Dumb Question

[Pretorien]

I was working with my DRO equipped X-2 yesterday, watching the numbers and thinking about how convenient it would be to finally get a CNC set-up to avoid all of the cranking when my mind wandered off on a tangent:

My DRO reads to half a thousandth which is more than adequate for my application. In use, it is a closed loop system with the operator as part of the feedback loop (eye>brain>hand)

Could not a similar system be implemented for CNC applications? Programming might be a bit different but it seems to me that the basic functions would be the same: Driving from coordinate (A) to coordinate (B) @ N increments/unit time with a ramp up from (A) and a ramp down as the system approaches (B)

The usual constraints of cutter speed, load, etc would apply but, there would be no issue of missed steps or loss of synch and backlash would not be a factor.

Seems obvious so there must be a flaw somewhere. What have I overlooked?

Thanks

[Bubba]

Not a dumb question at all, but if I am reading your statement correctly, there is a flaw. Correct me if I am wrong, but you want to use your dro as the feedback to the cnc program.
It will work, EXCEPT for the problem of backlash. If your using this for feedback to servos, they always work on a "following error" and backlash between the driven element (the lead screw) and the encoder (dro in this case) will cause the system to continually hunt and may go into distructive ossicilation.

As the saying goes, "been there, done that".
Sorry to burst your bubble.

[Andre' B]

With such a system backlash is the problem.
Works like this.

Control sees the axis is out of position and so sends a bit of power to the servo motor.
Control looks and does not see the position getting better so bumps up the power a bit more.
Repeat until finally all the backlash is taken up and things start to move.
Problem is now the servo motor is moving a lot faster then the control thinks and so the desired position goes by before things can come to a stop.
The cycle then repeats in the opposite direction.

This can result in anything from a constant humming noise to a movement that builds up to the point things start breaking.

You can detune the servo system and kind of make it work but you get a very slow sluggish machine, almost like turning the cranks by hand.

Most all machines that do have linear scales are build with the idea of reducing mechanical backlash as close to zero as possible. Even then they will also have some kind of rotary velocity and/or position encoders directly on the servo motors. The control can then use both encoders to decide what to do next.

[Pretorien]

Let me preface this by saying that I do have some experience in the feedback control world although it was long ago with vacuum tube analog electronics -both in audio systems and magnetic suspensions for laboratory instruments.

Let's take a simple example - a straight cut at a 45 degree angle:

The logic would look something like this:

1. increment X - no movement
2. increment Y - no movement
(do this n times until there is movement - assume for this example that "X" moves first)
3. increment X - movement
4. increment Y - no movement
(hold X without moving until "Y" moves)
5. increment Y - movement
6. Increment X and Y simultaneously until the end point is approached with a ramp-up to preset cutting speed.
7. Slow rate of incrementing to bring the system to a stop at the predetermined position

Follow a similar scheme if the cut continues in a different direction.

A curved path would simply require a different and varying relationship between the rates of movement of "X" and "Y" and a similar "wait" function when the path is reversed - e.g. cutting a circle

It would seem to me that the worst case path error in this example would be equivalent to the resolution of the position reader.

?

[Bubba]

On the other hand, unless you remove the backlash, in the case of your circle, you will at some point progress into the position of "climb cutting" and your system will "jump" and remove any backlash that you have. This can have again disastrous effects on both the part and the machine and depending on the violence of the movement, your shorts:{(

[Pretorien]

Good point - on one or two occasions I have had something similar happen while operating manually so clearly, the system needs to be as mechanically sound as possible. When this has occurred, i was using a large (for me) cutter - 1/4 or 3/8". Most of my cutting is done with 1/16" or 3/32" cutters and, while one certainly does not want to regard cutters as a disposable "overload" control, I doubt that they are strong enough to move the workpiece. It would also seem that a combination of tolerance for missed steps and somewhat larger than normal motors would allow one to run the machine fairly "tight"

(Did I mention that when I get an idea I'm like a dog with a bone and will persist until I run into a wall or fall of a cliff?)