatupe of mosfet about 40°C (ambient about 16°C). Hottest component is D1 which have little over 50°C.
I am very surprised with low temperatures, because I used clock for some time with Z573 and temperatures was higher.
My first plan was use 12V power supply, but mosfet was very hot (with Z573). I tried 9V and it was much better.
C1 is 10uF. Not sure if this matter.
My rough calculation says you can go as high as 200K and the neons will still light, but maybe not as bright as you'd like. I see no harm in trying progressively higher resistors until you get a good result.
Look into it later when the dust is clearing off the crater.