Re: Re: Closed loop steppers

Ian Wright wrote:

> From: "Ian Wright" <Ian@...>
>
> Thanks for the explanation Ray. I still feel a bit uneasy about the idea of
> using a 'free running' motor as a servo really is to get ultimate accuracy.
> While I can see that missing steps on a stepper system could be easily
> overcome by using a suitably sized motor, I don't quite understand how a
> servo system would avoid stalling if it hit a tough bit as it was ramping
> down to a stop - or do the servo boards all use some kind of pulsed power
> feed to maintain maximum torque at such slow speeds

Yes, the servo loop is in velocity-servo mode at ALL times. The servo
amp is a 'four-quadrant' drive system, so it either applies power to the
motor, or extracts power from the motor, and can switch back and forth
thousands of times a second, if needed to. So, the servo amp is constantly
comparing actual velocity with commanded velocity, continuously between
velocity command updates, and changing motor current to keep the velocity
right on track. If there is a 'bump', the motor slows, the tachometer senses
the slowing, and the velocity error amp commands the motor drive to increase
motor current. There is maximum power available at ZERO speed, as you'd
see if you try to deflect the motor. Even with the CNC NOT looking at
position, but just the tach looking at velocity, you can't move the motors!
I've tried this, and was absolutely amazed! It would take me a minute to
move the leadscrew .005" with my bare hands. And, I couldn't keep up that
force for more than a few seconds, anyway.

> - if they do then surely
> they are going to jerk as much as a stepper in half or microstepping mode?

No, I just don't understand why you keep talking about the jerking!
It NEVER jerks! Motion is so smooth, I have to look at the screen
to see if the machine is really moving, sometimes (with a slow feed).
It accelerates so smoothly, you can have your hand on the table and not
even know it is moving.

What I think you are missing here is that the velocity servo is computing
velocity pretty much continuously even in between the intervals where
the CNC control computes position error. My servo amps compute a
new pulse width 100,000 times a second, and the velocity and torque
servo is all analog circuitry, with several kilohertz bandwidth. So, all
this is VERY smooth. The EMC control program I use reads position
and computes a new velocity 1000 times a second (you can go faster,
if you need to) so that is pretty smooth, too. Finally, the EMC software
ramps up and down on all moves. Maybe this is what confused you,
you think the CNC program will command the machine to immediately
go to whatever velocity is commanded. I actually had too much acceleration
when I first set up EMC, and realized it when I broke a timing belt on
my Y axis. I turned the acceleration down to a more gentle value, and
haven't broken any belts since.

Finally, the CNC positioning system is reading position 1000 (or more)
times a second, and even if the motor stalls, the position read is always
accurate. So, even if you have a crash, and the servo amps trip offline,
or the CNC control senses excessive following error and shuts down,
and you then manually back the machine away from the crash, it STILL
never loses position.

My system, which is a bit underpowered, has broken off 3/8" end mills
without even slowing down, by the way. That means that it didn't get
more than .005" of following error while breaking off the endmill.
(This is during a careless setup, when the tool was not spinning.)

Jon