Sie sind auf Seite 1von 25

Phase Noise in Including Comparisons to

GoldenGate SpectreRF and


Measurements

Ian A. Rippke, Ph.D.


Senior Applications Engineer
Electronic Measurements Group
Agilent - EEsof EDA

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Outline

Introduction
• Noise theory review
• Phase noise basics
Phase noise theory
Analyzing circuit phase noise in GoldenGate
Comparing GoldenGate to SpectreRF
Comparing GoldenGate to measurements
Conclusion

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Noise Theory Review
Random process
• Also called stochastic process
• Generates set of continuous, randomly varying functions of time

Probability density function


• Derivative of distribution function
• Gaussian distribution occurs frequently

Central-limit theorem
• Large sample of independent random variables with arbitrary probability
density, tends towards a Gaussian

Noise voltage is a random process


• Gaussian PDF
• Represented by noise power spectral density (PSD) in frequency domain
• This is a mean-square quantity Phase Noise in GoldenGate
Agilent Restricted
15 Sep 2006
Noise Theory Review
Types of Noise
• Flicker noise (1/f noise)
– Physical mechanism not known
– Suspected to be related to trapping at interfaces in devices
• FETs more susceptible than BJTs
– Low frequency noise effect
– Falls off at 30dB/decade
• Thermal noise
– Noise due to fluctuating charge in electrically lossy elements
• White, Gaussian noise process
• Can be modeled by “noise temperature”
– Determines noise floor of circuit
– High offset frequencies from carrier
• Shot noise
– Noise due to electrons crossing a barrier
– At RF frequencies, has PSD that decreases with frequency
Phase Noise in GoldenGate
Agilent Restricted
15 Sep 2006
Noise Theory Review

Stationary Noise
• A random process is stationary if it does not change over time
• Can be represented as a Gaussian distribution
• In frequency domain, noise is not correlated

Cyclostationary Noise
• Noise that varies as a function of time, but in a periodic fashion
• Time-varying operating points modulate noise
• Causes multiplication in time domain, convolution in frequency
• Mixes noise up to harmonics of oscillation frequency
• Resulting overlapping noise sidebands are correlated
– Needs to be accounted for in analysis
– Can still be analyzed in linear fashion
Phase Noise in GoldenGate
Agilent Restricted
15 Sep 2006
Phase Noise Basics

Definition: The time-dependent phase and amplitude


fluctuations of a signal due to random noise processes
Represents the frequency stability of an oscillator
Phase and amplitude noise cause distortion of received signals
– Modern digital communications carry data in phase and amplitude
– Noise in either corrupts received data

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Phase Noise Theory
The ideal output of an oscillator can be written as

v(t )  e Vˆ0e j0t 


The amplitude and frequency jitter added by noise can be written as

 
v(t )  e Vˆ0  vˆ(t ) e j  0t  p (t )  
t
where p(t )   0 ()d
0
Through mathematical manipulation, this can be re-written as

 
v(t )  e Vˆ0  a(t ) e j 0t  p ( t )  
where a(t ) is the amplitude modulation
and p(t ) is the phase modulation
Phase Noise in GoldenGate
Agilent Restricted
15 Sep 2006
Phasor Representation

Noiseless carrier phasor V0


a(t)
PM, AM noise
v(t)
Magnitude of
• Net carrier phase noise: p(t) perturbed signal

• Net carrier amplitude noise: a(t) p(t)

p(t)
Voltage noise
• Combination of phase and voltage jitter Magnitude of V0

• Shown as grey arrows, v(t), p(t)


• Resultant vector is defined by AM, PM noise
V0

AM, PM noise are real quantities


• Spectrum is even function of frequency
• Power spectral density of noise is double-sided

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Sideband Noise

Starting with the equation for a noisy oscillator:

 
v(t )  e Vˆ0  vˆ(t ) e j 0t p (t )  
t
where 
p(t )   0 ( )d
0

We can show that the difference between that and a noise-free waveform is

vˆ(t )  vˆ(t )e j0t

Taking the Fourier transform gives the sideband noise



Vˆ ( f )  vˆ(t )e j 2ft dt
0

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
AM, PM, and SB Noise Relationship

Sideband noise is combination of AM & PM noise.


Far enough from carrier, noise is given approximately as:

vˆ(t )  vˆ(t )  jVˆ0p(t )


Or more specifically
vˆ(t )
 a(t )  jp(t )
V0

In the frequency domain,

Vˆ ( f )
 A( f )  jP( f )

0

This shows that sideband noise is approximately a linear combination of AM and PM noise.

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
AM, PM and SB Noise Relationship

Upper and lower sideband noise


• Sideband noise is a complex variable
• Spectrum is not always an even function of frequency
• The sideband noise can then be expressed as
Lower Sideband Noise|dBc = 20log|V(f)/V0| , f < 0
Upper Sideband Noise|dBc = 20log|V(f)/V0| , f > 0

If PM noise is >> AM noise, sideband noise can be written as


|V(f)/V0| = L(f) = |P(f)|
Where |P(f)| is the peak PSD of PM noise, so in rms terms,
L(f) = PM Noise / 2, or in dBc, PM Noise – 3dB
This gives us the relationship that PM noise is 3dB higher than
sideband noise!
Phase Noise in GoldenGate
Agilent Restricted
15 Sep 2006
Phase Noise Analysis in GoldenGate

Solves for full nonlinear noise, no linearized models


AM, PM noise are state variables
Noise calculations are different based on circuits
• Autonomous (oscillators)
– Computes AM & PM noise directly
• Non-autonomous (driven circuits)
– Computes total noise and extracts AM & PM

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Autonomous Circuits

GoldenGate phase noise uniquely suited to oscillator noise


• Envelope technique
• No linearization
Compute AM, PM noise directly in circuit
• Includes correlations between AM & PM noise
• Allows computation of cyclostionary noise
Noise using envelope technique
• Treats noise as a modulation signal about a carrier
• Separate analysis for each frequency offset
Assumes noise is stationary for oscillators
• User can select between stationary and cyclostationary in options

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Non-Autonomous Circuits

Sideband (total) noise is calculated


AM, PM extracted using the equation

Vˆ ( f )
 A( f )  jP( f )
ˆ
V0

Which in this case is a true equality


This is the same calculation as SpectreRF
Noise is assumed to be cyclostationary

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Noise Contribution Tables

GoldenGate v4.0
• Three NCTs – AM, PM, and DSB
• AM, PM NCT
– Shows contributions of each element in dBc/Hz
• DSB NCT
– calculated from SB noise results
– Noise is computed as

 SB Noise3dB20log (peaksignal) 
Noise  10   [V 2 /Hz]
10
DSB

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
NOISECOR

New GoldenGate element “NOISECOR” in v4.0


• Stands for “Noise Correlations”
• Replaces “OSCNOISE”

Captures AM, PM noise from a simulation, including correlations


between the two
• Also captures oscillation frequency and waveform shape (harmonics)
• More accurate than using noise file in ideal source

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
NOISECOR
LO
Interfering
Channel

Desired
Channel

Interfering IF

2
Desired IF

LO VCO

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Comparing GoldenGate to SpectreRF

SpectreRF computes total noise in circuit


SSB noise is computed from total noise

SSB Noise = 10*log(total noise [V2/Hz]) –20*log(rms signal voltage) [dBc/Hz]

This is the reverse of how GoldenGate computes noise for autonomous circuits
SpectreRF uses a linearized model
• Based on PSS simulation
• Sources of error include lack of steady state, convergence error, limitations of linearized
model
SpectreRF always uses cyclostationary noise
• GoldenGate currently uses stationary noise for oscillators
• If results don’t match in flicker noise region, change type from “auto” to “cyclostationary”

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Comparing GoldenGate to SpectreRF

SB matches SpectreRF SSB,


PM, AM 3dB higher than SpectreRF
NCT includes Port noise specified by “noisetemp”
Modulated noise in SpectreRF
• Recent addition, claim AM & PM noise
• Results are presented as “SSB” quantities
• Gives AM, PM, USB, LSB noise

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Comparing GoldenGate to SpectreRF

Common questions
• Q: In SpectreRF I see sideband noise higher than 0dBc/Hz. Will I see that
in GoldenGate?
• A: YES, but we’re correct because we compute PM noise. PM noise can become very large
at low offset because of continued accumulation of phase close to the carrier. Technically,
units of PM noise are [dB rad/Hz]. Because our SB noise is an approximation combining AM
and PM noise, our SB will also go above 0dBc/Hz at low offset.

• Q: GoldenGate phase noise doesn’t match SpectreRF phase noise


• A:
1) try more harmonics
2) change noise from “auto” to “cyclostationary”
3) change flicker noise model in registry
– If this does not get results to agree, encourage user to increase
tolerances in SpectreRF because PSS solution may not be accurate
Phase Noise in GoldenGate
Agilent Restricted
15 Sep 2006
Comparing to Measurements

Spectrum analyzer
• Measures rms noise power in a single sideband, normalized to a 1Hz
bandwidth
• Compares this to the rms signal power measured at the carrier frequency
• Corresponds directly to single sideband noise

Measurement of SSB Phase Noise

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Comparing to Measurements

Spectrum analyzer caveats


• The units of measurement are essential
– Some units give power or voltage, others reference the carrier signal
• Some spectrum analyzers already compute the noise in both sidebands
– This effectively adds a factor of 2 to the total noise, or increases the phase noise
by 3dB
– This double sideband value should equal the PM noise results from GoldenGate,
assuming the AM noise is negligible
• The local oscillator may be too noisy for high performance measurements
• Instability of the center frequency
– Drift in measurement introduces “non-ideal” slopes
– Close-in phase noise may not be measurable
• Average of square is not same as square of average
– A Gaussian noise will have a 2.5dB error

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Comparing to Measurements

Phase noise measurement system (such as Agilent E5500)


• Directly measures AM, PM noise
• Wideband amplitude detector measures AM noise
• PM noise measured by demodulating phase fluctuation of device under test
– Uses a locked low-noise phase reference w/calibration
– Balanced mixer serves as phase comparator
– Resulting output is time representation of phase fluctuations

PM Noise Measurement Block Diagram

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Comparing to Measurements

Phase demodulation caveats


• Reference source must be very low noise
• PLL must have precise control, calibration

Phase demodulation advantages


• Carrier signal is suppressed, increasing dynamic range
• Sensitivity limited only by reference source
• Separates phase and amplitude effects
• Low PLL drift leads to more precise measurements
• Close-in phase noise measurements are possible
• Can compare directly to GoldenGate!

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006
Conclusion

Noise analysis is a constant source of confusion with customers


• Many make incorrect assumptions about phase noise
• Need to educate on difference between PM and SB noise
GoldenGate phase noise analysis is fundamentally different
than SpectreRF
• But results often match -> high confidence
Measurements also have sources of error
Education is key to success!

Phase Noise in GoldenGate


Agilent Restricted
15 Sep 2006