I wrote:
> > I am trying to implement control loops with the pwm output of an
> > 16C73. As I am not an engineer, I don't have the theoretical
> > background to do it properly myself. The following two projects are on
> > my mind:
> >
> > 1. A more or less stabile adjustable output voltage 0..4V that can
> > source a current of 0..1A. I thought of the following schematic:
> >
> > ---------------+
> >                |    feedback
> >      A/D input +------------------------------------+
> >                |                                    |
> >                |                   +5V              |
> > 16C73          |                    /               |
> >                |      +-----+     |<                |
> >     PWM output +------+  R  +-----|   PNP           |
> >                |      +-----+     |\                |
> > ---------------+                    \       L       |
> >                                      ----UUUUUUU----+------>
> >                                                    _|_
> >                                                    ___ C     output
> >                                                     |
> >                                                GND--+------>
> >
> > 2. A regulation of the RPM of a small dc motor. I thought of using the
> > pwm feature of one ccp module to generate a pwm signal that
> > drives the motor (buffered with a transistor) and use the capture
> > feature of the other ccp module to measure the RPM with some kind of
> > optical sensor.
> >
> > In both cases I would like to have a resolution of about 6 bits if
> > possible. The changes in voltage and RPM won't be faster than 1s for
> > 0%-100% or 100%->0%.
> >
> > I guess that somebody has done something similar already. I would
> > appretiate any ideas/corrections/code fracments/schematics.

Gvran Mvrk wrote:
> 1: Your circuit won't work. Add a rail-to-rail op like LMC7101, put
> an NPN emitter follower to the output and the RC-filtered PWM to the
> +input The -input goes to the emitter.

Do I understand you correctly? You mean that I should create the DC
before the Transistor? I can't do that because I want prevent the
transistor to heat up. The purpose behind the L-C-circuit was that I
don't need to cool the transistor.

Why do you think my circuit won't work?


Gvran Mvrk wrote:
> 2 You don't need the CCP or A/D to do that. We've used a 4MHz 16C620 to
> generate a 20 kHz PWM with 4 bit resolution. Speed feeedback is obtained
> from the EMF-voltage obtained when drive is removed,comparing it against
> the programmable reference. This gives 5 bits resolution and simple
> statistical proessing adds the remaining bit(s). The time constant is a
> problem, an ironless armature motor with short mechanical time-constant
> usually has too low inductance, so an external inductor will be needed.
> The EMF feeedback will need some RC filtering to remove the commutation
> spikes

My problem with EMF is that I don't know what moter will be connected
to the output of (1). The motor in (2) won't change but I maight not
have enough space to use EMF. How big is such a circuit? Do you have a
schematic to share? BTW: I use a very small Faulhaber-motor in (2)


--
Stefan Bormann  s.bormann@tu-bs.de

http://www.stud.uni-hannover.de/~toci/stefan_d.htm