PowerPoint-Pr

1
Title
1 kHz Glitches in S4
Max-Planck-Institut für Gravitationsphysik(Albert
-Einstein-Institut)
2
GEOs sensitivity for last 20 days of S4
Peak sensitivity drifted during S4 (much more
than at other frequencies)
3
BLRMS of H(t) vs glitch rate
Glitchrate 700-2000 Hz
H(t) BLRMS 990-1000 Hz
Found relation between glitchrate and peak
sensitivity.
4
Spectra for good and bad time
Glitches cause additional noise between 400 Hz
and 1.3 kHz

• Example times
• high glitch rate
• 2005-03-22 061000
• low glitch rate
• 2005-03-22 130000

5
Spectrograms of differential EP-P
High glitchrate
Low glitchrate
Glitches present in MI_diff_EP ? not an artifact
from calibration process
6
Timeseries of 1kHz Glitches
Filtered with bandpass, 950 to 1050 Hz, 6th Order
7
Noise projections
1 kHz glitches cant be explained by our current
set of noiseprojections
8
Gliches vs BLRMS _at_ 1 kHz
9
Timedomain distribution
10
SNR distribution
11
Number of pixels
12
Duration of bursts
13
Bandwidth of burst
Probably maximal bandwidth is limited to 400 Hz
(strong lines 822 and 1266 Hz)
14
Time-delay between glitches
15
Periodicity of the glitches
Periodicity of 1.45 mHz
Sometimes one can see a periodicity (5-10 times
per hour) of the glitchrate
16
1.45 mHz in autoalignment signals
1.45 mHz
21.45 mHz
17
1.45 mHz in MC_AA rotation
21.45 mHz
1.45 mHz
18
Temperatures in the central building
1.45 mHz
21.45 mHz
2.3 mHz
19
2.3 mHz from air conditioning system
AC system usually dominates vertical seismic rms
at the central building
2.3 mHz is produced by the outdoor unit of
central buildings airconditioning system
20
Temperatures in the central building
1.45 mHz
21.45 mHz
2.3 mHz from AC
1.45 and 2.9 mHz appear strongly in temperature
of the laser bench
21
Periodicity of the glitches
Periodicity of 2.5 mHz
Sometimes one can see a periodicity (5-10 times
per hour) of the glitchrate
22
2.5 mHz in MPR alignment
22.5 mHz
2.5 mHz
32.5 mHz
1.45 mHz
2.9 mHz
23
2.5 mHz in MC2-AA
22.5 mHz
2.5 mHz
32.5 mHz
24
2.5 mHz in central buildings temperatures
2.5 mHz
2.3 mHz
25
How to go on ??
Analysis of GEO S4 data

• Is there any possibility to get vetoes for the
  1 kHz Glitches ?

• Do we have to make vetoes for the 1 kHz Glitches
  ?

Instrumental viewpoint
Some further effort is needed to investigate the
Glitches and find their origin.
Future issues

• How can we avoid such a problem in future?
• More time in front of a data run (???)

26
How to attack the problem
Thinking Which hardware of GEO600 could cause
glitches with these properties ? Coupling ? What
experiments do we need to check a potential
hypothesis?

• Data analysis
• Try to find a correlation between glitchrate
  during S4 and ???
• Compare detector status for good (low glitch
  rate) and bad time (high glitch rate). Anything
  different ?
• Maybe the periodicity leads us to origin of the
  glitches ?

27
HACR Burst and glitch analysis
Take in 32 seconds of data
Subdevide this into subsegments of 32 ms (Overlab
of 28 ms)
Windowing and FFT
Index i subsegment Index k frequency bin
Spectrogram (2-dim set of numbers / pixels)
mean
Clustering
standard deviation
high threshold 25
low threshold 5
Triggers