Module and DiodeNoise Power Spectrum

and Correlations

Cuts

Data from W-band CES #3 Run ID# 3010

B-glitch correction using adc_anomaly_info.txt-w91-multi-v3.txt(default diode rms = 0.2 counts)

Not looking at TT modules (modules 85-90)

Removed bad modules:

7, 8, 28, 42, 81

Removed bad diodes:

module 4 U2module 38 Q1Module 40 Q1

FFT of Double Demodulated Diode Signals

mV2/Hz versus freq(Hz)

Averaged over all Modules and Diodes

Scan Synchronous Signal

Other glitches

TT-modules

FFT of Double Demodulated Diode Signals

mV2/Hz versus freq(Hz)

Some Modules show Stronger Glitch Signals(could be ground pickup)

Scan Synchronous Signal

Other glitches

Q1 Q2

U1 U2

Typical Fit Result to function P1/f + P2 Typically cannot fit the first bin

Definition of White Noise Correlation (from I. Buder)

F are the frequency components from the FFT

F-bar denote complex conjugate of F

i,j are diode indices (Q1, Q2, U1, U2)

< > refer to average over all frequencies > 10 Hz

DDmod Ideal Combination Correlation Coeff

Q1 – Q2 0

U1 – U2 0

Q – U 0.5

From: Chicago Memo on Noise Correlations

Assuming NO polarization at input

Q1-Q2, U1-U2, and Q-U correlation dependence on power difference between Leg A and Leg B

Q1 - Q2U1 - U2

Q – U

Correlation Dependence on Power Imbalance, and Hybrid Imperfection

Typical Diode-Diode Noise Correlation versus FrequencyCij versus freq (Hz)

Bad Diode-Diode Noise Correlation versus Frequency from module 9Cij versus freq (Hz)

Noise Correlation Between Diodes (within same module)

* I haven’t checked, but this result seems to agree with study by I. Buder

What next ?

Implement B-glitching correction (done)

Analyze 91 modules from one CES (done)

Write fitter to fit for 1/f knee and white noise level (done)

Write fitter to fit for correlation coefficients (done)

Analyze more CES

Examine Glitch patterns in FFT (may be due SSS)