Kai,

I can't believe you went to the length of drawing the schematic! The concept of using a transformer puzzles me - apart from the size and cost. So for the AC side, we have isolation via the transformer and via the opto!
The other problem is holding down small transformers to pass the shake test - it's a pain having to use wire ties in production.

As I mentioned in another post, the Sharp pc900v has an input current of just over 1mA. So the circuit I'm envisaging is diode,constant current source,opto (I don't know how to display schematics like you do!) and the obligatory fuse and MOV protection. The constant current source will be the usual transistor (high voltage) ,a handful of resistors and a couple of diodes. I have to sit down, cobble up a circuit and do some calcs on the worst case dissipation.

I was hoping someone had come across this type of application - it seemed like one of those tricky problems that someone had an elegant solution to.

I am well aware of exploding varistors - so for everyone reading this beware - they are quite violent when hit with an overload and make a mess.