Doddy, Neale,

Thanks for the words of encouragement and I'm open to any advice anyone wants to offer! More decoupling of the BOB power supply is worth looking at but the VFD noise is all over everything so there's something more fundamental at work here.

The problems I've always had with this machine show up as a limit switch trip on the Z axis. It all started when I changed from microswitches to proximity sensors. Oscilloscope measurements showed lots of spikey noise on the limit inputs of the cheapo Chinese BOB but I ruled out the individual proxy sensor early on. There's a whole thread about this from last year. http://www.mycncuk.com/threads/13519...-problem/page2 post #20 explains what I discovered about the input circuits to the BOB.

The solution then was to use electro-mechanical isolators. Sounds much better to young ears than 'relay' but that's what I used. Miniature relays hold the inputs solidly to ground except when the proxy sensors trigger. Excelent noise supression and low-pass filtering in one simple package. The relays and driver transistors are mounted on a board which sits just above the BOB with only a few mm of wiring between them.

There is only any noise present when the VFD starts and the noise is all over everything. The earthed box of the controller and all it's contents including the BOB power supply rails and the earthed cases of the motor PSUs as well as the BOB inputs/outputs show it. I even tried shorting the scope probe to it's own earth screen clip and you can still see the noise when the earthed probe touches anything metal! It's a PC scope and I use it on my laptop running on batteries so the scope is entirely isolated from ther mains.

last year I rebuilt the controller with the BOB in it's own ally box with screened cables for all signals in and out and individual screened cables for each proxy sensor and each stepper motor. I really thought I'd done evrything right and it was working OK until the recent rebuild. The noise is very short pulses of about 1uS each. Closer inspections shows each pulse to be something ringing at about 6MHz for that short period. It's not just the VFD output waveform to the spindle windings and I'd love to find out what it is but I'd much rather be making stuff with the machine than fault-finding it again.

I really thought the mains filter would fix it, especially after plugging the VFD into a different mains socket fixed everything but it seems the filter doesn't work as reliably as a few metres of 3-core wiring. The only other possibility is that particular input of the parallel port into the PC is dodgy but that seems a long shot, though I'm convinced it is not spurious triggering of the solidly earthed limit switch input so something else is triggering that specific port input to the PC. It's always the same one.

Sorry it's ended up such a long rant. Today the LinuxCNC PC decided to throw a fit as well, so that now has a new system battery and a re-seated CPU. I think it may be telling me something. I think it might be yearning for a one way trip to the dustbin. It's working again at the moment but I'm not going to put too much more time into this current controller if it plays up again.