Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/nicmos/documents/isrs/tir_98_001.pdf Дата изменения: Tue Feb 12 14:43:37 2002 Дата индексирования: Sun Dec 23 01:09:38 2007 Кодировка: Поисковые слова: п п п п п п п

Technical Instrument Report NICMOS 98-001

NICMOS/NCS EMI Test Data Results
Louis E. Bergeron November 4, 1998

ABSTRACT In an attempt to better understand how the cryo-cooler induced noise (Schneider, 1998) would affect future NICMOS calibration, I repeated Glenn's analysis and looked at a few additional properties of the test data.

Summary
The observed induced frequencies, including the 60 Hz line noise and the 5-9 kHz induced signal (with the cooler running, at various turbine speeds) were recovered in the power spectra as expected. The maximum amplitude of the 5-9 kHz signal is ~7 DN PtoP, and has an RMS of ~5 DN. This signature is seen in each readout of a MULTIACCUM sequence, and the pattern and its amplitude remains in a fixed position on the array through all the reads of a given MULTIACCUM. By performing a 0th read subtraction, which is the standard procedure in the NICMOS calibration pipeline, the induced signal, including the 60 Hz line noise and 5-9 kHz noise, is COMPLETELY removed from all subsequent reads, in both the NCS-ON and NCS-OFF configurations. The pattern in the image does shift between MULTACCUM sets, but not within each MULTACCUM. Power spectra of 0th read corrected data confirm that all induced power is removed completely. It appears that this signal is either imprinted into the array during the array reset sequence, making it simply a part of the fixed BIAS of all the subsequent non-destructive array reads, or it is phase-locked to all the reads of the sequence. Given that the 60 Hz noise is seen to have this behavior even in the NCS-OFF data, the entire induced signal may just be an artifact of the test environment and may not reflect the true NICMOS/NCS signature at all. If this signal did not subtract out with a 0th read correction it would amount to a ~25% increase in the characteristic on-orbit per-readout noise, given a 35 e- readnoise and a gain of 5.4 e-/DN: sqrt(35^2 + (5*5.4)^2) / 35 ~ 1.25

1


Short Analysis Notes
To reproduce Glenn's power spectra, I repeated his calibration steps exactly. I built a median "dark" sequence using all the NCS-OFF MULTIACCUM sequences, taking the median-per-readout, without doing a 0th read subtraction. This is necessary to remove the shading and amplifier glow signatures from the data before measuring the power spectra. Subtraction of this dark leaves the quad's DC levels unbalanced, but quite flat. This is due in large part to the kTC noise which determines this DC level after each reset (one of the main reasons for doing a 0th read subtraction is to remove this). Then, the quads were balanced to each other by subtracting their medians. This is not strictly necessary, as each quad is read-out independently, and a DC does not change the measured power spectrum in any way. I then made power spectra for some of the NCS-OFF data, as well as NCS-ON data at various turbine speeds, and did the same after doing a 0th read subtraction. Power spectra were measured per-quad by taking the FFT of the sequentially clocked horizontal row, with the addition of 2 pixels at the end of each 128-pixel row as placeholders for the 21 microsecond vertical line address increment. The power spectra from the 4 quads were then averaged together to increase the S/N. Once it was found that the pattern was stable through all the readouts, I medianed all 26 readouts together and made the power spectrum from the median image of each MULTIACCUM. To produce images and plots of the induced signals, I applied narrow band-pass filters to the power spectra, removing the 1/f power contribution in the band, and then taking the inverse transform. The mean 1/f level was estimated by eye in each band.

Reference
Schneider, Glenn, NICMOS Project, "EMI Noise Properties of the NICMOS Cooling System as Seen by a NICMOS-3 Flight Spare Detector (or, Turning NICMOS into a Spectrum Analyzer)", May 1, 1998

Figures
(all figures and an additional mpeg of the noise are temporarily available on the web at: ftp://ftp.stsci.edu/outside-access/out.going/eddie/emi/ )

2


60 Hz Signal (stdev = 2.25 DN)
40 40

60 Hz Signal, zoomed

20

20

DN

0

DN 0 5.0·103 1.0·104 Sequential Pixel Number 1.5·104

0

-20

-20

-40

-40 3500 4000 4500 Sequential Pixel Number 5000

5-9 kHz Signal (stdev = 4.94 DN)
40 40

5-9 kHz Signal, zoomed

20

20

DN

0

DN 0 5.0·103 1.0·104 Sequential Pixel Number 1.5·104

0

-20

-20

-40

-40 3500 4000 4500 Sequential Pixel Number 5000

Sum of 60 Hz and 5-9 kHz Signal (stdev = 5.43 DN)
40 40

Sum of 60 Hz and 5-9 kHz Signal, zoomed

20

20

DN

0

DN 0 5.0·103 1.0·104 Sequential Pixel Number 1.5·104

0

-20

-20

-40

-40 3500 4000 4500 Sequential Pixel Number 5000

Actual EMI Test Data (stdev = 12.63 DN)
40 40

Actual EMI Test Data, zoomed

20

20

DN

0

DN 0 5.0·103 1.0·104 Sequential Pixel Number 1.5·104

0

bergeron Wed Nov 4 11:45:30 1998

-20

-20

-40

-40 3500

3

4000 4500 Sequential Pixel Number

5000


bergeron Fri Nov 6 14:01:47 1998

Positive Frequency Power 0.0 10 100 1000 Frequency (Hz) 10000
4

0.2

0.4

0.6

0.8

1.0

1.2

NCS-OFF


NCS-ON, Turbine Speed = 4500 Hz
1.2

1.0

0.8 Positive Frequency Power

0.6
5

0.4

0.2

bergeron Fri Nov 6 14:01:44 1998

0.0 10 100 1000 Frequency (Hz) 10000


Turbine Speed = 4500 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 4500 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 4800 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 4800 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 4800 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 5000 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 5000 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 5000 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 6000 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

Turbine Speed = 6000 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10
bergeron Fri Nov 6 11:03:45 1998

100

1000

10000

Positive Freqencey Power

Turbine Speed = 6000 Hz
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 100 1000 10000

6

Hz


Turbine Speed = 4500 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 4500 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 4800 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 4800 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 4800 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 5000 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 5000 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 5000 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 6000 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7000

8000

9000

Turbine Speed = 6000 Hz
0.700 0.525 0.350 0.175 0.000 4000
bergeron Fri Nov 6 11:01:17 1998

5000

6000

7000

8000

9000

Positive Freqencey Power

Turbine Speed = 6000 Hz
0.700 0.525 0.350 0.175 0.000 4000

5000

6000

7

7000 Hz

8000

9000


Mean of the power spectra of the 4 quads (NCS-ON, single readout)
1.0 Positive Freq Power 0.8 0.6 0.4 0.2 0.0 -0.2 10 100 Hz 1000 10000
8

Each quad's power spectrum minus the mean spectrum, same scale as above
1.0 Positive Freq Power
bergeron Fri Nov 6 10:11:12 1998

0.8 0.6 0.4 0.2 0.0 -0.2 10 100 Hz 1000 10000


Turbine Speed = 4500 Hz, single readout
1.2 1.0 0.8 0.6 0.4 Positive Frequency Power 0.2 0.0 10 100 1000 10000
9

Turbine Speed = 4500 Hz, first difference image (0th read subtracted)
1.2 1.0 0.8 0.6 0.4 0.2

bergeron Fri Nov 6 11:43:19 1998

0.0 10 100 Hz 1000 10000


10


11


Phase lag from Q1 to the other 3 quads
Q2 = 1.05000 Q3 = 17.3250 Q4 = 18.9000

Total power in difference image, 6-9 kHz band

30

Q4 Q3 Q2 Q1

20

The upper figure shows that the quads, although measuring the same induced signal, measure it phaseshifted slightly because the quads are not read out perfectly simultaneously. By stepping through a set of lag times between quads and then measuring the power in the difference image, one can iterate to determine what the phase lags are, and thus the relative quad to quad readout timings. The sensitivity is at the sub-microsecond level thanks to the high spatial frequency of the induced signal. Note that there is a 16 microsecond gap between when the readout of the lower half of the array is started and when the top is started. This does not affect the pixel to pixel timings, its simply the way the quadrant timings work apparently. There is a 1 microsecond gap between the left and right-hand quads of a given half.

10

0
bergeron Tue Nov 10 17:39:31 1998

0

5

10 15 Delay in microseconds

20

25

12