Spindle encoder with transmission reduction

[bbatarelo]

Hi all,

after observing's Marty's most recent build, I started thinking about spindle encoder mounted on a motor rather then on spindle itself for the same reasons Marty did it that way. However, his belt gearing is 1:1 between motor and spindle where mine isn't. I have big pulley on motor side and small on the spindle side. This creates several kinds of problems - first I have 7k rpm spindle this way which is nice, but I'm finding difficult to source encoders that are made for more then 5 or 6k rpm. 25mm hollow encoders are usually very expensive and have low resolutions. So my question would be: if it turns out that I have a nice motor to spindle ratio that can be represented with integer (let's say 2:1), would I be able to place hypothetical 2000 line encoder on motor side and fool Acorn by telling it that it's running with 1000 line spindle encoder?

Thank you,
Bruno

[Dave_C]

Hi all,

after observing's Marty's most recent build, I started thinking about spindle encoder mounted on a motor rather then on spindle itself for the same reasons Marty did it that way. However, his belt gearing is 1:1 between motor and spindle where mine isn't. I have big pulley on motor side and small on the spindle side. This creates several kinds of problems - first I have 7k rpm spindle this way which is nice, but I'm finding difficult to source encoders that are made for more then 5 or 6k rpm. 25mm hollow encoders are usually very expensive and have low resolutions. So my question would be: if it turns out that I have a nice motor to spindle ratio that can be represented with integer (let's say 2:1), would I be able to place hypothetical 2000 line encoder on motor side and fool Acorn by telling it that it's running with 1000 line spindle encoder?

Thank you,
Bruno

If you are just wanting an RPM read out you can do lots of things to trick Acorn but if you plan on doing rigid tapping then NO! Spindle and encoder must be at a 1 to 1 ratio. This is the only way the Z pulse can arrive at the Acorn control at exactly the same point and time to index the start of the thread.

Dave C.

[bbatarelo]

Hi all,

after observing's Marty's most recent build, I started thinking about spindle encoder mounted on a motor rather then on spindle itself for the same reasons Marty did it that way. However, his belt gearing is 1:1 between motor and spindle where mine isn't. I have big pulley on motor side and small on the spindle side. This creates several kinds of problems - first I have 7k rpm spindle this way which is nice, but I'm finding difficult to source encoders that are made for more then 5 or 6k rpm. 25mm hollow encoders are usually very expensive and have low resolutions. So my question would be: if it turns out that I have a nice motor to spindle ratio that can be represented with integer (let's say 2:1), would I be able to place hypothetical 2000 line encoder on motor side and fool Acorn by telling it that it's running with 1000 line spindle encoder?

Thank you,
Bruno

If you are just wanting an RPM read out you can do lots of things to trick Acorn but if you plan on doing rigid tapping then NO! Spindle and encoder must be at a 1 to 1 ratio. This is the only way the Z pulse can arrive at the Acorn control at exactly the same point and time to index the start of the thread.

Dave C.

Can you be more specific about this precise timing in case of rigid tapping and why this wouldn't work? I mean, mathematically my suggestion (in that particular scenario I mentioned) seems sound to me, and the worrisome part then should be maybe in some sort of mechanical delay, belt stretching, etc. But on the other hand, systems with 1:1 ratio shouldn't be immune to it also, but they apparently work.

Formally speaking, I am proposing a sort of mechanical (custom) frequency divider without reminder where if resultant frequency is high enough, things should work.

Bruno

[eng199]

The index (aka marker or Z) pulse from the encoder occurs once per revolution. This is the issue. An exact once per revolution pulse is required.

The counts per revolution is not an issue. That can be easily changed by a parameter.

[cncsnw]

In bbatarelo's scenario, it should work just fine.

He proposed to have the large pulley on the spindle motor, and an exact integer ratio (e.g. 2:1). In this case, the spindle encoder index pulse would come around once every other turn of the spindle, always at the same spindle orientation. Tapping should work just fine with that setup. The belt between the spindle motor and the spindle would need to be synchronous (i.e. a timing belt).

[bbatarelo]

In bbatarelo's scenario, it should work just fine.

He proposed to have the large pulley on the spindle motor, and an exact integer ratio (e.g. 2:1). In this case, the spindle encoder index pulse would come around once every other turn of the spindle, always at the same spindle orientation. Tapping should work just fine with that setup. The belt between the spindle motor and the spindle would need to be synchronous (i.e. a timing belt).

This has suddenly become interesting again
I can’t help myself but to wonder what exactly is the purpose of a index pulse then - or better to say what Acorn does with it and how it treats it? I realize that it’s a sync signal which helps with occasionally improperly detected encoder lines, but if encoder has 2000 lines and I tell Acorn that it has 1000, clearly indexing signal will arrive 1000 lines after Acorn was expecting it. Last thing I would want is for Acorn to be confused because of some weirdly defined state. It would be interesting to see what cnckeith would have to say about this.

Bruno

[Muzzer]

That scheme sounds as if it should work from a control point of view, as you say. However, I can't help wondering what sort of tap size and motor torque we are talking here. The motor torque will need to be twice the tapping torque and a very low spindle speed. On my machines, I can't sensibly tap unless I select a low (reduction) ratio. As both my machines are turret mills where I can't mount a spindle encoder, there is then a reduction ratio between spindle and (motor) encoder which rules out rigid tapping. My solution is to use a tension-compression head or thread milling.

It's worth pointing out that the CNC12 controls seem to require a relatively high PPR count for the encoder due to the way it determines speed and position. The recommended PPR seems to be around 1000, which pretty much rules out sensing gear teeth or using optical targets, even if you manage to generate an index pulse.

[eng199]

I can’t help myself but to wonder what exactly is the purpose of a index pulse then - or better to say what Acorn does with it and how it treats it? I realize that it’s a sync signal which helps with occasionally improperly detected encoder lines, but if encoder has 2000 lines and I tell Acorn that it has 1000, clearly indexing signal will arrive 1000 lines after Acorn was expecting it.

The index pulse gives an absolute position within one revolution. The count derived from from A/B is incremental, so it can not indicate the exact starting point of a thread for example. The index pulse is a known location within the rotation of the encoder.

ACORN does not compare the counts per revolution setting to the index pulse location to verify the counts per revolution. It will not generate an error if encoder pulses are more or less than expected or if an index pulse is missed or spurious. This means cncsnw's suggestion will work. However, it is not the expected configuration and future changes could be made to firmware that do not support less than one index per revolution.

[Dave_C]

It will not generate an error if encoder pulses are more or less than expected or if an index pulse is missed or spurious.

IF your A/B pulses do not come in in the correct order you will get an error and the spindle will stop! I have an encoder that needs a proper shielded cable installed because it gets a false pulse once in a while. I had to disable the error tolerance from a default of 3 to a 256 in order to make it run.

All of this seems a bit confusing as Centroid as been adamant about the encoder being belted at a 1 to 1 ratio from the beginning and I've never seen any documentation that says "full" rigid taping routines can be done with an encoder set up any other way. Yes, you can do a single pass in and out but Centroid says you can't do multiple pass threading back into the same hole . In a 2 to 1 overdriven setup you are getting 2 Z pulses per one turn of the spindle. So how can you hit the thread start on the second pass? You might and you might be 180 degrees off!

Dave C.

[tblough]

In a 1:2 pulley setup you are getting 2 encoder pulses for every turn of the spindle. That will not work as the re-entrant tap could start at 0 or 180 degrees out from the first pass. With a 2:1 pulley setup you are getting 1 pulse every two spindle revolutions. That pulse is always in the same spot so that will work.

I.e. if you are doing a speed reduction between the motor and spindle, you must attach your encoder to the spindle. If you are doing a speed increase, you can attach the encoder to the motor only if the increase is an integral amount (2:1, 3:1, 4:1, ...) and if you use a synchronous belt and pulley setup for the encoder. You have to use a toothed belt for the encoder because v-belts, orings, etc have some slip and therefore can't guarantee that the encoder pulse will occur in the exact same spindle orientation each time.