Re: Low cost Servo Controller

True, I was assuming from the previous conversation that the system would be
built using both linear encoders for position detection, and rotary
encoders. The rotary encoders would be just for motor velocity generation.
On each cycle, an difference error would be generated between the current
position, and the previous cycle goal. This error signal would be used to
throttle the feed rate when tracking errors start to occur. The system
would then generate target velocities for the current cycle goal, and send
them to the 629's. This is would not be the most accurate way to do things,
I suspect that the positioning error inside a single cycle will be
unacceptable for moderate to high precision applications.

I am not advocating this approach, I believe that it would be much easier to
develop an integrated system using the TI motion DSP's (TMS320 series). The
chips are cheaper, the hardware simpler, and the development exercise
becomes almost totally a software one. It's much easier to do basic
research (incremental design) in software than hardware (particularly when
you have DSP's with flash memory).

I am building a servo/encoder based system. I have built a table-top test
x-y table (using threaded rod and joiner nuts), and am developing the
software in x86 assembler (software encoder decode and pwm generation). The
x86 is too slow for real machine control, but it provides an opportunity to
solve all the problems I will face with a full side/ball screw machine.
Once the software is right, I will port it to the TMS320x and build an EMC
interface.

The application is a gantry wood router.

Dean

> From: Jon Elson <jmelson@...>
>
> We're talking about the LM628/LM629 series, here.
> Yes, but if you put it in velocity mode, it won't ever stop (unless you

set

> velocity to zero). So, what's the advantage here? Making an outboard

chip

> do a couple of small calculations? A pentium CPU can do the same
> calculations, in full floating-point, too, 100 x faster than it could

communicate

> what it needs to over the ISA bus! And, it STILL is NOT looking at
> motor current, which is significant, because the inductance of a typical
> medium-power servo motor introduces SUBSTANTIAL lag in the
> control loop. SO, you change the pulse width of the PWM amp
> NOW, and the motor current, which is almost identical to torque,
> changes sometime later. Oops, too late, now we're going the wrong
> way!
>
> Jon
>