understanding low phase noise signals · each type of noise process has distinct characteristics...
TRANSCRIPT
Agilent Copyright 2014
Understanding Low
Phase Noise Signals
Presented by: Riadh Said
Agilent Technologies, Inc.
Agilent Copyright 2014
Introduction
Instabilities in the frequency or phase of a signal are caused by a number of different effects.
Each type of noise process has distinct characteristics that can be measured using time
domain and frequency techniques.
In many cutting edge radar and communication systems, phase noise is the characteristic
that limits the system performance. In radar systems, phase noise degrades the ability to
process Doppler information in radar. And, in digitally modulated communication system,
phase noise degrades error vector magnitude.
I
I
Frequency
Instability
Effects
Q
3 Agilent Copyright 2014
Phase Noise can Matter:
Getting the Best Possible Performance
– Phase noise may be limiting parameter in measurement/sys. performance
Balancing Phase Noise with Cost, Other Factors
– How to identify, quantify effects
– Identifying When it’s Not Important
Phase Noise Tradeoffs
– Acquisition cost
– Switching speed
– Frequency and offset noise
3
•A Lot
•Some
•Not At All
4 Agilent Copyright 2014
Agenda
• Frequency Stability
– Terminology & representation
– Architecture implications
• When Phase Noise Matters
– Example: Doppler Radar
– Example: Phase noise and EVM in OFDM signals
– Degrading phase noise performance for real-world signal substitution
• Signal Generator Choices, Performance & Tradeoffs
• Getting the Most from the Signal Generator You Have
4
Agilent Copyright 2014
Random perturbations that
appear as instabilities in
frequency can be represented in
either the frequency domain or
the time domain.
• Frequency Domain
- Frequency Offsets of a few Hz
to tens of MHz
• Time Domain
- For close-in noise (<<1 Hz)
- Short-Term Stability
5
Frequency Stability: Frequency & Time Domain Measurements
6 Agilent Copyright 2014
Frequency Stability: Short & Long Term
Terms, Measurements, Displays
Short Term Stability
• Short term = seconds
• Terminology: Phase Noise, Jitter
• L(f) curves, integrated totals, spot
measurements, jitter (p-p)
• Can be a function of both signal
generator and frequency reference
Long Term Stability
• Long term = minutes - years
• Terminology: Accuracy, drift, aging
• Often determined by frequency
reference
Understand Your System and
its Sensitivity vs. Frequency
6
Detail References Phase Noise Measurement Methods
and Techniques
7 Agilent Copyright 2014
Pedestals, Slopes & Bumps: Signal
Generator Architecture & Phase Noise
Example: Agilent PSG Microwave Signal Generator
7
Reference Section
Synthesizer
Section YIG
Oscillator
Output
section
Detail References Reducing Phase Noise at Microwave
and RF Frequencies
8 Agilent Copyright 2014
Agenda
• Frequency Stability
– Terminology & representation
– Architecture implications
• When Phase Noise Matters
– Example: Doppler Radar
– Example: Phase noise and EVM in OFDM signals
– Degrading phase noise performance for real-world signal substitution
• Signal Generator Choices, Performance & Tradeoffs
• Getting the Most from the Signal Generator You Have
8
9 Agilent Copyright 2014
When Phase Noise Matters
Matters More
• Doppler RADAR
• OFDM
• Oscillator substitution
• ADC testing
Matters Less
• OFDM (matters more and less, depending on offset)
• Wideband single-carrier modulation
• Harmonic distortion testing
• Wideband ACPR testing
• Amplifier gain testing
9
Agilent Copyright 2014
Example: Phase Noise Doppler RADAR
10
Agilent Copyright 2014
Phase Noise and RADAR Applications
Doppler Example
11
F 0 Fd1
BW = 1/t
STALO phase noise sidebands
with no target clutter
STALO
LNA IFA
PA
F0
F0r + Fd1
Pt
Pr
Detail References Reducing Phase Noise at Microwave
and RF Frequencies
12 Agilent Copyright 2014
Doppler Frequency Shift and
Phase Noise Offset Frequencies
Close-In (example: Airport Surveillance Radar)
• 100 mph, S-band (3 GHz): fd = 900 Hz
Wider Offsets (example: Military Radar)
• Mach 1.5, X-band (10 GHz) fd = 33,000 Hz
12
R
Doppler Frequency fd :
Rvc
fRvfd
022
Detail References Radar & Electronic Warfare Threat
Simulation
Agilent Copyright 2014
Consider Noise Contributions from
Amplifiers, Including Broadband Noise
13
Additive noise of the transmitter
is present in the large reflected
signal from the clutter and can
mask the targets response
F0r
BW = 1/t
STALO
Pulse Mod
PA
PA
STALO
Fd2
Agilent Copyright 2014
Also Consider Spurious Performance
for RADAR and EW Applications
14
Noise and Spurious Effects
F 0 Fd1
BW = 1/t
Fs1 Fs2 Fd2
System Spurious Signals within the Receiver BW
STALO
LNA IFA
PA
F0,Fs1,Fs2
F0 + Fd1
Agilent Copyright 2014
F0r
BW = 1/t
Transmitter phase noise sidebands with clutter
Example: 1 GHz Doppler radar
Doppler BW = 10 kHz
Target = 1 m2
Range = 100 km
Min Det Vr = 80 knots
Clutter visibility = 80 dB
Transmitter Noise Sidebands:
L(f) < -120 dBc/Hz @
200 Hz offset
STALO
PULSE MOD
PA
PA
STALO
Fd2
Estimating Required Phase Noise
Performance
15
Detail References Introduction to Radar Systems
Skolnik
Agilent Copyright 2014
Example: Phase Noise and OFDM
16
17 Agilent Copyright 2014
OFDM and Pilot Tracking
Pilots Shown in Time (I/Q) and Frequency
17
18 Agilent Copyright 2014
OFDM and Pilot Tracking
BPSK Pilots: Demodulation Reference, No Data Transmitted
Pilot Tracking “Tracks Out” Some Freq/Phase Instability
Instability Not Just Constellation Rotation, also FFT Leakage or
Inter-Carrier Interference
18
No Pilot Tracking Pilot Tracking Enabled Common Pilot Error: Phase
19 Agilent Copyright 2014
Translating Phase Noise to EVM
in OFDM
Assumptions
• Estimating effects of only phase noise and broadband (wide offset) noise
• Spurious and other nonlinearities are not significant
• Equalization effective in removing linear errors
• Pilot tracking is effective to ~10% of subcarrier spacing
– Common phase error removed in demodulation
Example: WiMAX
• 10 kHz subcarrier spacing
• 10 MHz channel bandwidth
19
20 Agilent Copyright 2014
Integrate Power
for
Total EVM
Phase
Noise
Tracked
Out
10 MHz Channel Bandwidth
10 kHz Subcarrier Spacing
Phase Noise
Filtered Out
Example: Phase Noise Contribution to
EVM in OFDM
EVM a function of
integrated phase noise
beyond tracking BW
and inside channel BW (and correct SSB to DSB)
20
-95 dBc/Hz integ. over ~100 kHz
& Convert SSB to DSB: add 3 dB
-95 dBc/Hz + 10log(100 kHz) + 3 dB
EVM = -95 dB + 50 dB + 3 dB
EVM = -42 dB (conservative)
Error power calculated on log scale:
95 dBc/Hz pedestal
21 Agilent Copyright 2014
Signal Generation and Signal Analysis
for Design & System Integration
Q: Generate with no phase noise or representative amount?
A: Yes, both
– Understand ultimate performance and residual error (error budget)
– Understand phase noise tolerance, design for “just good enough”
performance
Q: Measure with pilot tracking enabled or disabled?
A: Yes, both
– Even when tracked out, phase noise can reduce demodulation margins
21
22 Agilent Copyright 2014
Adding Known Phase Noise Using FM
Signal Modulated with Uniform Noise
Simple Technique for -20 dB/Decade
22
Example: Agilent PSG
• Set up an signal generator for FM modulation, by selecting:
– FM path 1
– FM on
– FM deviation, as specified
– FM waveform to noise, uniform
• Ensure that noise of the PSG in FM off is at least 10 dB less than the desired calibrated noise at a desired offset frequency, to ensure accuracy.
Detail References Phase Noise Measurement Methods
and Techniques
23 Agilent Copyright 2014
Degrading Phase Noise for Accurate
Signal Substitution
Simulate VCOs, Lower-Performance Synthesizers
When “Representative” is Better than “Perfect”
Use Baseband Real-Time Processing
23
Agilent Copyright 2014
Selectively-Impaired Phase Noise
Performance
24
Llow
Lmid
f1
Lhigh
f2
Llow
Lmid
f1
Lhigh
f2
•User sets f1, f2, and Lmid
•Slope -20dB/decade below f1 and above f2
f1=5kHz, f2=500kHz,
Lmid=-80dBc
f1=10kHz, f2=1MHz,
Lmid=-90dBc
25 Agilent Copyright 2014
Agenda
• Frequency Stability
– Terminology & representation
– Architecture implications
• When Phase Noise Matters
– Example: Doppler Radar
– Example: Phase noise and EVM in OFDM signals
– Degrading phase noise performance for real-world signal substitution
• Signal Generator Choices, Performance & Tradeoffs
• Getting the Most from the Signal Generator You Have
25
26 Agilent Copyright 2014
Sig. Gen. Choices: Frequency Stability Narrowing the Hardware & Software Choice
CW or “Analog” vs. Digital Modulation or “Vector”
• Phase noise performance choices may interact with other capabilities
– Internal and/or external digital modulation
– Pulse modulation
– Internal/external software
– Memory and/or real-time baseband signal generation
– Power, distortion
Single-Loop vs. Multiple Loop
Phase Noise Performance Levels as Options
VCO (voltage-controlled oscillator) vs. YIG (yttrium iron garnet)
• Switching speed, phase noise, cost
26
New X-Series
Single- & Multiple-Loop Synthesizers
Single Loop (example: EXG)
• Less complex, less expensive
• Moderate performance, can
provide very good ACPR
• Simpler to design, optimize
Multiple Loop (example: MXG)
• Typically fine loop + offset/step loop
+ sum loop
• Better phase noise
• Lower spurious
• More optimization choices and
flexibility
27
EXG N5173B Microwave Analog
EXG N5172B RF Vector
MXG N5183B Microwave Analog
MXG N5182B RF Vector
Agilent Copyright 2014
1Hz 10Hz 100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz
Single vs. Multiple-Loop Architecture
New Agilent X-Series (MXG, EXG)
28
MXG enhanced low
phase noise option
MXG standard
EXG SSB Phase Noise @ 10 GHz
L(f) [dBc/Hz] vs. Frequency
29 Agilent Copyright 2014
MXG N5183B Analog, N5182B Vector
Spurious and Broadband Noise May be as Important
as phase noise
29
MXG std. -134 dBc/Hz typ. (20 kHz offset @ 1 GHz)
kHz
MXG opt. UNY -146 dBc/Hz typ. (20 kHz offset @ 1 GHz)
Spur/Non-Harm. -96 dBc typ.
(opt. UNY @ 1 GHz)
Broadband Noise -163 dBc typ.
(10 MHz offset @ 1 GHz)
30 Agilent Copyright 2014
-160
-140
-120
-100
-80
-60
-40
-20
0
1 10 100 1,000 10,000 100,000 1,000,000 10,000,000
Ph
ase
No
ise
(dB
c/H
z)
Frequency Offset from Carrier (Hz)
PSG (15 GHz)
60 GHz (15 GHz x4)
90 GHz (15 GHz x6)
120 GHz (15 GHz x8)
180 GHz (15 GHz x12)
270 GHz (15 GHz x18)
450 GHz (15 GHz x30)
Generally Worse with Increasing Frequency
Delta = 20*log(N), where N = multiplication factor
Phase Noise vs. Frequency
Agilent PSG Microwave/Millimeter Signal Generator
Agilent Copyright 2014
Phase Noise vs. Frequency
31
Agilent MXG RF Signal Generator (opt. UNY)
100 Hz 10 kHz 1 MHz 10 MHz
Not always a Simple Relationship
Agilent Copyright 2014
Oscillator and PLL Technologies Affect
SSB Curve Shape, Tradeoffs
32
New MXG vs. EXG vs. ESG @1GHz
100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz
33 Agilent Copyright 2014
Tradeoffs
Switching Speed
• Generally faster with VCOs (vs. YIGs)
• Faster with wider loop bandwidths
• But wider loop BW may move noise to undesirable offset region
Phase Noise
• YIGs generally better than VCOs
• Multi-loop significantly better than single-loop
• Move problem energy to a non-problem region
• Optimize for wide offsets vs. optimize for close-in
33
34 Agilent Copyright 2014
Other Capabilities, Tradeoffs vs. Phase
Noise Performance
Modulation Capability
• Maximum modulation bandwidth
• Modulation quality
– In-band: EVM or MER, etc.
– Out-of-band: ACPR
• Modulation inputs & outputs (bandwidths can be different)
Pulse Capability
• Complex pulse scenarios, pulse performance
– Modulation on pulses, Doppler, PRT
• Software/application support
Multi-Channel Synchronization
34
35 Agilent Copyright 2014
Agenda
• Frequency Stability
– Terminology & representation
– Architecture implications
• When Phase Noise Matters
– Example: Doppler Radar
– Example: Phase noise and EVM in OFDM signals
– Degrading phase noise performance for real-world signal substitution
• Signal Generator Choices, Performance & Tradeoffs
• Getting the Most from the Signal Generator You Have
35
36 Agilent Copyright 2014
Methods for Improving Signal
Generator Phase Noise
Connect to a Low-Noise External Frequency Reference
Advanced Architectural Improvements to the Reference and
Synthesizer Design
Adjustment of the Reference Oscillator & VCO PLL Bandwidth
Use of Dividers at Lower Frequencies Instead of Heterodyne
Mixing
36
Detail References Reducing Phase Noise at Microwave
and RF Frequencies
37 Agilent Copyright 2014
Optimizing Signal Generator Phase
Noise
Choice of External Reference Oscillator
Choosing External Reference Bandwidth
Optimizing for Close-In or Wide Offset Noise
Optimizing for Broadband Noise
Optimizing for low frequency coverage
Noise Subtraction, Including Phase Noise Subtraction
Read Documentation, Specs for Product-Specific Optimization Choices
37
Agilent Copyright 2014
1Hz 10Hz 100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz L(f) [dBc/Hz] vs. Frequency
Residual Phase Noise: Comparisons
with Perfect Reference
38
Using perfect external10 MHz reference
up to 20 dB
improvement @ 1 Hz
Agilent Copyright 2014
Residual Phase Noise: Comparisons
with Perfect Reference
39
Agilent PSG UNY
up to 20 dB improvement @ 1 Hz
40 Agilent Copyright 2014
Optimize Frequency Reference Choice
and Settings: Signal Analyzer Example
Quality of external 10 MHz ref. oscillators can make a big
difference in phase noise measurements (& signal generation)
40
• If an external reference
oscillator has better close-in
phase noise than the PXA the
wide locking bandwidth can
be used to obtain better close-
in (< 30 Hz) measurement
performance.
• If higher offset frequencies
are more important the narrow
setting can be used.
External reference
lock wide
External reference
lock narrow
Detail References Phase Noise Measurement Methods
and Techniques
Agilent Copyright 2014
Reference Oscillator PLL Bandwidth
Adjustment: PSG Signal Generator
41
The 10 MHz
reference signal
can be internally
generated or
externally supplied
Adjusting the
PLL integrator
bandwidth
Detail References Reducing Phase Noise at Microwave
and RF Frequencies
Agilent Copyright 2014
PLL Bandwidth Adjustment: PSG
Microwave Signal Generator Example
42
300 Hz
650 Hz
125 Hz
55 Hz
25 Hz
Agilent Copyright 2014 43
Optimizing Pedestal Phase Noise: PSG
Microwave Signal Generator Example
44 Agilent Copyright 2014
Optimizing Signal/Noise vs.
Phase Noise
44
10 kHz 1 MHz 10 MHz
45 Agilent Copyright 2014
Generating Lower Frequencies:
Freq. Division vs. Heterodyne Mixing
Frequency division reduces the phase noise sidebands of the
signal by a factor of 20 dB/decade or 6 dB/octave
• Example: Dividing by 4 reduces phase noise by 12 dB. (20*log(1/4) )
Divider tradeoff: Maximum FM and PM deviations are reduced
by the same factor as the division number
Heterodyne mixing provides fine frequency adjustment while
retaining full bandwidth FM and PM
Heterodyne mixing tradeoff: Heterodyne mixing to lower
frequencies provides no phase noise reduction
45
46 Agilent Copyright 2014
Subtracting Signal Generator Phase
Noise Power In some scalar (power) measurements the phase noise
contribution from the signal generator can be subtracted from
the measured result
46
Raw measurement Reference measurement Result after subtraction
Agilent Copyright 2014
Subtracting Signal Generator
Broadband Noise Power
47
B M a r k e r 4 4 3 0 5 9 0 . 3 8 1 H z - 1 3 5 . 7 7 5 d B m / H z
T R A C E A : F 1 P S D 1 / K 1
A M a r k e r 4 4 3 0 5 9 0 . 3 8 1 H z - 1 2 3 . 6 d B m / H z
C e n t e r : 4 . 4 2 0 7 7 7 8 8 1 M H z S p a n : 5 0 k H z
- 5 0
d B m / H z
- 1 5 0
d B m / H z
L o g M a g
1 0
d B
/ d i v
48 Agilent Copyright 2014
Noise Sidebands May Not Be Entirely
Phase Noise
AM Noise may be Significant Fraction of Noise Sideband Power
AM Noise may Affect a System or Not
• Example: Pilot Tracking Usually Removes Close-In AM Noise
48
• AM noise is included in direct
noise sideband measurements
• X-Series phase noise meas.
app (N9068A) can reject AM
noise < 1 MHz
• AM rejection can improve phase
noise measurements
• Understand system sensitivity,
signal generator AM noise
AM Rejection
off
AM Rejection
on
Detail References Phase Noise Measurement Methods
and Techniques
49 Agilent Copyright 2014
Phase Noise can Matter:
Getting the Best Possible Performance
– Phase noise may be limiting parameter in measurement/sys. performance
Balancing Phase Noise with Cost, Other Factors
– How to identify, quantify effects
– Identifying When it’s Not Important
Phase Noise Tradeoffs
– Acquisition cost
– Switching speed
– Frequency and offset noise
49
•A Lot
•Some
•Not At All
50 Agilent Copyright 2014
References, More Information
“Techniques for Improving Noise and Spurious in PLLs” Eric Drucker,
Microwave Journal, May 2012
“Reducing Phase Noise At Microwave and RF Frequencies” Agilent A/D
Symposium 2011
“Signal Generators Provide Perfect and Precisely Imperfect Signals” Ben
Zarlingo, Microwave Journal, May 2012
“Testing Low-Noise Components in Pulsed or Moving Target Radars” Agilent
A/D Symposium 2009
“Phase Noise Measurement Methods and Techniques” Agilent A/D
Symposium 2012
“Radar & Electronic Warfare Threat Simulation” Agilent A/D Symposium 2011
“Introduction to Radar Systems” by Merrill Skolnik, 2002,
ISBN 978-0072881387
Agilent Copyright 2014 51