Boxford 190VMC conversion with Mesa 7i96S

Trawling the internet time machine I found a document with the manufacturer's original spec for the version of the 190VMC that I believe I may have (see attached). It is referred to as the 190FMS and unfortunately has a shorter X travel range than other models. The spec claims tapping of sizes up to M4 should be possible. (The spec seems very conservative - it claims only 600mm/min but in reality can easily do the 2000mm/min specified for the 190VMCxi, perhaps my machine was upgraded or is an intermediate model).

Regarding the multiple slotted switches... Yes I realise I'd have to tune the separation to get the quadrature signals synchronised with the correct phase offset. I just meant a pre-made pcb that would provide power and convert the output signals into a useable format. I.e. Only wired up for electrical signal processing, not for mechanical positioning.

Attached is the spindle torque curve, they claim 0.93 N.m at low speed.

Also uploaded the spec for the standard 190VMC in case others want it. 

Apart from rigid tapping are there any other significant benefits with having a spindle encoder? Accurate rpms I guess?

I did some testing this weekend and unfortunately I'm seeing large shifts on the X and Y axes whilst machining. Milling aluminium with only quite a thin cut, so very disappointing. (I.e. 6mm cutter, 2500rpm, 100-200mm/min feed, 2.5mm depth, but only 0.2mm sideways stepover).

The leadscrews are belt drive and I did my best to tighten up the belts. I can't totally discount physical slippage yet but I'm begining to think my stepper wiring may be a bit suspect.

The toolpath I was using did a helix ramp down into the surface then had lots of small circular paths as it cleared out the pocket. It sometimes sounds rough at very low speeds - resonances? backdriving?

The drivers are ST GS-D200S (see attached). If my reading of the manual is correct, they expect active low signals but then step on the rising edge of that signal.

I currently wired it as though they were differential, with step - connected to the driver gnd and step + connected to the step pin. I then selected the invert checkbox in pncconf.

Note: The driver manual doesn't mention opto-isolators at all so there probably aren't any.

I used really conservative timings...
Step Time 30000ns
Step Space 30000ns
Direction Hold 50000ns
Direction Setup 50000ns

The driver manual suggests that 500ns step width and 1000ns for the other timings should be possible.

I'm thinking of trying to make the stepper wiring single ended. Step + to 5V, step - to the driver step input, join the grounds. Is that correct, and likely to help?

Any advice would be most welcome.