'^liiiiiiiS

PUBLICATIONS

Nisr
National Institute of
Standards and Technology
U.S. Department of Commerce

NIST Special Publication 250-90

Calibration Uncertainty for the NIST

PM/AM Noise Standards

AM

0) 0)
'l_
L_
CD CD

Archita Hati O o
Craig Nelson
Neil Ashby
David Howe
AM PM
'

100
.U57
# 250-90
2012
 rhe National Institute of Standards and Technology was established in 1988 by Congress to "assist industry in the
development of technology ... needed to improve product quality, to modernize manufacturing processes, to ensure
product reliability ... and to facilitate rapid commercialization ... of products based on new scientific discoveries."
NIST, originally founded as the National Bureau of Standards in 1901, works to strengthen U.S. industry's
competitiveness; advance science and engineering; and improve public health, safety, and the environment. One of the
agency's basic functions is to develop, maintain, and retain custody of the national standards of measurement, and provide the
means and methods for comparing standards used in science, engineering, manufacturing, commerce, industry, and education
with the standards adopted or recognized by the Federal Government.
As an agency of the U.S. Commerce Department's Technology Administration, NIST conducts basic and applied
research in the physical sciences and engineering, and develops measurement techniques, test methods, standards, and related
services. The Institute does generic and precompetitive work on new and advanced technologies. NIST's research facilities
are located at Gaithersburg, MD 20899, and at Boulder, CO 80305. Major technical operating units and their principal
activities are listed below. For more information visit the NIST Website at http://www.nist.gov, or contact the Publications
and Program Inquiries Desk, 301-975-3058.

Office of the Director Chemical Science and Technology

•National Quality Program Laboratory
International and Academic Affairs Biotechnology
•Process Measurements
Technology Services
•Surface and Microanalysis Science
•Standards Services
•Physical and Chemical Properties'
•Technology Partnerships
•Analytical Chemistry
•Measurement Services
Information Services Physics Laboratory
•Weights and Measures Electron and Optical Physics
•Atomic Physics
Advanced Technology Program
Optical Technology ;
•Economic Assessment
Ionizing Radiation
•Information Technology and Applications
•Time and Frequency'
•Chemistry and Life Sciences
•Quantum Physics'
•Electronics and Photonics Technology

Manufacturing Engineering
Manufacturing Extension Partnership
Laboratory
Program •Precision Engineering
•Regional Programs Manufacturing Metrology
•National Programs •Intelligent Systems
•Program Development •Fabrication Technology
•Manufacturing Systems Integration
Electronics and Electrical Engineering
Laboratory Building and Fire Research Laboratory
•Microelectronics Applied Economics
•Law Enforcement Standards •Materials and Construction Research
•Electricity •Building Envirormient
•Semiconductor Electronics •Fire Research
•Radio-Frequency Technology'
•Electromagnetic Technology' Information Technology Laboratory
Optoelectronics' •Mathematical and Computational Sciences"
•Magnetic Technology' •Advanced Network Technologies
•Computer Security
Materials Science and Engineering •Information Access
Laboratory •Convergent Information Systems
•Intelligent Processing of Materials •Information Services and Computing
•Ceramics •Software Diagnostics and Conformance Testing
•Materials Reliability' •Statistical Engineering
•Polymers
•Metallurgy
NIST Center for Neutron Research

;At Boulder, CO 80305

"Some elements at Boulder, CO
 NIST Special Publication 250-90

Calibration Uncertainty for the NIST

PM/AM Noise Standards
Archita Hati
Craig Nelson
Neil Ashby
David Howe

Time and Frequency Division

Physical Measurement Laboratory
National Institute of Standards and Technology
325 Broadway
Boulder. Colorado 80305

July 2012

U.S. Department of Commerce

Rebecca Blank, Acting Secretary

National Institute of Standards and Technology

Patrick Gallaher, Director

 Certain commercial entities, equipment, or materials may be identified in this
document in order to describe an experimental procedure or concept adequately. Such
identification is not intended to imply recommendation or endorsement by the
National Institute of Standards and Technology, nor is it intended to imply that the
entities, materials, or equipment are necessarily the best available for the purpose.

National Institute of Standards and Teciinology Special Publication 250-90

Natl. Inst. Stand. Technol. Spec. Publ. 250-90, 44 pages (July 2012)
CODEN: NSPUE2

U.S. GOVERNMENT PRINTING OFFICE

WASHINGTON: 2001

For sale by the Superintendent of Documents, U.S. Government Printing Office

Internet: bookstore.gpo.gov — Phone: (202) 512-1800 — Fax: (202) 512-2250
Mail: Stop SSOP, Washington, DC 20402-0001
Contents
1 Introduction 1

2 AM and PM noise - Basic concepts 2

2. 1 Characterization of AM and PM noise 2

2.2 Overview of amplitude, phase and single sideband modulation 3

2.2. 1 Amplitude modulation 3

2.2.2 Phase modulation 3

2.2.3 Single-sideband modulation 5

3 Description of the noise standard 7

4 Measurement setup for calibration of the PM/AM noise standard 9

5 Calibration procedure at NIST 10

5. 1 Measurement of the carrier power via down-conversion from the upper sideband 10

5.2 Measurement of the carrier power via down-conversion from the lower sideband 10

5.3 Measurement of the noise power spectral density down-converted from both the
upper and the lower sidebands 11

5.4 Measurement of the noise floor power spectral density 12

6 Measurement equation 13

7 Estimation of uncertainty 15

8 Uncertainty budget 18

9 Validation of the measurement 20

10 Acknowledgement 20
Appendix A Verification of the baseband measurement equation 21

Appendix B Measurement of correction factors, K^l, Krf, Kbw and Kp 25

Appendix C Error of signal power measurements in the presence of background noise 28

REFERENCES 31

iii
Notations
Acronym Meaning

AM Amplitude Modulation
BPF Bandpass Filter
BW Bandwidth
dB Decibel
dBm Decibel relative to 1 milliwatt
DDPNMS Direct Digital Phase Noise Measurement System
FFT Fast Fourier Transform
IF Intermediate Frequency
LO Local Oscillator
'

LPF Low-Pass Filter

LSB Lower Sideband
PDLMS Photonic Delay Line Measurement System
PM Phase Modulation
PSD Power Spectral Density
rf Radio Frequency
RF Reference Frequency
SNR Signal-to-Noise Power Ratio
SSB Single Sideband
USB Upper Sideband
VSWR Voltage Standing-Wave Ratio

Symbol Meaning

o(t) Instantaneous amplitude fluctuations

Phase modulation index
Maximum error of Knl from unity
Maximum error of Krf from unity

£fi
Correction factor due to the small angle modulation approximation
(/)(t) Instantaneous phase fluctuations
^(ppeak Peak phase deviation
A0rms Root-mean-square phase deviation
7 Noise power in 1 Hz bandwidth
Vo Carrier frequency
Vlo Carrier frequency at the LO port of the mixer
Vrf Carrier frequency at the RF port of the mixer

<7fj
Fractional uncertainty due to the small angle modulation approximation
(TBear- Fractional uncertainty in the lower-sideband beat measurement
Fractional uncertainty in the upper-sideband beat measurement
(^B-FFTAve Uncertainty due to the number of FFT averages for beat measurement
(Tbw Fractional uncertainty in the estimated measurement bandwidth for the power

iv
 spectral density function

<5c Combined fractional uncertainty

ctlr Uncertainty due the long-term reproducibility of the PM/AM noise standard
between calibration cycles
(^N-FFTAve Uncertainty due to the number of FFT averages for noise measurement
CfNL Fractional uncertainty due to the FFT analyzer and mixer nonlinearity at baseband
(^NoiseOn Fractional uncertainty in the noise measurement with the noise source on
Orf Fractional uncertainty due to the frequency response of the down-convertor
(includes offset signal generator and mixer) at rf

(^SR Uncertainty due to the short-term repeatability between measurements

Oh Angular frequency of the carrier signal
rn Angular modulation freouencv
BWaActual
^ f .1 il.V' ILJ-Cll hanHwiHth
Artiij^l \J iXxWX VV lU-Lli

'
' est Estimated bandwidth
J Fourier or offset freouencv
Lower frequency limit of the integrated phase noise
T Innpr frpniipnrv limit of thp intpprafpH nhasp noisp
rrain of thp
VJCilll Wl lliw Hown-ronvprtpr
VV 11 ^Wll V wl Lwl
VJv-' fit
Cll tllw hasphPinH
thp L/ClowU/ClllU

SRF VJCllll of
frain v^l thp
lliw Hown-pon
W 11 Wll vprtpr
V 1^1 Mt its two
iXv ILO Vy\ \J onpratinp
Cll. 1 1 V^vl Ll^llV^ i^O^ Vnc
W lywl Ill £l frpniipnpips Kr
' anrl
t^A'vl Vi/-v
V ]^(_)

gVSWR Gain of the down-converter due to VSWR mismatch

gLVL Gain of the down-converter for a specific power level at the RF port
G Gain of the down-converter
Jn n-th order Bessel function
k Coverage factor
Kp Correction factor applied to £(f) due to the small angle modulation approximation

Kbw Correction factor for incorrect estimation of the actual measurement bandwidth
for the power spectral density function
Correction factor due to the FFT analyzer and mixer nonlinearity at baseband
Krf Correction factor due to the frequency response of the down-convertor (includes
offset signal generator and mixer) at radio frequencies

m
tn
Single sideband phase noise equal to one half of

Amplitude modulation index

if)

iMUlllUCI Ui IcpCdLcU lllCaSUrcillCIll aCLS 111 d glVCIl CallDrdlUjn

i^UlllUC'l \JV 111 dVCld^Co \.\jV UCCIL lliCd^Ul ClliClll

Noise l>IUillUCl yjV JTIT^ 1 aVClagCJ^ lUl llVJlwiC lliCdc>UlCillCilL

n T-Ti CTnPf-r^fHpf
riigiici ulU-ci tPTTTic
Lciiii^ c\\
ui TTd^/lr^T*
1 ayiui PYt^cincir^n
CAUdii?>iuii
p
^p earner Powpr IpvpI at thp RP nort of thp mixpr HiiHncr rarripr nowpr mp?i<iiirpmptit
Carrier power afterdown-conversion to baseband
PH01S€ .

Power level at the RF portof the mixer during the noise measurement
PN-Beat Noise power after down-conversion to baseband
PSSB-AM Single-sideband AM signal power
PsSB-PM Single-sideband PM
power signal

Sa(f) One-sided double-sideband power spectral density of amplitude fluctuations

One-sided double-sideband power spectral density of random phase fluctuations

V
uc Combined uncorrelated uncertainty
U Expanded uncertainty
Y(t) Instantaneous voltage fluctuations
Vb Peak voltage of the carrier signal
yBaseband Foldcd doublc-sideband down-converted voltage of the PM/AM noise standard at
baseband
VBeat- Voltage of the lower-sideband beat frequency
yBeat+ Voltage of the upper-sideband beat frequency
VBeatAve Avcragc voltagc of the upper- and lower-sideband beats
VBeat-.M Mcasurcd voltage of the lower-sideband beat
yBeat+, M Mcasurcd voltage of the upper-sideband beat
VNoiseOff Voltage of the additive noise at baseband with noise source off
VNoiseOn Voltagc of the additive noise at baseband with noise source on
M
VNoiseOff. Mcasurcd voltage of the additive noise at baseband with noise source off
VNoiseOn, M Mcasurcd voltage of the additive noise at baseband with noise source on
VpNST Signal voltage of the PM/AM noise standard at rf
VpNST(CamerOn) Signal voltagc of thc noise standard when the additive noise is off and the carrier
is enabled
VpNST(NoiseOn) Signal voltage of the noise standard at baseband when the additive noise is

enabled and the carrier is off

VssB Peak voltage of the single-sideband tone

vi
1 Introduction

In this document, we provide the total (or combined) uncertainty of phase modulation fPM) and
amplitude modulation (AM) noise measurement of the NIST portable noise standard calculated
from the individual measurement uncertainties. NIST provides an on-site calibration service
(77 BSC) of the accuracy of PM and AM
noise and noise floor of the measurement systems at
the customer's site by use of two separate portable PM/AM secondary noise standards [1]. One
noise standard is for carrier frequencies of 5 MHz, 10 MHz, and 100 MHz, designated as model
510100 [2], [3], and the other for carrier frequencies of 10.6 GHz, 21.2 GHz, and 42.4 GHz,
designated as model 102040 [4].

The PM and AM noise of the noise standard are usually measured twice at NIST, once before
sending it to the customer (the "out" measurement) and a second time after receiving it back (the
"in" measurement). The customer uses its PM/AM noise measurement system and measures the
calibrated noise level of the NIST secondary noise standard. Customer results are compared with
the average of the "in" and "out" NIST.
measurements If the results are within the
at

measurement uncertainty then the calibration is certified and a "PASS" calibration report is
issued. When there is significant discrepancy in the results, the customer is advised to recheck
the measurement configuration and re-measure. If the customer results cannot be achieved

within the measurement uncertainty, a "FAIL" calibration report is then issued.

The purpose of this document is to describe the calibration uncertainty for the NIST PM/AM
noise standard. The document is organized as follows. In Section 2, we first introduce the basic
concepts of AM and PM noise. We briefly discuss the characterization of AM and PM noise in
Sub-section 2.1, and an overview of different types of signal modulation is presented in Sub-
section 2.2. Section 3 describes the working principle of the PM/AM noise standard, and in
Section 4 the experimental set-up for calibrating the standard is presented. In Section 5 a
detailed step-by-step calibration procedure is provided. The measurement equation, a detailed
uncertainty calculation and the uncertainty budget for the noise standard are presented in
Sections 6, 7 and 8, respectively. Finally, in Section 9, a brief discussion for validating the
calibration results is presented. This document is supported by three appendices. Appendix A
provides verification that the baseband measurement equation is equal to the original definition
of single sideband (SSB) PM noise at the radio frequency (rf). Measurement of different
correction factors involved in the calibration of the noise standard isdescribed in Appendix B.
Finally, Appendix C presents a calculation for the error of a signal power measurement in the
presence of background noise

1
2 AM and PM noise - Basic concepts

In this section we briefly describe the basic concept of phase-modulated (PM) noise and
amplitude-modulated (AM) noise and their characterization in the frequency domain [5-8]. We
also discuss the basic theory behind the PM/AM noise standard.

2.1 Characterization of AM and PM noise

In the real world, an oscillator outputs a signal that fluctuates in both amplitude and phase. This
signal can be mathematically represented by

vit) = V,[l + a{t)']cos[a),t + ^(t)'], (1)

where Vq is the peak amplitude, a(t) is the normalized instantaneous amplitude

fluctuation, co^ = Itwq is the angular frequency of the carrier, (jit) is the instantaneous phase
fluctuation, V,,,,^ is the root-mean-square (rms) voltage, and Vq is the carrier frequency.

In the frequency domain, the amplitude stability of a signal is characterized by the power spectral
density (PSD) of instantaneous amplitude fluctuations Sa (f)- given by

Sjf)=^, (2)

where ^cir(/)^'is the mean-square normalized amplitude fluctuation at an offset or Fourier

frequency /(O <f<oo) from the carrier and BW is the bandwidth of the measurement system.
The unit of Sa (/) is 1/Hz. Similarly, in the frequency domain, the phase instability of a signal is

characterized by the PSD of instantaneous phase fluctuations (/), given by

where (a^,.„,^ (/)) is the mean-square phase fluctuation at an offset frequency/(0 </< <») from

the carrier frequency Vq. The unit of 5^ if) is rad"/Hz. Since / ranges from zero to infinity and
since both Sa if) and 5^ if) include the fluctuations from upper and lower sidebands of the carrier,
these two quantities are single-sided double-sideband units of measure.

The PM noise unit of measure recommended by the IEEE [5] is £{f), defined as

£{f)^^^. ,, (4)

2
£{f) includes the fluctuations from only one sideband of the carrier, and hence it is a single-

sideband unit of measure. When !£{f) is expressed in the form 101og[/(/)], its unit is dBc/ Hz,
that is dB below the carrier in a 1 Hz bandwidth. The logarithm is computed to the base 10.

Also, when the integrated PM noise for offset frequencies 0 < /< cxd is less than 0.01 rad", /(/)

can be viewed as the ratio of phase noise power per unit bandwidth in a single sideband to power
in the carrier [See later Eq.(12)].

2.2 Overview of amplitude, phase and single sideband modulation

Before we discuss the working principle of the NIST PM/AM noise standard, it is worthwhile to
briefly discuss different types of signal modulation.

2.2.1 Amplitude modulation

In amplitude modulation, the carrier amplitude varies with the modulating baseband signal while
the frequency and phase are not modulated. The simplest AM signal is produced when the
modulating signal is a sinusoid at frequency co,n [9], [10]. The resultant modulated signal also
called single-tone AM, is given by

(r ) = [1 + m cos COj] cos 0)^t

(5)

where m is the amplitude modulation index. The two sidebands are symmetrical about the carrier
frequency and are equal in magnitude. The ratio of SSB AM power {Pssb-am) to carrier power
{Pc) is

For AM, the resultant vector sum of the sidebands is always in phase with the carrier (Figure 1).

2.2.2 Phase modulation

In the case of phase modulation, the modulating signal leaves the amplitude of the carrier
unchanged. Instead, the phase of the carrier varies with the modulating signal [9], [10]. A single-
tone PM signal can be represented by

V [t) = VQCos[coQt + P sin coj]

3
where P is the phase-modulation index (or peak phase deviation, also denoted by Aippeak)- The
second exponential term in the curly bracket is a periodic signal with period lTdcOm and can be
expanded by the exponential Fourier series

(8)

JniP) is the - order Bessel function of the first kind [11]. For small modulation index (/?«
1),

(9)

\n\ > 1.

Also, from symmetry, J-niJ^ =(-1)" JniP)'-, therefore, Eq. (7) becomes

(10)

This resembles the AM case in Eq. (5), except that in PM, the phase of the lower sideband is

reversed, and the resultant vector sum of the sidebands is always in phase quadrature with the
carrier (Figure 1).

Further, the ratio of SSB PM power {Pssb-pm ) to carrier power is

(11)

As mentioned in Section 2.1, when the integrated PM noise for offset frequencies 0 </< <» is

less than 0.01 rad", U^(f) can be viewed as the ratio of phase-noise power in a single sideband to
the carrier power. This can also be shown analytically by use of Eqs. (3), (4) and (11), as
follows:

SU-PM if)
(12)
^ ' 2 ' 2BW 4BW \ P^
BW=lHz

when

4
 ^2js,(/)d/«0.1 rad^,,,. (13)

However, the practical values for the lower and upper integration limits, fi and fu, are
respectively the reciprocal of the measurement time and system half-bandwidth.

The condition of Eq. (13) follows by defining a higher order correction, , as a normalized

departure of Eq. (11) from the small angle modulation approximation given by

(14)

Then the error (^^^ -1) in X{f) for 0. 1 is 0.3% (0.01 dB) or less.

2.2.3 Single-sideband modulation

In this Section, we discuss briefly single-sideband modulation, as this is necessary to explain the
working principle of the PM/AM noise standard. A single-tone upper SSB signal is represented
by
V (r) = cos 6^0? + y,,^ cos ( CO,, + )t

1 +
(15)

V,Re

AM PM

where Vssb is the peak voltage of the single-sideband tone at (ftb+ (t^n)- Now comparing Eq. (15)
with (5) and (10), the second term within the square bracket represents AM, and the last term
represents PM. Therefore, the SSB signal can be envisioned as a combination of four signals of
equal amplitude; two of them consist of pure AM, and the other two consist of pure PM. For a
SSB signal, half of the power is present in AM
and other half is present in PM. Eq. (15) can also
be explained in other words as follows: the two lower sideband terms (^""'') are 180 out of

phase, and thus cancel, and the upper sideband terms ) are in phase, and thus add

coherently, resulting in a SSB modulation.

Now if instead of a single tone, there is additive white noise, the theory of single-tone SSB
modulation can still be applied [12], [13]. Consider an upper noise sideband at "Vo+ /" with
associated noise power ;^ in 1 Hz bandwidth. It will produce upper and lower PM and AM noise
5
 2
sidebands, each having average power y /4. If there is a similar amount of uncorrelated white
noise power in the lower sideband at "Vq-/", it also will produce four signals, two AM and two
PM of average power y^lA. The contribution from the lower and upper sidebands produce on
average equal amounts of PM and AM noise; however, the phase difference between them varies
randomly with time.

Figure Phasor diagrams of the amplitude, phase and SSB modulation. LSB is the lower
1.

sideband and USB is the upper sideband. The gray vector indicates the resultant of the carrier
and sidebands.
3 Description of the noise standard

The block diagram of the noise standard at 5 MHz, 10 MHz, and 100 MHz is shown in Figure 2.
The detailed description of this noise standard can be found in [2]. The noise standard has two
outputs, one clean unmodulated reference and another modulated. The powers of the two output
signals are approximately +15 dBm for each carrier frequency.

5 MHz Phase
Shifter
_^ +18 dBm
Power
Carrier Reference
10 MHz Divider
Off T
5 MHz 2X
Coupler +15
On
Modulated
100 MHz
0.
Noise Off
50 ii
External
Input
f
5 MHz BPF

1-900 MHz
Noise Source 10 MHz BPF

100 MHz BPF

Figure 2. Block diagram of 510100 PM/AM noise standard, which selects one of three
frequencies of 5 MHz, 10 MHz, and 100 MHz. BPF is the bandpass filter

A calibrated level of bandpass-limited Gaussian noise "Modulated Output" of the

is added to the
standard to create equal levels of PM and AM
noise. The level of added noise is selected such
that it satisfies Eq. (13), and is at least 40 dB above the noise floor of most measurement
systems. This calibrated noise level is typically constant in magnitude to better than ± 0.5 dB for
Fourier frequencies from dc to 10 % of the carrier frequency. For the 5 MHz standard, the
calibrated noise level is approximately -110 dBc/Hz with a 3.5 MHz bandwidth. The residual
PM and AM noise between these two outputs is exceptionally low [2]. Typically, the differential
PM noise between the two output signals is less than -190 dBc/Hz for Fourier offsets of 10 kHz
and above.

7
The modulated output has precisely equal PM and AM noise, since there is no phase coherence
between the signal and the noise, as described in 2.2.3 and [12]. The power spectral density

(PSD) of the SSB PM noise, [/(/)], and AM noise, [1/2 Sa(/)], generated by the noise standard is

given by

1 P^D[Noise (K,-/)] 4- P5D[Noise Vq+D]

(

^[f)=-Sa[f)= An '
^^^^
2 4 Carrier Power

when Eq. (13) is satisfied [2]. One factor of two in the denominator arises because half of the
additive noise power appears as PM and half as AM. The other factor of 2 is due to the equal
distribution of noise power between the upper and lower sidebands.

^(f ) is constant with frequency from dc to approximately the half-bandwidth of the filter,

assuming that the added noise is constant in amplitude from Vq -/ to Vq +/.

8
4 Measurement setup for calibration of the PM/AM noise standard
To measure ^{f) at multiple offset frequencies, /, a digital fast Fourier transform (FFT)
spectrum analyzer is used to take advantage of its speed, high dynamic range and flexibility in
measurement bandwidth. To measure beyond the upper frequency capability of the FFT, a
heterodyne down-converter is used to convert the output of the PM/AM noise standard to an
intermediate frequency (EF) that can be conveniently measured by the FFT analyzer.

Ref.
AM/PM Noise AMr>
Calibration 50 Q -33 dB
Standard i—WV— -
Mod. Optional
RF
Amp
Double
Offset Spectrum
Balanced Analyzer
Signal
Mixer
Generator "-PF
LoT
+-vy/\H-
-3 dB
+10 dBm

Figure 3. Block diagram of a heterodyne measurement system used to calibrate the 510100
PM/AM noise standard

The block diagram of the heterodyne down-converter used to calibrate a PM/AM noise standard
at NIST is shown in Figure 3. It is composed of a double-balanced mixer (LO level +7 dBm), an

offset signal generator, and a low-pass filter (LPF). The modulated port of the noise standard is
connected to the reference frequency (RF) mixer port via a large attenuator. The large attenuator
at the RF port ensures that the signal from the PM/AM noise standard is far below the saturation
point of the mixer and that it is operating as a linear frequency down-converter. The small
attenuators on the local oscillator (LO) and IF ports are used to stabilize the impedance of the
mixer. A LPF is used after the mixer to reject the sum of the two mixed frequencies and pass the
difference signal.

9
5 Calibration procedure at NIST

The offset generator and the PM/AM noise standard are turned on and allowed to stabilize for at
least24 hours. The offset signal generator is connected to the LO port of the down-converter
and adjusted to +10 dBm. If possible, the offset generator and the PM/AM standard are also
locked to a common frequency reference. The reference port of the PM/AM noise standard is

terminated with 50 Q, and the modulated output port is connected to the highly attenuated RF
port of the down-converter. The noise and the carrier of the PM/AM noise standard can be
independently enabled. This allows the carrier and the noise to be independently characterized,
utilizing the full dynamic range of the FFT device.

A typical calibration consists of a set of individual calibration points at different offset

frequencies from the carrier. This set of offset frequencies is typically logarithmically spaced, 3
to 5 points per decade, spanning from 1 Hz to 10 MHz or a maximum of 10 % of the carrier
frequency. For each offset frequency, /, in the calibration set, four measurements are made and
used to calculate the phase noise generated at that offset frequency. These four measurement
steps are graphically shown in Figure 4 and described in the following four sub-sections.

5.1 Measurement of the carrier power via down-conversion from the upper
sideband

The carrier power of the PM/AM standard is turned ON and the noise is turned OFF. The
frequency of the offset generator is set to The carrier at Vq from the PM/AM standard and
Vo+/
the signal from the offset generator at Vo+ /are mixed in the down-converter creating a baseband
beat at frequency /. The power in the baseband signal at / will be proportional to the carrier
power of the PM/AM noise standard
at Vq. The FFT analyzer range is set to auto to select the
maximum dynamic measurement of the signal power at / is made and recorded in
range, and a
units of V^rms, Utilizing a Flattop window. Also a measurement of the signal-to-noise power ratio
(SNR) between the beat signal and the background noise is recorded. This measurement is
graphically shown in Figure 4(a).

5.2 Measurement of the carrier power via down-conversion from the lower
sideband

Similar to the previous step, the carrier power of the PM/AM standard stays ON and the noise is

OFF. The frequency of the offset generator is set to Vo - / The carrier at Vq from the PM/AM
standard and the signal from the offset generator at Vq - / are mixed in the down-converter,
creating a baseband beat at frequency / The power in the baseband signal at / will be
proportional to the carrier power of the PM/AM
noise standard at Vq. The FFT analyzer is auto-
ranged for maximum dynamic measurement of the signal power at / is made and
range, and a
recorded in units of V^„K by use of a Flattop window. Also, a measurement of the signal-to-
noise ratio (SNR) between the beat signal and the background noise is recorded. This
measurement is graphically shown in Figure 4(b).

10
5.3 Measurement of the noise power spectral density down-converted from
both the upper and the lower sidebands

The carrier power of the PM/AM standard isOFF and the noise is turned ON. The
then turned
frequency of the offset generator The double-sideband noise spectrum about Vo is
is set to Vq.

down-converted to baseband by mixing with the offset generator at Vq. The PSD of the baseband
noise at/ is proportional to the PSD of the double-sideband noise of the PM/AM noise standard
about Vq. The FFT analyzer is auto-ranged for maximum dynamic range, and a measurement of
the down-converted PSD at / is made and recorded in units of V^nn/Hz, by use of a low-leakage
Manning window 114]. This measurement is graphically shown in Figure 4(c).

RF Signal Baseband

(a) 1 * t I
SNR A
I— ^ I

^0 %+f /

vLM)
(b)i

I
— I »
/

\ \

(d)|

V,-f V, v,+f f
Frequency [Hz] Offset Frequency [Hz]

Figure 4: This diagram graphically represents the four measurement steps involved in

calibrating the PM/AM noise standard. In (a) and (b), the down-converted signal power at 'f
'
is

proportional to the noise standard carrier power at Vq.

11
5.4 Measurement of the noise floor power spectral density

The carrier power of the PM/AM noise standard is turned OFF and the noise is turned OFF. The
frequency of the offset generator kept at Vq. Next, a noise floor measurement of the down-
conversion process is made. The FFT analyzer must NOT be auto-ranged, and the range setting
should be identical to that used in Section 5.3. Keeping the auto-range at the same level ensures

that the noise floor of the FFT analyzer is measurements 5.3 and 5.4. The PSD at
the same for
offset frequency / is measured in units of V\m.JHz, utilizing a Banning window. The PSD of the
noise floor at /will contain noise contributions from the down-converter components, such as the
mixer, IF amplifier, the noise of the offset generator, and the input noise of the FFT spectrum
analyzer. This measurement is graphically shown in Figure 4(d).

The measurement procedure described above is repeated at least three times for each carrier
frequency. All measurements are made with a single one-time connection to the measurement
system, and the same analyzer is used for all measurements. The absolute gain or frequency
response of the down-converter and the spectrum analyzer are unimportant, because they are
canceled due to the ratiometric relationship between the measurement of the carrier power and
additive noise. This same characteristic also removes any inconsistency resulting from voltage
standing wave ratio (VSWR) mismatch or connector interface non-repeatability.

The noise-voltage data combined with the beat-power measurements described above
are
according to the baseband measurement Eq. (18) to yield the relative power in either AM, 1/2
Sa(f), or PM noise.

12
6 Measurement equation

£{f) in Eq. (16) represents a ratio of power measurements performed at radio frequency (rf). A
higher order correction, AT//, due to the small angle modulation approximation is introduced and
Eq. (16) is rewritten as

^ ^^^[Noise(Vo-/)]+P5D[Noise(F,+/)]
£[f) = ls " (f)= (17)
2 K„ 4 Carrier Power

The correction factor, Kp, is introduced only to propagate the error due to the approximation in
the measurement equation. A value of unity will be used.

The actual measurements described in the above section are made at baseband by use of a down-
converter. The following equation is used to calculate the AM or PM noise generated from
these baseband measurements:

£if)
\J
= -S a if) = - ^
(18)
/ ^ \ A
-J J
J/' X/' V V

^woMfO«(/) the additive noise down-converted from both the upper and lower (v,j±/)

sidebands to baseband. is the carrier power down-converted via either the upper or

lower sideband to a baseband beat of /. Kj^i, Krf, and Kbw are correction factors needed to
account for nonlinearity, frequency response, and bandwidth errors in baseband
the
measurement. Appendix A shows the equivalence of Eqs. (17) and (18), and Appendix B
describes the techniques to measure these correction factors.

When the bias due to background measurement noise is removed, Eq. (18) can be written as
follows:

PSD V.NoiseOff,M (/)

PSD '^NoiseOn,M (/)J
1-
PSD V,NoiseOn , M (/)
(19)
p + v2 + .M\J L_
y Beat-.M ff)
fy2 \J )^^Beai if\\\\
)
SNR

^NoiseOffMif) SNR are respectively the noise floor of the measurement system for the
additive noise measurement and the signal-to-noise power ratio of the carrier measurements.

13
The same SNR is assumed for both VBeat- and VBeat+ measurements. The logarithmic form of Eq.
(19) is given by:

PSD[vLeojr.Af)S
101og^(/) = 101ogP5D[v^_,,„^(/)] + 101og 1

V dBc/Hz (20)

-101og(vi.,,„^ (/) + V^L.>; (/))-l01og 1

The first term is the measured noise level with noise source ON, the second term is the bias on

the measured noise due to the noise floor (measured with the noise source OFF), the third term is
the average of the measured carrier power down-converted via both the upper and lower
sidebands, the fourth term is the bias on the carrier power measurements due to the background
noise, and the last term contains the correction factors.

14
7 Estimation of uncertainty

Normalizing the PSD to a 1 Hz bandwidth and rewriting the measurement Eq. (19) gives

^NoiseOn.M (/) ^NoiseOff.M if)

^(/) = (21)
p iy'-
^
V
if\ + V'^
Beat-, M \ J 1^ ^ lieat + M {f)\\\
\J )
L
.

SNR

From the Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement
Results [15] [16], the equation for the combined uncorrelated uncertainties (wc) for a
measurement result, y, is written as

y— f {x^ , Xj ,
Xf^f )

a/
ir(x,). (22)
ydx^
Sensitivity
Coefficients

In terms of fractional uncertainties Eq. (22) is

Total ith Component

Fractional ,, , Fractional
Multiplier
Uncertainty Uncertainty

iiUy) a/ X II' (x)

2 2
.=1 J y
'th Component (23)
Multiplier
A ^ Fractional
Uncertainty

(=1 y

cr^ is used here to represent the fractional variance. The combined uncertainty for is

therefore

' ay ay
NoiseOn.M
" ^ NoiseOnM )
+
3VNoiseOff
" {^NoiseOff ,M )
J V J

(24)
«'(^^z.) +

d/
+
dV

15
The individual multipliers of (23) are

V,NoiseOn. M 4y,NotseOn. M
(25)
dV iv-
\
* Nois.eOn.M ^ NoiseOff . M

V NoiseOJ] .M
(26)
\^ ^ NoiseOJf M .
J NoiseOn , M -VNoiseOJf
' , M

K NL K BW
(27)

^ Beal-,M 4V,Beat-, M
(28)
\^ ^ Heal-.M J Beal-.M ^ ^ Beui+.M

^
4V^
Eeai+,M
(29)
av ^ Becil-.M ^ ^ Becil+.M
\ J

V SNR- 1

(30)
dSNR (SNR-\y

The multipliers are simplified by making the following assumptions:

VNoiseOn. M »V
"^-^ * NoiseOff .M

VBecii+.M -V
^ — lieai-.M * (31)

SNR » 1

The combined fractional uncertainty, (7c, obtained by combining the individual fractional
uncertainties, whether arising from Type A evaluation or Type B evaluation [15], is

A Z I I
NoheOn ^ Heat- ^ Beat +
(32)

where riset is the number of repeated measurement sets in a given calibration. (7^ is the
uncertainty characterizing the long-term reproducibility of the PM/AM noise standard between
calibration cycles. This includes tiie combined effect of temperature variations, connector
mismatch, long-term variations of noise or carrier power, effects due to shock during shipping,
and other unknown effects. GNoiseOn and Gseat include uncertainties due to the average of a
random process given by ,

16
 2 — j-rr^ — (33)

2 _ 2 2
^Bea/ ~ ^It-FFTAve ,

'^iK (34)

iVA?o/\se and Nbccv are the number of FFT averages used for the noise and beat measurements
respectively. Gsr is the short-term repeatability of the PM/AM noise standard during a single
calibration. This includes the variation due to temperature fluctuations and variability in the

insertion loss of the internal relays used to enable and disable the carrier and noise source for the
calibration. Appendix C presents a derivation of (32) describing the error of signal power
measurements in the presence of background noise.

The expanded uncertainty (U) is equal to toe, where k is the coverage factor and is chosen to be 2
for a 95.45 % level of confidence [15].

17
8 Uncertainty budget

As stated in the Introduction, the AM

and PM noise of the noise standard are usually measured
two times at NIST, once before sending it ("out" measurement) to the customer and a second
time after receiving it back ("in" measurement). The combined uncertainty includes the
contributions of different estimated errors associated with the noise-standard calibration. In
practice, all estimated errors are expressed in the same domain [linear (%) or log (dB)] for the
calculation of combined uncertainty. Typically, the final measurement results for PM and AM
noise are reported in the log domain (dBc/Hz). However, in this calibration, not all errors are
available in the same domain. For example, the uncertainty due to a FFT average is statistically

calculated in the linear domain, whereas the uncertainty in the linearity of the FFT analyzer is

provided by the manufacturer in dB. Other errors associated with the noise standard and/or
calibration that are experimentally measured can be calculated either in dB or %. We chose to
write the measurement and uncertainty equations in the linear domain and convert any errors
expressed in dB to the linear domain. When converting dB uncertainties to linear, the upper
and lower confidence range can be unequal for larger en-ors. In this document, we will convert
all the dB errors to their linear values and consider the largest value if they are unequal for the

uncertainty calculation. Certain assumptions concerning distributions in dB cannot always be

properly converted to linear, and vice-versa.
The combined uncertainty is calculated entirely in terms of linear fractional uncertainty and the
final expanded uncertainty (U) result is converted to dB. The uncertainty budget for the
calibration of the PM/AM noise standard is given in Table 1.

Table 1 - Uncertainty budget: PM/AM noise measurement of the noise standard

Sources of Error Estimated Effect Type Distribution Divisor Standard Symbol

Fractional Fractional
Error, % Uncertainty
(CT), %
Number of FFT averages 1.0 Random A Normal 1 1.0 (^N-Ff'TAve

for noise measurement

=
A^A,„„, 10.000
Number of FFT averages 0.0 Random A Normal 1 0.0
for beat measurement
A^«„„= 100,
SNR= 1000
Uncertainty in the 1.2 Systematic B Rectangular V3 0.7
estimated measurement
bandwidth for the PSD
function. (0.05 dB)
Frequency response of the 2.3 Systematic B Rectangular V3 1.3

down convertor (includes

offset signal generator
and mixer) at rf

FFT and mixer linearity at 4.7 Systematic B Rectangular V3 2.7 Oni.

baseband
Small angle modulation 0.0 Systematic B Fixed value 1 0.0 Op
approximation. yS = 0.02
Short term noise standard 2.3 Random A Normal 1 2.3 OSR
repeatability
Long term noise standard 7.2 Systematic B Rectangular V3 4.1 Olr
reproducibility

18
By use of Table 1 and Eq. (32) the expanded fractional uncertainty for values shown in this
example is ± 11 .5 % or ± 0.5 dB for User equal to 6.

It should be noted that this calibration standard creates equal amount of AM and PM noise.
When this type of standard is used to evaluate AM/PM measurement systems, the AM-to-PM or
PM-to-AM conversions in the detector cannot be ignored 117]. In a typical PM noise
measurement, AM-to-PM conversion can typically be 15 dB or worse leading to an error of 0.3
% (0.1 dB).

19
9 Validation of the measurement

To further validate the cahbration method, we use two different approaches. In the first

approach, the noise of the standard is measured with different PM noise measurement systems

and with different measurement techniques, for example, a direct-digital phase noise
measurement system (DDPNMS) [18], or a photonic delay-line measurement system (PDLMS)
[19]. The agreement between the results of the DDPNMS and/or PDLMS and the calibrated
result validate the calibration method. The second approach is to calibrate a phase noise
measurement system by use of a calibrated noise standard and compare the results with that
obtained by other calibration methods such as a single-sideband modulation, PM/AM
modulation, beat frequency, or static phase shift [7], [20]. All of these methods should give
identical results within the measurement uncertainty.

10 Acknowledgement '

The authors thank Fred Walls for useful discussions and thoughtful comments on this

manuscript, and Tom Parker, Mike Lombardi, David Smith, and Danielle Lirette for their helpful

suggestions in the preparation of this manuscript.

20
 Appendix A. Verification of the baseband measurement equation

The phase noise, £{f) and amplitude noise, Sj(f) generated by the PM/AM noise standard is

given by

X
^(/>-5.(/) =
1

2
, ,
— 1

AKp
P5D[Noise(v,-/)l +P5D[Noise(F„+/)l
^
; p ^

Carrier
^.

Power
[dBc/Hzorl/Hz] (Al)

The noise represented by Eq. (Al) is not measured directly at rf, but rather indirectly via down-
conversion to baseband. The following equation, called the "measurement equation," describes
the measurement at baseband:

^NoiseOn if)
1 1

(A2)
4 ^NL^KF^HW^fi ^UniiAw if)

A proof follows that shows that the baseband measurement Eq. (A2) is equivalent to the
definition (Al) when appropriate approximations and correction terms are defined. V^,,,,^^^
(/)
the measured additive noise down-converted from both the upper and lower (
v^,
+/ ) sidebands

to baseband. Vg"^,,,^,,,
(/) is the average of the measured carrier power down-converted via both

the upper and lower sideband to a baseband frequency of /. BWEst is an estimate of the
measurement bandwidth used to calculate the PSD function. Knl, Krf, Kbw and Kfj, to be defined
later, are correction factors to account for nonlinearity, frequency response, bandwidth and small
angle approximation errors in the baseband measurement.

The simplified function of the heterodyne down-converter is described as follows. An offset

signal generator is connected to the LO port of a double-balanced mixer, and the PM/AM noise
standard is connected to the RF port of the mixer via a large attenuator, ensuring linear
conversion from the RF port to baseband. The gain of the down-converter for a given LO
frequency (Vlo) and RF frequency (Vrf) to a baseband IF frequency {f= \Vlo- ^rf\) is

G(y,o ' V- '

le^'^l) = 8VSWH§,F (\^L0 - ^HF 8 KF
\) LO ' V )^LVZ.
(A3)

where, g's are different gain terms. gyswR represents any mismatch error due to VSWR [21],

gifif) represents the baseband frequency response at IF, 8rf(^lo^^rf'^ represents the rf

frequency response of the down-converter at its two operating frequencies, and the 8lvl^^^
represents the response of the down-converter at a specific power, P, at the RF port.

21
When the signals from the offset generator and the noise standard are mixed, two sidebands at

± / are translated to a baseband frequency at / The folded double-sideband down-

conversion is represented by Eq. (A4) and is shown graphically in Figure 5 [1 1].

^/Lw(/) = G^(v/„,v„ -/,P)V;^,,(v„ -/)-f-G^(v,,Vo + /,P)V4,,(v/o +/), (A4)

where Vpnst is the signal voltage of the PM/AM noise standard at rf.

RF Signal ^ f i, ^
Baseband

V,-f V, V,+f f
Frequency [Hz] Offset Frequency [Hz]

Figure Graphical representation of the baseband calibration measurements. The red and blue traces
5.

represent the ojfset generator and the PM/AM noise standard, respectively. The green spectrum is the
down-converted baseband signal. The dark gray trace is the noise floor of the measurement system. The
dotted lines indicate the down-conversion gains, where the light gray dotted line indicates the undesired,
negligible cmd hence ignored, down-converted image.

By use of Eq. (A4) the numerator of Eq. ( A2) can be written as

NoiseOn)
(v^ + f). (A5)

VpNST(NoiseOn) reprcscnts the signal voltage of the standard when the additive noise is enabled and
the carrier is off. Pnoise is the power level at the RF port of the mixer during the noise

22
measurement. The upper and lower sideband gains are factored out into a single term and
simplified by a first order Taylor expansion when both sidebands are approximately equal as

follows:

c(x+ y) -ax + by
ax + by
c = (A6)
x+y
a+b
c = + O when x ^ y,

where O contains the higher-order terms of the Taylor expansion. Therefore,

'G\v„v,-f, P„,J+G\v,,v, +/, /^_)

+0 V.
FNST(Noisenn) '(NiiiscOn)
(A7)

Similarly, in the denominator of (A2) can be rewritten, when the contribution from
the unwanted and negligible down-converted image is ignored, as

_ (Vq - /, Vq , P,,.„„ ) + G' {V, + f,v„ P,,.,.,,.

^Bra/Alt' (/) FNST(CaiTierOn
(A8)

where VpNSTiCamerOn} represents the signal voltage of the standard when the additive noise is off
and the carrier is enabled, and Pcamer represents the power level at the RF port of the mixer
during measurement of carrier power. Substituting the numerator and denominator in Eq. (A2)
with Eq. (A7) and Eq. (A8) gives

^PNST{NoiseOn) / ) ^PNST(NoiseOn) /)
1 K
(A9)
4 ^NL^RF^BW^fi ^PNST[NoisL'On) ^

All common and similar terms are collected in the variable K, given by

(AlO)

This clearly shows that the baseband gains as well as VSWR mismatches cancel for the
measurement Eq. (A2).

Finally, the four correction factors K^i, KRf, Kgw and Kp are defined. Knl defines the change in

gain that occurs due to nonlinearity in the down-converter and FFT analyzer due to a difference
in power level between noise and carrier power measurements:

23
 (All)

K^r is the ratio of the effective power gain between noise and carrier power measurements.
Ideally, this number should be unity;however, it is affected by the non-flat amplitude response
of the offset generator. It also contains the asymmetrical frequency response of the mixer at its

LO and RF ports as well as the higher-order terms of the Taylor expansion of example (A6):

§If (^0 ,Vo-f) + 8i(Vo,Vo + f) + 20

(A12)

Kbw is a correction for an imperfect power spectral density (PSD) function given by

K (A13)

where BWAcmai is the true measurement bandwidth.

Using Eqs. (Al 1), (A12) and (A13) in Eq. (A9), we see that Uiif) at the baseband measurement

equation is equal to the original definition of£{f) at rf (Al) as follows:

V, PNST(NoiseOn
PNST(NoiseOn)^ 0

1
BW Acliiiil

4K (A14)
,2
1
PSD PNST{Noi.ieOn)^ 0 (Vo+/)
PNSTiNoiseOn)''-'' 0

4K, yp.SriCarr.r0.^y^

24
 Appendix B. Measurement of correction factors, K^l, Km,^ K^w and Kp

(i) Set-up for measuring nonlinearity of the down-converter and the FFT analyzer -

As discussed earlier, K^l is defined as the change in gain that occurs due to nonhnearity in the

down-converter and FFT analyzer due to a difference in power level between noise and carrier
power measurements and given by Eq. (All). The carrier power is measured with additive noise
off, while the noise-power measurements are made with the carrier off.

Noise Standard/
Synthesizer

e Power
-33

Double
dB

RF
Optional
Amp
Spectrum
Meter
Balanced
Mixer (g)-&gH LPF
Analyzer

LO

-3 dB
Offset
Signal
Generator

Figure 6. Measurement setup for the estimation ofK^Land Krf.

The measurement system used to measure uncertainty due to the FFT and the mixer nonlinearity
at baseband is shown in Figure 6. The frequency of the offset signal generator is set at Vq + /and
the noise standard at Vq. A portion of the signal from the noise standard is coupled out and
connected to a NIST-calibrated power meter with very low noise floor (-60 dBm). First, the

carrier power (Pcamer) and noise power (Pnoise) of the standard are measured with the power
meter. The noise standard is then replaced by a rf synthesizer at frequency Vq and power equal to
Pearlier- The corrcspondiug down-converted power (Pc-Bear) is measured on the FFT analyzer by

use of a flattop window. Next, the synthesizer power is adjusted to equal the integrated noise
power of the noise standard (P„o,^e), and the down-converted power {PN-Bead is measured on the
FFT analyzer. From these measurements Knl can be calculated as:

^ 2
OLVL\iPimise J _ 'PN-Beal /IPC-Beal
-*

(Bl)
SlVL i^ainier )

25
These measurements are repeated for all the specified offset frequencies ( +/) at each of the three
carrier frequencies. Because the mixer is operating in linear mode, K^l should ideally be equal

to 1. The maximum observed error from unity, t?vL, can be used to construct a rectangular
distribution for Knl- Therefore,

^NL^'^^^='^±(^NL , (B2)

where Gnl is the fractional uncertainty of K^l.

(ii) Set-up for measuring frequency response of the down convertor at rf - Krf

Krf is the ratio of the effective power gain between noise and carrier power measurements and
given by Eq. (A 12). The experimental set-up for determining KRf is thesame as shown in Figure
6. In this case, the beat power, Pc-BeatiVLO, ^rf), is measured for four different LO and RF
frequency combinations. For all four measurements, the power of the synthesizer is adjusted so

that its power stays constant at Pcamer- KRfCm then be calculated from

^C-Becil ^C-Beal
(B3)
^C-Beat

These measurements are repeated for all the specified offset frequencies (+/ ) at each of the three
carrier frequencies. Krf should ideally be equal to 1 . The maximum observed error from unity,

5rf can be used to construct a rectangular distribution for Krf. Therefore,

,

^Rr-^±^-^±(^RF , (B4)

where cTrp is the fractional uncertainty of Krf.

(iii) Measurement of PSD correction factor - Kbw

Kbw as defined in Eq. (A 13) is power spectral density function.

a correction for an imperfect
Since the EFT is digitally implemented, the resolution bandwidth of measurement and associated
windows are well defined mathematically. Exact details of the EFT implementation depend on
the manufacturer. The EFT manual should be checked for possible bandwidth uncertainty for a
given instrument. Verification of the PSD function can also be implemented by use of a well
characterized filter and a calibrated power meter 114]. In this case, we use Kbw equal to unity
with a fractional uncertainty, (7bw = 0.7 %.

26
(iv) Calculation of small angle modulation correction factor - Kfi

Kp is a correction to the small angle modulation approximation used to calculate J'if ) from a
single sideband power ratio. An effective peak phase modulation P is obtained by integrating the
phase noise and utilized to determine Kji as follows

2js,(/)d/. (B5)

Typically, /u is the half-bandwidth of the band-pass filter in the noise standard.

Kp is chosen to be unity with a fractional uncertainty dp.

'
fj;(^)/r-{j3)
K^=i±{£^-i) = i±\ -1 ±(Tr, (B6)

27
Appendix C. Error of signal power measurements in the presence of
background noise

Consider the fast Fourier transform (FFT) of a signal of amplitude A and frequency / , such that
the signal occurs in a single-frequency FFT bin (no leakage to neighboring bins) with white
uncorrelated noise in that bin (and in every bin). Measurements are repeated A'' times and
averaged. Let the noise in the measurement be represented by n. , a Gaussian sequence of

values with zero mean and covariance . Then if the signal amplitude in the frequency bin of

interest is S- , the average of the measurements is

^=^tiS,+^h)- (CI)

We can regard the expectation value of a limited set of measurement as the simple average:

The expected rms error of the measurement due only to the noise is

(C3)

Given the Gaussian uncorrelated property of the noise, we can write

(/2,«.) = C7^4., (C4)

which is zero unless / = j There

. are N such terms in the sum in Eq. (C3), so

E = — 4ng'=-^. (C5)

>

Thus the measurement can be reported with a statistical error given by

-TS,±^. , (C6)

Suppose, however, that the FFT is squared and then averaged. We then have for the
measurement

The middle terms drop out, because there is no correlation between the signal and the noise. In
the last sum, the average of the noise is cr' by Eq. (C4). The measurement is then

28
 M=-y\s,f+(7\ (C8)

If the noise power is known, it can be subtracted off and the measurement of the average signal
power reported as

M-a'= — y\s\\ (C9)

Apart from the propagation of error that occurs in this subtraction, there will be some error
associated with the measurement itself. This can be denoted by E, where

=1(m-{m))' (CIO)

Writing this out,

E' = (Cll)

There is a rather subtle change in the order in which the operations are performed. If we again
assume no correlation between signal and noise, the terms involving the signal in the above
equation all cancel out or vanish. Then

E'=' (C12)

Evaluating each of these terms one by one, in the first sum there are N terms with / = j. These
will contribute the amount
J_ (C13)

There are additional contributions in the first sum with / ^ j, where there are

N~ - N = N(N -I) terms, each contributing (o"" ) • The first sum therefore contributes a total of

) Ml M — -w
A^(A^-1)
, ^4 (C14)
H ; <7 .

In the second term of Eq. (C 1 2) , there are N terms each contributing a'^ ^
so the second term

contributes

(C15)
N
The last term contributes
(C16)

Collecting all the terms, the error is given by

(T -2a +<7 =- — 1
(7 . (CI 7)
N N N
Thus, we have to estimate the fourth moment. In a normalized Gaussian distribution that is a
function of x, the probability of getting a value x within the increment dx is

29
 P{x)dx (CI 8)
iTtO
The fourth moment is

e -'"'dx^Za' (CI 9)

Thus the error is

2 3a' la'
N N N (C20)
E = ^IjNa-
The measurement of signal power can then be reported with an uncertainty as:

M -a- ±—i=-. (C21)

^|N

Expressing Eq. (C21) in terms of SNR gives

SNR =
M
(C22)
M
SNR

30
REFERENCES
[ 1 ] "Calibration Services User Guide,"
http://ts.nist.gov/MeasiirementServices/Calibrations/pm_am_noise. cfm#pm_am. [Online].
Available: http://ts.nist.g0v/MeasurementServices/Calibrati0ns/pm_am_n0ise.cfm#pm_am.
[2] F. L. Walls, "Secondary standard for PM and AM noise at 5, 10, and 100 MHz," IEEE
Transactions on Instrumentation and Measurement, vol. 42, no. 2, pp. 136-143, Apr. 1993.
[3] F. L. Walls, "Calibration system for determining the accuracy of phase modulation and amplitude
modulation noise measurement apparatus," U.S. Patent 5,172,06415-Dec-1992.
[4] F. G. Ascarrunz and F. L. Walls, "A standard for PM and AM
noise at 10.6, 21 .2 and 42.4 GHz," in

Frequency Control Symposium, 1996. 50th., Proceedings of the 1996 IEEE International., 1996, pp.
852-853."
[5] IEEE, "IEEE Standard Definitions of Physical Quantities for Fundamental Frequency and Time
Metrology —
Random Instabilities," IEEE Std 1139-1999, 1999. [Online]. Available: ISBN 0-7381-
1754-4. [Accessed: 06-Aug-2010].
[6] D. B. Sullivan, D. W. Allan, D. A. Howe, and F. L. Walls, "Characterization of Clocks and
Oscillators," in NIST Technical Note 1337, 1990.
[7] F. L. Walls and E. Ferre-Pikal, "Measurement of Frequency, Phase noise and Amplitude noise," in

Wiley Encyclopedia of Electrical and Electronics Engineering, 1st ed., vol. 12, 24 vols., Wiley-

Interscience, 1999, pp. 459-473.

[8] E. Rubiola, Phase Noise and Frequency Stability in Oscillators. Cambridge University Press, 2010.
[9] S. I. Baskakov, Signals and circuits. Mir Publishers, 1990.

[10] B. P. Lathi, Modem Digital and Analog Communications Systems, International 2 Revised ed.
Henry Hoh & Company, 1983.
[11] M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions: with Formulas, Graphs,
and Mathematical Tables. Dover Publications, 1965.
[12] F. L. Walls, "Correlation between upper and lower sidebands," IEEE Transactions on Ultrasonics,
Ferroelectrics and Frequency Control, vol. 47, no. 2, pp. 407-410, Mar. 2000.
[13] R. P. Scott, C. Langrock, and B. H. Kolner, "High-dynamic -range laser amplitude and phase noise
measurement techniques," Selected Topics in Quantum Electronics, IEEE Journal of, vol. 7, no. 4,
pp. 641-655, 2001.
[14] F. L. Walls, D. B. Percival, and W. R. Irelan, "Biases and variances of several EFT spectral
estimators as a function of noise type and number of samples," in Frequency Control, 1989.,
Proceedings of the 43rd Annual Symposium on, 1989, pp. 336-341.
[15] B. N. Taylor and C. E. Kuyatt, "Guidelines for Evaluating and Expressingthe Uncertainty of NIST
Measurement Results," NIST Technical Note 1297, 1994.
[16] BIPM, "Evaluation of measurement data — Guide to the expression of uncertainty in
measurement." 2008.
[17] E. Rubiola and R. Boudot, "The effect of AM
noise on correlation phase-noise measurements,"
IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, vol. 54, no. 5, pp. 926-
932, May 2007.
[18] J. Grove, J. Hein, J. Retta, P. Schweiger, W. Solbrig, and S. R. Stein, "Direct-digital phase-noise
measurement," in Frequency Control Symposium and Exposition, 2004. Proceedings of the 2004
IEEE International, 2004, pp. 287-291.
[19] E. Rubiola, E. Salik, S. Huang, N. Yu, and L. Maleki, "Photonic-delay technique for phase-noise
measurement of microwave
oscillators," Journal of the Optical Society of America B, vol. 22. no. 5,

pp. 987-997, May 2005.

[20] F. L. Walls, "Method and apparatus for wide band phase modulation," U.S. Patent 4,968,90806-
Nov-1990.
[21] Hewlett Packard, "Fundamentals of RE and Microwave Power Measurements," Application Note,
no. 64-1, 1978.

31
 Technical Publications

Periodical

Journal of Research of the National Institute of Standards and Technology— Reports NIST research and
development in metrology and related fields of physical science, engineering, applied mathematics, statistics,
biotechnology, and information technology. Papers cover a broad range of subjects, with major emphasis on
measurement methodology and the basic technology underlying standardization. Also included from time to time are
survey articles on topics closely related to the Institute's technical and scientific programs. Issued six times a year.

Nonperiodicals

Monographs—Major contributions to the technical literature on various subjects related to the Institute's scientific

and technical activities.

Handbooks— Recommended codes of engineering and industrial practice (including safety codes) devel- oped in

cooperation with interested industries, professional organizations, and regulatory bodies.

Special Publications— Include proceedings of conferences sponsored by NIST, NIST annual reports, and other
special publications appropriate to this grouping such as wall charts, pocket cards, and bibliographies.
National Standard Reference Data Series—Provides quantitative data on the physical and chemical properties of
materials, compiled from the world's literature and Developed under a worldwide program
critically evaluated.

coordinated by NIST under the Law 90-396). NOTE: The

authority of the National Standard Data Act (Public
Journal of Physical and Chemical Reference Data (JPCRD) is published bimonthly for NIST by the American
Institute of Physics (AlP). Subscription orders and renewals are available from AIP, P.O. Box 503284, St. Louis,

MO 63150-3284.
Building Science Series—Disseminates technical information developed at the Institute on building materials,
components, systems, and whole structures. The series presents research methods, and perfonnance
results, test
criteria related to the structural and environmental functions and the durability and safety characteristics of building
elements and systems.
Technical Notes—Studies or reports which are complete in themselves but restrictive in their treatment of a subject.
Analogous monographs but not so comprehensive in scope or definitive in treatment of the subject area. Often
to
serve as a vehicle for final reports of work performed at NIST under the sponsorship of other government agencies.
Voluntary Product Standards— Developed under procedures published by the Department of Commerce in Part 10,
Title 15, of the Code of Federal Regulations. The standards establish nationally recognized requirements for
products, and provide all concerned interests with a basis for common understanding of
the characteristics of the products. NIST administers this program in support of the efforts of private-sector
standardizing organizations.

Order the following NIST publications—FIPS and NISTIRs—from the National Technical Information Service,
Springfield VA 22161.
Federal Information Processing Standards Publications (FIPS PUB)— Publications in this series collectively
constitute the Federal Information Processing Standards Register. The Register serves as the official source of
information in the Federal Government regarding standards issued by NIST pursuant to the Federal Property and
Administrative Services Act of 1949 as amended. Public Law 89-306 (79 Stat. 1 127), and as implemented by
Executive Order 1 1717 (38 FR 12315, dated May 1 1,1973) and Part 6 of Title 15 CFR (Code of Federal
Regulations).
NIST Interagency or Internal Reports (NISTIR)—The series includes interim or final reports on work performed
by NIST for outside sponsors (both government and nongovernment). In general, initial distribution is handled by
the sponsor; public distribution is handled by sales through the National Technical Information Service, Springfield,
VA 22161, in hard copy, electronic media, or microfiche form. NISTIR's may also report results of NIST projects of
transitory or limited interest, including those that will be published subsequently in more comprehensive form.
U.S. Department of Commerce
National Institute of Standards and Technology
325 Broadway
Boulder, CO 80305-3337

Official Business
Penalty for Private Use $300