Deciding on boards for a knee mill retrofit

[Moray]

I've got a Matchmaker knee mill that's reached the top of the retrofit list, but am still umming and arring over what boards to use.

Requirements are -
XYZ running servos.
Option to add an A axis (might be stepper, might be servo..).
Spindle powered by VFD, but it is also varispeed currently fitted with a stepper (it's a 3 phase stepper, so it will probably get replaced, as a 3 phase driver is as expensive as 2 phase stepper and driver), along with a hi/lo gearbox. I'd like to keep the varispeed working for the rough speed adjustment, then use the VFD for fine tuning and keeping the spindle motor speed up for cooling purposes.
Eventually motorise the knee. I'm thinking suitably geared servo, but feedback done by glass scale (the ones with the coded index strip, so it doesn't need fully homed every time - I've forgotten who does them, so if anybody can prompt me, I'd be very grateful!)
SSO, and FRO.


So with that list in mind, I think I've got the following needs.
4 Servos, that could be Analog or Step/Dir.
A axis is probably best left as step/dir to keep options open.
One analogue output for the VFD.
One step/dir for the varispeed.
Encoder feedback for 5 possibly 6 motors (XYZ, Knee, & Spindle, with possibly the A)
Two analogue inputs for SSO and FRO.

So the potential totals are-
4 +/-10V and 2 step/dir outputs OR 6 step/dir outputs
1 Analogue output (only 0-10V)
5 or 6 encoder inputs (every axis will have a differential index as well)
2 Analogue inputs

All that is ignoring the required digital outputs, which I'd like to handle via a Konnect or two.

If I go the Kanalog option, that would give me all the DACs and ADCs I'd need, but would need encoder indexes connected to conventional inputs via suitable buffers (not a big issue, but something to consider), as I'll need to use one of my adapter boards to utilise two off the encoder channels for step/dir outputs.

If I go KNozz, that could take care of the 0-10V and ADC requirements. Then have step/dirs and encoders split between JP7/JP5 on the KFlop, but then if my understanding/counting is correct, I'm going to run out of high speed digital inputs for all the index pulses (I'm assuming that any reasonable homing speed will be too fast to be reliably caught by a Konnect input, especially if two boards are in use?)


Any body got any thoughts?
I had been hoping to go closed loop step/dir for a bit off simplicity, but I'm now swaying back towards analogue, so as not to limit myself.

[TomKerekes]

Hi Moray,

Then have step/dirs and encoders split between JP7/JP5 on the KFlop, but then if my understanding/counting is correct, I'm going to run out of high speed digital inputs for all the index pulses (I'm assuming that any reasonable homing speed will be too fast to be reliably caught by a Konnect input, especially if two boards are in use?)

Two Konnects update every 360us which is about as fast as a User Program can sample Anyways (270us with 2 User Threads active). Konnect Inputs do require 4.75V to guarantee to turn on. An Encoder may not supply that.

With Kanalog it has 8 3.3V GPIOs that update every 90us.

[Moray]

Mmm.
Time for some maths.

My favoured servos are 2500 line, or 10'000 counts. IIRC an index pulse lasts for 2 counts, so I can assume it lasts for 1/5000 of a rotation.
Now I'd need it to last for a minimum of 360uS, but I'll round up to 500uS to give a bit margin of error.

500uS times 5000 gives 2.5 seconds, or 150RPM.
Which isn't actually that slow for a two stage homing sequence, so that is a possibility.

Off course, the other possibility I just thought of, is use the servo drive built in homing sequence, which would just need a suitable output to trigger once the initial home switch has been triggered, and an input to be notified once the homing sequence is complete.
But then I'd like to use differential signalling for everything, so I'd have to design a suitable interface board, whereas the Kanalog option I already have a suitable board designed.

I'll have to think about this some more.

[TomKerekes]

Hi Moray,

500uS times 5000 gives 2.5 seconds, or 150RPM

I think that actually works out to 24RPM. But anyways less than 2.5 seconds seems reasonable to me.

[Moray]

Hi Moray,

500uS times 5000 gives 2.5 seconds, or 150RPM

I think that actually works out to 24RPM. But anyways less than 2.5 seconds seems reasonable to me.

I'm glad there is somebody around to catch my glaringly obvious mis-calculations!