AMIGOS workshop, October 15 2005

Noise Characterization Using Transfer Functions: Examples from Spacecraft Doppler Tracking

John Armstrong for the Cassini Radio Science GW Group*

* R. Ambrosini, J.W. Armstrong, B. Bertotti, L. Iess, P. Tortora, H.D. Wahlquist

AMIGOS workshop, October 15 2005

Noise Characterization Using Transfer Functions

• The Doppler technique & its signal and noise transfer functions

• Examples of noise identification using transfer functions– Frequency & timing system– Antenna mechanical noises– Phase noise due to propagation through irregular media

• Model example for LISA: identifying proof mass noises

AMIGOS workshop, October 15 2005

DSS25 and Cassini

AMIGOS workshop, October 15 2005

Cassini Radio System

AMIGOS workshop, October 15 2005

Three-Pulse GW Response

AMIGOS workshop, October 15 2005

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

AMIGOS workshop, October 15 2005

Frequency/Timing Glitch

AMIGOS workshop, October 15 2005

Antenna Mechanical Event

AMIGOS workshop, October 15 2005

Plasma Events

AMIGOS workshop, October 15 2005

AMIGOS workshop, October 15 2005

Isolating Proof-Mass Noises with Transfer Functions

• LISA unequal-arm geometry• Nomenclature for proof

mass disturbances vi and vi*

• Idea: isolate which proof mass glitched (Braginsky, ≈2001) -- or which proof mass is noisier -- through their transfer functions

AMIGOS workshop, October 15 2005

Isolating Proof-Mass Noises with Transfer Functions

• Temporal transfer fucntion of 6 PM’s to the three unequal-arm Michelson combinations (X, Y, Z) + symmetrical Sagnac ()

• First generation TDI

AMIGOS workshop, October 15 2005

GW Transfer Function to TDI “X”

• Source/detector geometry dependent

• “8-pulse” response in general for first-gen TDI [equal arm lengths assumed in this example so it degenerates to 7-pulse response; ApJ 527, 814 (1999)]

• Will not be confused with PM glitches

AMIGOS workshop, October 15 2005

Isolating Proof-Mass Noises with Transfer Functions (Statistically)

• Noise-only spectra of first-generation TDI combinations X and Z (expressed here as spectra of fractional Doppler fluctuations, vs. phase)

• Proof mass 1 assumed 10X noisier than nominal

• PM1 does not enter in Z, so it is OK -- spectrum of X affected, however

CQG, 20, S283 (2003)

AMIGOS workshop, October 15 2005

Concluding Comments

• In spacecraft Doppler tracking, at least, the data themselves are much more sensitive than any of the engineering housekeeping channels which monitor experiment health -- so you have to use the data themselves, in general, to understand the noises

• I would expect this to be true for LISA, too

• Transfer functions of the signal and noises to the TDI combinations offer a way to use the “science” data to understand the noises and instrument performance