# Pointing model gradients

#### 18jan08

For some
applications you need to know how fast the model correction changes
with az, za so you can estimate the error you make if you
evaluate the model at a "nearby" point. To check this, Model 17 was
used to compute how fast the pointing model changes.

The steps tested were:

- If you move by the slew speed (maximum velocity) in 1 second, how
much does the model change.

- This can be useful if you want to evaluate the model correction
for the current az,za but you have the az,za position of the
previous second.

- Measure the model correction at az,za, subtract the model
correction, and then recompute the model correction for the new az,za.
How much does the model correction change?

- This is useful if you want to remove the model correction from
then az, za measured by the encoders.
- The encoder position already has the model correction included.
You really want to know the model correction for the az, za before the
correction is added.

An az,za grid of 1 degree steps in za and 5 degree steps in az were
used to evaluate the model.

The plots show the changes in the
model correction for various steps (.ps) (.pdf) :

- Top: the model errors vs za. Black is error in az, red is
err in za. The vertical spread comes from the az going 0 to 360.
- 2nd: The change in the model Az,Za errors when the az, za motion
is .4deg and .04 degrees. This is the maximum that the az,za can move
in 1 second.
- 3rd: The change in the az, za errors if you move the az,za
position by the model error at that position.
- Bottom: The total error for the 2nd and 3rd plots. The az, za
errors have been added in quadrature.
- Page two shows the same changes plotted versus azimuth.

#### Conclusions:

- You can probably ignore the changes at za=2,3 degrees since you
spend little time their.

- If you evaluate the model 1 slew step away, the largest error is
around 2 asecs.
- If you evaluate the model without removing the model correction,
you can be off by about 1 asec.

*processing: x101/080117/chkmodcor.pro*
page up

home ~phil