Spectrum Analyzer Fundamentals

Chris Gillis
Application Engineer
Signal/Spectrum Analyzers & Signal Generators [email protected] +1.438.863.5760

October 29, 2013 University of British Columbia

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Frequency vs Time Domain

l

Fourier Transform links time and frequency domain

For periodic signals, this is a Fourier Series

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Frequency vs Time Domain

Periodic

vs

Non-periodic

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Frequency vs Time Domain

Looking at the time or frequency domain can reveal different information about the signal

Oscilloscope: look at amplitude vs time Spectrum Analyzer: look at power vs frequency

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Frequency vs Time Domain

l

For example, harmonics could easily be missed

Time Domain

Frequency Domain

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

FFT Analyzer
l

As time and frequency are linked by the Fourier transform, we could just capture time data and compute the Fourier transform Instead of capturing infinite time, we can compute the Discrete Fourier Transform, which transforms discrete time data into discrete spectrum data

Use Fast Fourier Transform algorithms

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

FFT Analyzer
l

According to Nyquist, you need a sampling frequency at least twice the highest frequency component to properly recreate a signal

>10 samples

Alias

Encounter problems with bandwidth, range

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

10

Filter Bank Spectrum Analyzer

l

Problem: not very practical

f1 f2 f3 f4

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

11

Tunable Filter Spectrum Analyzer

Problem: bandpass filter changes bandwidth depending on center frequency

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 12

Simplified Swept Tuned Block Diagram

Mixer

IF Amplifier

Resolution BW Filter BPF

Log Amp

Envelope Detector

Video BW Filter LPF

Input Atten

Local Oscillator

y x
Display

Sawtooth
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 13

Input Mixer
Mixer

IF Amplifier

Resolution BW Filter BPF

Log Amp

Envelope Detector

Video BW Filter LPF

Input Atten

Local Oscillator

y x
Display

Sawtooth
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 14

Types of Mixing
RF
l l

IF

Fixed RF, Swept LO and IF Fixed LO, Swept RF and IF Upconversion

l

l
l l

LO Fixed IF, Swept LO and RF (used in spectrum analyzers)

IF frequency is higher than RF and LO frequency

Downconversion
l

IF frequency is lower that RF and LO frequency

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

15

Mixer Example
|mfLO nfRF| = fIF
RF 1 GHz IF

Possible frequencies on IF portto name a few: LO-RF=100MHz LO+RF= 2.1GHz LO=1.1 GHz RF=1 GHz 2LO-RF=1.2 GHz 2RF-LO= 900 MHz

LO
1.1 GHz

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

16

Resolution Bandwidth
Mixer

IF Amplifier

Resolution BW Filter BPF

Log Amp

Envelope Detector

Video BW Filter LPF

Input Atten

Local Oscillator

y x
Display

Sawtooth
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 17

Resolution Bandwidth
l l

Sets IF Bandwidth of Spectrum Analyzer Filter types:

Standard sweep filters: digital Gaussian filters Channel filters

EMI filters (available with Quasipeak detector)

FFT filters RRC

Determines frequency resolution and noise floor

Sweep Time is function of Resolution Bandwidth and Span

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

18

IF Filter Types
* RBW

20 kHz
50 kHz

* RBW

20 kHz
Ref
-20

* RBW

20 kHz
50 kHz

VBW

VBW

Ref
-20

-20 dBm

Att

5 dB

SWT 2.5 ms

Ref
-20
A

-20 dBm

Att

5 dB

AQT 2.5 ms

-20 dBm

Att

5 dB

SWT 50 ms

-30 1 AP CLRWR
-40

-30 1 PK CLRWR
-40

A
1 AP CLRWR

-30

-40

-50

-50

-50

-60

-60

-60

-70

-70

-70

-80

-80

-80

-90

-90

-90

-100

-100

-100

-110

-110
-120

-110

-120

-120

Center

1 GHz

10 kHz/

Span

100 kHz

Center

1 GHz

10 kHz/

Span

100 kHz

Center

1 GHz

10 kHz/

Span

100 kHz

Date: 7.NOV.2006

12:17:44

Date: 7.NOV.2006

12:17:11

Date: 7.NOV.2006

12:16:44

Normal (Gaussian)
* RBW

FFT
18 kHz
50 kHz
Ref -20 dBm Att 5 dB
* RBW

Channel
20 kHz
50 kHz VBW

VBW

Ref
-20

-20 dBm

Att

5 dB

SWT 65 ms

SWT 2.5 ms

-20

-30 1 AP CLRWR
-40

A
-30 1 AP CLRWR
-40

-50

-50

-60

-60

-70

-70

-80

-80

-90

-90

-100

-100

-110

-110

-120

-120

Center

1 GHz

10 kHz/

Span

100 kHz

Center

1 GHz

10 kHz/

Span

100 kHz

Date: 7.NOV.2006

12:15:43

Date: 7.NOV.2006

12:16:17

RRC
Default Setting for standard spectrum analyzing tasks

5 Pole

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

19

Resolution Bandwidth
200 Hz

2 kHz

Signals separated by 1kHz cant be resolved by 2kHz RBW

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

20

Resolution Bandwidth and DANL*

RBW

1 MHz
300 kHz 100 kHz

*DANL: Displayed Average Noise Level

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 21

Envelope Detector
Mixer

IF Amplifier

Resolution BW Filter BPF

Log Amp

Envelope Detector

Video BW Filter LPF

Input Atten

Local Oscillator

y x
Display

Sawtooth
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 22

Envelope Detector
RMS detector (power average)
Samples / pixel is determined by sweep time and sample rate
pixel n (8 samples) s1 s2 s3 s4 s5 s6 s6 s8 pixel n+1 (8 samples) s1 s2 s3 s4 s5 s6 s6 s8

Vrms

1 N

s
i 1

2 i

RMS detector reports the true noise power. (The RMS value)

Ave detector (voltage average)

A/D samples (linear range)

1 Vave N

s
i 1

Averages the noise voltage, then converts to power. This is lower by 1.05 dB.
(squaring the ave is not equal to averaging the square)

displayed pixels
posit ive peak sample rms ave negative peak

Sample detector

Takes the first sample Randomly located between peaks

Sample detector & trace averaging

Noise averaging is done on a log scale, introducing a new error of 2.51 dB Total error is now 2.51 dB
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 23

Spectrum Analyzers How to measure noise

l

Measure Noise with different detectors

l

RMS detector measures true noise power

RBW 200 kHz
VBW 500 kHz

Ref

-90 dBm

Att

5 dB

SWT 2.5 ms

RMS detector

-90

*
A

-91 1 RM * VIEW
-92

2 AV * VIEW
-93

3 SA

Average detector

VIEW

-94

Delta: 1.05 dB
3DB

-95

Delta: 2.51 dB

-96

Sample detector & trace ave (Log)

-97

-98

-99

-100

Center

1 GHz

1 MHz/

Span

10 MHz

R&S at UBC: Spectrum Date: 13.MAR.2009 15:06:16

Analyzer Fundamentals, October 2013

24

Spectrum Analyzers How to measure noise

l

Measure Noise with different detectors

l l

RMS detector measures true noise power We can apply linear or power trace averaging to an RMS detector.
RBW 200 kHz
VBW 2 MHz

Ref

-90 dBm

Att

5 dB

SWT 2.5 ms

RMS detector

-90

*
A

-91 1 RM * VIEW
-92

2 RM * VIEW
-93

RMS detector & trace ave (Lin) or (Pwr)

-94

-95 3DB
-96

-97

-98

-99

-100

Center

1 GHz

1 MHz/

Span

10 MHz

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Date: 13.MAR.2009 17:29:46

25

Spectrum Analyzers How to measure noise

l

Measure Noise with different detectors

l l

RMS detector measures true noise power Sample detector & linear or power trace averaging yields the same results
RBW 200 kHz
VBW 500 kHz

Ref

-90 dBm

Att

5 dB

SWT 2.5 ms

RMS detector

-90

*
A

-91 1 RM * VIEW
-92

2 SA AVG
-93

Sample detector & trace ave (Lin) or (Pwr)

-94

-95

SWP
-96

1000 of

1000

3DB

-97

-98

-99

-100

Center

1 GHz

1 MHz/

Span

10 MHz

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Date: 13.MAR.2009 15:13:07

26

Spectrum Analyzers How to measure noise

l

Measure Noise with different detectors

l l

RMS detector measures true noise power Ave detector plus any trace averaging does not yield the same result
Do not use trace averaging with the average detector
RBW 200 kHz
VBW 2 MHz

RMS detector
1 RM * VIEW 2 AV * VIEW

Ref
-90

-90 dBm

Att

5 dB

* SWT

2 s

*
A

-91

-92

-93

Average detector Average detector & Log trace average

3 AV * VIEW

-94

Delta: 1.05 dB
SWP 2 of 1000
3DB

-95

-96

-97

-98

-99

-100

Center

1 GHz

1 MHz/

Span

10 MHz

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Date: 13.MAR.2009 17:57:02

27

Spectrum Analyzers How to measure noise

l

Measure Noise with different detectors

l l

RMS detector measures true noise power Ave detector plus any trace averaging does not yield the same result
Do not use trace averaging with the average detector
RBW 200 kHz
VBW 2 MHz

RMS detector
1 RM * VIEW 2 AV * VIEW

Ref
-90

-90 dBm

Att

5 dB

SWT 2.5 ms

*
A

-91

-92

-93

Average detector Average detector & power trace average

3 AV * VIEW

-94

Delta: 1.05 dB
3DB

-95

-96

-97

-98

-99

-100

Center

1 GHz

1 MHz/

Span

10 MHz

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

Date: 13.MAR.2009 18:01:57

28

Detector and Trace Usage

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

29

Video Filter
Mixer

IF Amplifier

Resolution BW Filter BPF

Log Amp

Envelope Detector

Video BW Filter LPF

Input Atten

Local Oscillator

y x
Display

Sawtooth
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 30

Video Filter

500kHz

500Hz

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

31

Local Oscillator
Mixer

IF Amplifier

Resolution BW Filter BPF

Log Amp

Envelope Detector

Video BW Filter LPF

Input Atten

Local Oscillator

y x
Display

Sawtooth
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 32

Local Oscillator
l l l l l l

Tunable Sweeps across measurement Span Linear sawtooth drives LO and X-position on Display Repetition rate (sweep time) determined by RBW Sweep time can be manually adjusted (for certain measurements) Not perfect, has phase noise

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

33

What is Phase Noise?

Ideal Signal (noiseless)
V(t) = A sin(2t)
Level

where A = nominal amplitude = nominal frequency

Real Signal

V(t) = [A + E(t)] sin(2t + (t))

Level

where E(t) = amplitude fluctuations (t) = phase fluctuations

Key Point: Phase Noise is unintentional phase modulation on a carrier that spreads its spectrum
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 34

Phase Noise Unit of Measure

Phase Noise is expressed as

L(f)

L(f) is defined as single sideband power due to phase

fluctuations in a rectangular 1Hz bandwidth at a specified offset, f, from the carrier

L(f) has units of dBc/Hz

AMPLITUDE

LOG A

(f)
O LOG f

1 Hz

V0

V0 + f

FREQUENCY

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

35

Phase Noise Who cares?

Modulator designers
Phase noise degrades EVM

Transmitter designers
Phase noise degrades adjacent channel power (ACPR)

Receiver designers
Phase noise degrades receiver sensitivity and selectivity

Radar designers
Phase noise degrades sensitivity to small return signals in the presence of clutter
R&S at UBC: Spectrum Analyzer Fundamentals, October 2013 36

Why do we measure Phase Noise?

l

Especially relevant: phase noise impacts the ability to detect small signals near larger interfering signals

IF

Wanted signal mixed to IF by the LO

IF

But an interferer can mix with phase noise of the LO to the same IF

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

37

What happens if you sweep too fast?

l l

Frequency error Amplitude error

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

38

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

39

Diagram of Superheterodyne Spectrum Analyzer

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

40

Simplified Model vs Actual Implementation

Why we have multiple IF stages
l

If we do straight downconversion, our input, LO and image frequencies overlap. This would require complex filtering to eliminate

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

41

Simplified Model vs Actual Implementation

Why we have multiple IF stages
l

If we use a high IF, filtering becomes much easier!

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

42

Simplified Model vs Actual Implementation

Why we have multiple IF stages
l l

However we cant simply downconvert to DC as we still have filtering issues Creating a very narrowband filter at a high frequency is difficult

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

43

Diagram of Superheterodyne Spectrum Analyzer

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

44

Simplified Model vs Actual Implementation

Local Oscillator
l l l

Use a synthesized signal for the LO Locked to reference signal (internal or external) Use multiplication and division factors

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

45

Simplified Model vs Actual Implementation

Higher Frequencies
l

Tracking pre-selection and switch which side of the LO your input signal is using to increase range

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

46

Simplified Model vs Actual Implementation

Higher Frequencies
l l

YIG filter allows for excellent selectivity Overcomes our problem with filters at high frequencies with wide bandwidths

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

47

Simplified Model vs Actual Implementation

Higher Frequencies Harmonic Mixers

|mfLO nfRF| = fIF

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

48

Simplified Model vs Actual Implementation

Diagram of FSW IF stages

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

49

Simplified Model vs Actual Implementation

Diagram of FSW
l

Different paths for different frequency ranges and bandwidths Pre-amplifier option for looking at weaker signals Signals are digitized higher and higher up the chain FFTs are used in combination with heterodyne principle

l l l

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

50

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

51

Important Settings
l l l l l l

l
l l

Center frequency and span Number of points Resolution Bandwidth Video Bandwidth Sweep Time Detector Trigger Reference level Attenuation

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

52

Important Settings

Reference Level + Attenuation

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

53

Important Settings

Reference Level + Attenuation

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

54

Important Settings

Reference Level + Attenuation

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

55

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

56

Important Specifications
l

Phase Noise

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

57

Important Specifications
l

Displayed Average Noise Level (DANL)

l

There are typically processing techniques to lower the noise floor

With preamp. With preamp. + noise correction

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

58

Dynamic Range: Internal Distortion

The difference (in dB) between the Input Level that produces distortion products equal to the noise floor and the noise floor level (DANL)
But, what type of distortion?
Compression Point Second Order Third order

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

59

Dynamic range: Intermodulation and Harmonics

level intermod. 2nd order Intermod. intermod. 3rd order harmonics

2nd order

3rd order

f2 -f 1

2f 1 -f 2

f1

f2

2f2 - f1

2f1 f2 +f1 2f2

3f1

3f2

frequency

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

60

Dynamic Range: WCDMA ACLR

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

61

Other Important Specifications

l

Speed
l

Sweep speed and processing speed

Measurement uncertainty

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

62

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

63

Measurement functions
l Time domain power

l CP / ACP (Single and Multi-Carrier)

l Spectrum Emission Mask l Occupied bandwidth

l Spurious search
l Noise l Statistics (CCDF) l TOI l Harmonics

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

64

Agenda
l l

Frequency vs Time Domain Spectrum Analyzers

l l

FFT Analyzer Superheterodyne Spectrum Analyzer

l l l l l

Implementation Important Settings Important Specifications Common Measurements Additional Functionality

l l

Vector Signal Analysis Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

65

Vector Signal Analysis

l l l

Digitize RF signal Bandwidths as high as 320MHz are possible Phase information is obtained (which is discarded in spectrum analysis) I and Q data: signals can be demodulated

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

66

Vector Signal Analysis

QPSK

BPSK

GMSK

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

67

What is real-time
l

A Real-Time spectrum analyzer shows the spectrum without any loss of data:

No Blind Time !
Time

FFT

FFT

FFT

FFT

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

68

How is it implemented?

Diagram of FSVR Real-Time implementation

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

69

Real-time Spectrum Analysis

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

70

References
Christoph Rauscher, Roland Minihold, Volker Janssen. Fundamentals of Spectrum Analysis (2008). Rohde & Schwarz.

R&S at UBC: Spectrum Analyzer Fundamentals, October 2013

71