If you've done some reading and perhaps looked over my CNC Dictionary (http://www.thewarfields.com/MTCNCDictionary.htm), you have at least an intellectual idea of the concepts of steppers, servos, backlash, closed loop, and open loop operation. You will have heard that backlash is very bad, and that a closed loop or servo system is much better than an open loop or stepper based system for CNC. What's lacking is an intuitive feel for why? Or, how bad is it really? What will happen if I choose a stepper based open loop system with lots of backlash?

In order to answer those questions, I developed a simulator in Excel that may be used to explore the concepts. The spreadsheet model is pretty simple. It assumes you want to command the machine to cut a circle. I chose the circle because they're inherently a bit of a torture test for this problem because the axes change direction as you move around the circle.

If you'd like to play with the model, visit my page for more detail:

http://www.thewarfields.com/MTStepperServoBacklash.htm

Meanwhile, here are some graphs showing the results. First, a graph showing the effects of 0.020" backlash on each axis when drawing a 1" diameter circle:

http://www.thewarfields.com/img/Toys/MachineTools/CNCCookbook/BacklashGraph.jpg

The plot shows the path followed as well as the ideal "As Commanded" circle. You can see the classic backlash "ears" or glitches each time an axis must reverse direction.

How about what happens in an open loop stepper based system when 2% of the steps are lost?

http://www.thewarfields.com/img/Toys/MachineTools/CNCCookbook/Stepper2PctGraph.jpg

As you can see, 2% produces a completely unacceptable result. You begin to see why folks turn down the speeds and turn up the torque on these open loop stepper systems to try to avoid lost steps at all costs!

Let's compare that against a servo based system, assuming no backlash:

http://www.thewarfields.com/img/Toys/MachineTools/CNCCookbook/ServoVsStepperGraph.jpg

The servo system plot (dark blue) is almost an exact match for the ideal curve (yellow), but our little stepper is way out of sync. The assumptions used were again 2% of steps lost, but the servo system had the ability to correct 10% of the error with each step using it's closed loop. Amazing what a difference that made!

There are a lot of different conclusions to be reached looking at a simple simulation like this. I've written some of it up on the web page and don't want to reproduce it all here in a message that would be too long.

I thought being able to see, compare, and play with the parameters in such a simplified simulation testbed might help people understand a little better the differences in these systems.

Best Regards,

Bob Warfield

Your simulation is being overly kind.

In reality, we found that the backlash actually results in "chatter" at the direction change points, and/or it caused severe flat spotting at these same points.

This due to the fact that although the stepper or servo is trying to move to a commanded signal, cutter forces can force the table to either bounce around thus generating chatter.

If the table is heavy enough, although no chatter occurs, the table isn't responding to the requested/commanded move so it sits there and simply cuts flats.

Backlash come "addresses" the issue, it doesn't FULLY COMPENSATE for the issue. If you want to solve the backlash issue for good, get rid of it.

None the less, the simulation is GREAT for showing folks what these "mechanical vagaries" can cause with respect to actual machine motion versus commanded machine motion.

....In reality, we found that the backlash actually results in "chatter" at the direction change points, and/or it caused severe flat spotting at these same points.....

I had posted a picture of an axis creating what I guess you could call chatter;

http://www.cnczone.com/forums/showthread.php?t=22589

My intepretation was that it was due to the axis getting ahead of itself during acceleration; sort of creating its own 'forward'lash??

Your simulation and explanation is very good; it brings out how impressive CNC machines are when you consider there are hundreds of thousands in routine operation doing circles that are accurate to within less than a thousandth of an inch in most cases.

Geof: the chatter in your earlier post could be the result of something else besides backlash.

First some defininitions to help the nooby's:

displacement is distance traveled (inches, feet, whatever).

rate of change of displacment is velocity (inches per minute of infeed/travel).

rate of change of velocity is acceleration (to go from 9"/min to 12"/min involves a velocity change of 3 BUT but if you do that in 0.5 sec, you see a change of 3"/min/0.5 sec or roughly 360"/min/minute normalized for a full minute. Pretty high factor, no??)

Sometimes it aint the velocity that gets you, its the acceleration that sends the machine spazoid.

The motion you're asking for is coming at a unique intersection of relative constant velocity required (the flat cutting). This is dumping out into a point on the circle where the motors are immediately entering a region where the X and Y velocities are simultaneously changing or will be asked to change QUITE rapidly so as to make a circle.

You could have inadvertently found a situation that the machine controller could have had a real tough time keeping up with - sort of doing something mindless and then IMMEDIATELY asking it to rub it's tummy and pat it's head at same time. I'm sooo confused!!!

First remedly: I'd try is to slow down the inches/min, DRASTICALLY. This simple fix is something we learned via trial and error. It helps eliminate follower error when cutting profile shapes. PITA to productivity but it sure did wonders for the finish.