Spectrum Analyzer CW Power Measurements and Noise 2012 NCSL International Workshop and Symposium Spectrum Analyzer CW Power Measurements and Noise 1.

Page 1

Spectrum Analyzer CW Power Measurements

and Noise

2012 NCSL International Workshop and Symposium


Spectrum Analyzer CW Power Measurements and Noise

Overview

• Signal + Noise Model

• Bias & Variance of Noise and Signal + Noise Measurements

• Variance of Averaged Measurements

• Spectrum Analyzer Block Diagram and Measurement Recommendations



2

Signal + Noise Model

Signal

Noise = Ni + jNq

Signal+Noise

Ni distribution

Nq distribution



3

Spectrum Analyzer Averaging Methods

• Averaging Types:

• Voltage

• Power

• Log Power



5

Voltage Power Log Pwr(mV) (mW) (dBm)

Meas 1 16 0.01 -22.81Meas 2 79 0.12 -9.09Meas 3 84 0.14 -8.48Meas 4 762 11.62 10.65Meas 5 178 0.64 -1.97Meas 6 817 13.36 11.26Meas 7 964 18.60 12.70Meas 8 648 8.39 9.24Meas 9 923 17.02 12.31Meas 10 541 5.85 7.67Average 501 7.57 2.15Voltage 501 615 286Power 5.02 7.57 1.64Log Pwr 7.01 8.79 2.15

Noise – Measurement Bias

Averaging Bias (dB) Notes

Power 0 by definition

Voltage -1.05 -

Log Power -2.51 -



6

= Euler’s constant: 0.577216…

Noise – Standard Deviation of Discrete Measurements

Averaging Std Dev (dB) Notes

Power 4.34

Voltage 4.54

Log Power 5.77 Monte Carlo



7

Signal + Noise – Measurement Bias



8

Averaging Bias (dB)

Power

Voltage

Log Power

m = signal-to-noise ratio (W/W) = Euler’s constant: 0.577216…

Signal + Noise – Measurement Bias



9

Signal + Noise – Standard Deviation of Discrete Measurements



10

Averaging Std Dev (dB)

Power

Voltage (model based on Monte Carlo)

Log Power (model based on Monte Carlo)

m = signal-to-noise ratio (W/W)

Signal + Noise – Standard Deviation of Discrete Measurements



11

Standard Deviation of Averaged Measurements

• Averaging Discrete Measurements – reduced by

• N = number of samples.

• Time Averaged Measurements – reduced by

• tint = integration (measurement) time

• NBW = noise bandwidth



12

Filter Type Application NBW/RBW

4-pole sync Most SAs analog 1.128

5-pole sync Some SAs analog 1.111

FFT/digital FFT/digital IF swept SAs 1.056

Spectrum Analyzer Block Diagram

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator

Processing and Display

IF (Resolution Bandwidth) Filter

Video Filter

Envelope Detector



13

Spectrum Analyzer Block Diagram – Input Attenuator and Preamplifier

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



14

Attenuation = 0 dBPreamplifier On

but… don’t overdrive the mixer!

Spectrum Analyzer Block Diagram – Preselector

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



15

set based on impact to noise floor

Spectrum Analyzer Block Diagram – Frequency Span & Sweep Time

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



16

span = zerosweep time = based on noise uncertainty

Spectrum Analyzer Block Diagram – RBW

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



17

Signal + Noise: set RBW to give at least 8 dB SNR. Reducing RBW beyond this does not give any benefit (assumes log power averaging)

Noise: set RBW wide

(power averaging)

Spectrum Analyzer Block Diagram – RBW (Signal + Noise)



18

SNR > 8 dB produces negligible bias when using log power averaging

Spectrum Analyzer Block Diagram – RBW (Signal + Noise)



19

SNR approximated by

Spectrum Analyzer Block Diagram – RBW (S+N)• Standard Uncertainty of discrete measurements

approximated by:

•

• m = SNR (W/W) =

• For time-averaged measurements:

• But

• Therefore: (time-averaged, SNR > 8 dB)



20

For time-averaged measurements, changing RBW does not change uncertainty due to noise

Spectrum Analyzer Block Diagram – Envelope Detector

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



21

detector = average

Spectrum Analyzer Block Diagram – Video Filter

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



22

video filter = “wide open”

Spectrum Analyzer Block Diagram – Averaging Type

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



23

averaging typenoise: powersignal + noise: log power

Spectrum Analyzer Block Diagram – Trace Averaging & Trace Points

Inputsignal

InputAttenuator

Preamplifier

Preselector

Mixer

Local Oscillator

Sweep Generator



Video Filter

Envelope Detector



24

trace averages = 1trace points = anyfunction: average entire trace

Spectrum Analyzer Block Diagram – Summary of Recommendations

Function Noise Signal + Noise

Input Attenuation 0

Preamplifier On

Preselector based on impact to noise floor

Frequency Span Zero

Resolution Bandwidth widenarrow enough for SNR > 8 dB

Detector average

Video Bandwidth wide

Averaging Type power log power

Trace Averages 1

Trace Points n/a

Function average entire trace



25

Spectrum Analyzer CW Power Measurements and Noise 2012 NCSL International Workshop and Symposium Spectrum Analyzer CW Power Measurements and Noise 1.

Documents

noise voltage

noise floor slide

ncsl international workshop

voltage power log power

noise ratio ww slide

rayleigh distribution

log power5

measurement recommendations