Control system for Harrison M250 CNC

[Chaz]

If you check the RS part number for that encoder on RS, it is only a 500 ppr encoder.
I've just tried finding an official spec for suitable encoders, and although I've found it has to be a 5V differential output encoder, all they say is they recommend 2000ppr, not what the actual minimum or maximum possible is.

Chaz, what ppr is your encoder?
And do you know what limits they put on the appropriate setting in the software?


2000 is high for a spindle encoder on a controller aimed more at the hobby/light industrial end of the market. I've got an old mill that has a 200 line spindle encoder, and it would of been able to rigid tap using it's original controller, which is far more demanding on accuracy than single point threading on a lathe.
For a lathe, even a 50 line encoder would be more than adequate for most reasonable sized threads. I can guess why Centroid recommend such a high count, but that's just because I'm a bit of a geek.

If you're existing encoder is differential output with all the required outputs (A,B, Z, plus their compliments - should have at least an 8 core cable if it is), I'd be inclined to wire it in and try it. Worst case you have to change it, best case you get a functioning spindle encoder that does the job without having to change anything.

Mine is 8192, so 2048 X 4. i think Acorn can read much higher than that.

[AdrianW]

One question that's been nagging me - I'm using a lot of speculation from a position of relative ignorance so please bear with me!

- On a lathe, the absolute position of the spindle (and I mean absolute in the sense of knowing e.g. where a specific chuck jaw is from one work session is to the next) isn't important. What's important is knowing where the spindle is relative to the first cut that was made on a specific workpiece.

- If the above is correct, then the only need for a Z (index) output on the spindle encoder is for pulse-count verification.

- If both the above are correct, then there is no reason you couldn't increase the effective pulse count of the encoder by having a different drive ratio to the encoder e.g. having a 1:4 drive would increase a 2000p/r encoder to be 8000p/r of the spindle.

Does this make sense, or am I missing something? I guess you could run into non-linearity problems if one of the pulleys was too small and the belt had any 'set' in it.

[m_c]

It all depends on the controller.
In the Centroid case, as they specify an encoder with an index pulse, I'd guess they are relying on the index pulse to synchronise moves, which if you're not threading, it likely doesn't serve much purpose. Without it, unless the controller continually counts the encoder, the controller wouldn't be able to synchronise for multiple threading passes.
The controller relies on you setting the encoder count in settings, and I would doubt if there is in any verification between the counts and the index pulse. There might be, but it's more likely to just show up as threading moves failing to start due to the lack of index pulse.

If you were to have more than one index pulse per revolution, then it's likely to cause synchronisation problems. If the controller is relying on the index pulse to start a synchronised move, then if there are four index pulses per spindle revolution, odds are not good for success.


I wouldn't worry about non-linearity problems on a lathe spindle. For basic turning, even if the spindle location varies by 10% over a revolution, it's going to have minimal effect on normal thread pitches. Even on a 2mm pitch thread, that's only potentially 0.2mm, which on a 2mm pitch thread, is neither here or there unless you're aiming for some kind of precision fit thread. A generic nut and bolt will likely have more play than 10% of pitch.

[AdrianW]

if there are four index pulses per spindle revolution, odds are not good for success.

Yeah - I had been assuming that the rotations would be counted by a running pulse total i.e. (pulses/pulses per rev) but as you say it's entirely possible that the index is used to reset the position - this could have odd effects!

[m_c]

Yeah - I had been assuming that the rotations would be counted by a running pulse total i.e. (pulses/pulses per rev) but as you say it's entirely possible that the index is used to reset the position - this could have odd effects!

It's something that the designers will know, but they don't always release the information, as it's not something you really need to know for implementation. You just need to know if an index pulse is needed or not.

I know the KFlop continually counts the pulses and you don't need an index pulse, but it has a lot of the internal workings exposed so you can customise exactly what it does. If you don't have an index pulse, it relies on using wherever the encoder count started as the index point for spindle orientation. Even if you do add an index pulse, all it can do is reset the zero point to the index pulse location (or any offset you'd like).
On my lathe, although I have the encoder index wired up, I only ever used it for confirming the encoder count when I was setting things up, it's never been used since.