There was a series of azimuth motion failures
on 07dec09 and again on 31jan10. A motion failure can be caused by
encoder showing that the system moved opposite to the direction
requested. It can also occur if requested motion does not occur after a
specified time interval (assuming the system is active).
07dec09:
Aeronomy had been running for a few days and had
been getting the motion failures. The moved the azimuth from 168 to
168+90 degrees at slew speed for the entire time. The errors occurred
when:
azimuth moving at .4 deg/sec and passes 168 degrees.
azimuth then commanded back to 168
azimuth overshoots 168 to about 168.06 (this is normal)
azimuth then approaches 168 but never quite arrives (gets to
about 168.02).
azimuth sits for a 5-90 secs and then moves away from az=168.
This causes the motion failure.
The plots show the azimuth position for each azimuth swing as it
moves toward az=168 degrees. The vertical axis is the azimuth position.
The horizontal axis is the number of seconds since the azimuth crossed
az=169 degrees.
Top: This shows the azimuth swings when a motion failure did not
occur.
Bottom: This has the azimuth swings with the motion failure. The
* and dashed vertical line occurs at the motion failure. At each motion
failure the azimuth position jumps up away from azimuth=168 (the
requested position).
Resolution:
The speed monitor output is used by the PI loop to
control the azimuth. All 8 of the amp velocities are summed to get the
final control command. Looking at one of the amplifiers, it's speed
monitor output had a large dc offset (when the azimuth was stationary).
After replacing the amplifier the az motion failure problem went away
(at least till jan10). processing:
x101/091207/agc.pro
31jan10:
The az motion failure occurred 3 times in the early
morning. The first time was during an astronomy. The last 2 times was
the aeronomy azimuth swings.
Page 1: azimuth position vs hour of day. The failures are flagged
with red,green,blue lines. The last two failures occurred at 180
degrees when the azimuth was turning around.
Pages 2-4: azimuth failures 1,2,3
Top: azimuth position vs time around the failure
Bottom: az velocity during failure. The red line is computed
from the encoder. The black line is the velocity from the amplifiers
(all 8 amplifiers values are averaged).
Page 4: Amplifier velocity vs Encoder velocity
Top: all data 5am to 10am
bottom: blowup showing abs(vel)< .02 deg/sec.
There is a large vertical spread in the amplifier velocity
when the encoder velocity is 0.
Page 5: Encoder Velocity - amp velocity vs time when encoder
velocity < .01 deg/sec.
The plot shows how velocity difference when the encoder
velocity is small. motion failures 1,2,3 are marked in red,green,
and blue.
Failures 2, 3 (green,blue) have a large amplifier offset and
then jump down to 0 when the failure occurs and the motors are shutdown.
Summary:
failures 2,3 (the az failures while swinging the azimuth arm)
show motion in the wrong direction. Failure 1 shows no jump.
All 3 failures show encoder velocity=0 while the amp velocity is
non-zero prior to the failure.
Probably need to check the individual amplifier speed outputs to
see if one has a dc offset.