Date: Tue, 5 Nov 2002 07:50:46 -0800
From: Russell E Owen
Subject: Re: 2.5m obslog 20021102 (MJD 52581)
>Problem #1 - Secondary actuator B did not find reverse limit switch first try
>-----------------------------------------------------------------------------
>
>While homing the secondary we saw the following time-out while homing axis B:
>
>:A=0;: B=0;: C=0;: XQ#HOME
>: 0363.1,  0363.1,  0363.1,  0000.0,  0000.0 max sec to find reverse limit
>switch
>Reverse limit switch not depressed for axes:  0,  1,  0,  0,  0 - trying again
>  0000.6,  0363.1,  0000.6,  0000.0,  0000.0 max sec to find reverse
>limit switch
>  0030.5,  0030.5,  0030.5,  0000.0,  0000.0 max sec to find home switch
>  0008.6,  0008.6,  0008.6,  0000.0,  0000.0 sec to move away from home switch
>Finding next full step
>
>Re-homing a second time, we had no problems.  Note that we started the first
>homing from scale=1, focus=0 (for inst=spectro), set up to take mirror numbers
>(see mirrros section).  We _hope_ this was an isolated time-out.
The first homing worked; there was no need to home again (though it's
nice to home twice if you have the time). Homing makes two tries to
find the reverse limit switch, if necessary, and in this case it
succeeded on the second try (thus the message "trying again").
-- Russell
More details on the limit switch problem:
The axes have a bit of crosstalk, such that motion in A or B will
cause slight motion (relative to the reverse/home limit switch) in C.
Homing consists of running each axis towards home until it touches
the reverse limit switch, at which point that axis halts with full
acceleration (as you'd want in such a situation). The axes accelerate
quickly, so each switch is barely depressed. Suppose B reaches the
switch first; A and C keep moving and this may cause B to release its
reverse/home limit switch. I believe that's exactly what happened in
the case above. The automatic second attempt almost always fixes the
situation--all home switches are almost always still depressed after
all axes halt. Very rarely I've even even that fail, in which case
issuing the home command again does the trick.
Note: if the automatic second attempt takes a long time to find the
home switch, then the axis in question is almost certainly stalled
(not moving), and there is no point in trying to home again. You can
also monitor the Mitutoyos to see if the axes are moving.
I wish my code could tell the difference between these two cases, but
the Galil makes this nearly impossible (and believe me, I tried!). I
use soft limits as a time limit, so each axis will either stop when
it hits its reverse/home limit switch or after a reasonable time has
passed. This is safe and simple. Unfortunately, there is no to find
out whether an axis halted because it hit a limit switch or because
it reached a soft limit--the stop code is identical for the two
cases. Thus my code cannot easily say "home switch hit, but then
released; trying again" vs "timed out while trying to reach the home
switch; giving up". One could use multitasking and try to set up
timers, but in the Galil this would be messy and less reliable.