Looked at improving the reference cavity transmission by making adjustments to the mode matching.  I moved the last
lens before the reference cavity L12.  The reference cavity transmission was ~1.65 V when I started and was ~2.87 V
when I finished.
  L12 original micrometer position 3.48
  L12 new micrometer position 3.00

    When the reference cavity was unlocked and blocked there was a -1.6 mV offset in the output of the RFPD.  When the
reference cavity was simply unlocked, the RFPD output was 475 mV.  When locked, 80 mV.  This yields a cavity visibility
of ~83%.  Whis not bad but not great.  L12 could not be moved back any further because its base would hit a mirror mount.

    Previously the common gain was set to 20 dB.  The attached plot "tf.png" shows the transfer function for a few values
of the fast gain.  The higher the fast gain is increased, the more pronounced is the dip in the transfer function, which
in turn means the cross over between the fast actuator and the Pockels cell is not quite right.  I found that dropping
the fast gain to 3 dB improved this a little.  It could be that a finer adjustment would fix things but this is the best
I could do in the time available.  Finally the common gain was increased to 21.5 dB in order to push the unity gain up.
With a common gain of 21.5 dB and fast gain of 3 dB, the unity gain frequency is ~518 kHz with a phase margin of ~60
degrees.

    I'm not quite sure how robust these settings are since there's a large-ish resonance out at ~762 kHz which may
cause some problems.

    There is still some room for improvement.  The increased reference cavity transmission should be enough for ALS from
the end stations (hopefully).