RE: dec drive data?

["Creager, Robert S" <CreagRS@LOUISVILLE.STORTEK.COM>]

Tom,

This is a piece of the information I was looking for.  If the DEC drive is
not moving for 1000 steps, then the brake is binding or the large sprocket
is moving.  I had to move the carriage off the break so I could see
daylight, then open the clamp ~400 steps until it was fully open (flush
against the stepper housing).  Now, ROB's DEC drive moves immediately, and
smoothly, until the balance point shifts.  This is why I don't currently
believe that the DEC drive requires a re-design.

Using the limit switch to measure, I'm getting typically 2 steps of backlash
(drive out 1 step at a time, checking for limit each step, then drive back 1
step at a time, checking for the limit at each step).  This does not account
for chain slop as the balance point is not changing.  This does account for
gearbox slop, or lack there of.

I know that near the limit switch, ROB is overbalanced away from the limit
switch, so all chain play is taken up.  I also know that when ROB
overbalances the other direction, both chain play and sprocket play will
cause a jump in the position, but I have not measured that yet.  I hope to
remove the sprocket play by epoxying the sprocket to the set screw, as that
is where the sprocket movement is coming from, and quite a bit of it.

Overall, I expect to be able to point DEC to within 1/10 degree (~120 steps
per degree), as long as the balance point does not shift.  Once it does
shift and the shift is classified, I will again expect 1/10 degree accuracy.
My only question is how much chain slop adds, which will occur at the
overbalance point, and persist for some number of degrees.  I'm thinking
that if the DEC carriage can be biased (with a spring?), that even the chain
slop won't create problems.

Now, I aim to prove the above statements of smoothness and potential
accuracy this month, and we'll see if I'm full of it or not ;-)  I may yet
be eating humble pie.

Cheers,
Rob

> 
> to some position.  I find that it does not move for the first thousand
> steps or so.  But once you make a table of angle vs the number of steps it
> takes to get there by a backlash removing rout, then it is pretty
> good.  You might set to a half degree.
>