In ,
Steve Taylor typed:

Just to say thank to all who sent suggestions for improvements.

How many turns does the final drive from the motor make with the
actual focussing mechanism ? You could just detect position by
inference from the rotation of the drive wheel- use a multi-turn pot
( 6 quid)


I have a multi-turn pot around somewhere. It is quite big and looks like a
precision device.

Linear plastic pot ?

Shaft encoder from a mouse ? I'm also pretty sure that there are
encoders in VCRs. Have one fixed stop in your focusser.

The nice thing about the current system is that it is an independant module
that I can remove easily for manual focusing and I am reluctant to start
adding limit switches to the Meade focuser. However, there is a way around
this requirement. The controller can simply make some assumptions about the
initial position of the focus tube and I will always do an initial manual
focus before starting. If it is a few mm off, so what?


Drive onto the
stop and set your zero reference. That would need get nasty if you
stick with an analogue servo though - you'd need a counter and DAC to
get a voltage to drive the loop....not for the faint hearted.

You got me thinking, see below...


What kind of positional resolution would you need I wonder ?

Not sure about required resolution at the focuser tube..but I do know that
one turn of the DC motor makes no visible difference to the focus, it takes
quite a few turns before any optical changes occur..and that is after taking
precautions that I'm not looking at backlash in the gear train.. Here is my
plan:

I don't need absolute positional feedback, just relative feedback so I can
tell the remote drive "Move in/out a bit from the current position" Bear in
mind that the focus tube will always be initially set somewhere near optimum
focus before switching the controller on.. The "move" command translates to
"Move in/out xx motor turns" Where xx will always be 1. For this I need to:
1) Count DC motor turns.
2) Stop when the required number of turns has been reached..but not
suddenly.

1) I could use a optical encoder disk/led setup to count DC motor revs.
Alternatively, I may be able to count revs by monitoring the DC feed to the
motor and watching for commutator breaks. In other words, use the motor
itself as a tachometer. There is the complication that I'm currently pulsing
the feed to the motor for speed control but a combination of zener
thresholds and suitable time constants might allow me to see the commutator
open circuits as the motor shaft rotates.

2) Given that I can get a supply of tachometer pulses back from the motor in
step 1) I need to count them. A digital counter is one solution but I have
thought of something simpler. I integrate the pulses and compare the
integrated signal with the output from my multiturn pot that is tied to a
stabilised reference voltage. The comparitor output is then used to
determine the pulse width and polarity of the drive to the motor.

In the above positional resolution will depend on:
a) Motor turns per unit movement of the focuser tube.
b) "Goodness" of the integrator circuit and comparitor.
c) Resolution of the multiturn potentiometer.
d) Stability of the reference voltage.

I need to do a *lot* of prototyping to get the above running...should be
fun. What do you folks think?

Jo