Encoder failures during 2007

06apr07

Sections:
Intro
11apr07 encoder 1 failure
08apr07 encoder 1 failure

plots:
11apr07:azimuth summary plots (.ps) (.pdf)
08apr07: azimuth summary plots (.ps) (.pdf)
08apr07: pointing errors caused by the failure (.ps) (.pdf):

Intro

        The azimuth arm has 2 encoders. They are situated on opposite sides of the azimuth arm by the encoder rack gear. The encoder on the gregorian dome side (az encoder 1) is used for pointing. The encoder on the carriage house side (az encoder 2) is used to measure the bending of the azimuth arm.
    If az encoder 1 fails then the system switches to az encoder 2. This allows continued operation with degraded pointing accuracy. As the azimuth rotates, the azimuth arm bends slightly. The pointing is servoed to the encoder on the dome side. The encoder on the ch side will be off from encoder 1 by the bending amount. This changes depending on the direction the motors are driving the azimuth.
    We have been having a number of azimuth 1 encoder failures recently. The system has continued running using az encoder 2 (and degraded pointing). Some observations are:

Often times the encoder 1 error looks to be a glitch (since the encoder difference remains reasonable after the failure).
When a glitch occurs the system does not automatically switch back to az encoder 1.
A reset command will sometimes clear the failure and switch back to az encoder 1.
Occasionally we have had to reboot the vertex plc.

The Failures. (top)

11apr07: Az 1 encoder failure

The azimuth 1 encoder failed twice during the morning of 11apr07. The failures occurred at 2:35 and 5:30 am. The operator was able to reset the errors remotely. The two failures occurred at: az=203,za=5.5 and az=332.4,za=14.999.
The azimuth summary plots (.ps) (.pdf) show the failure:

Page 1: has the az position and the encoder differences. The dashed vertical lines are where the errors occurred.
Page 3: The red line at the bottom shows the encoder 1 faults.
Page 8: This shows the azimuth and dome za for the day. The dotted red lines show the failure positions.

Processing: x101/070411/chkaz.pro

08apr07: Az 1 encoder failure

The failure occurred at 12:33. The azimuth was sitting at 270 degrees and the dome was at za=10 degrees.

The azimuth summary plots (.ps) (.pdf) show the failures:

Page 1: the azimuth position and the az1-az2 value for the day. The dashed line at 12.5 is when the failure occurred. You'll notice that after the enc1 fault, the azimuth encoder difference was still reasonable. This means that az encoder 1 was still working after the failure. This makes me think that the problem was a glitch and not a catastrophic failure.
Page 3: The red trace at the bottom shows the enc1 fault bit. It stayed high until around 22:24. The glitch in all the bits at 22:24 was a power dip. It looks like the power dip caused the system to reset, removing the encoder 1 failure.

The pointing errors caused by the failure (.ps) (.pdf):

The pointing model was made with encoder 1. When we switch to encoder 2, the encoder difference becomes the pointing error. The error on the sky is the encoder difference times the sine of the zenith angle. The plot shows the az encoder error on the sky caused by using encoder 2. The error was present between the two dashed red lines. It got up to about 25 arc seconds.

Processing: x101/070408/chkaz.pro

<- page up
home_~phil