Automated Coil Winder (in the works)

[chevalij]

Cool setup. I just lucked into a bunch of 80-20 section cutoffs, too, and that's a good way to use them.

I was actually going to put the bobbin pulley right on the stepper motor shaft with a shaft coupler.

Not a great idea. Steppers aren't made for constant rotation at that speed. Get a DC motor and program the speed. There are digital speed controllers that will do that.

A stepper will work fine on a winder, just like on the CNC router as long as you keep the motor torque and resonance-bounce down to where the motor does not lose a step. The same concepts apply - step size, oscillation and resonances, torque, etc.

Remember, CNC routers use the stepper to position the motor, not drive it.

Frankly, I'm not sure that G-codes are needed. The task will always be the same - spin the bobbin in a single direction at a coordinated position/speed to the traverse. Move the traverse in coordination with the bobbin position on its turning.

And you plan on repeating that task how?

One of my objectives has been to build in the ability to "record" a bobbin being either wound or unwound into a long series of rotational/traverse position pairs.

yada yada

Impossible with a stepper, possibly with a linear encoder. However, the problem is, to properly record the position the wire comes off the bobbin, it would have to be pulled perpendicular to the bobbin at all times and the friction of the measuring device would have to be almost nil so as to not affect the position reading. Any deviation between the location that wire is on the bobbin and the measured position would make the whole exercise useless. To read the encoder, you'd need DAQ hardware and software. Even if you could record the correct position of every wrap, how would you duplicate it? You'd have to program EVERY wrap position back into your stepper software.

[NB70]

Not a great idea. Steppers aren't made for constant rotation at that speed. Get a DC motor and program the speed. There are digital speed controllers that will do that.

Its true that a DC motor would be better to rotate the bobbin, that's the way any commercial machine would do it.
EMC2 can control the rotation speed of a DC spindle with PWM. With a rotary encoder on the spindle the linear traverse could be synchronised using G33 codes (spindle synchronised motion). We then have a single axis machine with a variable speed spindle.
Having said that, my first machine will use steppers out of convenience. Version 2 will have a DC motor driving the rotation, but I need to learn all about the PWM stuff first! Of course this is all completely overkill, but thinking about and making the machines is the part I enjoy.

Nathan

[chevalij]

It's easier to just use a standard stepper controller and have an optical switch or similar provide the step command to the controller. One rotation, one step. I intentionally bought a linear stepper with a 0.004 step. Average wire dia. is about 0.003. This would leave about 0.0005 between winds. Not ideal, but I think better than stacking the wires at a 0.002 step. Using this method, you can vary the speed and the stepper will traverse according to speed. Of course, I haven't got it all together yet, so it may not work at all.

[R.G.]

Not a great idea. Steppers aren't made for constant rotation at that speed. Get a DC motor and program the speed. There are digital speed controllers that will do that.

It's not a question of what they're made for. It's a question of what we can pervert them to do. Actually, steppers are a derivative of the synchronous motor, and there are linear stepper drivers that drive the step sequence to approximate a two phase sine wave drive to the stepper, making it ... a DC motor.

Yes, I'm very aware of DC motors, controllers and encoders. I'm not trying to just barely grasp the beginnings of motion control. It's an attempt to minimize the precision mechanics for a DIY bobbin winder that will never make hundreds, much less thousands of bobbins. I think we can be allowed a little motor abuse, if it comes to that.

Remember, CNC routers use the stepper to position the motor, not drive it.

I do remember that! Oddly enough, the traverses on CNC routers last seconds at a time, and are repeated. It comes very close to spinning a bobbin if you calculate the feeds and speeds.

And you plan on repeating that task how?

I plan on storing the position pairs in a list and playing them back from that list. There are computers here to help, remember?

Impossible with a stepper, possibly with a linear encoder. However, the problem is, to properly record the position the wire comes off the bobbin, it would have to be pulled perpendicular to the bobbin at all times and the friction of the measuring device would have to be almost nil so as to not affect the position reading. Any deviation between the location that wire is on the bobbin and the measured position would make the whole exercise useless. To read the encoder, you'd need DAQ hardware and software.

Bummer. You mean you'd actually have to cobble together some more hardware, of about the difficulty of a phono turntable arm on a rotary encoder to read the position and angle that the wire is coming off the bobbin? Well, I guess I'll never get to there; tsk, tsk. I guess us dumb old plow boys from Alabama never will understand these here machines.

Even if you could record the correct position of every wrap, how would you duplicate it? You'd have to program EVERY wrap position back into your stepper software.

That did look like a huge task to me. My first urge was to store the rotation/traverse steps in a huge list. Then write a list processor to run through the list again, converting the position to phase movements with a stepper phase driver routine. Actually, converting a position list like that to even something like step/direction is simpler than, say writing an assembler.

We have these here computers to help us, right?

I'll tell you what. You're absolutely, positively 100% percent correct. Nothing about this scheme, from the largest concept to the smallest detail is possible in this universe. Natural laws forbid them from ever working. So don't worry yourself a bit about it. I'll just waste my time on it on my own, OK? And I'll periodically stop and remember that just like you told me, it can't be done. That way you don't need to keep reminding me. I'll remind myself and save you the trouble.

[chevalij]

Well, I'm glad the pressure is off me to remind you. Should make the weekend go a little smoother. I'm not saying it's impossible, just extremely difficult and going to be expensive to do it right. You'll still need G-code or some C routines to program output steps. If you ever get it done, send me a pic. Oh yeah, include one of the plow

[RedeemerbassesScott]

Scott,
is the linear actuator for the traverse one of these?
http://www.mpja.com/prodinfo.asp?number=17284+MS

Yes David it's similar to that one. Here is the exact one I have http://cgi.ebay.com/Airpax-digital-l...QQcmdZViewItem