1. Measurements of digital signals with spectrum analyzers
Thomas Hasenpusch
Federal Network Agency
Germany

2. Types of available Spectrum Analyzers
Sweeping Analyzer
Scans the desired frequency range with a narrow filter
FFT Analyzer
Captures the time signal and calculates spectrum mathematically
Thomas Hasenpusch, Bundesnetzagentur

3. Sweeping Analyzer: Principle in Theory
Filter is swept through the desired frequency range (Span) f A
Thomas Hasenpusch, Bundesnetzagentur

4. Sweeping Analyzer: Realization of Principle
IF signal is swept through fixed frequency filter (RBW) f A IF
Thomas Hasenpusch, Bundesnetzagentur

5. Sweeping Analyzer: Simplified Block Diagram
Detector
Video Bandwidth (VBW)
Reference level
Input
Mixer
IF Amp.
IF Filter
Log. Amp.
Envelope Detector
Video Filter
Detector
Display
Local Oscillator
Sawtooth Generator
Resolution Bandwidth (RBW)
Centre frequency
Detector
Trace Mode
Span,
Sweep time
Thomas Hasenpusch, Bundesnetzagentur

6. Sweeping Analyzer: RBW
3dB „dip“
RBW = frequency spacing is not always sufficient to separate two signals
Optimum (best frequency resolution): RBW = span / display pixels
Thomas Hasenpusch, Bundesnetzagentur

7. Sweeping Analyzer: Envelope Detector
„Filters“ out the RF, leaves only modulation component A
Video signal
Thomas Hasenpusch, Bundesnetzagentur

8. Sweeping Analyzer: Detectors
Analyzer measures much faster than it can display
Multiple measurement results lie behind each display pixel
Detector determines which of the measured values is displayed s1 s2 s3 s4 s5 s6 s1 s2 s3 s4 s5 s6 A t
Average level
RMS level
Pixel 1
Pixel 2 A t
peak AV
RMS
sample
Thomas Hasenpusch, Bundesnetzagentur

9. FFT: Theory
Fourier says: Each random time signal is the sum of discrete, unmodulated sinewaves
sum
FFT
Thomas Hasenpusch, Bundesnetzagentur

10. FFT Analyzer: Principle
Fourier formulas allow calculation of the spectrum of each time signal
Fast Fourier Transform (FFT) greatly reduces calculations, but work only under certain assumptions
Time signal is captured (acquired) for a certain time, digitized and stored in memory
FFT spectrum is then calculated from the stored time samples by a Digital Signal Processor (DSP) X
A/D
DSP
RF in
FIF
Low Pass
Display
FFT
Memory
fRF
Thomas Hasenpusch, Bundesnetzagentur

11. FFT: Problems and issues
Usually no seamless acquisition (blind times during calculation)
Spectrum 1: Display
Spectrum 2: Display
Block 1 acquisition
Block 1 processing
Block 2 acquisition
Block 2 processing
time
blind time
blind time
Solution: Deploying two separate processing lanes with alternate timing (one lane acquires while the other one processes previous block)
Thomas Hasenpusch, Bundesnetzagentur

12. FFT with pulsed signals
FFT analyzers are usually fast enough to show the spectrum of even very short pulses (e.g. Radar) A t
FFT window
FFT window
FFT window A f A f A f
Thomas Hasenpusch, Bundesnetzagentur

13. Important levels of digital signals
Peak: maximum possible level over a long meas. time
Applies when assessing interference potential
RMS (continuous signals): average power a over long meas. time
Applies when checking reception capability, coverage and licence conditions
AV burst (pulsed signals): average power during burst only
Applies as RMS, but in case of bursted signals
AV burst level A t
burst duration
Thomas Hasenpusch, Bundesnetzagentur

14. Bandwidth Measurement (Direct Method)
Most important for monitoring stations: 99% bandwidth (equal to occupied bandwidth)
Definition: bandwidth in which 99% of all transmitted energy lies f A
100%
99%
Span (100%)
0.5%
0.5%
OBW
With analyzer: narrow RBW, MaxHold, OBW function
Thomas Hasenpusch, Bundesnetzagentur

15. Level Measurement: Procedure With Sweeping Analyzer (1)
Peak level:
Span ≥ signal bandwidth or zero span
RBW ≥ signal bandwidth
Detector: peak
MaxHold
Read highest level with Marker
RMS-level:
Span ≥ signal bandwidth
Narrow RBW (span/display points)
Detector = RMS or sample
ClearWrite
Channel Power measurement function
If reading is instable: increase sweep time (never use MaxHold!)
Thomas Hasenpusch, Bundesnetzagentur

16. Level Measurement: Procedure With Sweeping Analyzer (2)
AV-burst level:
Span = zero span
RBW ≥ signal bandwidth
Detector = RMS or sample
ClearWrite, trigger on burst
Sweep time ≥ burst time
Time domain power measurement A t 1 d B C L R W 3 D k H z V e f - 2 m n r 4 . 5 M s / S T G * 9 8 7 6 a [ ] P O E
Average level:
Span = zero span
RBW ≥ signal bandwidth
Detector = Average or sample
Trace = linear average
Thomas Hasenpusch, Bundesnetzagentur

17. Level Measurement: Procedure with FFT Analyzer (1)
Peak level:
Capture bandwidth = signal bandwidth
Time domain analysis
Select shortest possible acquisition time
MaxHold over multiple acquisitions or amplitude vs. time together with long analysis time
Read highest value
RMS level:
Capture bandwidth ≥ signal bandwidth
Channel power function
Long acquisition time or average over multiple short acquisition times
If reading is instable: increase acquisition time or number of averages
Thomas Hasenpusch, Bundesnetzagentur

18. Level Measurement: Procedure with FFT Analyzer (2)
AV-burst level:
Capture bandwidth ≥ signal bandwidth
Trigger analysis on burst start
Channel power function
Acquisition time (or analysis time) = burst time
acquisition time
analysis time
Thomas Hasenpusch, Bundesnetzagentur

19. Level Measurement Under Low S/N Ratios
For accurate Pk measurement, S/N ≥ 20 dB is necessary
For accurate RMS, AV, AV-burst measurement, 10 dB S/N is sufficient
Corrections to indicated level for measurements under lower S/N values:
Thomas Hasenpusch, Bundesnetzagentur

20. Literature
ITU Spectrum Monitoring Handbook (2011): Chapter 4.3: RF level measurements
Thomas Hasenpusch, Bundesnetzagentur