\documentclass[12pt]{ligodoc}
\usepackage{amsmath}
\usepackage{graphicx}
\usepackage{wasysym}
\usepackage{ifsym}


\newcommand{\murad}{\ensuremath{\mu \textrm{rad}~}}
\newcommand{\haz}{\textcolor{red}{\textbf{Hazardous~}}} 
\newcommand{\tla}[1]{\textcolor{green}{\textbf{#1:~}}} 
\renewcommand{\labelitemi}{\Square}
\addtolength{\itemsep}{-0.75\baselineskip}

\author{S J Waldman}
\ligodccnumber{T}{09}{00157}{v2}{I}
\title{Proposed changes for the Advanced LIGO Output Mode Cleaner}
\begin{document}

\section{Introduction}
The Advanced LIGO Output Mode Cleaner enables DC readout of the
differential arm error signal.  DC readout was a central component of
the Enhanced LIGO program,  enabling extended prototyping and
commissioning of low noise DC readout.  Although the Enhanced LIGO
Output Mode Cleaner performs as designed,  there are a few points
where design changes may improve performance and usability.   

DC readout, also known as homodyne detection, uses carrier light at
the anti-symmetric (AS) port as the local oscillator against which the
signal sidebands interfere.  In contrast to RF heterodyne detection
for which the photodiode signals occur at RF frequencies, in DC
homodyne detection the photodiode signals occur directly at the signal
frequency.  For Advanced LIGO, this requires shot noise limited
detection of the carrier light from a differential arm offset across
the entire signal frequency range from 10 Hz to 10 kHz.  The OMC
spatially and chromatically filters the AS port light so that only
differential arm signals fall on the photodiode.   

The OMC consists of a suspended,  semi-monolithic four mirror bow tie
cavity with round trip length of approximately 1 meter.
Figure~\ref{fig:zemax} shows the as-built optical layout for the H1
Enhanced LIGO OMC.  The basic design will remain unchanged:  two QPDs
determine the input alignment,  four mirrors define the filter cavity,
two actuators (one fast PZT, one slow thermal transducer) match the
cavity length to the carrier frequency,  and two DC photodiodes detect
the mode cleaner transmitted light.   

The following section will enumerate the proposed changes. 
\begin{figure}[hbt]
  \centering
  \includegraphics[width=\textwidth]{OMC_AsBuiltAnnotated.pdf}
  \caption{As built optical layout for the H1 output mode cleaner.} 
  \label{fig:zemax}
\end{figure}

\newpage
\section{Proposed changes}

\subsection{Monolithic tombstones}

The eLIGO OMC uses standard REO 1 inch optics bonded to custom designed
fused silica ``tombstones''.    The advantages of this scheme,
parallel procurement and  well understood tolerances,  are outweighed
by the disadvantages,  difficult tombstone manufacturing, restricted
access to optic surfaces,  and tight spatial constraints.  For the
advLIGO OMC we will procure custom tombstones for all of the fixed
optics consisting of optical quality fused silica prisms with
coatings applied directly to the prism surfaces.  

\subsection{Fused silica breadboard}

The eLIGO OMC optics base plate,  also known as the breadboard,  was
constructed of Zerodur in order to minimize the effect of thermal
drifts in HAM6 and from the heat of the electronics and thermal length
actuator.  In practice,  thermal drifts have not been a problem and
the manufacturing in Zerodur was more difficult than anticipated.
Future breadboards will be manufactured out of fused silica (not
optical quality) for which the thermal properties are practically
equivalent and the manufacturing better understood.

\subsection{Increase payload weight}

Fully loaded, the eLIGO OMC weighs 6.9 kg (higher than the intended 6
kg). We consider 7~kg the baseline mass for the advLIGO OMC.

\subsection{Improving the length actuation}

The fast PZT and slow thermal actuator combined to provide an adequate
dither error signal, high bandwidth control loops, and long actuation
range.  We will retain the two actuator design for advLIGO. However,
both actuators were designed with the mirror high reflector surface as
the bonded surface recessed behind a small aperture.  This aperture
caused two problems: it was the limiting aperture in the OMC,
potentially causing clipping, and it made access to the mirror
surface for cleaning difficult.

Consequently,  for advLIGO we will bond the optics to their back
surface with  the high reflector as an unobstructed front surface.  We
will continue to extract the beams through the high reflectors for
diagnostic purposes. For the PZT actuator,  this design change is
mainly an issue of tolerances:  can the PZT mounted mirror be aligned
in pitch taking into account the  wedge in the substrate?  

In addition to the small aperture, the eLIGO thermal actuator
introduced a pointing into the OMC that should be eliminated. The
actuator also had an inadequate temperature read back that a)
introduced an additional time constant and b) was sensitive to pick up
from the PZT drive signal. For advLIGO, a full redesign of the thermal
actuator will be required.

\subsection{AFWFS alignment}

The Audio Frequency Wave Front Sensors (AFWFS) allow readout of the
OMC cavity alignment without large angular dither lines.  The AFWFS
alignment scheme will be the baseline advLIGO alignment,   allowing a
reduction in the high frequency modulation requirements for the
tip/tilt suspensions.  However,  we should retain the   capability for
low frequency dither alignments,  and the co-located QPDs will
definitely be maintained.   Because of the alignment requirements of
the AFWFS diodes,  Pico motors will be included in the AFWFS path.

\subsection{Analog mod / demod}

The eLIGO OMC senses length offsets using a  10 to 12~kHz dither on
the OMC PZT and demodulation of the DCPD signals.  The high frequency
dither was difficult to produce digitally and introduced noise via the
digital-to-analog conversion process.   The advLIGO OMC will use
an analog sine-wave synthesizer and demodulator (a.k.a. lock-in amplifier)
for the OMC length error signal.   Note that this requires 
demodulators for the AFWFS as well.  

\subsection{16 kHz front end clock}

The eLIGO front ends run at 32 kHz (30 microseconds).  This high
speed is motivated by the high frequency digital demodulation which
will be deprecated for advLIGO.  Furthermore,  the 30 microsecond
timing is restrictive.  Consequently,  we will run the advLIGO front
ends at 16 kHz (60 microseconds). 

\subsection{Other electronics changes}

The electronics changes are minor.  The advLIGO OMC will incorporate
independent switching of the DCPD whitening and trans-impedance.
Pending a full noise analysis, the high frequency roll off of the
DCPDs will be extended from 16~kHz to a 100~kHz (or higher) to allow
high frequency modulations and observations of the test mass internal
modes.  Before advLIGO, the DCPDs will be tested at the required shot
noise limited sensitivity in a bench top experiment.

\subsection{Scattered Light Control}

The eLIGO OMC doesn't include very many beam dumps and baffles.  The
advLIGO OMC design will include AR coated black glass beam dumps
wherever appropriate,  particularly in front of the DC photodiodes.

\subsection{Construction Fixtures}

Both the H1 and L1 eLIGO OMC builds were more dramatic than
necessary.  In both cases,  the problems arose because of inadequate
fixtures during the epoxy process.   The advLIGO fixtures should
positively position the tombstones with 3-axis control (x, z, and
yaw).  Ideally all tombstones will be fixtured.  This will require
some design since the clearances are tight.  All four OMC mirrors
should be held simultaneously during bonding.  

\subsection{Varying G-factor}

The H1 eLIGO OMC changes the mode spacing as a function of the thermal
actuator temperature.  This is believed to be caused by a 7~mm radius
of curvature change on the 2~m radius of curvature mirror bonded to
the thermal actuator.  This ``bug'' may be considered a ``feature'' --
the temperature dependence was used to change the OMC g-factor
\emph{in situ} and minimize the noise.  If this feature is
incorporated into the advLIGO OMC , it should be explicitly modelled
and deliberately designed.  Otherwise the new thermal actuator design
should take pains to leave the cavity g-factor unaffected.

\subsection{DCPD path lengths}

The eLIGO OMC has unequal path length between the two DCPDs.  If
possible,  this situation will be avoided for advLIGO.


\section{Conclusion}

Although the list of changes in OMC design for advLIGO seems long,
the OMC has been a successful part of the eLIGO program. The changes
are,  on the whole,  modest and should be easy to implement.  

\end{document}