e>Help for Webmasters « back to results for ""
Below is a cache of http://www.ligo.caltech.edu/docs/T/T040028-00.pdf. It's a snapshot of the page taken as our search engine crawled the Web.
The web site itself may have changed. You can check the current page or check for previous versions at the Internet Archive. Yahoo! is not affiliated with the authors of this page or responsible for its content.
Parameter Variations in Quadruple Pendulum Design for ETM/ITM Investigation of Wire Lengths in Advanced LIGO Quadruple Pendulum
Design for ETM/ITM

Norna A Robertson, Mark Barton, Calum Torrie
26 Jan 2004

DCC: T040028-00-R
Draft
1. Introduction.

An updated conceptual design for the quad pendulum suspension for the test masses in
Advanced LIGO has been presented in the document Advanced LIGO Suspension
System Conceptual Design, T010103-03-D, 21 October 2003. Since that design was first
put together, we have been considering alternative parameter sets for various reasons. In
this document I address the overall length of the pendulum and in particular the
individual lengths of the first and third stages (counting from the top).

The general design which has been developed is such that either a sapphire or a silica test
mass of the same mass could be suspended see NARs talk at the LSC meeting in
Hannover, Aug 2003: "Quadruple Pendulum Design Update" G030437-00-Z (not yet
posted on the LSC website). However for the purposes of this document, I have assumed
that sapphire is the material used for the test mass, and that the penultimate mass has the
same mass and density as sapphire.


2. Changes to wire lengths in quad pendulum.

2.1 Increasing space for catcher

The lengths of the wires quoted in the conceptual design, from top to bottom, were 0.54,
0.304, 0.302, 0.6 m. Note that the wires in the first three stages are not vertical, and that
the attachments are not exactly at the centres of mass so the overall length from top
blade to centre of optic is not simply the sum of these values. The full calculation is done
in the MATLAB program. The first change which was made to these lengths was to
increase the length between the 2
nd
(upper intermediate) and 3
rd
(penultimate) masses by
~4 cm to allow more room for fitting in the catcher structure which will be used for the
assembly and installation of the penultimate and test mass. i.e. wire lengths become 0.54,
0.304, 0.342, 0.6 m. With this revised length, and holding all other parameters as given in
the conceptual design, the overall vertical length of the pendulum from the tip of the top
blades to the base of the optic becomes 1.894 m, and allowing for 0.211 m divided above
and below the pendulum for the overall supporting structure which attaches to the optical
table, leads to an overall vertical length including structure of 2.105 m. These numbers
are captured in the document ETM Controls Prototype: - Mass estimate of an ETM
suspension layout, T030137-04, written to allow mass budget and centre of mass to be
estimated for use by the SEI subsystem. Note: in that document, due to rounding errors,

1 the overall length is quoted as 1.893 m and the lengths allowed above and below total
0.212 mm to give the overall vertical length of 2.105 m.

2.2 Shortening overall length for ease of installation.

The overall vertical length including support structure is in general constrained by several
factors: the height of the crane above the tanks, the size of the SEI structure, the method
of installing both SEI and SUS into the tank, the position of the SEI structure within the
tank and the height of the optical beam. The prime consideration for the quad is to come
up with an acceptable length which meets performance requirements and can be installed
in LIGO. A secondary consideration is installation at LASTI. We are currently planning
to install the SEI and SUS in LIGO as a single unit, with the intention that the quad will
be built so that it can be split into an upper and lower section, the lower section
comprising the catcher structure containing the penultimate and test masses. With such a
split structure, we would have the facility of removing and replacing the lower part
without removal of the SEI and upper part of the SUS, if and when required.

The SUS team has been asked to consider whether we could reduce the overall length to
2.005 m to allow LASTI installation without major structural changes at the facility.
There are two issues with installing the quad pendulum structure at LASTI due to
restricted headroom. See document D030715-00-D (L Jones). The first issue arises due to
the required addition of a spreader beam to allow connection of two cranes to a single lift
pin on the top of the SEI. With the additional height this implies we will require the quad
to be installed in two pieces whatever its length. The second problem arises with putting
the dome in place once the SEI and SUS are installed. The dome will not clear the top of
the SEI structure if a 2.105 m SUS is in place and the optics is hanging at the correct
height for the optical beam, unless it is split into two pieces. However the dome can be
installed if the SUS structure is 10 cm shorter. (Note: there is potentially another fallback
position the optical beam height could be changed.). If the quad could be 10 cm shorter
this would be the simplest solution for LASTI installation.

We believe that the ~0.2 m total added to the basic quad pendulum length for the
supporting structure is close to the minimum we can use to allow for clamps, catcher, etc.
Thus we have considered taking 10 cm off the quad itself. Noting the wire lengths as
given above, and recalling that the lowest length (0.6 m), corresponding to the silica
ribbons/fibres, has been carefully chosen from thermal noise considerations, the obvious
place to consider removal of 10 cm is in the top length. Thus we have run the MATLAB
model with wire lengths 0.445, 0.304, 0.342, 0.6 m, other parameters remaining
unchanged, to give an overall vertical length from tip of top blades to base of optic of
1.793 m. (slightly less than 10 cm being taken off the top wire length, which is at a small
angle to the vertical, to give 10 cm off the vertical length).


2 2.3 Results of changes to wire lengths.

We will look at the effect of doing both the changes outlined in 2.1 and 2.2 taken
together. We consider mode frequencies, transfer functions for seismic and sensor noise
and impulse decay times. Since the thermal noise behaviour is dominated by the final
stage (unchanged) at frequencies around and above 10 Hz, we do not reconsider this here.

The MATLAB model used for these investigations is available as QUAD_2003NOV14.
The full parameter set is given in Appendix A.

The frequencies of the modes for the two sets of wire lengths are given in Table 1.

longpitch1: [0.372 0.433 0.867 0.960]
longpitch2: [1.98 2.00 3.41 4.16]
yaw: [0.646 1.37 2.39 3.14]
transroll1: [0.441 0.820 0.976 2.03]
transroll2: [2.73 3.56 3.93 12.5]
vertical: [0.687 2.73 4.40 8.76]

longpitch1: [0.370 0.440 0.867 0.987]
longpitch2: [1.92 1.99 3.42 4.10]
yaw: [0.666 1.40 2.42 3.07]
transroll1: [0.447 0.841 1.00 2.02]
transroll2: [2.78 3.56 3.96 12.5]
vertical: [0.686 2.73 4.40 8.76]

Table 1. Mode frequencies in Hz for two sets of wire lengths: upper set 0.54, 0.304,
0.302, 0.6 m, lower set 0.445, 0.304, 0.342, 0.6 m, as given by MATLAB model.

One can see that the changes are very small as expected. The lowest longitudinal mode at
around 0.44 Hz is slightly higher due to the slightly shorter overall length, but the change
is less than 2% and similarly other changes are of this magnitude. These frequencies were
also checked using M Bartons Mathematica program which more accurately models the
effect of non-vertical wires attached to the tips of the blades. The frequencies in general
agree well to those given in Table 1, and again show changes between the different wire
lengths only at the few or less percent level. We should point out here one frequency
the second highest transroll mode, is not well modeled with the MATLAB code at
present. This mode corresponds to the movement of the two top masses in antiphase in
roll and for this mode the non-vertical wire/blade interaction does play a significant role.
The Mathematica program shows that this mode is in fact around 5.5 Hz for both sets of
wire lengths, compared to the value of ~ 4 Hz which the MATLAB model predicts. A full
listing of the modes predicted using the Mathematica code is given in Appendix B.

Checking on vertical and longitudinal transfer functions, we find that the vertic