[G59]

It would be a simple matter to mount an encoder to the spindle end if that is practical.

That's what I thought, but not true. Being that an ATC spindle is being implemented, you cannot have anything that prevents the gripper in the spindle from being pushed down. Hollow shaft encoders are expensive.

[RCaffin]

Hi Ian

why a second belt driven pulley with the encoder.......why not have the encoder directly on the spindle end.....the bigger diam for the encoder the better the lines of resolution are spaced.

Oh, quite possible, but:
A typical (small) encoder would block the end of the spindle so you could not machine long rods any more.
A larger commercial hollow-shaft encoder around the spindle would be a whole lot more expensive - probably close to adding a zero to the cost. But see below.

how many lines on the encoder minimum do you need to do screw cutting?

Well, I manage to do quite good screw cutting with just the Index pulse - that's effectively a 'one-line' encoder. The momentum of the chuck helps a lot. But if you have, say, 60 lines that would give you really wonderful control. However, you would still need the Index pulse on the spindle to guarrantee absolute sychronisation.
Could you make up an encoder wheel yourself with good enough resolution for this? For sure. The optical sensing would be only a few dollars.

two flat ways just aching to have linear rails mounted on them to carry a slab of aluminium for the new carriage.

Would work fine, but one would need to watch out for the loss of mass in the carriage. Big lumps of cast iron do make a difference sometimes.

If you change to a toothed belt you could put a conventional encoder on the motor instead, but you would still need the Index pulse on the spindle itself unless you made it a 1:1 ratio ONLY. That is of course possible with a full servo motor.

Cheers
Roger

[Tkamsker]

Handlewanker i think Encoder should be configurable i would Start with 2 * 360 or 3 Times depending if you do all flanks of Signal which is an factor of 4 so try 720 Thomas


Gesendet von iPad mit Tapatalk

[RCaffin]

There's a problem with a high encoder count on a spindle.
Imagine 720 pulse/rev at 3,600 rpm: that's 720*60 = 43,200 Hz.
That's rather a high update rate for a PC doing threading. You might get away with it using an external FPGA engine like an ESS - have to ask Greg@Warp9 what speed an ESS could handle. (It can't, at the moment.)

Cheers

[handlewanker]

GASP.........screwcutting at 3,600 rpm.......not blinking likely Mrs Jones.........I would think that a rpm of 200 to 300 even 500 would do nicely without breaking the sound barrier.....LOL.

You'd only need the encoder for screwcutting.....normal turning under CNC control is done via the leadscrew (ballscrewed) and the cross feed ballscrew........spindle rpm is as the program sets it for the material and diam, and is not critical to the single rev, which is critical for screwcutting to give the relationship between the movement of the tool along the bed and the pitch of the thread.

If, as Thomas states, the encoder needs 720 lines, that can be accomplished with a disc on the spindle end (not in the bore) with 720 lines on the face or rim or wherever the read head can access it.

Encoders are a strange new world for me, so if I have misinterpreted the device....please enlighten as it's starting to get exciting.

I could scratch 720 lines on a disc face 100mm diam quite easily, just once, with my dividing head........using a twin tool scriber you only need to scratch 180 times to get the 720 lines and that would be easy peasy.

BTW, the capstan lathe idea I thought of could have a solid 25mm steel crosslide as the weight factor for turning is not like milling with rapid feeds back and forth all the time, so inertia is not a factor for crosslide lightness.

It would most likely be better to have the carriage and crosslide in steel with really big HIWIN rails, 45mm on the bed and 25mm on the crosslide comes to mind.

With that idea a lot depends on how much you want to use it, so production doesn't enter the picture, rather more to have the opportunity to make under CNC conditions with repeatability instead of re-works etc.....I think it's easier to run the part a second time if the first time is not quite good enough.......a real pain if you have to do it again the manual way.
Ian.

[Tkamsker]

Hi
ok yes i am spoiled by the fpga board i use (general mechatronics i think they can handle 500 kHz ,.. )
But i think for a lathe it makes sense to use the index signal for RPM and you will cut (even in alu ) threads only with low speeds so the 720 should be able even if the number you calculated is close to the limit of an ordinary lpt board.
i would have to look at the blate i did probably a steel gear with 2 90° shifted hall sensors will also work
i was referring to the AMT 102 encoders which are configurable by dip switches
so you are flexible.
And i am now really thinking of the best way to solve that issue because i want an reliable stable save solution
option 1)
somehow (i will test different methods ) the spindle drives the encoder
best way i think
option 2 ) mount an plate with holes where the light bridges are giving the wave signal
should also work fine
option 3 )
use an steel gear where you have 3 reed contacts or npn switches
this may be very stable but also verry expensive

so i will do some tests then we see
thx thomas

[handlewanker]

Back to the mill and rigid tapping.......the only time you want an encoder in the picture is when you do tapping, so why not have the encoder disc attached to the chuck as a disc above the ATC disc so that the ATC can still grip the tool below the encoder disc.

With the encoder disc in this position the sensor will be able to read it if it's attached to the bottom of the head casting.

I don't think holes would be any good as the pitch would mean a huge disc diam just to get enough pin holes in.

What I have in mind, and it may be a totally wrong idea, is to have a clear disc attached to the chuck with lines engraved on it so that the sensor head can get a signal from the line spacing as it passes the lens, or isn't this way practical.

I assume that the disc would be read by the lines passing the lens and as soon as the reqd number of lines had passed the Z axis stepper would get a signal to move one step down etc for the pitch you want to cut......... after so many steps down the signal would reverse the stepper to move it back up at the same rate......?

I assume that with the tap at a set height (zero) above the job the spindle starts rotating and the encoder gives the reqd pulse rate to make the Z axis stepper move down until it penetrates the job and after the reqd step count down to depth it would reverse out.

The starting point above the job would have to be the same for all holes with the same tap on the same level during a set of tapped holes so that the depth could be reached and backed out without hitting bottom, but of course, different tapping depths are just a shift in the program for each hole.

It seems simple, unless there are other factors that need to be in force.

One question......when does the tapping start operation......as soon as the spindle rotates and the count begins?

If an ATC is going to be the main reason for the encoder, it could still be attached to the bottom of the spindle under the head and maybe as an integral part of the spindle nose stick out area and engraved directly on the spindle itself..

How a trigger pulse is generated to indicate the starting point for a count is a mystery to me as with rigid tapping you need to have continuous pulses for the complete rotation of the chuck for the entire depth of the tapping and that means the lines on the encoder are evenly spaced from start to finish and so endless.

Does this mean one single line on the disc is extended for a second sensor to trigger and start the count sequence?

I assume that to do a tool change when the spindle was revving at 7,000 rpm or so, the spindle would have to be slowed down to allow the trigger pulse to start the count so that the spindle can be stopped reliably at the tool change position and for when the new tool is positioned and inserted into the spindle end drive dogs.
Ian.

[G59]

And i am now really thinking of the best way to solve that issue because i want an reliable stable save solution

It would help if we knew what motor you have chosen and belt type, positive drive or friction.
With positive drive, timing belt, encoder on the motor is the easiest and cheapest.

the only time you want an encoder in the picture is when you do tapping, so why not have the encoder disc attached to the chuck as a disc above the ATC disc so that the ATC can still grip the tool below the encoder disc.

Not really, if you use a spindle that has drive dogs, and an ATC, you need to index the spindle to facilitate the automation of the tool changes.

I assume that to do a tool change when the spindle was revving at 7,000 rpm or so, the spindle would have to be slowed down to allow the trigger pulse to start the count so that the spindle can be stopped reliably at the tool change position and for when the new tool is positioned and inserted into the spindle end drive dogs.

A microprocessor with a 4Mhz clock frequency, divided by 4, = 1Mhz. 8000 rpm divided by 60 seconds, x 1024 pulse encoder = 136532.992 pulses or 136khz at 8000RPM. No problem for todays microprocessor or PC to work with.

Attachment 261420

Just an example. This wheel has 180 teeth. If you read the high and low of a tooth, you have 360 pulses per revolution. Add another sensor, whether it be optical or hall etc...at 90 degrees, and you now have 720. 720 pulses per revolution x 133.33 revolutions a second = 95999.99 pulses at 8000 rpm. Totally doable, but then why reinvent the wheel, when you could probably go buy an encoder ready made for same results, and probably cheaper.

[handlewanker]

So, as I'm learning fast, how does the set-up know when to start the count.......there must be a starting point or an indication of which tooth is no 1 on the disc.

How will the PC know when the spindle is in the stop position for tool changing?

BTW, as I have never seen an encoder, I assumed that it was a disc of some sort with lines etc scribed on it to give the pulse rate to the optical sensing head.....whatever.

Would the toothed wheel be an interrupter to a light source or a magnetic pulse by the tooth tips to a sensor as it rotated?
Ian.

[G59]

The example given above is only good as an incremental encoder. For absolute encoders that have a start point, these encoders use definite code on a glass or a plastic disc.Attachment 261424

Multi turn encoders(absolute), have more than one disk to get reference from.
They also have a small battery or capacitor in them to remember where there position is even if you have movement during power off.

Like I said, just better off buying one than trying to make your own. Incremental encoders are good and yes, in a spindle application you could have a start pulse and count how many turns it did. The problem with that is it would always look for that start cog or space, whatever, and would start counting from there.

[handlewanker]

Mysteriouser and mysteriouser.........as Alice said.....for me, buying one would be like buying an integrated circuit without knowing what the pin outs were for or what it actually did, but I'm learning fast.

By definite code I take it that means the PC can determine which direction the spindle is going in at any time by nature of the layout of the blocked in squares and their sequence as they pass the sensor.......bump bump skip bump means it is going clockwise, and bump skip bump bump means it's going anti clockwise?

I take it that is an optical solution, possibly a magnetic one too?.......whatever, it definitely is a buy in package....no home brew....and the price is only......gasp gasp as the moths flutter out of my swag bag.

So, would the encoder disc and sensor be a complete assembly, plugged into the control box and the PC just gets the signal from it and controls the spindle via Mach 3 or other?

Would that mean an extra plug in card for the control box to accept the encoder signals etc........price?
Ian.

[RCaffin]

use an steel gear where you have 3 reed contacts or npn switches

Reeds have a switching time of about 1 mS at best.

I think Option2 is pretty good for sensing, and having the spindle on the encoder is most desirable.

Cheers
Roger

[RCaffin]

have the encoder disc attached to the chuck as a disc above the ATC disc

Theoretically possible, but I doubt either Mach or other hobby controls can handle a spindle which only sometimes has an encoder.

[quote]I don't think holes would be any good as the pitch would mean a huge disc diam just to get enough pin holes in.[/quote
The HEDS encoders are 38 mm diameter. Photo-etched though!

when does the tapping start operation......as soon as the spindle rotates and the count begins?

Actually, when the Z axis starts to move. But that has to be sync'd with the spindle rotation.

one single line on the disc is extended for a second sensor to trigger and start the count sequence?

Exactly.

the spindle can be stopped reliably at the tool change position

I am not aware of any ATC which requires spindle orientation. But there may be.

Cheers
Roger

[G59]

I am not aware of any ATC which requires spindle orientation. But there may be.

Actually, any spindle that makes use of drive dogs, does.

[RCaffin]

Hi Ian

Every small CNC system I know of uses a simple A/B/I optical encoder. Originally these were HEDs encoders by HP, and they altered the industry, taking a zero off the price of the older generation. You need to read up about them. They are not expensive, but they do work extremely well. The A&B channels give you speed and direction; the I(ndex) channel gives you absolute position (rarely used though except for syncing spindles).

The version G59 shows is called an absolute encoder. They are expensive, and I don't think any small system uses them. In fact, I don't think many modern large systems do so either these days. Maybe some older telescopes might?

While optical encoders are the norm, one can also use magentic encoders on exactly the same principle. But they are more $$.

These days most drivers ASSUME the use of an A/B encoder. Those little blocks of electronics (eg Gecko) take step pulses and Dir line as commands, and in the case of servo motors expect A&B channels back from the motor. In the case of stepper motors there is no feedback used. Each step pulse is the equivalent of saying 'advance 5 microns' or something similar. (Don't take the '5' literally!)

So - no extra control card. A PC running Mach, ethernet to an ESS hardware controller, through an optically isolated 'Break out Board' (prevents nasty electrical interference), to Step/Dir motor drivers. But you don't have to use an ESS, as there are a number of alternatives. Oh, the fun....

Cheers
Roger

[RCaffin]

Hi G59

any spindle that makes use of drive dogs, does.

Oh, for sure. But do any small/home/hobby systems do this?
Drive dogs are not really needed for connections like BT30 etc, afaik.

Cheers
Roger

[G59]

But do any small/home/hobby systems do this?

Nope, and I agree.

As far as absolute encoder use, almost all commercial units use them for the axes, but few for the spindle. They use incremental with index pulse. So they get speed, direction and last position.
It is fairly easy to make your own, but the time and parts costs would add up fairly quickly. I still say use a positive drive belt, aka timing belt and an encoder on the motor shaft. No slippage and reliable.

This would be the simplest and most cost effective way. If you have 1:1 pulleys, you don't have to do any calculations. Whatever the motor spins = what the spindle does.

[RCaffin]

I'd go with the 1:1 idea too, with the encoder off the side.

Cheers
Roger

[handlewanker]

Hi, is it practical to have the encoder on a flexible connection whereas you really need the encoder to read the spindle as it rotates as opposed to the spring in a belt that it gets from a sudden load like tapping.

I think the encoder should be driven from the spindle side so that it gets exact information......drive belts that are subject to loads while milling have a stretch factor and aren't always at their utmost tension, so I imagine the spindle "could" lag behind the motor input if the tap was big enough to put a load on the system........smaller taps probably won't even be noticed.

So, hobby mills don't put enough torque on the drive to warrant having drive dogs on ISO 30 spindles.....wouldn't that then be overkill in the tooling as the TTS system with a parallel shank would more than fill the gap in that case.....draw bar pressure in good order that is.

So the encoder would only be needed for rigid tapping then?
Ian.

[G59]

wouldn't that then be overkill in the tooling as the TTS system with a parallel shank would more than fill the gap in that case.....draw bar pressure in good order that is.

Yes BT30 is overkill for hobby, but it makes an ATC much easier to implement.

So the encoder would only be needed for rigid tapping then?

Not quite. You get a closed loop system when you can monitor spindle RPM. It is necessary for rigid tapping.

I think the encoder should be driven from the spindle side so that it gets exact information......drive belts that are subject to loads while milling have a stretch factor and aren't always at their utmost tension

Timing belts have very little slack and most have zero stretch in them. Certainly not enough to affect tapping.