Constant Surface Speed implementation

Hi everyone,

I have what I think is an unusual application, but I think that linuxcnc has the ability to provide an elegant solution to it, and I wanted to get some feedback before committing to the build. Briefly, I need to build a roll-to-roll paper coating machine. I have built several CNC machine tools around linuxcnc, so I am reasonably familiar with the system, but I am about to venture into the unknown.

The crux of the problem is this: as you draw paper off the supply roll and onto the takeup roll, the diameter gets larger, which means that if you run the takeup roll at a constant rotational speed, the speed of the paper through the system increases with time. I realized, though, that this is the analogous (though reversed) situation to cutting large diameter parts down to small ones on a lathe. As the diameter changes, so too does the surface speed. Enter linuxcnc, with its G96 constant surface speed command capability. What I want to do is put a rotary encoder on the motor that drives the takeup roller and a linear encoder that bears on the roller/paper itself with a freewheel on the end. As the roll diameter increases, the encoder registers the change in position and feeds the new position back to the motion control system and the spindle speed is automatically adjusted for the surface speed I specify.

All that sounds perfect in theory, but here is where I run into my own ignorance: does the spindle-synchronized routine that makes the G96 code work require that the machine is actively controlling both axes, or can it be used to passively correct spindle speed as in the above description? By analogy, what I am asking is this: can you configure linuxcnc to control the spindle of a manual lathe for constant surface speed control if you have an encoder that reports the position of the cutting tool?

Thanks in advance for any input/advice! Also, if you know of a better way to skin the cat, I'd welcome that advice too. I'm not married to this idea, it just seems like the best solution that I could implement.

By experience:
Add a roll that is flying and use its position into a pid.
It's a kind of force regulation.
You will never know the correct diameter or thickness of your foil.

Such a flying roll can be on a long lever or a top down sliding mechanism. Very important is the squareness against both media rolls.
Mike

Hi Mike,

Thanks for your suggestion. I think I see what you mean about not knowing the thickness: if the control is at the takeup roller, you are correcting for variations in thickness that have already been coated, not the bits that are currently being subjected to the process. So I am sure I am understanding you correctly, are you suggesting that I put an encoder on this tension roller and use that to control the surface speed?

No, don't try to measure the bypassing length of your foil. (that is what you do by any sensor measurement over a rolling movement).
Give your construction a device that can measure the tension in the foil.

Example. Take your two rolls and let the foil hung down between them. Just lay an additional roll in that hanging part so you get a U. If you fix one roll and turn the other this middle roll will go up and down. Use this position change to control the winding roll.
Mike

Is there any way to control the feed rate with rollers that would pull the paper off the supply roll but be ahead of the coating section? If so then you could just use a servo in torque mode to load the take up roll. In photo labs the take up roll flanges were 18" diameter and just rested on a couple of rotating steel rods that turned at a constant speed. The slipping friction would supply the necessary tension on the paper regardless of the diameter of the paper on the roll. Occasional spraying the rods with silicone would adjust the tension.

I would suggest controlling roll speed with a PID component and then measure the web speed with a friction wheel and encoder.

But you probably could use G96. The spindle speed of a lathe does changes as you jog in and out.

But I think direct PID control of web speed just makes more sense.