Re: Servo-PWM-DSP

David,
The reason for the tach on high performance servos is to get an analog voltage that is proportional to speed and can be used directly in the feedback loops to stabilize the system. Once the analog voltage is differentiated, it is now a proportional to acceleration. None of the above is possible with steppers as feedback.

Normally, there is only one encoder and if the DSP approach is taken, it is used both for position and as a tach substitute.

There has been several extended articles in IEEE on pros and cons of "software tachs" and different algorithms for the math.
Bertho
========================================

David Howland wrote:

> From: David Howland <dhowland@...>
>
> I am considering putting the control functions in the interface which will eliminate the need for an analog board. Of Course, when the encoder is mounted with the motor, rather than on the table itself, the encoder will supply the tach (speed) data as a part of the position data. I understand your sugustion to consider a tach with the motor as it applies to the case where you mount the encoder on the table. I appreciate your suggustion and think it is a good one. To reduce the difficulty of locating a tach, I believe a stepper motor may be used as a tach by reading it like a generator. No one seems to have any difficulty finding stepper motors. A small stepper is all that is required and I've seen them for under two dollars.
>
> I beleive stepper motors may also be used as encoders in a similar manner. To overcome the possibility of loosing possition while they are at rest (not generating any signals), I have a scheme that works pretty well on paper, but I don't believe it is a good idea because there are plenty of encoders around, and encoders are already encoders and don't need a scheme. Why climb out the window where there is a door.
>
> The embeded control firmware person I may employ in this potential effort, is not well experienced in DSP. He is good though and given a clear understanding and an envirnoment to test the code in, he never fails. To make sucess likely, my thinking is that the project performance is 3/4 of the way up from a stepper, but not equal to an off the shelf servo system. I hope to keep the cost under control for one thing. It may be difficult to develop something which is stable across many applications and super high performance at the same time.
>
> What do you think about getting say 3 times the speed of a stepper system? 4 times the speed? Steppers are slow; they have resonance; and some discussions I've read suggust that you can tell by looking at the tooled part if a CNC system was controlled by a stepper based or servo based controller.
> My thinking is that smooth, fast, quiet, and reasonable cost are major bullet items?
>
> Thaks for some good information.
>
> >From your sugustion, it is easy enough to have both an encoder input for position, as well as an optional input for the motor speed (with a second encoder or tach) when the position encoder has backlash involved. The PIC family of microcontrollers has the required PWM and A/D inputs to reduce the component count (and cost). It's really not all that scary of a project. With a multi-channel parallel port board for a PC (which is very fast), some DC motors, some encoders, and some details it should be easy enough to out pace stepper systems. I'm am hopeful some one can spell out the software side of the concept before I smile that smile.
>
> David Howland