Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Test Strategies for High-
Frequency Wideband
Vector Analysis
Walt Schulte
Applications Engineer
Microwave and Communications Division
Agilent Technologies
10/16/2012
2
Agenda
• Nonlinear Distortion
• Methods to Characterize Nonlinear Distortion
• Two-Tone Measurements
• Multitone Measurements
• CCDF Measurements
• Noise Power Ratio (NPR) Measurements
• Making Wideband High Frequency Measurements
• Making Vector Based Measurements
10/16/2012
Linearity Concerns
4
Signals can interact
Non-linear power amp
Spectral re-growth
Unusable bandwidth
Need to characterize
H V
H V
10/16/2012
5th order
IMD
Nonlinear Distortion: Intermodulation
7
Amplifier
In Out
(3A-2B)
(2A-B)
(3B-2A)
2nd
harmonics
3rd
harmonics
2B 2A
f A B
3rd order
IMD typical
channel
bandwidth
(2B-A)
3A 3B
A B
2nd order
IMD
A B
(B-A)
Intermodulation
distortion
3rd order
IMD
5th order
IMD
10/16/2012
Non-Linear Distortion: Spectral Re-Growth
Page 8
Pt
Pi
f
Spectral
‘Re Growth’
Non-Linear
Effect
8
10/16/2012
IMD and Spectral Re-Growth
9
Pt
f
A B
A+
A
A+
B
B+
B
A+
A–
B
B+
B–
A
Pt
Pi
IMD3
Distortion
TOI or IP3
IMD3
Fundamental
Pt
f
A B
A+
A–
B
B+
B–
A
IMD3
1st, 2rd, 3th, etc.
harmonics mix together
forming IMD
10/16/2012
Two Tone or TOI Test
10
Advantages
• Quick test
• Simple to perform
• Uses common test equipment
• High signal to noise
Disadvantages
• Essentially narrow band
• Requires wideband matching
• Does not simulate correct loading conditions
• Probably needs components to increase source
to source isolation
10/16/2012
PXA performance
spectrum analyzer
PSG
CW signal generators
Traditional Two-Tone Measurements
11
DUT
LPF
Combiner
Isolator AMP
Attenuator
2nd & 3rd
harmonics
IMD
products
+
10/16/2012
Three Tone Testing
12
Advantages
• Reasonably quick test
• Simple to perform
• Uses common test equipment
• Better Intermodulation test
Disadvantages
• Essentially narrow band
• Requires wideband matching
• Open to interpretation
• Averaging required
3rd 5th 7th
Note:
Uses uneven spacing for tones
10/16/2012
>10 Tone Testing
13
In practice, using analog signal sources and combining techniques
proves impractical for the majority of users.
Why is this?
• < +20 dBm is typically maximum out of signal sources, but usually not
able to use this due to distortion due to the source levelling circuits
• Combiners with good port isolation have high attenuation
• Use of filters and isolators creates limited flexibility
• Post combiner amplification either not available or too expensive.
• Cost and physical size of signal sources
10/16/2012
Modern Multi-tone IMD
14
• For wideband components two-tone measurement
results vary depending on tone spacing
• Simulate real-world operating conditions
• Stress device with higher peak-to-average ratio
• Test with multiple phase sets
Why use multi-tone test signals?
10/16/2012
• Well established test procedure
• Common test equipment
• Complicated test setup
• Signal parameters are not easily modified
• Manual tuning
• Difficult to generate random phase sets
• Equipment and capital intensive
Conventional Analog Test Stimulus
15
Disadvantages of analog test approach
Advantages of analog test approach
10/16/2012
Analog I/Q modulation vs. digital I/Q up-conversion
Page 16
Conventional I/Q
modulation -
Analog I and Q
signals are
generated using an
AWG. An I/Q
modulator
generates the IF or
RF signal
Digital up-
conversion –
I/Q modulation is
performed digitally
- either in real-time
(in hardware) or
up-front in software
Mixer /
Multiplier
/ LO
AWG
D/A
D/A
Memory
Memory
Analog IQ Modulator
X
X
+ ~
90
AWG
Digital signal Analog signal
D/A
X
X
+ ~
90
Memory
Memory
X
~
10/16/2012
Vector Test Stimulus up to 1 GHz
17
I Q
DUT
Isolator
Dual LXI arbitrary
waveform generator
(N8241A)
(Differential)
PSG vector
signal generator
(E8267D)
PXA performance
spectrum analyzer
(N9030A)
10/16/2012
Vector Test Stimulus
18
Advantages of vector test approach
• Simple test setup and procedure
• Easily modify signal parameters
• Apply pre-distortion to improve signal quality
• Repeatable and accurate test results
• Save time and capital equipment cost
Disadvantages of vector test approach
• Available output power
• Carrier feed through
• Images
• Relative tone spacing
10/16/2012
• Up to 4097 tones
• Vary tone power
• Change phase settings
• Randomly spaced tones
• CCDF plot
• COM-based API
LAN/GPIB
PSG PXA
N7621A Signal Studio for Multi-tone Distortion
19
…and After
• Improved IMD
suppression
• Correct with additional
devices in the loop
• 80 MHz correction BW
(internal AWG)
• 1 GHz correction BW
(Agilent’s WB AWG)
Before…
Option 203
10/16/2012
Achieving the Right Signal Statistics
20
22 Pv
Average random power?
Noise vs. pseudo noise
Peak to average ratio
CCDF signal statistic
PSG NPR adjustable
Generator seed selectable
20
t
Average
v
Power
% Time
Power
% Time
Power
% Time
Power
% Time
2
cdf1
cdf1log10
CCDF
10/16/2012
Complementary Cumulative Distribution Function
21
Link- CCDF demo video
10/16/2012
IMD products
from DUT
Low IMD reduces
test uncertainty
Tone correction
Minimize test stimulus IMD…
even at the output of an
external power amplifier!
Non-linear distortion measurement
E8267D PSG
Enhanced Multi-tone Measurements
22
DUT
N9030A PXA
10/16/2012
Complicated Loading Scenarios: NPR
23
Multiple signal bandwidths, amplitudes & modulation types
3rd’s, 5th’s, etc… Can add together
Difficult to predict, so measure with NPR test
?
f
10/16/2012
NPR Measurement Overview
24
Notch
Depth
No
tch
Frequency
Amplitude
Broadband Noise
Notch
Depth
Frequency
Amplitude Notch Bandwidth
Spectral Re-Growth
NPR
10/16/2012
Traditional NPR Test Setup
25
LO
RF IF
Up converter
Noise Source
Band stop
filter
Spectrum analyzer
CW signal generator
DUT
10/16/2012
Traditional NPR Test Problems
26
Notch filter shape issues
Depth & bandwidth
Filter Q requirement
Cavity filters
Fixed center frequency
Noise flatness
Test station correlation
Notch
Depth
Frequency
Power Notch Bandwidth
NPR
Band Stop
Filter
10/16/2012
PXA Performance
Spectrum Analyzer
(N9030A)
I Q
DUT
(Differential)
27
Arbitrary Waveform Generator for NPR
Dual LXI arbitrary
waveform generator
(N8241A)
PSG Vector
Signal Generator
(E8267D)
10/16/2012
Using a Vector Signal Generator and an AWG for NPR
28
Repeatable, flat noise power, square notch….
Deeper notch with easily adjusted center frequency
Up to 80 MHz bandwidth in PSG & 1 GHz with the N8241A!
Conventional Analog NPR Stimulus Digitally Synthesized NPR Stimulus
10/16/2012
Signal Studio for Multitone Distortion
Option 204
29
LAN/GPIB
PSG
Digitally Synthesized NPR
Stimulus
PXA
Synthesize NPR stimulus
Vary notch depth
Vary notch bandwidth
Vary notch frequency
Distortion correction
Vary stimulus’ CCDF
10/16/2012
NPR Measurement
30
Amplifier Measurement
Notch Creation
38 GHz!
Using Noise and Band Power Markers
10/16/2012
Arbitrary Waveform Generator Requirements
31
• Wide bandwidth (1 GHz)
• High dynamic range
• Low distortion products
• Flatness
• Repeatability
• IF and IQ generation
capabilities
N6030A (PXI) N8241A (LXI)
10/16/2012
New ARB – M8190A
16 QAM Example with Analog IQ Modulation using
Vector Signal Generator
32 10/16/2012
High-Precision AWG Example: Analog IQ Modulation,
Fc=10GHz
Wideband digital
modulation:
QAM16, 1.76G Sym/s
Fs = 7.2 GHz
with amplitude
correction
EVM=1.17%
33 10/16/2012
High-Precision AWG Example: Digital Upconversion,
Fc= 1GHz (without PSG RF Sig Gen)
Wideband digital
modulation:
QAM16, 1G Sym/s
Fs = 7.2 GHz
with amplitude
correction
EVM=0.89%
34 10/16/2012
Page 35
Vector Modulation Analysis - 89600 VSA Software Use the SAME measurement tool at ALL stages of your block diagram!!!
DSP
Digital (SSI) IF/RF BB (I-Q)
Logic Analyzer Oscilloscope Signal Analyzer
DUT
Wideband High-Frequency Vector Measurement Options
Page 36
Bandwidth
Dyn
am
ic R
an
ge
PXA 160MHz
@78 dB
Wide Band VSA
(PXA + Infiniium)
900 MHz @ 40 dB*
X93204A Infiniium scope
32 GHz @ 40 dB
MXA 25MHz
@78 dB
10/16/2012
Instrument and System Calibration
Page 37
Calibration
Amplitude Flatness Phase linearity
Minimum Error Vector Magnitude EVM
I
Q
Ideal Signal
Measured
signal
θ
Amplitude error
Phase linearity error
The goal is to measure the EVM of the DUT not the
EVM introduced by the measuring system
10/16/2012
0-3.6 GHz low band
3 Hz-26.5 GHz
Input
Cal input
2 dB-step mech atten
μW converters
8.3-14 GHz LO
10.9M
.3M
4.8 GHz LO
RF converter
3.8-8.73 GHz LO
2nd converter
FPGA
300 MHz LO
200 MHz CK
100 MHz CK
ADC
ADC
Switched filters,
F0=22.5 MHz
X1 3.6-13.6 GHz
X2 13.6-26.5 GHz
140 MHz
3.5-26.5 GHz high band
FPGA
140 MHz Front End
Swept IF & 10 MHz BW
& 25 MHz BW (option B25)
25 MHz
966K
303K
79K
9K
Switched filters,
F0=322.5 MHz
140 MHz BW (option B1X)
2 2 6 10 20 30
RF preamp
40 MHz
400 MHz CK
40 MHz BW (option B40)
ADC
F0=250 MHz
F0=300 MHz
F0=322.5 MHz
Linearity
Corrections
Low noise path
μW preamp
YIG filter with
bypass relay
1 dB-step electronic atten
F0= 5.1225 GHz
4 GHz
ASIC
2Gbyte
SDRAM
ASIC
2Gbyte
SDRAM
PXA Simplified Block Diagram
38 10/16/2012
Example of Quality of Magnitude and Phase Corrections
on 140 MHz BW
EVM Results for 138 MHz OBW QPSK Signal vs. Center Frequency in Band 0
39 10/16/2012
PXA Simplified Block Diagram (900 MHz IF Path)
40
3 Hz-26.5 GHz
Input
8.3-14 GHz
3.6-13.6 GHz Path
3.5-26.5 GHz high band
Front End Low noise path
μW preamp
YIG filter with
bypass relay
13.6 - 26.5 GHz Path
Aux IF out
Option CR3 Option MPB
Rear Panel
900 MHz IF BW centered at 600 MHz
10/16/2012
Setting up the Vector Signal Analyzer
Page 44
1. Connect a source to
the PXA
2. Connect the VSA the
scope over the LAN, and
the PXA’s wideband IF
output to channel 1 of
the scope’s input
10/16/2012
Setting up the Vector Signal Analyzer
Page 45
3. Configure the VSA for use
with a downconverter.
10/16/2012
Correcting Downconverter Frequency Response
Page 46
1. Configure your
source for a 0 dBm
CW at the desired
carrier
2. Set VSA averaging
to “continuous peak
hold”
10/16/2012
Correcting Downconverter Frequency Response
Page 47
1. Program your source to
slowly sweep across the
desired band
2. Be sure that the ADC is not
being overdriven during the
sweep.
3. Use an external leveling
loop if necessary
10/16/2012
Applying Frequency Response Corrections
Page 48
1.
2.
10/16/2012
Applying Frequency Response Corrections
Page 49 10/16/2012
Amplitude Corrections
Page 50 10/16/2012
Vector Analysis with Wideband QPSK, APSK, and
SOQPSK Agilent 89601B VSA
Page 51
10 GHz QPSK signal
BW = 900 MHz and
EVM = 1.4%
APSK and SOQPSK
10/16/2012
Summary
52
Nonlinear behavior must be characterized and addressed
Common test signals include two-tone, multitone, and NPR signals
Digital signal generation approach provides repeatability and cost advantages over analog
generation approaches including notch depth, bandwidth, and adjustability
Multiple techniques and measurements non-linear behaviors in the PXA
Using the PXA as a down-converter is a technique for lower cost high-frequency
wideband measurement
VSA software enabled is a flexible platform that allows for easy input of corrections as
well as a platform to make a large amount of vector/modulation quality measurements
10/16/2012
Thank You!
53
10/16/2012