Hopefully you read this JazzCNC, as I think I have fallen foul of the "Pulse edge being on the wrong side to what the drives are set to" issue you described above.

I did run a test as suggested above and all looked good and I've even run a couple of jobs (nothing too stressful), but today I decided to use the machine in earnest and ran a job which was a multi-level adaptive clearing toolpath (just facing off a chunk of hardwood), with lots of curved cuts.

The first level looked OK i.e. nice spiral ramp down to depth then expand out and curved passes for the length of the stock. The next level down again spiralled down and ran fine, but I then notice that level is offset from by -3.8mm in X and -1.8mm in Y, like its losing steps or zero. I have verified I have an error by running a much longer X -230 X -130 rinse and repeat toolpath over a steel rule for about 10 minutes @4000mm/min and lo and behold I have a "slip" in the minus on the X axis of about 5mm (more if I run it longer).

Is that consistent with the Pulse Edge issue?

(I'll try and track down the Dir LowActive in Motor Outputs in MACH4 and give it a try in any case, as nothing's changed mechanically i.e. no binding on axis etc. so it must be either electronic or software).