How can I measure position uncertainty of conveyor(belt) encoder?

You will need to know how much the conveyor belt moves between encoder
steps.  Do you know how many steps are in the encoder wheel?  Then you
will need to know how much the belt moves per motor revolution.  If
you have this information, you can use the equation below.  If you
don't have any of this information, you can get this empirically.  I
assume that you must have some PLC reading the encoder.  If I had to
do this, I would write a program which would run the conveyor belt for
a short time, enough time that the conveyor moves a few revolutions.
Put a mark on the belt and on the frame before you start, so you can
see how many feet the belt has moved during this test. Then in the
PLC, you probably have some high speed counter or an accumulator which
keeps track of the encoder count.  Have the encoder input increment a
counter each time the encoder value changes.  Use a one shot rising in
front of the counter to prevent multiple counts per encoder step.  You
will need a timer as well.  The timer should be prompted by whatever
begins your test, and stopped by whatever ends your test.  Once you
have run this test, you should have the total distance traveled by the
belt, the number of increments by the encoder wheel, and the total
time elapsed between starting and stopping the belt (note: you will
have some error in the timing because the belt will not stop
immediately).  The belt distance traveled divided by number of steps
is your distance resolution.  The belt distance divided by the number
of seconds elapsed is your belt rate.  If you need to estimate where
you are in between encoder steps, you can set up your PLC to begin a
timer each time an encoder value changes.  Your estimated position
will be the position as last noted by the encoder, plus the belt rate
times the time elapsed since the last encoder step.

1 Revolution of motor     Conveyor dist/rev (ft)
-------------------------------- * ------------------------------
# Encoder  Steps         1 Revolution of motor

leo wrote:

How fine are the encoder steps? How much backlash is there between the
encoder and the belt? How much can the belt stretch? Are there other
potential sources of error or uncertainty? There's too much missing
information.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

On Wed, 04 Mar 2009 23:24:34 -0800, leo wrote:


How can you _measure_ the uncertainty?  Or how can you _compute_ the
uncertainty?  Which uncertainty -- the encoder shaft position
uncertainty, the uncertainty of the belt position vs. the encoder shaft,
or the overall uncertainty of the position of the belt vs. where you
think it is?

If you just want the encoder's shaft-to-reading uncertainty then it's
pretty much the step size, which you can find out either by rocking the
shaft and seeing how much movement before you see a step, or by looking
up the encoder by mfg part number (assuming that you can get your hands
on the data, of course).

--
http://www.wescottdesign.com