After the initial success of having set the mill up, worked out what I was doing with the basics of NativeCAM and not breaking too many cutters, I noticed strange things afoot when the mill ran for longer periods.
At first I thought I had messed something up, pushed too hard or otherwise just had a loose connection, but then it became very obvious when the machine started moving by itself when it was supposed to be sat there doing nothing.
My first thought on this (as touching the machine made it worse) was either a floating ground or other bad earthing / poor connection. The Proxxon, as well as the CNC controller, are both isolated from ground, so I dug out some extra wire, stripped and rebuilt the machine adding an earth wire to the body as well as some other stuff that I've forgotten since.
However, it was not to be - there was still a ghost in the machine, and it was at the Beetlejuicey end of the ghost spectrum.
The CNC driver board is already (mostly) isolated by opto-isolators from the PC parallel port, but being Chinese and quite old technology / basic design using TB6560 driver chips I didn't entirely trust it. It was also operating quite near the edge of its happy current limit for the motors, and has no fan to cool it.
For peanuts I bought a replacement one that was a different design with a cooling fan & heat-sink but that didn't fare any better straight out of the box.
So, I bit the bullet and did what any sane and logical person would do - I made my own fully-isolated CNC driver board, from scratch.
The basics of parallel-port CNC control boards are very simple - the PC waggles some of the parallel port pins up & down and these are buffered and fed directly into the ENABLE, STEP and DIRECTION pins of some driver chips that move the stepper motors.
I decided to try making my board to accept the pseudo-standard StepStick footprint driver boards which would give me the option to use a whole range of cheap & available driver boards from a variety of suppliers.
I also (just to be different) used RF isolator IC's rather than traditional opto-isolators, as an experiment as much as anything. They claimed to be capable of higher speeds, and they certainly take up less board space.
Ah, no. It turned out the StepSticks just don't have the capcaity to drive my steppers reliably, which is a real shame as the Trinamic-based ones were as smooth as a greased otter and super silent.
Instead I bit yet another bullet and bought some external single-channel stepper driver modules and wired those to my isolated board… so far, so good.