Convert from Anilam Crusader M to EMC 2

Even with ability to read the tachometer,you are still 'dead reckoning' between scale counts and with enough idle time you will drift until you hit a scale count. So when idle I suspect all you will accomplish is to lower the frequency of the +-1 count dither that you normally get with digital encoder feedback systems. It should work well when moving if implemented correctly though.

As Andy mentioned, a negative tachometer reading in a positive direction is expected since the tachometer is negative feedback
(the negative tachometer output is summed with the positive command voltage when moving in a positive direction)

Its easy to extend the 7I87 input range (adding a 100K input resistor will increase the range to +-20V)

PCW wrote: Even with ability to read the tachometer,you are still 'dead reckoning' between scale counts and with enough idle time you will drift until you hit a scale count

I wonder if you could integrate tacho data and add it to the position, rather like the "position interpolated" in the software encoder?

With some D-term shouldn't the drift be spotted and nulled?

Sure thats what dead reckoning is, but there will always be accumulative errors so the position will drift when idle.

It may actually be that you dont care about the tachometer voltage above a volt or so, since you can use encoder velocity signals for interpolation when moving at any reasonable rate.
So maybe a +- 1V full scale range might be better (with some input clamping method)

All in all I wonder if adding a high resolution rotary encoder to the ballscrew end is not going to be better ultimate solution

Previous attempts at this conversion have suggested that there is a problem with the loose coupling of the machine position to the motors, and the relatively high granularity of the position feedback.

In practice 10um error in absolute machine position is pretty reasonable for a milling machine as long as it can avoid dithering between counts.
The tacho feedback ought to be able to help with the loose coupling problem, even if we don't know exactly how yet.

It might actually be a good plan to run two PID loops, a velocity loop controlling the motors with tachos as feedback, then a position loop using the scales as feedback and making velocity requests.

I have to wonder how the original hardware handled it. Did the tack signal go to just the drive or the drive and control? If it only went to the drive, and the drive is handling the velocity loop, then might the corser scale feed back to the control be enough?

Just the ponderings of a person who doesn't realy know what they are talking about.

I'm going through almost exactly the same upgrade path myself - 0.01mm resolution linear scales, Anilam crusader II control, analogue drives with tach feedback.

I can confirm that on my Crusader control the tach feedback only went to the drives. The encoders connected straight back to one of the Crusader boards while the analogue control signals came from a different board to the drives.

On careful inspection of the wiring schematic and detailed testing of what pin went where I managed to deduce that the majority of the pins on the massive round multi pin plug did nothing at all. One impressive looking 26 conductor ribbon cable in the control box turned out to only actually have 5 wires doing anything at all.