seminar hanodut cover -...
Transcript of seminar hanodut cover -...
安捷倫科技高頻元件量測研討會
時間: 2006年 2月23日地點: 高雄金典酒店
1
安捷倫高頻元件量測研討會
2/23/2006Page 1
Packaging Development Trend of Integrated Analysis
Sung-Mao Wu
安捷倫科技高頻元件量測研討會2/23/2006
Page 2
OutlineDevelopment Trend for PKG
-- PKG Technology Trend-- Why SiP and POP/PIP
Challenges to PKG Integrity Design-- Design Challenges on Simulation, Measurement and Design-- Case I : Effective DK-- Case II : Impedance Control verify by TDR-- Case III: TDR FA Application-- Case IV : Substrate Ball Pad Design-- Case V : PDS Analysis
Integrated Design-- Components of Optimization PKG Design-- Advanced PKG Analysis Flow-- Plan and Actions for PKG Design
2
安捷倫科技高頻元件量測研討會2/23/2006
Page 3
Bio tech / New form chip
IC/module/System DesignHigh speed/high frequency
High thermal / Stress solution
Fab/fablessCopper wafer
Low K material
12” wafer
Naro meter tech
Front/Back end solutionEnvironment friendly
Compact Size
Low power consumption
Integrated SystemLow cost
Short Time to Market
Worldwide strategy
SemiconductorIndustry
Interesting SemiconductorWorld
安捷倫科技高頻元件量測研討會2/23/2006
Page 4
Packaging Technology Trend
Single Chip Package
Stacked Package
Flip Chip
Stacked Die
1995 2000 2005 2010
QFP
Multiple Chip Package
BGA
MCM
FC+WB
System in Package
Laminate 2 & 4 Layer
Build-Up SubstrateMulti-Layer
PCB
Chip
Functional Substrate(Active & Passive Chip)
PiP
PoP
Wire Bond Wire Bond + FC Bond New Interconnection?
3
安捷倫科技高頻元件量測研討會2/23/2006
Page 5
Why SIP ?
Single Chip Solution
SoC IC
Build-up Substrate
Sub-System Module
Board Assembly
Passives Micro-component
Memory IC
SIPCompare to Board Assembly:Compare to Board Assembly:
Performance EnhancementPerformance EnhancementThinner, Smaller, and LighterThinner, Smaller, and LighterLow CostLow Cost
Compare to SOCCompare to SOCLow Cost Low Cost Time to MarketTime to MarketFlexibleFlexible
SIP SIP PlatformPlatform
Die Stacking Platform
Package Stacking Platform
Flash SDRAMASIC
Flash SDRAM
ASIC
ASIC
Flash SDRAM
安捷倫科技高頻元件量測研討會2/23/2006
Page 6
1.4mm
1.2mm
0.8mm
1.0mm
0.5mm
2 die 3 die 4 die 5 die 7 die
Die Stacking Package Trend• Smaller & Lighter Package Size
• High Density Device in Package
4
安捷倫科技高頻元件量測研討會2/23/2006
Page 7
PoP and PiP
Available 2006 2007
Package/ Die Count
Package Thickness
Package Structure
Package/ Die Count
Package Thickness
Package Structure
2 PKG/ 3 Chip 2 PKG/ 3 Chip 2 PKG/ 4 Chip 3 PKG/ 7 Chip
1.6 mm Max 1.4 mm Max 1.2 mm Max 2.0 mm Max
ASIC
Flash SDRAM
ASIC
Flash SDRAM
Flash SDRAM
ASIC
Flash SDRAM
ASIC
1.4 mm Max 1.2 mm Max 1.0 mm Max
2 PKG/ 3 Chip 2 PKG/ 3 Chip 2 PKG/ 3 Chip
W/B Type
F/C Type
PoP
PiP
安捷倫科技高頻元件量測研討會2/23/2006
Page 8
IC Feature Size (um) 0.25 0.18 0.13 0.09 0.065
• Wire Bond pad Pitch (um)
(Single-In-Line)
• Wire Bond Pad Pitch (um)
(2 Row Staggered)
• Wire Bond Pad Pitch (um)
(Tri-Tier)
•Wire Bond pad pitch (um )
( Quad-Tier )
60 50 45 40 35
80/40 70/35 60/30 50/25 40/20
90/45 80/40 70/35 60/30 50/25
100/50 90/45 80/40 70/35 60/30
Leading-Edge Fine-Pitch Capabilities
In-Line: 45 um; Staggered: 60 um; Tri-Tier: 70 um ; Quad-Tier: 80um
Low k & Copper wafer capabilities available.
Fine Pitch Wire Bonding
5
安捷倫科技高頻元件量測研討會2/23/2006
Page 9
Focus Packages - Bumping, WLCSP, FCBGA, SIP, SCSP and modified LF package
Small & Light - Thin ThicknessThin Thickness in Wafer, Substrate, Package- Fine PitchFine Pitch in Wire Bonding, Flip Chip Bond and
Solder Ball- High DensityHigh Density by Stacked Die, Package, Multi-
Substrate Layer,Substrate Stacked, Staggered Viaand Small Trace Via Hole Size
Good Thermal and Electrical Performance
- High Speed and Low Thermal ResistanceHigh Speed and Low Thermal Resistance- Cu / LowCu / Low--K waferK wafer, Nano-technology
Green - Green SolutionGreen Solution
Low Cost & Fast Time-to-market - 1212’’’’ Wafer CapacityWafer Capacity- Total Turnkey Solution - Matrix design, Multi-die, package Design - LAB Design SupportLAB Design Support
Trends of Packaging Technologies
安捷倫科技高頻元件量測研討會2/23/2006
Page 10
OutlineDevelopment Trend for PKG
-- PKG Technology Trend-- Why SiP and POP/PIP
Challenges to PKG Integrity Design-- Design Challenges on Simulation, Measurement and Design-- Case I : Effective DK-- Case II : Impedance Control verify by TDR-- Case III: TDR FA Application-- Case IV : Substrate Ball Pad Design-- Case V : PDS Analysis
Integrated Design-- Components of Optimization PKG Design-- Advanced PKG Analysis Flow-- Plan and Actions for PKG Design
6
安捷倫科技高頻元件量測研討會2/23/2006
Page 11
Challenges to PKG Integrated DesignAnalysis Challenges
Multi-port parameters analysis and broadband calibration skillDouble-side calibration and probing technologySignal-integrity, SSN/SSO and IP drop Mixed-signal analysisSubstrate On-line testing, like via, bump and ball……
Design Challenges to SiPPKG selection for different thermal/electrical requestSubstrate Design integrated thermal/electrical solution.
-- Embedded Die/passive, IPD, decoupling cap……PKG IP development.Design Guideline and constraint
安捷倫科技高頻元件量測研討會2/23/2006
Page 12
Analysis Case I : effective dielectric constant
Er = 3.2
Er = 6.0
What is the effective Er of this mixture material ?
7
安捷倫科技高頻元件量測研討會2/23/2006
Page 13
VNA
Probe station
Sample
Long trace
Short traceshort trace
DUT: (Microstrip/Strip TL)same cross-section with different length
Analysis Case I : Measurement and ADS setup
安捷倫科技高頻元件量測研討會2/23/2006
Page 14
m4freq=Er_preg=3.828
2.358GHz
0.160.320.480.640.800.961.121.281.441.601.761.922.082.242.402.562.722.883.043.203.363.523.683.844.004.164.324.484.644.804.965.125.285.445.605.765.926.086.246.406.566.726.887.047.207.367.527.687.84
0.00
8.00
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
-1.0
4.5
freq, GHz
Er_p
reg
m4
m5freq=loss_tan1=0.011
4.985GHz
1 2 3 4 5 6 7 8 90 10
0.1
0.2
0.3
0.4
0.0
0.5
freq, GHz
loss
_tan
1
m5
2 4 6 8 10 12 14 16 18 20 22 240 26
-100
0
100
-200
200
freq, GHz
phas
e(S
(2,1
))ph
ase(
S(6
,5))
Effective dielectric constant
Loss tangent
Good Correlation between simulation and measurementOnce we have the accurate the dielectric constant of mixture material of substrate, we can achieve good electrical substrate design.
Sim/mea comparison
Analysis Case I : extraction result
Mea
Sim
8
安捷倫科技高頻元件量測研討會2/23/2006
Page 15
Analysis Case II : Impedance Control Verify by TDR/TDR System
Design Condition:Substrate Type : 2 Layer
Impedance Control : Differential
Simulation Structure : Micro-Strip Line
Trace Width (W): 0.39mm
Trace Thickness (T): 22 um
Dielectric Thickness (T1): 200um
Separation (s): 0.15mm
Dielectric Layerε= 4.0
Signal Trace in 1st Layer
GND Plane in 2nd Layer
W
T
T1
S
TDR Measurement Result
安捷倫科技高頻元件量測研討會2/23/2006
Page 16
Trace width(um) separation(um) Zodd(ohm) Zeven(ohm) Single ended(ohm)Q2D 388.95(from drawing) 150(from drawing) 41.2 58.8 50.9
Q2D 429um 116um 38.1267 55.6689 47.5768/435um
TDR 429um 116um 42.4~42.8 61.7 52.3~53
Impedance comparison
Impedance Control -- Manufacture Result and comparison
SEM-single ended trace width measurement
SEM-differential trace width/space measurement
Design / Measurement Comparison
9
安捷倫科技高頻元件量測研討會2/23/2006
Page 17
Chip capacitance
Open waveform
bare substrate
2202 T4failure good unit 2
good unit 1
Analysis Case III FA Application TDR of bare, good, and failure sampleTDR of bare, good, and failure sample
Slight difference on the chip capacitance charge curve –A possible reason of this phenomenon is the IMC makes the interface resistance between wire and bond pad growing, which causes the charge current different.
IMC
安捷倫科技高頻元件量測研討會2/23/2006
Page 18
Analysis Case IV : Ball pad effect analysis
Ball_Pad1
Ball_Pad3
Ball_Pad2
2 4 6 8 10 12 14 16 180 20
-20
-15
-10
-5
-25
0
freq, GHz
dB(B
all_
Pad1
_UP_
1112
..S(2
,1dB
(Bal
l_Pa
d2_U
P_11
12..S
(2,1
dB(B
all_
Pad3
_UP_
1112
..S(2
,1
Measurement Result Measurement Result Void PWR/GNG plane above the ball land on Layer3
No Void
Void layer3
Void layer2&3
Void the plane above ball pad area at PWR/GND will reduce the capacitance,
10
安捷倫科技高頻元件量測研討會2/23/2006
Page 19
Analysis Case V : PDS Analysis -- Measurement setup
Top view Bottom view
27mm
27mm
安捷倫科技高頻元件量測研討會2/23/2006
Page 20
Analysis Case V : Measurement and Simulation Result (Test Board only) Port 1
S21_dBS21_dB
Port 2
Field propagation @750MHz
11
安捷倫科技高頻元件量測研討會2/23/2006
Page 21
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5Frequency (GHz)
-55
-50
-45
-40
-35
-30
-25
-20
-15
-10
-5S 2
1 (d
B)
Bare PCBpackage onlyBGA and PCB
Effect of package
Noise coupling between PKG&PCb?
Analysis Case V : comparison with bare PCB, Package and BGA+PCB
安捷倫科技高頻元件量測研討會2/23/2006
Page 22
OutlineDevelopment Trend for PKG
-- PKG Technology Trend-- Why SiP and POP/PIP
Challenges to PKG Integrity Design-- Design Challenges on Simulation, Measurement and Design-- Case I : Effective DK-- Case II : Impedance Control verify by TDR-- Case III: TDR FA Application-- Case IV : Substrate Ball Pad Design-- Case V : PDS Analysis
Integrated Design-- Components of Optimization PKG Design -- Advanced PKG Analysis Flow-- Plan and Actions for PKG Design
12
安捷倫科技高頻元件量測研討會2/23/2006
Page 23
Design Rule/Specfor substrate layout
Characterization Labcapability on Electrical,Thermal, Stress and Material
Components of Optimization Package Design
Pkg options:structure,cost,thermal,board level...
Substrate:laminate,build-up,ceramic,RLC embed
Wire &Bumping
Leadframe:L/F for SOP/QFP etc,L/F for BCC/QFN etc.
•Knowing ElectricalCharacteristics
deep inside•Provide package solutions
from advanced pkgtechnology
安捷倫科技高頻元件量測研討會2/23/2006
Page 24
Integrity Design Flow for Advanced PKG
Quasi-State EM Simulator•Whole PKG modeling extracting
High Frequency/Speed Simulator•3D Field solver•FTDT simulator•2.5D Momentum•PDS Simulator
Frequency Domain•VNA/PNA with PLTS/ADS
8722ES (40GHz, 2-port)8364B (50GHz, 4-port)
•Impedance Analyzer•Spectrum analyzer
Measurement Plane•Probe Station
Single-side 2-portDouble-Side Multi-port
•High Performance Probe (50GHz)•High Performance cable (50GHZ)
Time Domain•Time Domain Reflectometer
with TDA/•High Speed Pattern Generator
POST-Analysis Capability & Integrated plane•Broadband modeling•Signal Integrity (SI)•PDS analysis•EMI/EMC analysis•data flow control plane•PKG IP Development
System Integrity Analysis and Design Guide for
Advance Package
Software HardwareSubstrate Design / Pre-simulation•Design Guideline and Constraint
Simulation
Transfer interface
Design Constraint & PKG IP
13
安捷倫科技高頻元件量測研討會2/23/2006
Page 25
Plan and Actions for PKG Design
Substrate Design Integrity Electrical Performance-Design Integrity Electrical performance flow -Design Rule and Constraint setting for SiP PKG Application,like RF Module, optical and wireless PKG
Active/Passive Device analysis capability-Embedded passive (RF-MEMS, Substrate embedded RLC) analysis and IP-development
-Sub-system measurement capability setting for advanced PKG
Co-Design and Co-development with key partners-Co-working with key partners for more close and detail study in Wireless, Optical or RF module
安捷倫科技高頻元件量測研討會2/23/2006
Page 26
ContactsContactsSamuel_Wu, Leader of Electrical LabE-mail: [email protected]: 886-7-3617131 ext. 15290/85290Fax: 886-7-3613094
Mark_Li, Project Engineer of Electrical LabE-mail: [email protected]: 886-7-3617131 ext. 15291/85291Fax: 886-7-3613094
ASE, Inc (Kaohsiung)26, Chin 3rd Rd., 811, Nantze Export Processing ZoneKaohsiung, TAIWAN
Website: www.aseglobal.com
14
安捷倫科技高頻元件量測研討會2/23/2006
Page 27
THE END
Thank You For Your Listening
1
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 1
PNA Based Solutions
- Pulsed RF S-Parameter Measurements- Multiport Test Solutions- Physical Layer Test Systems
Agilent Technologies Ltd.
Ming-Fan, Tsai Project Manager
Feb, 23, 2006
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 2
Pulsed-RF S-Parameter Applications Using The Agilent PNA Series Network Analyzer
2
Page 3
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda• Why Measure in Pulsed Mode and the DUTs We Test
– Wafer Test– Power Amplifiers– Antenna RCS– T/R Modules
• Review of Pulsed Measurements• Wideband synchronous• Narrowband asynchronous
• Evolution of Pulsed VNAs from Agilent• 8510, 85108, 85120, CTS Platform to PNA
• Test Sets Available
Page 4
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Why Test Under Pulsed Conditions?
• Device may behave differently between CW and pulsed stimuli• Bias changes during pulse might affect RF performance• Overshoot, ringing, droop may result from pulsed stimulus• Measuring behavior within pulse is often critical to characterizing
system operation (radars for example)• CW test signals would destroy DUT
• High-power amplifiers not designed for continuous operation• On-wafer devices often lack adequate heat sinking• Pulsed test-power levels can be same as actual operation
3
Page 5
安捷倫科技高頻元件量測研討會
Feb.23, 2006
On-Wafer Amplifier Test and Modeling
•Most applications are at microwave frequencies
•Devices lack adequate heatsinking for CW testing, so pulsed-RF used as a test technique to extract S-parameters
•Arbitrary, stable temperature (isothermal state) set by adjusting duty cycle
•Duty cycles are typically < 1%
•Often requires synchronization of pulsed bias and pulsed RF stimulus
Page 6
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Wireless Communications Systems
• TDMA-based systems often use burst mode transmission
• Saves battery power
• Minimizes probability of intercept
• Power amplifiers often tested with pulsed bias
• Most of wireless communications applications ≤ 6 GHz
4
Page 7
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Pulsed Antenna Test
• About 30% of antenna test involves pulsed-RF stimulus
• Test individual antennas, complete systems, or RCS
• RCS (Radar Cross Section) measurements often require gating to avoid overloading receiver
Page 8
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Component-Level Characterization
• Accurate characterization of components such as amplifiers, mixers,filters, and antennas is critical for effective system simulation
• Pulsed-RF stimulus crucial for many components in A/D applicationssuch as AESA Phased arrays and the individual T/R Modules that are used in such systems
Transmit / Receive Module Receive Module
5
Page 9
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Radar and Electronic-Warfare
•Biggest market for pulsed-RF testing
•Traditional applications ≤ 20 GHz
•Many now include Multi-Mode Ka-Band
•Devices include
• amplifiers• T/R modules• up/down converters
Pave-Paws AN/APG-81 JSF
AN/APG-79 F16C Block 60
Page 10
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda• Why Measure in Pulsed Mode and the DUTs We Test
– Wafer Test– Power Amplifiers– Antenna RCS– T/R Modules
• Review of Pulsed Measurements• Wideband synchronous• Narrowband asynchronous
• Evolution of Pulsed VNAs from Agilent• 8510, 85108, 85120, CTS Platform to PNA
• Test Sets Available
6
Page 11
安捷倫科技高頻元件量測研討會
Feb.23, 2006
VNA Pulsed-RF Measurements
Average Pulse
Magnitude and phase data averaged over duration of pulse
Point-in-Pulse
Data acquired only during specified gate width and position within pulse
VNA data display
Frequency domain
Frequency domain
Time domainPulse Profile
Data acquired at uniformly spaced time positions across pulse (requires a repetitive pulse stream) Magnitude
Phase
data point
Note: there may not be a one-to-one correlation between data points and the actual number of pulses that occur during the measurement
CWdB
deg
Swept carrier
Page 12
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Pulsed-RF Data Acquisition (Time Domain View)
t
carrier freq
Trace point 1
Trace point 2
Trace point 3
Trace point 4
Trace point 5
Trace point 6
. . .
Freq 1Freq 2
Freq 3Freq 4
Freq 5Freq 6 . . .
Point-in-Pulse
Pulse Profile
t
gate delay
Trace point 1
Trace point 2
Trace point 3
Trace point 4
Trace point 5
Trace point 6
. . .
Delay 1Delay 2
Delay 3Delay 4
Delay 5Delay 6 . . .
Note: the number of pulses per data point varies with
PRF and IF bandwidth
Pulsed-RF
Gate
7
Page 13
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Pulse-to-Pulse (Single Shot) Measurements
• Carrier remains fixed in frequency
• Measurement point in pulse remains fixed with respect to pulse trigger (requires wideband detection technique)
• One data point for each successive pulse, no pulses skipped
• Display magnitude and/or phase versus time
VNA data display Time domain
P1 P2 P3 P4 P5 P6 …
CW pulses
One data point for each successive pulse, no pulses skipped
Page 14
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Pulsed S-parameter Measurement ModesWideband/synchronous acquisition
• Majority of pulse energy is contained within receiver bandwidth• Incoming pulses and analyzer sampling are synchronous
(requires a pulse trigger, either internal (8510) or external (PNA)• Pulse is “on” for duration of data acquisition• No loss in dynamic range for small duty cycles (long PRI's),
but there is a lower limit to pulse width
Receiver BW
Pulse trigger Time domainFrequency domain
8
Page 15
安捷倫科技高頻元件量測研討會
Feb.23, 2006
10 MHz Ref
Typical Hardware Setup For Wideband DetectionPoint-in-Pulse and Pulse-to-Pulse
Output 1
Src OutRef In
Cplr Thru
DUT
Output 2
To TRIG IN (rear panel)
External pulse generator (e.g., 81110A/81111A)
Z5623A H81 2-20 GHz RF modulator
PNA (20, 40, 50, or 67 GHz) with:• 014 Configurable test set• UNL Source attenuators• 080 Frequency offset mode
Note: pulse generator controls timing
Additional PNA setup:• step sweep• frequency offset on (0 Hz)• Auto IF gain = off
Page 16
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Alternate Hardware Setup For Wideband Detection
Output 1
Src OutRef In
Cplr Thru
DUT
Output 2
TRIG OUT (rear panel) to pulse gen EXT INPUT
External pulse generator(e.g., 81110A/81111A)
Z5623A H81 2-20 GHz RF modulator
PNA (20, 40, 50, or 67 GHz) with:• 014 Configurable test set• UNL Source attenuators• 080 Frequency offset mode
Additional PNA setup:• step sweep• frequency offset on (0 Hz)• Auto IF gain = offNote: PNA controls timing
10 MHz Ref
9
Page 17
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Minimum Pulse Widths for Point-in-Pulse Measurements Using Wideband Detection
No2 us600 kHzPNA-L models(2-port, 6, 13.5 GHz; 4-port, 20 GHz)
Yes10 us250 kHzPNA-L models(2-port, 20, 40, 50 GHz)
Yes50 us40 kHzPNA models(20, 40, 50, 67 GHz)
IF auto-gain mode*
Minimum pulse width
Maximum IF bandwidth
* Note: for point-in-pulse measurements, the IF auto-gain mode should be turned off (i.e., set IF gains manually)
Page 18
安捷倫科技高頻元件量測研討會
Feb.23, 2006
P1 P2 P3 P4 P5 P6 …
CW pulses
Fastest PRI/PRF for Pulse-Pulse Measurements Using Wideband Detection
25 kHz40 us600 kHzPNA-L models(2-port, 6, 13.5 GHz; 4-port, 20 GHz)
12.5 kHz80 us250 kHzPNA-L models(2-port, 20, 40, 50 GHz)
5.9 kHz170 us40 kHzPNA models(20, 40, 50, 67 GHz)
Maximum PRF
Minimum PRI
Maximum IF bandwidth
Conditions: point sweep; external trigger, CW sweep; IF autogain=off
PRI
10
Page 19
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Pulsed S-parameter Measurement Modes
Narrowband/asynchronous acquisition• Extract central spectral component only; measurement appears CW• Data acquisition is not synchronized with incoming pulses (pulse trigger not required)• Sometimes called “high PRF” since normally, PRF >> IF bandwidth• “Spectral nulling" technique achieves wider bandwidths and faster measurements• No lower limit to pulse width, but dynamic range is function of duty cycle
IF filter
IF filter
Time domain
Frequency domainD/R degradation = 20*log[duty cycle]
Page 20
安捷倫科技高頻元件量測研討會
Feb.23, 2006
-3 -2 -1 0 1 2 3
x 104
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Filtered Output Using Spectral Nulling
Pulsed spectrum Output
X
-3 -2 -1 0 1 2 3
x 104
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Digital filter (with nulls aligned with PRF)
• With “custom” filters, number of filter sections (M) can be chosen to align filter nulls with pulsed spectral components
• With spectral nulling, reject unwanted spectral components with much higher IF bandwidths compared to using standard IF filters
• Result: faster measurement speeds!
11
Page 21
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Typical Hardware Setup(Narrowband)
Output 1
10 MHz Ref
Src OutRef In
Cplr Thru
DUT
Output 2
Pulse 2 drive to PULSE IN B (for point-in-pulse measurements)
External pulse generator(e.g., 81110A/81111A)
Z5623A H81 2-20 GHz RF modulator
PNA (20, 40, 50, or 67 GHz) with:• 014 Configurable test set• UNL Source attenuators• 080 Frequency offset mode• 081 Reference switch• H11 IF access• H08 Pulsed-RF measurement capability• 016 Receiver attenuators (optional)
PNA
Option H08 VB application/DLL
GPIB
Page 22
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda• Why Measure in Pulsed Mode and the DUTs We Test
– Wafer Test– Power Amplifiers– Antenna RCS– T/R Modules
• Review of Pulsed Measurements• Wideband synchronous• Narrowband asynchronous
• Evolution of Pulsed VNAs from Agilent• 8510, 85108, 85120, CTS Platform to PNA
• Test Sets Available
12
Page 23
安捷倫科技高頻元件量測研討會
Feb.23, 2006
85108A
First True pulsed Network Analyser
Still the most widely used system for TR Module R&D and manufacturing test (>100 off world-wide still in use)
Wholly COTS solution
Affordable
Low cost of ownership
Low risk
Point-on-pulse
Pulse profile, repetitive
Page 24
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Pulsed PNA is the New Generation 85108A
COTS stand alone hardware can perform:
– Point-on-pulse– Pulse-to-Pulse– Pulse profile
Timing generator can trigger custom user hardware e.g.
TR Module controller
Very much faster measurement speed
13
Page 25
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Comparing the 8510 and PNA8510 (85108A)• Dominant mode is wideband detection
• Detection is done BEFORE analog-to-digital conversion
• Analog synchronous detector produces baseband I/Q output(detector bandwidth = 1.5 MHz)
• Pulse profiling achieved by varying sample point of baseband pulses
• Trade off speed and dynamic range with averaging
PNA• Dominant mode is narrowband detection
• All processing (filtering and detection) is done digitally
• Widest bandwidth = 35 kHz
• Pulse profiling achieved with analog switches that gate IF (or RF) signals
• Trade off speed and dynamic range with variable IF bandwidths and averaging
PNA-L• Dominant mode is wideband detection• All processing (filtering and detection) is done digitally
• Widest bandwidth = 600 kHz
• Trade off speed and dynamic range with variable IF bandwidths and averaging
Page 26
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agilent TR Module Test Systems
85120A S10 Family 84000A Family
14
Page 27
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agilent TR Module Test Systemscontinued
CTS-I Family CTS-II Family
Page 28
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Pulse Summary
AveragePoint-in-pulsePulse profile
AveragePoint-in-pulse Pulse profilePulse-to-pulse
Measurements
H11/H08***None requiredPNA Options
PW > 20 ns (limited by IF gate)PW > 50 us/10 us/2 us**PNA
Limited* (High PRF Mode)PW > 1 us8510
Dynamic range loss with small duty cycles
No pulse-to-pulse
Lower pulse width limit
Elevated noise floorDisadvantages
Narrow pulse widthsConstant dynamic rangeAdvantages
Narrowband/AsynchronousWideband/Synchronous
* No nulling, no point-in-pulse** PNA/PNA-L 2-port 20, 40, 50 GHz/PNA-L 2-port 6, 13.5, 4-port 20 GHz*** Option H08 is usually used in conjunction with Option H11. Without H11, the user can perform average pulse, or point-in-pulse
measurements using external RF gates.
15
Page 29
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda• Why Measure in Pulsed Mode and the DUTs We Test
– Wafer Test– Power Amplifiers– Antenna RCS– T/R Modules
• Review of Pulsed Measurements• Wideband synchronous• Narrowband asynchronous
• Evolution of Pulsed VNAs from Agilent• 8510, 85108, 85120, CTS Platform to PNA
• Test Sets Available
Page 30
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Z5623A Hxx RF Modulators• Z5623A H81 (2-20 GHz)
• Expected to be most common configuration ($43K)• Contains pin switch, amplifier, directional coupler• Use jumpers to bypass internal amp or use high-power external amp
• Other quoted test sets:• Z5623A H83 1-20 GHz Bidirectional (two pin switches) $80K• Z5623A H84 20-40 GHz Bidirectional (two pin switches) $105K• Z5623A H85 20-40 GHz Unidirectional, no amplifier $35K• Z5623A H86 2-40 GHz Unidirectional, dual band (incl. band switch) $65K
• Customization via Agilent’s “Special Handling” group:• Different frequency ranges • No amplifier or higher power amplifiers• High power components
Z5623A H81
16
Page 31
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PNA Pulsed-RF Configuration Example 1
• User-supplied external modulator• Average pulse measurements
TTL
10MHz Ref
Cplr Thru
Src Out
+25 -25Com
DC(-)DC(+)
RF out
RF inPower Supply
Pulse out
Advantage: simplest – use any standard PNA with 014GPIB
81110A family pulse generator
Page 32
安捷倫科技高頻元件量測研討會
Feb.23, 2006
One output channel drives RF modulator
Src Out
Ref In
Three output channels drive internal receiver gates A, B, and R1 for point-in-pulse and pulse-profile measurements
10 MHz Ref
GPIB
Trigger
DUT
Z5623A H81 pulsed-RF test set
PNA
81110A family pulse generators
PNA Pulsed-RF Configuration Example 2
Advantages: • easily make point-in-pulse and
pulse profile measurements• more sophisticated RF
modulator boosts port power
• Modulator test set, internal receiver gates• Point-in-pulse, pulse profile of S21 and S11
17
Page 33
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PNA Pulsed-RF Configuration Example 3:Full Forward/Reverse S-Parameter Configuration
Three output channels drive internal receiver gates A, B, and R1/R2 for point-in-pulse and pulse-profile measurements
One output channel drives RF modulator
10 MHz Ref
GPIB
Trigger
DUT
PNA
81110A family pulse generators
Page 34
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PNA Pulsed-RF Configuration Example 3:Full Forward/Reverse S-Parameter Configuration a different view
18
Page 35
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Z5623A H83 – H84 Test Set Control Macro
Page 36
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Z5623A H83 RF Modulator 1-20 GHz
12
3 4
10/20/30
60 dB
12
3 4
12
3 4
1 2
34
10/20/30
60 dB
1 2
34
1 2
34SW1
SW2 SW3
ATN1 ATN2
CPLR1 CPLR2
16 dB 16 dB
SW4
SW5SW6
SW8SW 7
SourceIN
RCVRR1 IN
RCVRR2 IN
CPLRTHRU
CPLRTHRU
SourceIN
PulseOut
PulseOut
CPLRIN
AMPOUT
SourceOUT
AMPIN
Pulse 2IN
Pulse 1IN
CPLRIN
SourceOUT
AMPIN
AMPOUT
AMP 1TERM
AMP 1TERM
AMP 2TERM
AMP 2TERM
FilterIN
FilterIN
0
RCVR R1 OUT AMP IN
AMP OUTCPLR THRU
SOURCE IN
SOURCE OUT
CPLR IN
FILTER IN
PULSE OUT
PULSE 1IN
RCVR R2 OUTAMP IN
AMP OUT CPLR THRU
SOURCE IN
SOURCE OUT
CPLR IN
FILTER IN
PULSE OUT
PULSE 2IN
LINE
1
Z5623AH83 PULSE TEST SET, 1 GHz to 20 GHz
TTL 0,5 VDC, 10K OHM
PORT STATUS "GPIB ONLY"
AVOID STATIC DISCHARGE
CAUTION: SEE MANUAL FOR MAXiMUM POWER RATINGS CAUTION: SEE MANUAL FOR MAXiMUM POWER RATINGS
0
19
Page 37
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PNA Pulsed-RF Configuration Example 4
Pulse1 out
10 MHz Ref
Src Out
• Modulator test set, external receiver gate • Point-in-pulse, pulse profile
Ref In
+25 -25Com
Power SupplyTTL
DC(-)
DC(+)
Pulse2 out
GPIB
External switch in receiver B loop for external gating
Advantage: gate widths < 20 ns81110A family pulse generator
Z5623A H81 pulsed-RF test set
Page 38
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PNA Pulsed-RF Configuration Example 5
Pulse1 out
+25 -25Com
Power Supply
• User-supplied pulsed bias to amplifier, internal IF gate• Point-in-pulse, pulse profile
Pulse2 drive to internal receiver gates (for point-in-pulse)
CW Pulsed-RF
Advantage: pulsed bias to amplifier (with CW input)
10 MHz Ref
GPIB
81110A family pulse generator
20
Page 39
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PNA Pulsed-RF Configuration Example 6
Cplr Thru
Src Out
RF out
RF in
• Customer-supplied pulsed bias and pulsed RF, internal IF gate• Point-in-pulse, pulse profile
Pulse1 out
Power Supply
+25 -25Com
Trigger
Pulse3 drive to internal receiver gate B (for point-in-pulse)
Pulse2 outTTL
DC(-) DC(+)
Advantage: both pulsed bias and pulsed-RF stimulus
10 MHz Ref
GPIB
81110A family pulse generators
Page 40
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Summary
• Testing with pulsed-RF is very important for radar, EW, and wireless comms systems
• Narrowband detection:• Spectral nulling technique improves measurement speed
• For radar and wireless comms applications, offers superior dynamic range/speed
• No lower limit to pulse widths
• Although the PNA uses different hardware and detection techniques than the 8510, measurement results are essentially the same!
• PNA also offers numerous platform benefits:• Measurement flexibility (32 channels, 64 traces, 16 windows, 16,001 points)
• Connectivity (LAN, USB, …)
• Automation (open Windows®, COM, SCPI …)
• Ease of use (built-in HELP, Cal Wizard, ECal …)
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
7.74
7.99
8.24
8.49
8.74
8.99
9.24
9.49
9.74
9.99
10.2
4
10.4
9
10.7
4
10.9
9
11.2
4
11.4
9
11.7
4
11.9
9
12.2
4
12.4
9
12.7
4
Frequency (GHz)
S21
(dB
)
PNA8510
21
Page 41
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Resources• www.agilent.com/find/pulsedrf
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 42
Multiport - PNA based solutions
22
Page 43
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda
• Target Applications
• PNA Multiport Test Solutions
• Specials vs. Standard product
• Customer data required for new multiport specials
• Further Information
Page 44
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Target Applications
• Front End Module (FEM)• Cellular FEM (see next slide)• WLAN FEM (Base Band not included, 2.4G & 5G dual band)• Filter array (see next slide)
– Multiple filter in single package– Duplexer/Coupler array (see next slide)– Multiple Duplexer/Coupler in single package
• High Frequency Multiport devices– Triplexer, power splitter, multiport coupler etc.– RF module, RF switch IC
• High power device testing
23
Page 45
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Cellular FEM #1 (Dual band, 1xEV,GPS)
Diversity Rx
Page 46
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Cellular FEM #2 (Quad band, UMTS,GSM)
24
Page 47
安捷倫科技高頻元件量測研討會
Feb.23, 2006
WLAN FEM(2.4G・5G dual band)
2.4 GHz Rx
5 GHz Rx
2.4 GHz Tx
5 GHz Tx
Diversity SW
Diplexer
PA
BPF
LNA
2.4 GHz / 5 GHz WLAN FEM
Page 48
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport & High Power Testing Devices
• Typical High Frequency devices– Switches, Couplers, Power Splitter/Divider and etc..– Filter/Coupler array
• Multiport and high power– PA Cellular/WLAN FEM with PA– RF switch IC ( Switch filters)
25
Page 49
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Key Measurement Requirements
Cellular FEM without PA• IL, RL for each path up 12.75GHz• Isolation
– Frequency > 3x Carrier– Signal level 0dBm or –10dBm
• Switch distortion– Frequency up to 12.75GHz in R&D– Signal Level up +36dBm
UMTS850 UMTS1900
GSM Tx
Page 50
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Key Measurement Requirements
WLAN FEM with PA– Gain, 1dB Compression, Pout,PAE– RL– Isolation– Harmonics distortion
• Frequency Cellular > 3x Carrier Frequency• Frequency(WLAN): 17.4GHz• Power : 25dBm(WLAN)
36dBm(GSM/Mobile)
2.4 GHz Rx
5 GHz Rx
2.4 GHz Tx
5 GHz Tx
Diversity SW
DiplexerPA
BPF
LNA
26
Page 51
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda
• Target Applications
• PNA Multiport Test Solutions
• Specials vs. Standard product
• Customer data required for new multiport specials
• Further Information
Page 52
安捷倫科技高頻元件量測研討會
Feb.23, 2006
New PNA Rev 6.0 Firmware (Available December 2005)
Sets GP-IB address, or Test Set IO address
Set control lines (if test set has them)on a per channel basis
Set physical ports in ANY order
Select Testset Control File base on test set
New Functionality:• External test set control• “Any 2-port” capability
27
Page 53
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PNA/PNA-L Option 550
• New firmware option 550 for the PNA/PNA-L adds full 4-port capability and differential measurements to a two port network analyzer
• CPL Dec 1, 2005
• $7k RFP
ApplicationsReceiversBalancedSParameterMeasure
Page 54
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Hybrid
Variety Summary
Signal Conditioning
Switching
Special Multiport Test VarietiesSpecial Multiport Test Varieties
Extension
SCMM
28
Page 55
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Test Sets Overview
Platform Description Benefits87050A-Hxx/Kxx Switching test sets. Lower cost
No coupler inside test set
N4419/20/21B/H67 Extension test sets 4-port differential to 67GhzCoupler on each port
Z5623A-Hxx/Kxx Depend on the particular option Meets customer specific req’ts
Mixing of Switching and Extension
test sets
Page 56
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agilent’s Promoted Multiport Products & Specials
29
Page 57
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agilent PNA > 4-port Products offeringFreq. Coverage: PNA Based Unit & Ext. Test set Total Test Ports300 KHz – 20 GHz N5230A opt. 245 & Z5623AK64 6300 KHz – 20 GHz N5230A opt. 245 & Z5623AK66 14300 KHz – 20 GHz N5230A opt. 245 & Z5623AKxx 20.010 – 40GHz E8363/4B or E8361A & 87050A-K62* 6
* 2-port cal only
Page 58
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials – Variety
• Extension Test Sets– Extension test sets provide signal paths from the network analyzer access
ports to the test set. All test ports to the DUT are Coupler or Bridge base. Provides greater sensitivity.
30
Page 59
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials – Variety
• Hybrid Test Sets– Hybrid test sets are a combination of the switching and extension types.
Test ports can be either switched or coupler bridge based.
Page 60
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials – Variety
• Switching Test Sets– Switching test sets provide signal paths from the network analyzer test ports
to the DUT.
J15
A3 DriverDaughter
BoardA2 Controller
Interface MotherBoard
A1PowerSupply
Port 3
Port 1
Port 2
Reflection(Type-N)
C
A4 LCDController
Board
w22 w23
Transmission(Type-N)
J10-
J15,
J50-
J57
GPI
BJ1
w5
w4
w2
w3
w6w7
w8w9
w14w15
w16
w17
w21w20
w19
w18w13
w12w11
w10
w1 w1 w1 w1 w1 w1 w1 w1
J13
J12
J11
J10
J14
J10-J1587050-60055
J50-J57
1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2
C C C C C C C
Port 6
Port 4
Port 5
Port 8
Port 7
(Type-N)
Sw15
Sw13 Sw12 Sw11 Sw10
Sw14
Sw50 Sw51 Sw52 Sw53 Sw54 Sw55 Sw56 Sw57
2
3
6532 65326532
2
3
Z5623A Option H48
1
6532
Z5623-60013
J74-J79
2 3 4 5 6 7 8 9
Open / CollectorLines
87050-60053
J54
J53
J52
A3 DriverDaughter
BoardA2 Controller
Interface MotherBoard
A1 PowerSupply
R3R1 R2Reflection
12
C6C1
C
A4 LCDController
Board
Transmission
C
J50-
J55
GPI
BJ1
C2 C3 C4 C5
A T1 T2/3
12S50
2S55
C
S51
C
S5412
1 2S52
20052 20053
12S53
20054
20051
20057
2005620055
20061 20062
20058
20059
20060
C C
C
J50
J55
J51
Z5623-60013
Z5623-60012
J74-J79
2 3 4 5 6 7 8 9
Open / CollectorLines
1
(Type-N)
Z5623A Option H46
1
31
Page 61
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials – Variety
• SCMM– Single connection multiple measurement test sets allow the user to make
different types of measurements with out having to disconnect there DUT form the test set.
Power Supply
Display LCD Board
Controller Interface Mother BoardDriver Daughter Board
1 2 3 4 5 6 7 8 9 10 11 12Test Ports
Aux1 Aux2
Agilent 8720D Option K22Multi-Function Switch Matrix
B i/p 4-1A i/p 1-4
SW 50 SW 61
SW 12 SW 11
SW 13SW 10 SW 15 SW 14 SW 16 SW 17
1 1 1 1 1 1 1 1 1 1 1 1 222222222222
1 25 6 4 3 1 25 6 4 3 1 25 6 4 3 1 25 6 4 3
4 15 3 2 6 1 2 4 36 5
6 32 5 5 36 2
Page 62
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials – Variety
• Signal Conditioning– Test sets that require couplers, attenuators, amplifiers, filters and mixers that may or
may not directly connect to the DUT.
R2
R1
P1
A
B
RF IF
LO
RF
IF
LO
P2
M1
Z5623A Opt H61
DUT MIXERisolator
Normal Mode
Source OUT PNA
Coupler IN
B OUT
B IN
Source INPNA
CouplerOUT
B IN
B OUT
LO DUTPORT
Source
LO INPORT
Up/DownFilter/Mixer
IN
IN
OUT
OUTFilter/Mixer
LO EXTOUT
IN
32
Page 63
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda
• Target Applications
• PNA Multiport Test Solutions
• Specials vs. Standard product
• Customer data required for new multiport specials
• Further Information
Page 64
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials – Why Specials?
• Offer solutions for applications which can not be addressed with existing or standard products
• Increase sales of CTD standard products• Leverage standard platforms & OF capabilities to develop product extensions• Fastest development-shipment time for new opportunities & competitive
battles• Helps to identify market needs and trends of specific applications
33
Page 65
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials –Specials vs. Standard product
Multiport Special – Custom– Does not follow the NPI PLC process– Fast development to shipment– Typical supplemental performance– Provide just enough performance• Functional Certificate• Return to Factory support
Standard Product– Follows the NPI PLC process– Specifications with uncertainties. – Box and System level performance• Calibration Certificate• Return to Bench support• Verification test
Page 66
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Specials –Test Set vs. System PerformanceMultiport Test Set
• Typical supplemental performance specifications of the test set.
• Measures the Insertion Loss, Port Match, and path Isolation between ports.
• Provide just enough performance• Functional Certificate
System Level (PNA and Test Set)• Raw or Corrected system performance is not
provided.• Customers must take into consideration the
performance of the PNA, Test Set, cabling, fixtures, and other peripherals to characterize their system level performance.
• Once established that system level performance meets the application requirements. The customer can now define the PNA Multiport Test Solution and calibrate the system.
• The customer can define a system verification method by using either a golden standard or by establishing an additional type of verification process.
34
Page 67
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda
• Target Applications
• PNA Multiport Test Solutions
• Specials vs. Standard product
• Customer data required for new multiport specials
• Further Information
Page 68
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport Qualification Form
Qualification Form for Multiport Device
Device Description and Specifications Describe the device and application (please attach block diagram of device) ________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ Number of device ports: ______ Port impedance: ___ 50 ohms ___ 75 ohms Connector type(s):__ Type N __SMA __ 3.5 mm __ 2.4 mm __ 1.85mm __ other (if other, please explain) ___________________________________________________ Specify the frequency range of the device: _____________________________________ Specify the frequency range required for the device application: ____________________ Specify required test port power: _____________________________________________ Specify Insertion loss / gain of the device paths: _________________________________ ________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ Specify Isolation of other device ports: ________________________________________ ________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ Specify Port matches of the device: ___________________________________________ ________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ Specify desired measurement uncertainty: _____________________________________ ________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ Specify other measurement requirements: ______________________________________ ________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ If desired test system is complex (e.g. includes other analyzers, additional sources, etc), include block diagram of overall test setup.
• Most >4-port applications unique
• Define customer needs upfront
• Reduces time to quote and minimizes error
FE/AE fill out with customer and email back to CTD/Say Phommakesone
Page 1 of 4
35
Page 69
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agenda
• Target Applications
• PNA Multiport Test Solutions
• Specials vs. Standard product
• Customer data required for new multiport specials
• Further Information
Page 70
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Information Sources
Freq. Coverage: Ext. Test Set & #Ports300 KHz – 20 GHz Z5623AK64 2-port300 KHz – 20 GHz Z5623AK66 10-port.45 – 20/40/50 GHz N4419/20/21B 2-port.45 – 67.0 GHz N4421BH67 2-port
http://mktwww.soco.agilent.com/Product-Info/Network-Analyzers/PNA/multiport.htm
http://www.agilent.con/find/multiport
http://www.agilent.con/find/PNA
Agilent Test Solution for Multiport and Balanced Devices 5988-2461EN
Agilent PNA Series Configuration Guide 5988-7989EN
36
Page 71
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Summary
• PNA multiport test solutions provide: – Direct test set control made easy with PNA Firmware Rev 6.0s – Flexible test port configuration that enables customer to perform variety kinds
of multiport device measurements– Advanced measurement capabilities, such as the APE, result in accurate
measurements
Page 72
安捷倫科技高頻元件量測研討會
Feb.23, 2006
presented by:
Ming-Fan, TsaiAgilent Technologies Ltd.
Complete Characterization of Backplane Differential
ChannelsFebruary 23, 2006
© Copyright 2003 Agilent Technologies, Inc.
37
Page 73
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Overview
Backplanes
Measurement set up
Single-ended
Differential
Frequency & time domain
Eye diagrams
Model extraction
Page 74
安捷倫科技高頻元件量測研討會
Feb.23, 2006
All Next Generation High Speed Serial Links will use Differential Signaling
Serial ATA 1.25 Gbps
Hypertransport 1.6 Gbps
AGP8x 2.1 Gbps
Infiniband 2.5 Gbps
PCI Express 2.5 Gbps
Serial ATA II 2.5 Gbps
XAUI 3.125 Gbps
PCI Express II 5.0 Gbps
OC-192 9.953 Gbps
10 GbE 10 Gbps
OC-768 39.81 Gbps
38
Page 75
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Important Physical Layer Properties of Differential Channels
Differential impedance profile (diff return loss)
Transmitted differential signal quality (diff insertion loss)
Conversion of differential to common signal
Where conversion of differential to common signal occurs
Eye diagrams (1 Gbps 10 Gbps)
Page 76
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measurement System for Complete Physical Layer Characterization
GigaTest Labs Probe Station
Agilent Physical Layer Test System
Device Under Test(backplane)
39
Page 77
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential VNA/TDR Applied to All Passive, Linear Components and Interconnects
When an external precision signal is requiredApplies to any passive interconnect or component
BackplanesDiscretesPackagesConnectorsPCB structuresMaterial properties
Page 78
安捷倫科技高頻元件量測研討會
Feb.23, 2006
A Precision Instrument is Not Enough!
Component to characterize
Instrument Valuable information? ?
40
Page 79
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Complete Characterization System Solution
DUT + microprobes
GigaTest Probe Station
Physical Layer Test System: VNA + PLTS software
Page 80
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Microprobes Allow Precision Probing of Structures with Minimal Artifacts
Pitch ~ 50µ – 1000µ
Close up
41
Page 81
安捷倫科技高頻元件量測研討會
Feb.23, 2006
4 Port Differential VNA Techniques Applied to Tyco Electronics HM-Zd Legacy Backplane System
16 inches, 30 inches backplane
2 inches, daughter card
2 inches, daughter card
Total channel lengths: 26 inches, 40 inches
Page 82
安捷倫科技高頻元件量測研討會
Feb.23, 2006
4 Port Single-ended S-parameters
1
3
2
4(and their return paths!)
44434241
34333231
24232221
14131211
SSSSSSSSSSSSSSSS
Stimulus
Res
pons
e Interpreting single ended measurements:S11 : return loss, single endedS21= S12 : insertion loss, single endedS31= S13 : near end cross talkS41= S14 : far end cross talk
in
outin,out P
PS =
42
Page 83
安捷倫科技高頻元件量測研討會
Feb.23, 2006
DUTIncident wave
Reflected wave
Transmitted wave
S11
S21
Incident wave
Reflected wave
Transmitted wave
TDR
TDTt
t
DUT
TDR and VNA Techniques
Page 84
安捷倫科技高頻元件量測研討會
Feb.23, 2006
4 Port, Single-ended S-parameters: Tyco Backplane Example
Interpreting single ended measurements:S11 : return loss, single endedS21= S12 : insertion loss, single endedS31= S13 : near end cross talkS41= S14 : far end cross talk
43
Page 85
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Single-ended Return Loss and Insertion Loss: 26 inch channel length
2 GHz/div
0 dB
Input Single-ended Return Loss S11
2 GHz/div
-10 dB
-20 dB
-30 dB
-40 dB
-50 dB
Input Single-ended Insertion Loss S21
Page 86
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Microprobing on SMA Pads
Added ground pad
44
Page 87
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Bandwidth Limit of SMA vs. Microprobes
Measured with SMA connector
Measured with microprobe
Conclusions:
1. Microprobes can be higher bandwidth (important > 14 GHz)
2. Identical performance < 10 GHz for these SMA connectors
20 GHz full scale
S11- return loss0 dB
-10 dB
-20 dB
-30 dB
-40 dB
-50 dB
Page 88
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Design for Test (DFT):Optimized Pad Design for Micro-probing
Any signal via can be used as a probe point
Use a “copper fill” around the signal via with immediate connection to all adjacent ground vias
Every board should be designed with pads for optional microprobing- no impact on function
Ground vias shorted to the copper fill
45
Page 89
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Microprobing vs. SMA Connectors
• Probe station required
• Probes can be damaged
• Can use on any signal lines
• No constraints on how many or where
• Can be used on functional board
• Important for active probing
Micro Probes
• Can’t use on functional boards- loads the line too much
• Limited density
• No additional fixturing to VNA required
• Easy to use
• Mechanically robust
SMA Connectors
WeaknessesStrengths
Page 90
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Two Important Transformations Facilitate First Order Analysis
From single-ended S-parameters to differential S-parameters
From frequency domain to time domain
46
Page 91
安捷倫科技高頻元件量測研討會
Feb.23, 2006
4 Port Balanced Measurements:Frequency and Time Domain
Port 2Port 1Port 2Port 1
Common SignalDifferential Signal
Stimulus
22CC21CC22CD21CD
12CC11CC12CD11CD
22DC21DC22DD21DD
12DC11DC12DD11DD
SSSSSSSSSSSSSSSS
1
3
2
4
Single-ended
(and their return paths!)
44434241
34333231
24232221
14131211
SSSSSSSSSSSSSSSS
Stimulus
Res
pons
e
Diff pair port 1
Diff pair port 2
(and their return paths!)
Differential
Page 92
安捷倫科技高頻元件量測研討會
Feb.23, 2006
The Meaning of the Quadrants
22CC21CC22CD21CD
12CC11CC12CD11CD
22DC21DC22DD21DD
12DC11DC12DD11DD
SSSSSSSSSSSSSSSS
Differential in, differential out:Behavior of differential signals
Common in, common out:Behavior of common signals
Differential in, common out:Behavior of mode conversion
Common in, differential out:Behavior of mode conversion
Port 2Port 1Port 2Port 1
Common SignalDifferential Signal
Stimulus
47
Page 93
安捷倫科技高頻元件量測研討會
Feb.23, 2006
22CC21CC22CD21CD
12CC11CC12CD11CD
22DC21DC22DD21DD
12DC11DC12DD11DD
SSSSSSSSSSSSSSSS
Port 2Port 1Port 2Port 1
Common SignalDifferential Signal
Stimulus
Diff pair port 1
Diff pair port 2
(and their return paths!)
Signal quality of the common signal, time delay of common signalSCC21
Common impedance profileSCC11
Conversion of common signal to differential signal in transmission (susceptibility)
SDC21
Conversion of differential signal to common signal in transmission (emissions)
SCD21
Signal quality of differential signal, time delay of differential signalSDD21
differential impedance profileSDD11
Important Performance Terms
Page 94
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Single-ended to Differential S-parameters
Single-ended S-parameters Differential S-parameters
Note: One measurement with Physical Layer Test System yields above information
48
Page 95
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential Return Loss & Reflection Coefficient
Conclusions-Connectors create large impedance discontinuity-Daughter card differential impedance is 110 Ω-Backplane differential impedance is 102 Ω
0 dB--
-50 dB--
20 GHz full scale
Frequency Domain SDD11
1 nsec/divt = 0
Time Domain TDD11
100 Ω−80 Ω−
120 Ω−
Page 96
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Single-ended and Differential TDR
t = 0
Single-ended TDR(NOT odd mode impedance)
50 Ω
1 nsec/div
inside connector
60 Ω
Via fields on either side of connector
Backplane trace
1 nsec/div
Differential TDR
100 Ω120 Ω
Coupling brings differential impedance down
49
Page 97
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Important Design Feedback
Designing for 50 Ohm single ended line is not the same as a 100 Ohm differential line.
Characterizing with single ended TDR will not measure differential impedance.
Design the daughter cards with as much care as the backplane.
Most discontinuities from connectors are not from the connectors- they are from the via fields.
Optimizing connectors is all about optimizing the circuit board via field layout.
Design for test: add copper fills for microprobing
Page 98
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential TDR from Both Ends
TDD11 TDD22
similar connector, reduced bandwidth
1 nsec/div 1 nsec/divmirrored
Port 1 Port 2
100 Ω
50
Page 99
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential Transmitted Signal SDD21
Conclusions:
• Measurement system bandwidth > 40 GHz
• 26 inch traces have a 15 dB BW ~ 3.5 GHz
• 40 inch traces have a 15 dB BW ~ 2 GHz
10 GHz full scale
0 dB--
-100 dB--
26 inch backplane trace
40 inch backplane trace
-10 dB--
-20 dB--
Frequency Domain SDD21
Page 100
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential Transmitted Signal: Time Domain TDD21
1 nsec/div
40 GHz bandwidth, ~20 psec input rise time
Total of 26 inches
Total of 40 inches
51
Page 101
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Eye Diagrams: 26 inch Channel
1 Gbps, 200 psec/div 2.5 Gbps, 80 psec/div
5 Gbps, 40 psec/div 7.5 Gbps, 27 psec/div
Page 102
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Non-ideal Differential Signaling: Mode Conversion
Anything that affects one line and not the other will convert differential signal into common signal
Drive is asymmetrical between channels • skew• output impedance and launched voltage Signal environment in interconnect is asymmetrical• different characteristic impedance in each leg• length is different• loading from connectors, jags, pads, ground planes
Real problem of common signal is EMI from unshielded twisted pair
52
Page 103
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential Signal Input Common Signal Output
~7% of differential signal amplitude converted to common signal
May be a problem if it were on CAT5 twisted pair
TDD21
TCD21, 20x scale
1 nsec/div
26 inch channel length
Page 104
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Where did the Conversion Happen?
TCD11 x10 scale increase
Via field on daughter card
Via field on mother board
asymmetry of backplane traces
Conclusion: most mode conversion happens in the
via fields!
TDD11
53
Page 105
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measurement and Model Extraction
AGILENT TECHNOLOGIESPNA SERIES
VECTOR NETWORK ANALYZERS
TIME DOMAIN SIMULATORS(HSPICE®, SPECTRAQUEST®, SMARTSPICE)
FREQUENCY-DOMAIN SIMULATORS(ADS, ETC)
AGILENT TECHNOLOGIES86100-SERIES
TIME DOMAIN REFLECTOMETERS
AGILENT TECHNOLOGIESN1900-SERIES
PHYSICAL LAYER TEST SYSTEM
TDA SYSTEMS ICONNECT MEASUREXTRACTOR
S-PA
RAM
ETER
S TOPO
LOGI
CAL
MOD
ELS
DEVICE UNDER TEST
S-PARAMETERS
DEVICE UNDER TEST
BEHA
VRIO
ALM
ODEL
S
TIM
E DO
MAI
N
TDR or VNA
Note: TDR is NEW measurement engine for PLTS v1.2
Note
See Note
Page 106
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Modeling Example with PLTS & IConnect
54
Page 107
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Conclusions
Differential pairs will proliferate
Differential characterization requires • microprobes • probe station• 4 port VNA• Analysis softwareAbsolutely everything you ever wanted to know about the performance of a differential pair is contained in the 4 port balanced S parameters-displayed in either the frequency or time domain
Page 108
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Visit www.gigatest.com for..
• More than 100 application notes on high speed design• Schedule of signal integrity short courses• High-bandwidth measurement and modeling services• Complete signal integrity characterization systems
Technical Information Resources
• Visit www.agilent.com/find/plts for..• Physical Layer Test System data sheet & user’s guide• Signal integrity solutions brochure• XAUI backplane design case study• PCI Express tools brochure• N1900 series product flyer
Contact Gigatest Labs for more information....www.gigatest.com/about/ReqForInfo.jsp
55
Page 109
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Keep up to date on:Services and Support Information Events and Announcement
- Firmware updates - New product announcement- Manuals - Technology information- Education and training courses - Application and product notes- Calibration - Seminars and Tradeshows- Additional services - eSeminars
FREEFREE Agilent Email Updates
Subscribe Today!
Choose the information YOU want.Change your preferences or unsubscribe anytime.
Visit: www.agilent.com/find/eseminar-email
1
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 1
Enhanced TDR Channel Characterization Capabilities
Agilent Technologies Ltd.
Brian ChiSenior Project ManagerAgilent [email protected]
Feb, 23, 2006
Page 2
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Z Impedance?
High Speed Digital Design
Impedance in Time Domain
2
Page 3
安捷倫科技高頻元件量測研討會
Feb.23, 2006
E
R
PROBE
0Ω
E/2
0
E
What do you expect to see at the probe before, during, and after you close the switch?
Time
Short Termination
Impedance in Time Domain
Page 4
安捷倫科技高頻元件量測研討會
Feb.23, 2006
E
R
PROBE
∞Ω
What do you expect to see at the probe before, during, and after you close the switch?
E/2
0
E
Time
Open Termination
Impedance in Time Domain
3
Page 5
安捷倫科技高頻元件量測研討會
Feb.23, 2006
E/2
0
E
E
R
PROBE
R
What do you expect to see at the probe before, during, and after you close the switch?
Time
Perfect Termination
Impedance in Time Domain
Page 6
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Zero Ω
∞ Ω
ZL=Z0 ΩE/2
0
E
Vi
Vr
Vr
(∆V)
ρρ
−+
=11
0ZZLImpedance Calculated from
Source Impedance and Reflection Coefficient.
i
r
VV
=ρ Reflection Coefficient:How much was reflected?
Time
Impedance Mismatch Terms
Impedance in Time Domain
4
Page 7
安捷倫科技高頻元件量測研討會
Feb.23, 2006
E
R
PROBE
? ZL = ?
Z0 = 50 Ω Vi = 200 mV
Vr = 66.6 mV
200 mVolts
0 Volts200 mV
66.6 mVVi
Vr(ΔV)
What is the value of ΖL?
Time
Mismatch Exercise
Impedance in Time Domain
Page 8
安捷倫科技高頻元件量測研討會
Feb.23, 2006
ZL
Ei Er
STEP GENERATOR
OSCILLOSCOPE
TRANSMISSION SYSTEM UNDER TEST
Ei
Er
T
Oscilloscope display when Er ≠ 0Typical Step: 200 mV, 25 kHz square wave
with 35 ps rise time
Step Reflection Testing
5
Page 9
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Distance FormulaWhere
νp = velocity of propagation
T = transit time from monitoring point to the mismatch and back
2TD p ⋅=υ
ZL
Why is the transit timedivided by two?
Mismatches Location
Page 10
安捷倫科技高頻元件量測研討會
Feb.23, 2006
The shape of the reflected wave reveals the nature and magnitude of the mismatch
What is the nature of each of the loads shown at the right?
a) SHORTb) OPENc) IMPEDANCE > Z0d) IMPEDANCE < Z0
10
0 +=+−ZZZZ
L
L
10
0 −=+−ZZZZ
L
L
Z = ?
Z = ?
Simple Loads Reflection Analyzing
6
Page 11
安捷倫科技高頻元件量測研討會
Feb.23, 2006
The shape of the reflected wave reveals the nature and magnitude of the mismatch
What is the nature of each of the loads shown at the right?
a) SHORTb) OPENc) IMPEDANCE > Z0d) IMPEDANCE < Z0
100
0 +<+−
<ZZZZ
L
L
010
0 <+−
<−ZZZZ
L
L
Z = ?
Z = ?
Simple Loads Reflection Analyzing
Page 12
安捷倫科技高頻元件量測研討會
Feb.23, 2006
The shape of the reflected wave reveals the nature and magnitude of the reflection
Complex load impedances are also identified
Series R-LShunt R-L
LR
RL
Complex Loads Reflection Analyzing
7
Page 13
安捷倫科技高頻元件量測研討會
Feb.23, 2006
The shape of the reflected wave reveals the nature and magnitude of the reflection
Complex load impedances are also identified
Series R-CShunt R-C
R C
CR
Complex Loads Reflection Analyzing
Page 14
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Analyzing Reflections of Complex Loads
8
Page 15
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Analyzing Reflections of Complex Loads
Page 16
安捷倫科技高頻元件量測研討會
Feb.23, 2006
0 0 2incident reflected measured
DUTincident reflected incident measured
V V VZ Z ZV V V V
+= =
− • −
9
Page 17
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Required Parameters by Standard – S parameters
8RapidIO
XX1.3USB 2.0XXX8Firewire
X4.1DVIXXX4.1HDMI
XXN/AEIA-108XXN/AEIA-364-90XXX4.5Serial ATAXXX6.3Infiniband
XX?IEEE 802.3aeX2.4Fully Buffered DIMM
XN/AIPC
XX
Loss
S21
X7.5Serial Attached SCSIXXPCI-XXXX1.25PCI Express
XX5PCI Express Gen 2
CrosstalkReturn Loss
S11
ImpedanceMax Freq, GHz
Standard
Add: FSB, 1 GbEIPC was Institute of Interconnecting and Packaging Electronic Circuits
Page 18
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Introducing 86100C option 202- TDR S-Parameters capability
S-parameter display TDR -> S11 (Return loss)TDT -> S21 (Insertion loss)
Single-end & Differential
No external PC needed
Run on only 86100”C”
10
Page 19
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Introducing 86100C option 202- Corrected Impedance Profile - Peeling
Peeling mitigate measurement errors caused by multiple reflections at each impedance mismatch.
Blue: raw dataYellow : corrected
Page 20
安捷倫科技高頻元件量測研討會
Feb.23, 2006
E50 75 ∞
WHAT IF THE TRANSMISSION LINE OR CABLEDOES NOT MATCH THE SOURCE IMPEDANCE?
WHAT WILL THE RESPONSE LOOK LIKE?
Matching source Z0 to transmission line ZL
11
Page 21
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Peeling - Voltage Bounce Diagram
ΓR = = 1.0ΓS = - 0.2
240 mV
-48 mV
(192 –1.92)mV
240 mV
201.6 mV
383 mV
t=0
t=2
t=4
-Γl = - 0.2
192 mV
192 mV
75Ω 50Ω Open
Keeps on approaching steady state of 400 mV
9.6 mV
Raw Voltage
Peeled Voltage
50Ω
source
Page 22
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Corrected Z Profile into a 75 Ω Cable
Peeled Trace ImpedanceNotice that multiple reflections have been removed
Peeling assumes Loss-less device.
Loss, Resistance [R], degrade accuracy of peeling.
Initial Z mismatch is the most accurate.
Benefit is enhanced with TDR calibration
12
Page 23
安捷倫科技高頻元件量測研討會
Feb.23, 2006
What is Normalization?
• Built-in Firmware• Removes Test Fixture Error• Increases Accuracy• Allows Customer to Simulate
his own system risetime
Originally Licensed from Stanford University (BracewellTransform)
Critical for Rambus!
TDR normalization (As VNA’s Calibration)
Page 24
安捷倫科技高頻元件量測研討會
Feb.23, 2006
•First part of the calibration removes systematic errors due to trigger coupling, channel crosstalk, and reflections from cables and connectors by measuring the response with the DUT replaced by a short circuit.
•The second part of the calibration generates a digital filter. The filter removes errors by attenuating or amplifying and phase-shifting components of the frequency response as necessary.
•For TDR, this is done by replacing the DUT with a termination having an impedance equal to the characteristic impedance of the transmission line, all of the energy that reaches it will be absorbed.
The procedure of TDR Normalization
13
Page 25
安捷倫科技高頻元件量測研討會
Feb.23, 2006
The procedure of TDR Normalization
Page 26
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Effect of TDR Normalization Calibration
Reference plane by CalibrationFixture DUT
Remove Error caused by Test Fixture, Cable, connectors
14
Page 27
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Fixture Error Correction Techniques
Measurement Goals•Accuracy•Repeatability•High Dynamic range•Complete characterization
= Post-measurement process= Pre-measurement process
TDR Peeling
By VNA
Source : DesignCon 2005 “Designing Transceiver FPGA's Using Advanced Calibration Techniques”
Page 28
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Comparison of TDR and PNA
PNATDR Norm@20pSTDR with RPC
•TDR has wide band receiver•Higher noise floor•Lower S/N ratio•Lower dynamic range
•VNA has narrow band receiver•Lower noise floor•Higher S/N ratio•Higher dynamic range
Agree well to 9-10 GHz
15
Page 29
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Comparison of TDR and PNA (Error Correction)
RPC
@30pS
@20pS & PNA
PhasePNA
Norm@20pS
Norm@30pS
RPC
Magnitude
TDRWaveforms
Comparison: TDR Calibration Methods versus PNA SOLT CalibrationDevice Under Test: 3.5 mm Thru AdapterResults: The magnitude loss increases as a function of frequency and is dependent upon the calibration method for TDR-based measurements. Phase error due to timing jitter exists in TDR-based measurements.
Good Calibration yields ~1dB error below 10GHz
Page 30
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Mismatch Line Device #1 Insertion Loss S21 (Magnitude dB)
-35
-30
-25
-20
-15
-10
-5
05.00E+07 2.54E+09 5.04E+09 7.53E+09 1.00E+10 1.25E+10 1.50E+10 1.75E+10 2.00E+10
-35
-30
-25
-20
-15
-10
-5
0
VNA
TDR
TDR vs VNA comparison
50MHz 20GHz10GHz
16
Page 31
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Mismatch Line Device #1 Return Loss S11 (Magnitude dB)
-35
-30
-25
-20
-15
-10
-5
05.00E+07 2.54E+09 5.04E+09 7.53E+09 1.00E+10 1.25E+10 1.50E+10 1.75E+10 2.00E+10
-35
-30
-25
-20
-15
-10
-5
0
VNA
TDR
TDR vs VNA comparison
50MHz 20GHz10GHz
Page 32
安捷倫科技高頻元件量測研討會
Feb.23, 2006
N1024A TDR Calibration KitBest for Differential normalization 2 x Loads and 2 x Shorts
2 36” SMA cables2 3.5mm-f precision load2 3.5mm-m precision load2 SMA-f flush short2 3.5mm-m flush short2 3.5mm f-f adapter1 SMA-f to BNC-m1 Torque wrench
3 36” SMA cables1 SMA-f load1 SMA-m load1 SMA-f short
2 SMA-m to BNC-f5 3.5mm 20dB pads
Accessory Kit
N1024ACurrent 54754A-100
17
Page 33
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Diff. TDR Probe
Used for handheld Differential meas.
2 x Diff. and 2 x Single Ended Probes
2 SMA cables2 SMA-m precision load2 SMA-m flush short2 Size Diff. TDR probe2 Size Single Ended Probe1 Calibration Subtract1 Hand Held holder1 ESD protection (option)
3 36” SMA cables1 SMA-f load1 SMA-m load1 SMA-f short
2 SMA-m to BNC-f5 3.5mm 20dB pads
Accessory Kit
PS-X10-100Current 54754A-100
Page 34
安捷倫科技高頻元件量測研討會
Feb.23, 2006
TDR Accessories
Do you know we have ;
N1020A-K08 FireWire TDR Cable
(IEEE 1394)
N1020A-K09 HSSDC* TDR Cable
Gigabit Ethernet (IEEE 802.3
Standard)
N1020A-K10 DB-9 TDR Cable
Fibre Channel (ANSI x3.297-1997)
* High Speed Serial Data Connector InfiniBand 1x, 4x will be available
USB2.0 & HDMI are under investigation
18
Page 35
安捷倫科技高頻元件量測研討會
Feb.23, 2006
On Wafer & Packages measuring example
Page 36
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Interconnects – Using Excess L/C
19
Page 37
安捷倫科技高頻元件量測研討會
Feb.23, 2006
How close can two reflection sites be and still be seen as independent events?
The TDR edge needs time to reach its full height before the next event is encountered
So what determines the two-event resolution?
Answer: It’s not just the TDR step speed!
Page 38
安捷倫科技高頻元件量測研討會
Feb.23, 2006
A 35 picosecond step is insufficient to see closely spaced reflections
With a 35 ps step, all you know is the device is there
If there is more than one reflection, we can’t tell
35ps
20
Page 39
安捷倫科技高頻元件量測研討會
Feb.23, 2006
High resolution allows your customers to see what they could never see before
At 9 ps step speed, we see 5 separate reflections
Each event is easily seen and quantified
9ps
V-connectorpin-collette
V-connectorpin-collette
hermeticfeedthrough
coaxialfeed-
through
microstriptransmission line
coaxial-microstriplaunch
Page 40
安捷倫科技高頻元件量測研討會
Feb.23, 2006
86100/Picosecond Pulse Labs 4020/4022 Measurement capabilities
35ps<9ps
The Picosecond Pulse Labs 4020 modules takes the 35 Picosecond pulse from the Agilent TDR and increases the speed to under 9 picoseconds
Two-event resolution is improved by a factor of 4!
(1.5 mm ‘air’, less than 1 mm in common dielectrics)
21
Page 41
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Optimizing Measurements
You will lose your edge speed if you have:
• Excess or poor quality cabling to and from the DUT
• The scope receiver channel has insufficient BW
Recommend TDR with the 86118A ~75 GHz remote plug-in:
• Max. bandwidth• Minimum cabling distances• Connector recommend as
2.4mm
4020 RemoteTDR Head
SamplingPort
DeviceUnderTest
54754A TDR module
86118A
Page 42
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Configuring a system
86100 C mainframe
54754A TDR plug-in
86118A 70 GHz plug-in • Lower BW channels can be used, but edgespeed
and resolution will be reduced• Cabling between the DUT and the receive
channel degrades TDR speed
Picosecond 4020 (Single-end) or 4022 (differential) TDR or TDT enhancement module
22
Page 43
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Physical Layer Test System (PLTS) is the Most Complete for Differential S parameters
•Extensive Calibration
•Eye diagram simulation
•N5320A-225 20G VNA
•N5320A-240 20G VNA
•54754Ax2 TDR base
Page 44
安捷倫科技高頻元件量測研討會
Feb.23, 2006
1
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 1
Using ADS for Signal Integrity Design
Signal Integrity and Advanced Design System
Agilent Technologies Ltd.
Ming Chih, Lin Application EngineerFeb, 23, 2006
Page 2
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Overview
Part I
• Unified Environment for SI Design
Part II
• Application Guides for SI• Eye Diagram• IBIS Model• Momentum
2
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 3
Using ADS for Signal Integrity Design
Unified Environment for SI Design
Page 4
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Signal Integrity Problems are Everywhere!
WafersBackplanes
IC Packages
Cables
PC Boards
3
Page 5
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Signal Integrity (SI)- What is it?
It is the application of engineering principles to:Control impedance and reflectionsMinimize parasitic and unwanted coupling effectsControl skewAdjust for skin-effect and dielectric lossesTransmitter Pre-emphasisReceiver Equalization
So chips can communicate at higher data rates
“Signal integrity is a field of study half-way between digital design and analog circuit theory”
Dr. Howard Johnson
Page 6
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Projected Increase of Clock Frequencies
Source: ITRS 2003 SIA Roadmap
0
2500
5000
7500
10000
12500
15000
17500
1998 2000 2002 2004 2006 2008 2010 2012 2014 2016
Year
Clo
ck F
requ
ency
(MH
z) on-chip on-board
Microprocessor based products
Parallel busSerial bus
Production
R&D
4
Page 7
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Computer Interconnect Standards
XAUI
On Chip
PCI 32/33 & 64/66
Chip-to-Chip Local Bus SystemBackplane
CoreConnect SCSI
USB
Serial ATA
IEEE 1394
1Gb Ethernet
CompactPCI
VME
PCI-X 66 & 100
POS-PHY L3/L4
XAUI
3GIO/PCI-Express 2.5Gb/s
RapidIO 3.125Gb/s
3.125Gb/s
Fibre-Channel
InfiniBand 2.5Gb/s
1.5Gb/sHyperTransport 1.6Gb/s
Second gen PCI-Express (5-6.25Gb/s)
6Gb/s SATA III
6.25Gb/s double XAUI
VXS Backplane (VITA41)
AdvancedTCA (PICMG 3.x)
GigE Backplane (VITA 31.1)
StarFabric Backplane (PICMG 2.17)
Serial Mesh Backplane (PICMG 2.20)VME320
1.0Gb/s
2.5Gb/s
2.0Gb/s
3.0Gb/s
5.0Gb/s
6.0Gb/s
10Gb Ethernet
CSIX
Flexbus 4
10.0Gb/s
Page 8
安捷倫科技高頻元件量測研討會
Feb.23, 2006
High-Speed Signaling Standards
StandardData Rate
(Gb/s)Driver Edge
Rate (ps)Receiver
Sensitivity
Receiver Eye Opening or Setup/Hold
Serial PCI Express (3GIO) 2.5 RapidIO Serial 3.125 10GbE XAUI 3.125 50ps 200mVpp 100ps Fibre Channel 2125 2.125 to to to Infiniband 2.5 140ps 400mVpp 140ps Serial ATA 1.5Parallel RapidIO 8/16 2 HyperTransport 1.6
Edge rates are decreasing below 100psDifferential voltage swings are shrinkingThe model bandwidth required for accurate measurements is primarily
dependent on the signal’s risetime, not its data rate.
5
Page 9
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agilent’s Signal Integrity Solutions
Page 10
安捷倫科技高頻元件量測研討會
Feb.23, 2006
CapabilitiesCapabilities
High Speed High Speed Interconnect Interconnect
LibraryLibraryMomentumMomentum Layout Layout
ComponentsComponentsAdvanced Model
Composer
Circuit and NumericCircuit and NumericCoCo--Simulation Simulation
Phys
ical
Phys
ical
Ptolemy Ptolemy Synchronous Synchronous
DataflowDataflow
HFHFSPICE SPICE
(Transient)(Transient)
LinearHarmonic Balance
Circuit Envelope
ConvolutionHigh Frequency SPICE
PtolemyPtolemy Fixed Point
Planar EM (Quasi-static)Planar EM (Full-wave)
Custom EM-based Models
AC/AC/SS--ParametersParameters
ConvolutionConvolution
Harmonic Balance
Dom
ain
Dom
ain
Num
eric
Num
eric
Tim
eTi
me
Freq
uenc
yFr
eque
ncy
Circuit Envelope
Advanced Design System
Simulation Leadership
PtolemyPtolemyTimed Timed
Synchronous Synchronous Dataflow Dataflow (TSDF)(TSDF)
Vector SignalInstrument Links
6
Page 11
安捷倫科技高頻元件量測研討會
Feb.23, 2006
YEAR
TECH
NOLO
GY
Yield Optimization
Microwave System Simulation
Harmonic Balance Simulator
1980
1990
2000
Discrete-valued Optimization
Transient Convolution Simulator
GaAs Foundry Design Kits
Multilayer Interconnect Library
Circuit Envelope Simulator
High-speed Krylov Harmonic
Balance
Yield Analysis
ROOT MOSFET, Schottky and Varactor Models
Transient-assisted Harmonic Balance
2.5D EM Adaptive polygonal meshing
Ptolemy Time-synchronous
Dataflow Simulator
8510A Vector Network Analyzer
High-Frequency SPICE
PC-based Linear Simulator
MDS for Unix
Layout-driven simulation
Agilent – 20 years of Simulation InnovationFrom Touchstone to ADS
EEsof Touchstone & Libra, Academy, Series IV
Advanced Design System
MDS
Phase-noise Analysis
Optimization from Layout
Microwave
High-speed digital
RFIC
RFIC Foundry Design Kits
Standards-based Communications
Libraries
DesignGuides
Wireless Design Libraries
RF/ Analog co-simulation
Genetic Optimization
Fast 2.5D EM simulator for RF
Smart Simulation Wizards
Developer’s Studio
RFDE
2004
Advanced Model Composer
Automated thick-metal EM
Signal Integrity DesignGuide
4-PORT VNAand PLTS
Page 12
安捷倫科技高頻元件量測研討會
Feb.23, 2006
ADS for Signal Integrity
1924 pin LTCC Package Designed using ADSMeasured vs Modeled, 50 ps TDT response
2 4 6 8 10 12 14 16 18 200 22
-30
-20
-10
-40
0
freq, GHz
dB(S
(2,1
))dB
(S(4
,3))
Measured vs Modeled, Insertion Loss (S21)
7
Page 13
安捷倫科技高頻元件量測研討會
Feb.23, 2006
4-port Measurement of Connector
Page 14
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulated Step-response of Measured Connector
differential, true/complement skew +/- 50 ps.Both output steps shown Below.
0.4 0.5 0.6 0.7 0.8 0.9 1.0-600
-400
-200
0
200
400
600
time, nsec
Vpo
rt2, m
VV
port4
, mV
•Differential pair simulated their response to a voltage step.
•Main through line is held fixed during simulation
• The complement line is swept from +/-50ps
• magnetic effect in board is evident in measurement.
8
Page 15
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulated Eye Diagram of Measured Connector
Differential, true/compliment skew +/- 50 ps• Eye opening on left-side is smaller than on right-side of crossing point for –50, -40, -30ps.
• Jitter is higher for – 50, -40 and -30 ps eyes
• Slope of rise is shallower for + and – 50 ps eyes
• Eye opening is balanced and and rise is faster for 0, +10 and +20ps
• Physical line added to board to adjust delay based on these results
Page 16
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Typical SI Problem
IBIS or Spice model
Pattern Generator
Pre-emphasis / Driver
Encoder
DecoderReceiver
Equalizer
Physical Channel
Board Traces 2” (51mm) – 10” (254mm)Card
CardPackage
Die
Package
Die Driver
Receiver
CardHigh speed Connectors
IBIS or Spice model
Backplane Traces 10” (254mm) – 40” (1016mm)
Physical Channel
Channel Adaptation
Signal Recovery
9
Page 17
安捷倫科技高頻元件量測研討會
Feb.23, 2006
PackageCard
Card
Die
Package
Die
Driver
Receiver
ADS for Signal Integrity – Link Level Simulation
Simulators• Frequency-domain • Time-domain• Numeric Domain • 3-D Planar Electromagnetic• 3-D Electromagnetic Models• Optimized equivalent circuit models • Analytic transmission line models• Static field-solver based models• EM simulation models• Models from measurements• Matlab, VHDL, C++, SystemC, Verilog_AMeasurements• TDR and TDT• 2-port and 4-port VNA• Eye Diagram
Pattern Generator
Pre-emphasis / Driver
Decoder Receiver
Encoder
Equalizer
Physical ChannelChannel Adaptation
Signal Recovery
Page 18
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulators - Solve the Whole Problem
Each part of the Link can be modeled and designed separately, however…
What happens when the parts are brought together for the first time?
Would it be beneficial if the Link could be analyzed as a whole?
10
Page 19
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agilent Ptolemy
• Agilent Ptolemy is the Data Flow simulator
• System Level Simulation Kernel • based on UC Berkeley Ptolemy• Timed Synchronous Data Flow (TSDF)• Numeric Synchronous Data Flow (SDF) • Links to other simulators & instruments
Ptolemy Data Flow (discrete numeric/time-domain)• Signal Processing Verification – FEC, Encoding/Decoding• System Performance Verification – uncoded/coded BER
Page 20
安捷倫科技高頻元件量測研討會
Feb.23, 2006
• Statically Scheduled Simulation1. A component maps input tokens (current or numbers) onto
output tokens2. A set of firing rules specify when a node (component) runs3. A firing consumes input tokens and produces output token4. Nodes (components) connected by arcs (wires)5. Schedule is constructed once and repeatedly executed
• Suitable for synchronous multi-rate signal processing• Synchronous Data Flow has no concept of time, just
numeric data• Tokens can also be “time stamped” - then they
become samples. Now you can simulate time and frequency domain models and impairments such as reflections and dispersion TSDF
Enabled
Fired
What is Synchronous Data Flow (SDF)?
11
Page 21
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Agilent Ptolemy - The “IP Integrator”ADS Ptolemy is a solution for the following:• Design & verification of Communication Systems
(physical path).• Co-Simulation of Baseband (DSP) and Analog/RF (A/RF)
circuits in a single simulation:Ptolemy system with ADS circuit simulators (which can also contain Verilog-A models).Ptolemy system with 3rd party tools (Matlab, RTL-HDL simulators, C++, SystemC)
• ”Connected Solutions” connect Simulation, Design & Verification flows to Instrumentation and Measurements
• Circuit verification with System Test Benches (WTBs) to link ADS and Cadence (IC) design tools.
Page 22
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Link Level Simulation
Trace_Spacing sets the distance betwen line pairsas a multiple of intrapair spacing.
This upsample sets measurement resolution
Bipolar signal, +1, -1
The S ampleDelay of 3 will work for thedefault design. If the Channel is changedthis delay can be set via a slider in the Tkcontroller. Set the slider to the same delayas the E ye delay slider when the widest part of the eye is at t=0 on the eye plot.
The SampleDelay of 11 will work for thedefault design. If the Channel is changedthis delay can be set via a slider in the Tkcontroller. S et the slider to the same delayas the Eye delay slider when the widest part of the eye is at t=0 on the eye plot.
For the histograms to be correctthe phase needs to be set to the correct bit slice sample delay.(11 will work for the default design).
For the histograms to be correctthe phase needs to be set to the correct bit slice sample delay.(3 will work for the default design).
Typical Pre-emphasis is progrmaple in steps from 5% to 25%This equates to 1.05 to 1.25 on this controller.
Adding noise or other artefacts to thesignal can be done in this fashion.
Channel Model
PTOLEMY-SPICE CO-SIM SI CHANNEL WITH TX PRE-EMPHASIS AND RX EQUALIZATION
(Using the "drive_lines" channel example)Post Eq measurements
Receiver with Equalization
Bit-stream Transmitterwith Pre-emphasis
Adding Noise
Post-channel measurements
Post-channelInteractive Measurements
Post EqInteractive Measurements
Pre-channel measurements
Pre-channelInteractive Measurements
1
2
TkConstellationT20
Style=dotSampleDelay=10Amplitude=1.5NumSamplesPerSymbol=Samples_per_clockLabel="Bit S lice"
1
2
TkConstellationT18
Style=dotSampleDelay=6Amplitude=1.5NumSamplesPerSymbol=Samples_per_clockLabel="Bit S lice Eq"
1 2
RateLimiterR8RMax=4e10
1 2
drive_lines_cosim3_sub_s2pX7Trace_Spacing=3
I O
31
2
LMS_TkPlotL4
Identifier="LMS filter taps"SaveTapsFile="tapsout.tap"ErrorDelay=1StepSize=0.02DecimationPhase=0Decimation=1Taps="-2.3 0 0.4 1.0 1.3 1.3 1.2 0.8 -0.1 -0.8 -1.6"
1 2
FloatToTimedF3TStep=Sample_step sec
1 2
RepeatR5
BlockSize=1NumTimes=Samples_per_clock
1 2
LogicToNRZL5Amplitude=1.0
1
2
3
Add2A2
1
Tk Sl i de rSc al e 1
Gra nu la rity =1 00Pu tIn Co ntrol Pan el =YESIde nti fi e r= "Pre-e m ph as i s Ga in "Va l ue =1Hi g h= 2Lo w= 1
1 2
3
P re-EmphasisX 6
3
21 12
3
S plitterRFS 2
1
BitsB1
LFSR_InitState=1LFSR_Length=12ProbOfZero=0.5Type=Random
1
ConstC2Level=0.0
1
TkHistogramT19
DataPoints=10000NumberOfBars=32Bottom=-1.5Top=1.5Label="Bit S lice Histogram Eq"
1 2
DownSampleD3
P hase=11Factor=S amples_per_clock
1
TkHistogramT21
DataPoints=10000NumberOfBars=32Bottom=-1.5Top=1.5Label="Bit S lice Histogram"
1 2
DownSampleD2
P hase=3Factor=S amples_per_clock
1
TkEyeT14
Amplitude=1NumSymbols=2NumSamplesPerSymbol=S amples_per_clockLabel="Pre Channel Eye"
1
SpectrumA nalyzerPre_channel
WindowConstant=0.0Window=K aiser 7.865Stop=Data_Collection_time nsecStart=DefaultTimeStartRLoad=DefaultRLoadPlot=None
1
SpectrumAnalyzerPost_channel
WindowConstant=0.0Window=Kaiser 7.865Stop=Data_Collection_time nsecStart=DefaultTimeS tartRLoad=DefaultRLoadPlot=None
1
TimedS inkPost_E q_t
ControlSimulation=YESStop=Data_Collection_time nsecStart=DefaultTimeS tartRLoad=DefaultRLoadPlot=None
1 2
TimedToFloatT22
12
1
RR7R=50 Ohm
2
1
RR3R=50 Ohm
1
TkPlotT13
Style=connectyRange="-2 2"xRange="0 1000"yTitle="amplitude"xTitle="waveform"Label="Pre Channel scope"
1
TkEyeT5
Amplitude=1.5NumSymbols=2NumSamplesPerSymbol=Samples_per_clockLabel="Eye Post Channel "
1
TimedS inkPre_channel_t
ControlSimulation=YESStop=Data_Collection_time nsecStart=DefaultTimeStartRLoad=DefaultRLoadPlot=None
1 2 31
2
SubS9
2
1
RR6R=50 Ohm
1
1 2
TimedToFloatT12
1
TkPlotT7
Style=connectyRange="-1 1"xRange="0 1000"yTitle="amplitude"xTitle="waveform"Label="Post Channel scope"
12
Del a yD1N=1
1
ConstC1Level=0.0
1
TkEyeT17
Amplitude=1.5NumSymbols=2NumSamplesPerSymbol=Samples_per_clockLabel="Eye Eq"
12
3
SplitterRFS7
1
TimedS inkPost_channel_t
ControlSimulation=YESStop=Data_Collection_time nsecStart=DefaultTimeS tartRLoad=DefaultRLoadPlot=None
12
3
S plitterRFS 6
1
1 2
TimedToFloatT6
12
3
SplitterRFS8
12
3
S plitterRFS 4
DFDF2
VA RVA R1
Data_Collection_start=4Data_Collection_time=40Sample_step=1/S ample_rateSample_rate=Clock*S amples_per_clockSamples_per_clock=12Clock=12.8e9
EqnVar
1
2
3
Mpy2M3
1
TkSl id erSca l e2
Gran ul ari ty= 10 0PutInCon tro l Pa ne l= YESId en ti f i er="Add ed Noi s e"Val u e= 0Hi gh =1L ow=0
12
FIRF4
Interpolation=1DecimationP hase=0Decimation=1
1
IID_GaussianI1
V ariance=.1Mean=0
Pre-emphasis Channel Equalization
12
Page 23
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Transmitter with Pre-Emphasis
Page 24
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Channel Model
Aggressor Lines
Channel Subnetwork
13
Page 25
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Channel Subnetwork
Page 26
安捷倫科技高頻元件量測研討會
Feb.23, 2006
ModelsAccurate models of interconnects
Accurate models of the active devices
Robust simulator
Accurate Prediction of performance
+
+
=
The earlier in the design cycle that problems are found and designed out, the shorter the cycle time,
the lower the development costs
Accurate models of Tx and Rx Functions+
14
Page 27
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Interconnect Models
Account for impedance, delay, conductor loss, dielectric loss, and coupling
Momentum EM simulator for arbitrary planar structures. Has layout and schematic representations
Analytic models are fast, and have a layout and schematic representation
Multilayer Interconnect Models use a built-in field-solver, and have both layout and schematic representations
Page 28
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Current ADS Capabilities• Integrated Momentum EM Simulator• Momentum RF• Model Composer• Co-Simulation w/Layout Components• Co-Optimization w/Layout Components• Layout lookalike components
30 GHz transition
Layout editing and port
assignment
Layout look-alike component in the
schematic
EM Models - Momentum
15
Page 29
安捷倫科技高頻元件量測研討會
Feb.23, 2006
• near field / low freq approximation
L(ω) = L0 + L1ωR + L2(ωR)2 + …
C(ω) = C0 + C1ωR + C2(ωR)2 + …
• neglecting far field radiation
• [L0] & [C0] frequency independent
• [Z0] matrix reload very fast
Momentum RFQuasi-Static Electromagnetics
Low Frequency approximation :
|'|1'|| rrrr −−≈−− jke jk
[Z] matrix load is frequency independent !
[Z].[I]=[V]
[S]
Mesh strips, Vias and slots with rectangles and triangles (conformal surface mesh)
Performs mesh reduction while maintaining integrity of solution
Model surface current in each mesh cell (linear distribution)
Solve matrix equation for the unknown current coefficients
Calculate S-parameters
Electro- and magnetostatic Green’s functions
Quasi-static frequency scaling (jw, 1/jw)
L’s and C’s are real and frequency independent
R’s are complex (DC loss + skin effect √w)mesh topology
reduction
reduction 1 cell
4 cells
10 cells
Page 30
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Comparison of Models
• quasi-static inductance . . . . . . • quasi-static capacitance . . . . . • DC conductor loss (s) . . . . . . . .• DC substrate loss (s) . . . . . . . . • dielectric loss (tgd) . . . . . . . . . . . . . . . . . . . . . . . . . . .• skin effect loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Layout
S parameters
RF
Spice model
DC Spice Momentum
16
Page 31
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Models in Matlab, VHDL, C++, SystemC, Verilog-A
IP for “Link Level” blocks and functions often already exist, e.g.
Pre-emphasisEQEncoding/DecodingInterleaving/De-interleavingSource descriptions
Easy to include with Ptolemy, either natively or with Co-simulation
Page 32
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Models - MATLAB Co-Simulation in ADS Ptolemy
scripting code
MatlabSinkM6
NumberOfFirings=1MatlabWrapUp=""MatlabFunction=""MatlabSetUp=""DeleteOldFigures=YESScriptDirectory=""
MatlabCx_MM1
MatlabWrapUp=""MatlabFunction=""MatlabSetUp=""DeleteOldFigures=YESScriptDirectory=""
Matlab_MM8
MatlabWrapUp=""MatlabFunction=""MatlabSetUp=""DeleteOldFigures=YESScriptDirectory=""
MatlabSinkFM7
NumberOfFirings=1MatlabWrapUp=""MatlabFunction=""MatlabSetUp=""DeleteOldFigures=YESScriptDirectory=""
MatlabFCx_MM2
MatlabWrapUp=""MatlabFunction=""MatlabSetUp=""DeleteOldFigures=YESScriptDirectory=""
MatlabF_MM3
MatlabWrapUp=""MatlabFunction=""MatlabSetUp=""DeleteOldFigures=YESScriptDirectory=""
script files
MATLAB compiler – generate native shared librariesMatlabLibLinkCx
M5
Mode=AUTOFunction=""SetupParm=""Setup=""Library=""
MatlabLibLinkM4
Mode=AUTOFunction=""SetupParm=""Setup=""Library=""
Co-simulate with Matlab in ADS
Use full Matlab UI or as math function only
Components in “Numeric Matrix” library
17
Page 33
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Models - HDL Co-Simulation in ADS Ptolemy
In the HDL Blocks library
VxlCosimV1
HdlSimulatorGUI=OffHdlModelName=""OutputPrecisions=""Outputs=""InputPrecisions=""Inputs=""HdlSrcFile=""
OUT
IN
Clock
___ Set
NCCosimN1
HdlSimulatorGUI=OffHdlModelName=""OutputPrecisions=""Outputs=""InputPrecisions=""Inputs=""HdlSrcFile=""
OUT
IN
Clock
___ Set
HdlCosimH1
HdlSimulatorGUI=OffHdlLibrary=""HdlModelName=""OutputPrecisions=""Outputs=""InputPrecisions=""Inputs=""HdlSrcFile=""
OUT
IN
Clock
___ Set
Page 34
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Integration of other ModelsC++ Based Model Creation and SystemC
C++• Native Language for Ptolemy Kernel• Model Development Through GUI and Command Line
Interface• Full Debug Capability Through 3rd Party Development
Environments• Microsoft Visual Studio for PC Platforms
SystemC• Application Note 1482: Importing SystemC Designs into
Advanced Design System• http://eesof.tm.agilent.com/pdf/5989-0234EN.pdf
18
Page 35
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measurement-based Models Probing Solution + PLTS + ADS
ADS Design SW
S-parameters• Citifile• TouchstoneRLCG• Measured Parameters• ML2CTL Model
Device Under Test, Microprobes & Probe Station
4-Port TDR or VNAPLTS Software
View and analyze measurement data
Calibration and Measurements
View and analyze measurement data
using PLTS software
Page 36
安捷倫科技高頻元件量測研討會
Feb.23, 2006
TranTran2
MaxTimeStep=Sample_stepStopTime=Data_Collection_time nsec
TRANSIENT
DataAccessComponentDAC1
iVal2=Trace SpacingiVar2="Trace_Spacing"iVal1=freqiVar1="freq"ExtrapMode=Interpolation ModeInterpDom=RectangularInterpMode=LinearType=DatasetFile="drive_lines_s2p.ds"
DAC
2
3
4
1
S2P_EqnS2P1
Z[2]=fileDAC1, "PortZ[2]"OhmZ[1]=fileDAC1, "PortZ[1]"OhmS[2,2]=fileDAC1, "S[2,2]"S[2,1]=fileDAC1, "S[2,1]"S[1,2]=fileDAC1, "S[1,2]"S[1,1]=fileDAC1, "S[1,1]"
1 1
1
PortP1Num=1
1
PortP2Num=2
Link Level SimulationData based Channel Models can be derived from simulation models or measurements.
Radia l1
arbitrary4p_1
Radial2
CLin6
Slant 1
R4
R12
CLin5
CLin4
R10
R9
arbit rary4p
MLRADIAL2 MLRA DIAL2
MLSLANTED2
ML6CTL_V
ML4CTL_V ML4CTL_V
R
R
R
R
Channel Model
1 2
drive_lines_cosim3_sub_s2pX7Trace_Spacing=3
I O1
SplitterRFS2
2
3
SplitterRFS4
Direct co-sim
ulation
Parameterized Data
Simulated ChannelTDR or VNA Measurements
19
Page 37
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Bringing it All Together
•ADS has been used for SI design for over 10 years
•ADS can act as a scalable design whiteboard
•ADS encourages you to be curious about your design ideas
•ADS has a multitude of accurate built-in models
•ADS allows you to build accurate physical models
•ADS lets you use your existing models
•ADS agrees with measurements
•ADS shows results the way you want to see them
•ADS brings IP, simulations and measurements together
•ADS Ptolemy lets you work with the whole Link before you build it!
Page 38
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Signal Integrity Resources
Gigatest for probe stations and package modeling services www.GigaTest.com
CST for 3D EM Simulator www.cst.com
Fastest, Inc for designing high speed hardware and resolving signal integrity problems. www.fastestco.com
Admos for active and passive model extraction serviceswww.admos.com
20
Page 39
安捷倫科技高頻元件量測研討會
Feb.23, 2006
SI Resources on the Agilent EEsof website
Agilent EEsof EDA home pagehttp://eesof.tm.agilent.com
Signal Integrity Applications and Wireline Applications http://eesof.tm.agilent.com/applications/signal_integrity-b.htmlhttp://eesof.tm.agilent.com/applications/wireline-b.htmlMomentumhttp://eesof.tm.agilent.com/products/e8921a-a.htmlAgilent Signal Integrity eSeminar Series www.agilent.com/find/sigintNetSeminar: Challenges of Differential Bus Design -http://eesof.tm.agilent.com/news/news400.html#bus_design
Page 40
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Application Web site for 2005A
http://eesof.tm.agilent.com/applications/signal_integrity-b.html
21
Page 41
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Signal Integrity Training Class
http://www.agilent.com/find/educationSignal Integrity Class: N3215A Designing for Signal Integrity with ADS
Transient Simulation SetupConvolution and Frequency-Domain SimulationsTransmission Lines, Crosstalk and Resonances TDR/TDTNoise and Jitter2.5D EM Simulations (Momentum)Differential Circuits and Mixed-Mode S-Parameters
– Length – 2 days – hands-onPtolemy training: N3206A Signal Processing using ADS
Ptolemy SDF / TSDFCo-simulation
• Length – 3 days – hands-on
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 42
Using ADS for Signal Integrity DesignModels, Simulations and Measurements
22
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 43
Using ADS for Signal Integrity Design
Application Guides for SI
Page 44
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Amplifier
DesignGuides in ADS – Bridging the Gap
FilterSI
Linearization
Mixer OscillatorPLL
RF System
Passive
DesignGuides
Simulation TechnologyApplications
Linear, NonlinearCircuit EnvelopeTime DomainAgilent PtolemyElectromagneticOthers
Amplifier, FiltersMixers, OscillatorPassives, SystemMod/DemodsPackagingRadar, A-to-D, UWB,High-Speed Digital
23
Page 45
安捷倫科技高頻元件量測研討會
Feb.23, 2006
DesignGuides / Application Guides
IBIS model import and examples
Signal integrity simulations and examples
Highspeed circuits typical for wireline
Helps SI designer to use ADS for common tasks
Des
ignG
uide
sA
pplic
atio
n G
uide
s
Page 46
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Wireline ApplicationsPhotodiode
Bandgap
Transimpedance Amp
Limiting Amp
Laser Driver
VCSEL
TWA
Ring Oscillator
24
Page 47
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Wireline Applications
Multiplexer
Buffer, Divider
Latch, Selector
Page 48
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Wireline Applications Ckt-level Logic
Bandgap
Behavioral Logic
D Flip Flop
Phase Detectors
Multivibrator VCOClock Recovery
25
Page 49
安捷倫科技高頻元件量測研討會
Feb.23, 2006
IBIS Library
IBIS Model Import
Page 50
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Ideal Source BuffersOctal Pulse Source
Octal Loads
Oscilloscope Probe
Components from the IBIS Library Component Palette
Impedance Optimizer
Impedance Meter
26
Page 51
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Signal Integrity Applications
Eye Diagram measurements including jitter, FrontPanel and DCA file import
Single Ended TDR/TDT
Linear/Nonlinear Differential TDT
Mixed Mode S-ParametersImpedance Simulations
Pre-emphasis and Equalization
Page 52
安捷倫科技高頻元件量測研討會
Feb.23, 2006
TDR Simulation Instrument
27
Page 53
安捷倫科技高頻元件量測研討會
Feb.23, 2006
TDR Responses Using Time-domain Simulation
Page 54
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential Nonlinear Test Component
28
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 55
Using ADS for Signal Integrity Design
Eye Diagram
Page 56
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Qualitative vs. Quantitative: How close are these waveforms?“This 40 Gb/s eye diagram was then compared to the 40 Gb/s eye diagram
derived from the internal PLTS eye diagram generating algorithms. The qualitative correlation of these two simulated eye diagrams was very good.“
DesignCon 2004 paper “Utilizing TDR and VNA Data to Develop 4-port Frequency Dependent Models for Simulation” Mayrand, Resso, Smolyansky
29
Page 57
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Data Display strengths and weaknessesData Displays are flexible but require extensive use of equations and a
lot of user expertise.
Some measurements are calculated, but most characteristics are left up to the user to determine.
Page 58
安捷倫科技高頻元件量測研討會
Feb.23, 2006
FrontPanels: Eye DiagramConvenience – dedicated display and push-button measurements focused on
a common task. Provides over 30 data display pages and over 300equations.
Consistency – measurement algorithms checked against instrumentation.
30
Page 59
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Inspired by Agilent DCA-J Instrument
Oscilloscope and Eye Modes of operation
Measurements/TestsSummary of measurement results
Histograms
Pointers, masks
Page 60
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Oscilloscope Mode
31
Page 61
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Eye/Mask Mode
Page 62
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Convert DCA data (*.csv format) into Dataset
Data Parser available in Signal Integrity Application Guide
*.csv file from DCA
Dataset file
32
Page 63
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Compare Eye FP to DCA on Same Data
DCA Measurement
1G Data Rate through 20” trace
Eye Diagram FrontPanel on the DCA output file.
[NOTE: FrontPanel data is from a single trace. DCA measurements are averaged with 16 measurements.]
5.7/5.85.7pS5.8pSJitter rms22.2/22.222.2pS22.2pSJitter p-p
10.01/10.2310.0710.18Eye S/N273.9/274.9274.7mV274.7mVEye Height386.3/389.4389mV389.4mVEye Amp
211/219215pS212pSFall Time219/226222pS219pSRise Time
-177.5/-178.5-177.5mV-177.4mVLevel 0211.3/213.5211.3mV212mVLevel 1
DCA min/max
DCA typFrontPanel
[NOTE: FrontPanel results are preliminary until ADS 2005A final release.]
Page 64
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Compare Eye FP to DCA on Same Data
DCA Measurement
5G Data Rate through 20” trace
Eye Diagram FrontPanel on the DCA output file.
[NOTE: FrontPanel data is from a single trace. DCA measurements are averaged with 16 measurements.]
12.43/12.6512.5pS12.6pSJitter rms53.28/54.0254.02pS53.44pSJitter p-p
3.79/3.803.793.79Eye S/N60.8/61.561mV61.3mVEye Height
293.4/293.8293.7mV293.8mVEye Amp140.6/140.6140.6pS140.6pSFall Time136.2/136.9136.2pS136.3pSRise Time-136.6/-136-136mV-136mVLevel 0156.4/157.5157.5mV157mVLevel 1
DCA min/maxDCA typFrontPanel
[NOTE: FrontPanel results are preliminary until ADS 2005A final release.]
33
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 65
Using ADS for Signal Integrity Design
Example Measurements
Page 66
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Common SI Problem
Objective: 1m of “improved FR-4” through multiple high speed connectors.10” (254mm) Line Card > 20” (508mm) Backplane > 10” (254mm) Line Card. Check Eye Diagram at various points along the path.
Board Traces 2” (51mm) – 10” (254mm)
Backplane Traces 2-3 chassis/rack = 10” (254mm) worst case5-8 chassis/rack = 40” (1016mm) worst case
Card
CardPackage
Die
Package
DieDriver
Receiver
CardHigh speed Connectors
IBIS or Encrypted Hspice model
IBIS or Encrypted Hspice model
34
Page 67
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Example Test Board
Several trace lengths on “improved FR-4” six metal layers, 62.5 mil total thickness
10” (254mm) 15” (381mm) 40” (1016mm)
30” (762mm) 20” (508mm)
Launch detail
Page 68
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measurements on various length of trace
10” (254mm)15” (381mm)
20” (508mm)
30” (762mm)
40” (1016mm)
S-Parameter measurements
Differential S-Parameter measured with PLTS
Traces are lossy
Measurement shows multiple reflections
Data is band limited
35
Page 69
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Create Models from Data
Data-based ModelMeasure with 4-port NWA
Layout Look-alike component
Page 70
安捷倫科技高頻元件量測研討會
Feb.23, 2006
DCA/BERT Measurement SetupMeasure the input waveformBypass the Board, record the waveform
Capture waveform with Connection Manager or output *.csv file to translate to dataset.
E8251A+5dBm
1 GHz, 2.5 GHz, 5 GHz11636B
Splitter
86100C
Trigger
BERT
E8251A+5dBm
1 GHz, 2.5 GHz, 5 GHz11636B
Splitter
86100C
Trigger
BERT
PRBS Data Out
Measure the DUT (board and cable)
*.csv *.ds
PRBS Data Out
36
Page 71
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulation through Measured S-Parameter Data
SimulationSimulation of
measurement-based waveform through measured S-Parameter data for the board and cable.
Adjust gain to compensate for loss through resistor. May need to adjust gain slightly to match measurements.
Page 72
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulation compared to DCA Measurement
1G Data Rate through 20” trace
DCA Measurement of BERT
Simulation with Measured Data
Gain set to 1.95
5.7/5.85.7pS5.7pSJitter rms22.2/22.222.2pS22.17pSJitter p-p
386.3/389.4389mV389mVEye Amp211/219215pS192pSFall Time219/226222pS197pSRise Time
-177.5/-178.5-177.5mV-179mVLevel 0211.3/213.5211.3mV210.8mVLevel 1
DCA min/max
DCA typFrontPanel
[NOTE: FrontPanel data is from a single trace. DCA measurements are averaged with 16 measurements.]
[NOTE: FrontPanel results are preliminary until ADS 2005A final release.]
37
Page 73
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulation compared to DCA Measurement
2.5G Data Rate through 20” trace
DCA Measurement of BERT
Simulation with Measured Data
Gain set to 1.90
9.1-9.49.4pS9.39pSJitter rms37.1-38.537.1pS40.8pSJitter p-p344-345.8344mV346mVEye Amp172-176172pS157.9pSFall Time169-178178pS157.9pSRise Time
-158.6-157.4-157.7mV-158.9mVLevel 0186.8-188.6186mV187mVLevel 1
DCA min/max
DCA typFrontPanel
[NOTE: FrontPanel data is from a single trace. DCA measurements are averaged with 16 measurements.]
[NOTE: FrontPanel results are preliminary until ADS 2005A final release.]
Page 74
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulation compared to DCA Measurement
5G Data Rate through 20” trace
DCA Measurement of BERT
Simulation with Measured Data
Gain set to 1.90
12.43/12.6512.5pS12.3pSJitter rms53.28/54.0254.02pS54pSJitter p-p293.4/293.8293.7mV292mVEye Amp140.6/140.6140.6pS133pSFall Time136.2/136.9136.2pS133pSRise Time-136.6/-136-136mV-137mVLevel 0156.4/157.5157.5mV155mVLevel 1
DCA min/max
DCA typFrontPanel
[NOTE: FrontPanel data is from a single trace. DCA measurements are averaged with 16 measurements.]
[NOTE: FrontPanel results are preliminary until ADS 2005A final release.]
38
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 75
Using ADS for Signal Integrity Design
IBIS Model
Page 76
安捷倫科技高頻元件量測研討會
Feb.23, 2006
IBIS (Input/Output Buffer Information Specification) models enable IC vendors to communicate device characteristics without revealing circuit / process information.
Behavioral Level
Digital IC
Digital IC
Simulation requirements• Non Linear driver and receiver circuits• High speed board layout ( critical paths)
Circuit Level
Intellectual property
IBIS Model
IBIS Model
39
Page 77
安捷倫科技高頻元件量測研討會
Feb.23, 2006
IBIS Model
A basic IBIS model consists ofFour I-V curves: pullup & Power clamp, pulldown and Gnd
clamp
Two ramps (Rampup and Rampdown) and/or Two/Four Vttables
Die capacitance: C_comp
Packaging: RLC values
For each buffer on a chip
Ramp up (or Vt table)
Ramp down (or Vt table)
Pull up I-V
Pull down I-V
Power clamp I-V
GND clamp I-V
inputI/O pin
Vcc
Gnd
Page 78
安捷倫科技高頻元件量測研討會
Feb.23, 2006
For one pair of V(t) waveform,The other pair is created for a different value of V-fixture.
Red- V(t) waveformpair in IBIS data
Blue- SynthesizedV(t) waveformpair in IBIS data
40
Page 79
安捷倫科技高頻元件量測研討會
Feb.23, 20062.0E-8
2.2E-8
2.4E-8
2.6E-8
2.8E-8
3.0E-8
3.2E-8
3.4E-8
3.6E-8
3.8E-8
1.8E-8
4.0E-8
1.0
1.5
2.0
0.5
2.5
time
Am
plitu
de
Rise/Fall TimeVt Table
Comparison of IBIS model Vt table vs. Rise/FalltimeRise/Fall Time
uses an average value of risetime/falltime to set the time constant of a capacitor.
Usable back to IBIS v1.0
Uses SDD-based equivalent circuit
Can be used in HB as well as Transient simulation
Vt Table
uses voltage lookup table (Vt)
Usable starting with IBIS v2.x
Uses FDD-based equivalent circuit
Can be used only in Transient simulation
Page 80
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Importing IBIS Component
1. Select*.ibs file
2. Define design kit name
3. Define bitmap label
4. Create design kit
41
Page 81
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Installation of design kit
Select the design kit from $HOME directory
Install design kit
Page 82
安捷倫科技高頻元件量測研討會
Feb.23, 2006
IBIS design kit components
42
Page 83
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Simulation Setup and Simulation Results
Pin Number
V_fixture
Page 84
安捷倫科技高頻元件量測研討會
Feb.23, 2006
IBIS Model with Transistor Level Simulation
43
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 85
Using ADS for Signal Integrity Design
Momentum
Page 86
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Momentum RF for Digital Board Interconnect
full boardisolated trace
port 1
port 2
port 1
port 2
S(1,1)
isolated trace
S(1,2)
isolated trace
MomentumMomentum RF
S(1,1)
full board
S(1,2)
full board
44
Page 87
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Digital Board Interconnect - off resonance
isolated trace
port 1
port 2
harmonic signal 0.4 GHz
output
S(1,1)
isolated trace
S(1,2)
isolated trace
Page 88
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Digital Board Interconnect - on resonance
isolated trace
port 1
port 2
harmonic signal 2.33 GHz
resonanceblocks the signal
no output
S(1,1)
isolated trace
S(1,2)
isolated trace
45
Page 89
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Digital Board Interconnect
harmonic signal 2.33 GHz
harmonic signal is coupled to neighboring traces and spread around the board
Page 90
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Momentum RF for Package Models
S(1,3)
S(1,1)
Momentum
Mesh: 20 cells/wavelength, 5 GHz
Matrix size : 8244Process size : > 1 GBUser time : > 2 days
Momentum RF
Mesh: 20 cells/wavelength, 5 GHz
Matrix size : 1354Process size : 106.57 MBUser time : 5h 17m 53s
Rule of thumb: freq < 13.8 GHz
port 1
port 2
port 3
7.6 mm
7.6 mm
port 4
ref 3
ref 4
S(1,2)
S(1,4)
1 2
4 3
ref 3ref 4
GNDVboard
Vchip
epoxy
FR4
96-Ball Grid Array Package
PC-NT Pentium II workstation (330 MHz)
46
Page 91
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Transient Simulation from Schematic
Layout lookalike component used in the schematic
Page 92
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Bringing it All Together
•ADS has been used for SI design for over 10 years
•ADS can act as a scalable design whiteboard
•ADS encourages you to be curious about your design ideas
•ADS has a multitude of accurate built-in models
•ADS allows you to build accurate physical models
•ADS lets you use your existing models
•ADS agrees with measurements
•ADS shows results the way you want to see them
•ADS brings IP, simulations and measurements together
•ADS Ptolemy lets you work with the whole Link before you build it!
1
安捷倫科技高頻元件量測研討會
Feb.23, 2006Page 1
Advanced Calibration Techniques and Fixturing
Issues for VNAs
Agilent Technologies Ltd.
Ming-Fan, Tsai Application EngineerFeb, 23, 2006
Page 2
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
2
Page 3
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Introduction - Goals and Objectives
Course Goal• Understand new calibration paradigm and features of PNA• Get hands-on time on analyzer
Objectives• Upon completion of this course you will be able to:
– Calibrate a PNA and save a user cal set– Explain concept of Unknown Thru cal– Understand how a two-tier TRL cal is done– Recommend best cal approach for fixtured measurements
Page 4
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Welcome to PNA Firmware Revision A.06.0x
A.01.xx A.02.xx A.03.xx A.06.xx
A.04.xx
A.05.xxOriginal RF PNA code(released Sep 2000)
Added 3-port RF and 50 GHz PNAs(released Dec 2001)
Added 20, 40, 67 GHzand frequency offset(released Dec 2002)
Added PNA-L models(released Dec 2003)
4-port PNA-L ONLY(released Aug 2004)
Re-merged code set(released Dec 2005)
“Hawaii”
3
Page 5
安捷倫科技高頻元件量測研討會
Feb.23, 2006
What’s Totally New in 6.0…
• Target release date: 12 December, 2005• Features
– Calibrate using external trigger (e.g., during wideband pulse detection)– Calibrate with offset loads– External test set control– New FCA capability
• file embedding during cal(for wafer probes and fixtures)
• fixed input for up converters
– 1.1 GHz CPU and related capabilities– Agilent VEE Runtime installed
Page 6
安捷倫科技高頻元件量測研討會
Feb.23, 2006
What’s New in 6.0 From 5.0…
TRL calibration for 4-port PNA-LDedicated calibration window Calibration class label Data-based cal kits can now be modified Guided SmartCal supports ECal modulesOption H11 verification Safely shutdown the PNA without a mouse New *.csa save/recall file type 4-port fixture simulator functions
• 4-Port network embed/de-embed• Balanced conversion• Differential-/ common-mode port Z conversion• Differential matching circuit embedding
Automatic port extensions Interface control Agilent 5091A test set control 8 traces per window (previously 4)
C∆ on status bar .pdf version of Help file available New 4-port PNA Models:
• N5230A Options 240 and 245• Balanced measurements
Calibration registers Number of user ranges expanded to 16
(Stats and Markers) Port extension toolbar features Magnitude offset2-port fixture compensation Two LO sources for millimeter-wave measurements Material handler trigger controlGlobal pass / fail dialog Revised operator's checkRevised system verificationGuided calibration COM interface
Receiver power cal saved with cal set
Note: Highlighted text describes features that are new for 2-port PNA/PNA-L models. These features have already been released for 4-port PNA-L models.
covered in this module
4
Page 7
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Will A.06.xx Work on Discontinued PNAs?
“T1” RF PNA’s (E8356/7/8A): Yes, with upgrade to Windows XP
“T2” 2-port (E8801/2/3A): Yes, with upgrade to Windows XP
“T2” 3-port (N3381/2/3A): No
“M1” (E8364A): Yes, with upgrade to Windows XP
XP upgrade: Order new disk drive from Agilent
x
Order N8980A ($550)
Page 8
安捷倫科技高頻元件量測研討會
Feb.23, 2006
What’s New for the ENA
Rev 6.0 released 11/1/05
New features
• TRL/LRM & waveguide calibration• Automatic port extensions• 20,001 measurement points• Complex reference impedance for balanced matching• Flexible marker value display function• User preset• User recovery • 13/16-port configurable test set control function
More info at: kobemktg.jpn.agilent.com/field_eng/product/ena/customer_viewable/index.htm
5
Page 9
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
Page 10
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Old Calibration Paradigm
All calibrations were saved in a “calibration set”
“.cst” file pointed to data in a calibration set, but did NOT contain calibration data itself
“.cal” files contained calibration data but NOT instrument parameters
Saves instrument state using current state name(does not automatically select a .cst file if current state is .sta)
Opens a file dialog box to save instrument state with a new file name (choose .sta or .cst)
Saves instrument state with an auto-incremented name that won’t overwrite existing states (e.g. “at004.cst”)
Old file save dialog
6
Page 11
安捷倫科技高頻元件量測研討會
Feb.23, 2006
New Calibration ParadigmCal Registers
• Each channel has its own cal register• All cal data is saved to the channel cal register• Data in cal register is “temporary” –
data is saved on hard drive, BUT will be overwritten by the next calibration in that channel
User Cal Sets• Old “Cal Sets” are now called “User Cal Sets”• User must manually save data to a User Cal Set• User cal sets are stored on the hard drive as before
You no longer have a choice to save an instrument state!
Page 12
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Cal Registers Should Be Familiar…
Cal register model is similar to that used in 8720s and 8510s
Cal registers prevent accumulation of unneeded cal sets
7
Page 13
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Old Calset File Location
All calset data was contained in this file
Page 14
安捷倫科技高頻元件量測研討會
Feb.23, 2006
New Cal File Locations
8
Page 15
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Saving Instrument States and Cal Data*.cst - save instrument state and reference pointer to the cal set data
• data could be in cal register (not recommended)• data could be in a user cal set (recommended)*.csa - save instrument state and actual cal data (cal/state archive)
*.sta - save instrument state ONLY (no calibration data)
*.cal - save actual calibration data ONLY (no instrument state)
New!
Page 16
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Instrument States and Cal Data
Limited Instrument State• Frequency range• Number of points• IF bandwidth• Sweep type• Sweep mode
• Alternate sweep• Port powers• Source attenuators• Receiver attenuators• Testset map
Extended Instrument State• Channels/Traces• Windows• Triggering• Format• Scale
• Averaging• Markers• Math/memory• Limits• More…
Pointer to Calset• Name• Description• GUID
Error Terms• Crosstalk (1,2)• Crosstalk (2,1)• Directivity (1,1)• Directivity (2,2)• Load match (1,2)• Load match (2,1)
• Reflection tracking (1,1)• Reflection tracking (2,2)• Source match (1,1)• Source match (2,2)• Transmission tracking (1,2)• Transmission tracking (2,1)
.cst
.sta
.cal
.csa
9
Page 17
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Receiver calibrationReceiver calibration gives corrected absolute power reading (forunratioed traces as opposed to an S-parameter traces)
Often used with FOM for measurements like harmonics, TOI
Uses power-meter-corrected internal source as a reference (source power cal)
Now saved as a cal set
Multiple receiver cals requires separate channels for each cal
Source power calibrations• No change from current behavior• Saved as part of instrument states• Not saved in cal sets
New!
Page 18
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Correction Indicator on Status Bar
C - Full correction“C” is displayed immediately after a calibration is performed or when a valid Cal Set is applied. If you require optimum accuracy, avoid adjusting the analyzer’s settings after calibration.
C* - Interpolated correction"C*" appears in the status bar when interpolation is used during a measurement.
C∆ − Changed settings"C∆" appears in the status bar when one or more of the following stimulus settings change. The resulting measurement accuracy depends on which parameter has changed and how much it has changed. For optimum accuracy, recalibrate using the new settings.
• Sweep time• IF bandwidth• Port power• Stepped sweep enabled/disabled
LowestNo correctionNo CorUncertainChangedC∆UncertainInterpolatedC*HighestFullC AccuracyCorrection levelSymbol
10
Page 19
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
Page 20
安捷倫科技高頻元件量測研討會
Feb.23, 2006
TRL Calibration with 4-port 20 GHz PNA-L’s
Yes, it can be done! (requires firmware A.05.25 and later)
Is “true” TRL calibration, not an approximation like TRL*
Two-step (two-tier) calibration process:• Step 1: “Delta Match” calibration• Step 2: Normal TRL calibration
What is a “Delta Match” cal?• characterizes the difference (delta) between the source match
and load match of each port of the VNA• allows a VNA having a single reference receiver to
perform TRL-style calibrations (including Unknown-Thru)
Why 2 steps?
11
Page 21
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Assumptions for TRL (and Unknown Thru)
8-term error model assume match at each test port remains constant, independent of whether it is a source or receiver
With real hardware, assumption is poor due to port switch
Generalized S-parameters for a two-port device:
1
11
2122
12
22
22
11
2122
12
11
11
2221
1211
1
1−
•
=
aa
aa
ab
ab
ab
ab
SSSS
Ideal S-parameters Switch correction Note: a11 = a1f ; a12 = a1
r
a21 = a2f ; a22 = a2
r
b1 a1 a2 b2
Page 22
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Two-Port Four-Receiver VNAs (like PNA)Dual reflectometers replaced by splitters and 3-port couplers
Incident wave detection is still after the port switch
Switch corrections a12/a22 and a21/a11 can be directly measured
R1 R2
RF Source
LO
Test port 2
BA
Test port 1
12
Page 23
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Test port 1 Test port 2
C DBA
Test port 3 Test port 4
R
Single Reference Receiver VNAs (like 4-Port PNA-L)Reference receiver is prior to port switchSwitch corrections cannot be directly measuredTRL* ignores switch correction terms by setting them to zero
1
22
22
11
2122
12
11
11
2221
1211
1001 −
•
≅
ab
ab
ab
ab
SSSS
TRL* sets switch correction to zero
Page 24
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Alternative to Measuring Switch CorrectionsInstead of ignoring the switching terms, we can characterize the
match differences and then apply them during the TRL cal
To do so, we can restate the switch correction terms:
Gf and Gr are “delta-match” terms, and theycan be derived from an SOLT calibration
1
11
2122
121
21
21
11
2112
12
22
121
11
2122
12
1
1
1
1
1
1−−−
Γ
Γ=
=
f
r
ab
ab
ba
ab
ba
ab
aa
aa
Thru
SHORT
OPEN
LOAD
SHORT
OPEN
LOAD
13
Page 25
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Delta Match Cal
Delta match cal characterizes internal match differences,so it can be performed…• with mixed connector types• with or without test port cables, in any combination• with adapters that can remain or be removed for subsequent TRL cals
Page 26
安捷倫科技高頻元件量測研討會
Feb.23, 2006
When is Delta Match Used?
Definitions• Flush thru: insertable, zero-length• Known thru: characterized (all 4 S-parameters)• Unknown thru: reciprocal (S21=S12); usually low-loss and well matched
YesTRL - Adapter RemovalNoECal Internal ThruYesECal Unknown Thru
YesTRL - Known ThruYesSOLT - Unknown ThruNoSOLT - Adapter Removal
NoSOLT - Known Thru(Includes Flush-Thru, Known-Thru and Minimized Thru)
Uses Delta Match?Cal Method
14
Page 27
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Two Ways to Perform the Delta-Match Calibration
Global Delta Match
• Covers entire frequency range of instrument
• Interpolation used for TRL cals with smaller spans and/or less points
• Only one global delta match cal set can exist at a time
• Future: considering providing factory global delta match cal set
User Delta Match
• Perform a non-TRL calibration, such as mechanical SOLT, SOLR or ECal
• Save data as a normal user cal set
• Frequency range and number of points must be identical to that of the follow-on TRL calibration
• Interpolation is not used during the calibration, but can be used for measurements
Page 28
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Performing the Global Delta Match Cal
Default Settings:Points: 1601IF BW: 100 HzFreq: 300 kHz -
20.1 GHz
User cannot changethese settings!
15
Page 29
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Performing the Global Delta Match Cal
• Select DUT connectors (the only choice you have is in choosing the “DUT Port 1” connector)
• Add adapters to match those shown in the table
• ECal provides the easiest solution, especially with the new 4-port 20 GHz module (N4432/3A)
Page 30
安捷倫科技高頻元件量測研討會
Feb.23, 2006
With a 2-port ECal, it only takes two steps to complete the cal!Step 1 of 2: connect ECal to ports 1 and 2Step 2 of 2: connect ECal to ports 3 and 4
Using 2-Port ECal Modules
Note: when using a 2-port ECal module to perform the Global Delta Match calibration,
you must use an N4691B (300 kHz – 26.5 GHz)
16
Page 31
安捷倫科技高頻元件量測研討會
Feb.23, 2006
User Delta Match
Instead of selecting the Global Delta Match CalSet, select an available user CalSet that matches the current stimulus conditions
Page 32
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Performing the TRL CalibrationCalibration > Calibration Wizard
17
Page 33
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Here, as a test case, we created a cal kit definition using “probe” as connector type. In this cal kit, we defined Short, Thru and Line standards. Plus, we have given them TRL class assignments.
Use dropdown menu to select connectors.
Appropriate cal kits should appear based on connector type. Use dropdown menu to select Cal Kit.
Select this and the following screen will appear. Otherwise, the default uses the existing Global Delta Match Cal.”
Performing the TRL Calibration
Page 34
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Select this to choose the “delta match” calset.
Select to modify thru path choices and cal standards.
Default is “minimum thru’s” (known bug: TRL always measures all paths, regardless of the status of this box)
Performing the TRL Calibration
18
Page 35
安捷倫科技高頻元件量測研討會
Feb.23, 2006
• Selecting “Mod Stds” brings up this screen • In this case, the default is TRL with Defined Thru• If you select another thru choice, then cal type will be SOLT• Generally, you do not need to select “View/Modify”
Click here to view class assignments
Performing the TRL Calibration
Page 36
安捷倫科技高頻元件量測研討會
Feb.23, 2006
• All cal sets will appear, but not all will work for delta match• Highlight the file you want and click OK• If you pick a “bad” user
cal set, you get this error:
Selecting “Choose Delta Match” brings up this screen Performing the TRL Calibration
19
Page 37
安捷倫科技高頻元件量測研討會
Feb.23, 2006
• At this point, the calibration should start.• Based on the “probe” cal kit we created, we went through the
following steps for a 3-port TRL cal:Step 1 of 9, short to port 1Step 2 of 9, short to port 2Step 3 of 9, thru between ports 1 and 2Step 4 of 9, line between ports 1 and 2Step 5 of 9, short to port 3Step 6 of 9, thru between ports 1 and 3Step 7 of 9, line between ports 1 and 3Step 8 of 9, thru between ports 2 and 3Step 9 of 9, line between ports 2 and 3
Performing the TRL Calibration
9
Page 38
安捷倫科技高頻元件量測研討會
Feb.23, 2006
What About Calibrating On-Wafer?
Because delta-match cal can be performed at any interface,on-wafer customers have the following options:• Use ECal (or mechanical cal kit) at the end
of the cables, before connecting to the probes (probes tend to have female connectors, sosimply add adapters to make up two pairs of insertable cals)
• Use the probes with a 4-port with SOLT ISS
Cascade’s WinCal 2006• Will not support 4-port TRL calibrations initially
• Allows advanced 2-port calibrations, such as the LRRM family
20
Page 39
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Frequently Asked QuestionsCan I perform a 4-port TRL calibration?
-- Yes, simply follow the same steps covered in this presentation
Is TRL calibration limited to only 2 ports?-- No, it can be used with 3- or 4-port calibrations as well
Can I perform a 4-port TRL calibration on-wafer? -- Yes
Since the 4-port 20 GHz PNA-L has only one reference receiver, do I need a hardware upgrade before I can perform a TRL calibration?-- No, the 2-step calibration process does NOT require any hardware upgrade, but may require a firmware upgrade
Since the 4-port 20 GHz PNA-L has only one reference receiver, does this mean we can only do TRL*?-- No, the 2-step cal process provides true TRL calibration
Page 40
安捷倫科技高頻元件量測研討會
Feb.23, 2006
TRL Summary
TRL – first characterize the match differences with delta match cal and then use them during TRL calibration
4-port 20 GHz PNA-L with one reference receiver (switch occurs after the incident wave detector)
TRL* – ignore “switch correction” terms
Legacy 872x family with one reference receiver (switch occurs after the incident wave detector)
TRL – directly measure “switch correction” terms
PNA’s & 2-port PNA-L’s with dedicated two reference receivers (incident wave detectors occur after the switch)
New application note:On-Wafer Calibration Using a 4-Port 20 GHz PNA-L, 5989-2287EN
21
Page 41
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
Page 42
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Unknown Thru – Agilent’s Secret Weapon!
“Unknown Thru” (aka SOLR) algorithm is useful for both insertable and non-insertable calibrations
Traditional non-insertable methods:• Swap equal adapters• Use characterized thru (S-parameters known)• Perform adapter removal cal• Add adapters after cal, then, during measurement…
– use port extensions– de-embed adapters (S-parameters known)
22
Page 43
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Compromises of Traditional Methods
Swap equal adapters• Need phase matched adapters of different sexes (e.g., f-f, m-f)• Error results from loss and mismatch differences of adapters
Use characterized thru• Multi-step process• Need a non-insertable cal to measure S-parameters of characterized thru
Perform adapter removal cal• Accurate but many steps in calibration (two 2-port calibrations)
Add adapters after cal, then, during measurement…• Use port extensions – doesn’t remove adapter mismatch effects• De-embed adapters (S-parameters known) – similar to characterized thru
Page 44
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Non-Insertable ECal Modules
ECal resolves many, but not all non-insertable situations
23
Page 45
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Unknown Thru to the Rescue!Unknown Thru calibration makes 2-port calibrations much easier!
No need for matched or characterized thru adapters!
Works great for
• Non-insertable calibrations
• Fixed port positions
• Physically looooooooong DUTs
• Port orientations that are not in-line
• Multiport devices Note: Unknown Thru only works with guided calibrations
Page 46
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Unknown Thru Algorithm
Unknown thru algorithm uses “TRL” 8-term error model
α
EDF ESF
ERF/α
S11 S22
S21
S12b1m
a1m
b1
a1
a2
b2
β
ESR EDR
a2m
b2m
ERR/β
[ A ] [ T ] [ B ]a1
b1
b2
a2a2m
b2m
b1m
a1m
24
Page 47
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Unknown Thru Calibration Requirements
Systematic errors of all test ports (directivity, source match, reflection tracking) can be completely characterized
“Unknown thru” calibration standard:• Must be reciprocal (Sij = Sji)• Phase known to within a quarter wavelength
VNA signal-path switch errors can be quantified• Requires dual reflectometers on all ports
(e.g., a 2-port 4-receiver VNA) OR• Requires delta-match correction
Page 48
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Two-Port Unknown Thru Calibration SequenceMeasure open, short, load on port 1Measure open, short, load on port 2Measure insertable adapter (unknown thru)
between ports 1 and 2Confirm estimated electrical
delay of unknown thru
1-port calibrations
Unknown thru
Unknown Thru calibration is performed just like a “flush thru” 2-port calibration!
25
Page 49
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measuring Physically Long Devices (Usual Way)
2-PORT CALIBRATION PLANE
CABLE MOVEMENT
2-PORT CALIBRATION PLANE
DUT
Page 50
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Cable Movement Drift Error
-0.20
-0.18
-0.16
-0.14
-0.12
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
Frequency
dB
Good CableBad Cable
Cable Movement Error
26
Page 51
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measuring Physically Long Devices (Unknown Thru)
2-PORT CALIBRATION PLANES
DUT
1-PORT CALIBRATION PLANES
Unknown thru
No cable movement!
Page 52
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measuring Devices with Non-Aligned Ports(Usual Way)
2-PORT CALIBRATION PLANE
CABLE MOVEMENT
DUT
2-PORT CALIBRATION PLANE
27
Page 53
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measuring Devices with Non-Aligned Ports(Unknown Thru)
DUT
Unknown thru1-PORT
CALIBRATION PLANES2-PORT
CALIBRATION PLANES
Page 54
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport and Non-Aligned Case (Usual Way)
2-PORT CALIBRATION PLANE
3 PORT DEVICE
L
DUTA
C
B
2-PORT CALIBRATION PLANE
CABLE MOVEMENT
28
Page 55
安捷倫科技高頻元件量測研討會
Feb.23, 2006
OR
L
L
Multiport and Non-Aligned Case (Usual Way)
3 PORT DEVICE
L
DUTA
C
B
2-PORT CALIBRATION PLANE
CABLE MOVEMENT
DUT
A
C
B
DU
TA
C
B
CABLE MOVEMENTCABLE MOVEMENT
Page 56
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport and Non-Aligned Case (Unknown Thru)
1-PORT CALIBRATION PLANES
ANY THRU DEVICE 3 PORT DEVICE
L
DUTA
C
B
29
Page 57
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Multiport and Non-Aligned Case (Unknown Thru)
1-PORT CALIBRATION PLANES
ANY RECIPROCOL 3-PORT THRU
3-PORT DEVICE
DUTA
C
BA
C
B
Page 58
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Port 1
Port N
Port 2
Port 3
Port 1
Port N
Port 2
Port 3Unknown thru’s (adapters)
1-port calibrations, ECal or mechanical
Multiport Unknown Thru with Different Connectors
Different connector on each port
Finish multiport calusing unknown thru’s
30
Page 59
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Port 1
Port 4
Port 2
Port 3
Port 1
Port 4
Port 2
Port 3Unknown thru’s
and/or
TRL on-wafer cal
On-Wafer Calibrations Using Unknown Thru’s
Imperfect thru’s
Straight Thru’s
Page 60
安捷倫科技高頻元件量測研討會
Feb.23, 2006
1.85 f-f adapter comparison
-0.25
-0.23
-0.20
-0.18
-0.15
-0.13
-0.10
-0.08
-0.05
-0.03
0.00
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00Frequency GHz
Mag
nitu
de d
B
1.85 adapter removal cal 1.85 unknown thru cal
Unknown Thru and Adapter Removal Compared
31
Page 61
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Long (Aspect Ratio) Device, 3.5 inch x 1 mm cable, Test Comparison
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.0E+00 2.0E+01 4.0E+01 6.0E+01 8.0E+01 1.0E+02
Frequency in GHz
Mag
nitu
de (d
B)
Unknown Thru Flush Thru
Unknown Thru and Flush Thru Compared
Page 62
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Unknown Thru For Different Waveguide Bands
Watch out for these potential problems:1. Non-overlapping waveguide bands2. Attenuation near cutoff may be too high for thru calibration3. Higher-order modes
(longer adapters better attenuate undesired modes)
Higher cutoff frequencyLong enough?
Band X Band Y
Tapered or stepped adapter
32
Page 63
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Known Thru Versus Unknown Thru
Any passive device OKCan not use any device
Bad connection not a problemCan not tolerate bad connection
NARequires periodic calibration
NACharacterization error sensitive to cable movement stability
NAS-parameters can change
Accuracy sensitive to ES and ER errorsAccuracy sensitive to ED, ES, S-parameter errors and thru S21
Requires S21=S12Requires data or model and VERY stable device
Unknown ThruKnown Thru
Page 64
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
33
Page 65
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Offset Load Calibration OverviewOffset load calibration originated with 8510
Offset load is a compound standard – load is connected multiple times with differing offsets
In simplest and most common form, there are just two connections: the load by itself, and the load with an offset
Similar to a sliding load standard, except offsets are set by a known, precise transmission line (e.g., a waveguide section)
Not the same as a load standard with defined delay, which is a single standard
Offset (shim)
Load
1. Measure load by itself 2. Measure load plus offset
offset
Page 66
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Offset Load Calibration AdvantagesProvides higher directivity and load match accuracy when the definition of the offset is better known than the load definitionDoes not require a dual reflectometer VNA as it uses SOLT error model instead of TRL error modelIdeal for 1-port calibrationsAlso helpful in situations where calibration planes cannot physically move, such as fixed probe or waveguide positions, where TRL calibrations are difficultOffset standard can include loss term (especially valuable near cutoff frequency)
Probe Probe
Probe Probe
Not a good TRL standardAttenuation constant
in WR-159 waveguide
34
Page 67
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Offset Load Definition
Only available in guided calibration (SmartCal)
Math is enhanced over what 8510 did
New FW added offset load standards to waveguide cal kits
Page 68
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Offset Load Class AssignmentsSmartCal uses as many standards as it needs to complete calibration, based on frequency range of standards
In this case, only offset load is used
35
Page 69
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Thru Loss Definition
8510 did not use loss term
PNA includes loss for better accuracy, especially near cutoff and with poor raw source match
Enter waveguide loss here (different model than coaxial loss model)
×r
co
ev
hf ρπµoffset loss (waveguide) =
µo = free space permeabilityfc = waveguide cut off frequencyp = resistivity of waveguide metalh = waveguide height (short dimension)v = velocity of lighter = permitivity of dielectric (usually air)
Page 70
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Waveguide Connector Definition
New!
When defining waveguide cal kits,be sure to set Media to WAVEGUIDE
36
Page 71
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
Page 72
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Normalization (response cal)
Full 2/3/4-port corrections(SOLT, TRL, LRM...)
Port Extensions
DifficultyEasier HarderAccuracyLower Higher
Mod
elin
gD
irect
M
easu
rem
ent
De-embedding
Error Correction ChoicesPort extensions have gone APE!
37
Page 73
安捷倫科技高頻元件量測研討會
Feb.23, 2006
APE = Automatic Port Extensions
• First solution to apply both electrical delay and insertion loss to enhance port extensions
• First approach to give reasonable alternative to buildingin-fixture calibration standards or de-embedding fixture
1st
APE accounts for loss and phase of fixture transmission lines
Page 74
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Automatic Port Extensions – Step 1
• First, perform coaxial calibration at fixture connectors to remove errors due to VNA and test cables
• At this point, only the fixture loss, delay and fixture mismatch remain as sources of error.
Coaxial calibration reference planes
38
Page 75
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Automatic Port Extensions – Step 2• After coaxial calibration, connect an open or short to portion of fixture
being measured• Perform APE: algorithm measures each portion of fixture and computes
insertion loss and electrical delay• Values calculated by APE are entered into port extension feature• Now, only fixture mismatch remains as source of error
(dominated by coaxial connector).Coaxial calibration reference planes
Ports extended
Open or short placed at end of each
transmission line
Page 76
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Measurement Results
Without Automatic Port
Extension
With Automatic Port Extension
Delay & loss compensation (values computed by APE)
DUT Opens
Stepped Impedance
Adjacent Structure
ShortsBalanced
Unbalanced
Balanced
Unbalanced
Balanced
Unbalanced
“Quattro”demo board
39
Page 77
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Automatic Port Extension – Implementation
Measures Sii (reflection) of each portUses ideal open or short modelsComputes electrical delay using best-fit straight-line modelComputes insertion loss using a best-fit dielectric loss model
• Default setting uses frequencies in Active Channel Stimulus(results in loss at two frequencies)
• Active Marker choice uses frequency at Active Marker(results in loss at one frequency only)
Computed delay & loss values are automatically displayed via new port-extension tool bar
Values can be saved as part of instrument state or recorded for manual insertion next time
Page 78
安捷倫科技高頻元件量測研討會
Feb.23, 2006
How is the Loss Term Calculated?
Loss is calculated for each frequency data point (f) as follows:
If Use1 is checked and NOT Use2 then: • Loss(f) = Loss1 * (f/Freq1) ^ 0.5 (coaxial loss model)
If Use1 AND Use2 are checked, then:• Loss(f) = Loss1 * (f/Freq1) ^ n (printed circuit board dielectric loss model)• Where n = log10 [abs(Loss1/Loss2)] / log10 (Freq1/Freq2)
40
Page 79
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Which Standards Should I Use?For broadband applications, shorts or opens work equally wellChoose the most convenient standard (often an open) –this is a key benefit of Automatic Port Extensions!Will using both an open and a short improve accuracy?• Using two standards makes little difference for broadband applications, as
many ripples occur and calculated loss is the same for open or short• Using two standards improves accuracy
for narrowband applications, where a full ripple cycle does not occur
Broadband
Calculated loss is basically same for open or short
open
short
Page 80
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Broadband APE Example
Very little difference up to 10 GHz
APE with short
APE with open
DUT = short
No port extension applied
41
Page 81
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Narrowband APE Example
400 MHz span
Large variation betweenopen and short
DUT = short
APE with short
APE with open
APE with both open and short
Page 82
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Adjusting for Mismatch
Don’t adjust for mismatch
Adjust for mismatch
No port extension applied
Adjusting for mismatch keeps reflection below 0 dB
42
Page 83
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Active Span or Active Marker
Note: when using active marker, selecting or not selecting “adjust for mismatch” makes no difference
APE using active marker
APE using active span
Page 84
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Loss at DC
Offsets the entire freq span to account for a fixed loss (e.g. 3 dB attenuator)
Use positive value to compensate for loss added after APE
Fixture, no padFixture and 3 dB pad with DC loss compensation
Fixture and 3 dB pad
43
Page 85
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Summary of Automatic Port Extensions
Ideal for in-fixture applications where complete calibration standards are not available
Eliminate the need to design and build difficult load standards
Applicable to a wide range of fixture designs
Works with probes too
Easy to use and quick to get results
1st
50
Page 86
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
44
Page 87
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Fixturing for the Masses
Fixturing features are same or similar to 4-port PNA-L, ENA, and PLTS
Page 88
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Order of Fixturing Operations
First, single-ended functionsare processed in this order:• Port extensions• 2-port de-embedding• Port Z (impedance) conversion• Port matching / circuit embedding• 4-port network embed/de-embed
Then, balanced functions areprocessed in this order:• Balanced conversion• Differential- / common-mode port Z conversion• Differential matching / circuit embedding
Example circuit simulation
45
Page 89
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Port Matching
Port matching feature is same as embedding
Page 90
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential Port Matching
46
Page 91
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Port Z (Impedance) Conversion
Zs
Page 92
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Differential Port Z (Impedance) Conversion
Zs
Zs
Zs
47
Page 93
安捷倫科技高頻元件量測研討會
Feb.23, 2006
2-Port De-Embedding
In all cases, the assumption is:• Port 1 of the fixture is connected to the PNA• Port 2 of the fixture is connected to the DUT
1 2.s2p file .s2p file
2 1
Fixture A Fixture B
DUT
If “Fixture B” is measured in the forward direction, the columns in the .s2p file must be swapped:
S11 S22S21 S12
Page 94
安捷倫科技高頻元件量測研討會
Feb.23, 2006
4-Port Embedding/De-Embedding
For one-port balanced devices
For balanced to single-ended devices
For two-port balanced devices
48
Page 95
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Two Versus Four Port Embedding / De-Embedding
Question: On a balanced port, what is the difference between:• Two .s2p embedding/de-embedding files• One .s4p embedding/de-embedding files
Answer: Crosstalk terms!
.s2p file
.s2p fileDUTPNA
1
2
.s4p file DUTPNA1
2
Cannot simulate fixture leakage between PNA ports 1 and 2
Can use full leaky model to simulate fixture
Page 96
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Don’t Forget to Turn Fixturing On!
49
Page 97
安捷倫科技高頻元件量測研討會
Feb.23, 2006
On-wafer SMC MeasurementsPrevious versions of FCA (Option 083) did not allow on-wafer measurements using Scalar Mixer Cal (SMC)
A.06.xx allows embedding of probe data files during SMC
Perform power-meter and S-parameter calibrations in coax
After coax calibrations, reference plane is at probe tip
Power-meter and S-parameter
calibration plane
Extended (de-embedded) measurement plane
.s2p data
Page 98
安捷倫科技高頻元件量測研討會
Feb.23, 2006
AgendaOverview of PNA FW Rev. 6.0
Cal Registers
TRL Calibration with 4-port 20 GHz PNA-L
Unknown Thru Calibration
Offset Load Cal
Fixture And Probe Techniques
• Auto port extensions
• Embedding/de-embedding
• Frequency Converter Applications – SMC
• Measuring fixtures/probes
Q&A
50
Page 99
安捷倫科技高頻元件量測研討會
Feb.23, 2006
How Do I Get My Probe De-Embedding Data?
Perform an SOLT or TRL cal using wafer probes
Measure thru device and save data in .s2p file for de-embedding in later step
Probe
Measure thru after 2-port calibration
Probe
2-port calibration planes
Thru
Page 100
安捷倫科技高頻元件量測研討會
Feb.23, 2006
How Do I Get My Probe De-Embedding Data?
Perform an adapter removal calibration using coaxial and on-wafer standards
Probe Probe
Final 2-port calibration planes
Probe Probe2-port cal
“Adapter”
coaxial cal
Probe Probe2-port cal
wafer cal
51
Page 101
安捷倫科技高頻元件量測研討會
Feb.23, 2006
How Do I Get My Probe De-Embedding Data?Measure thru plus probe
De-embed swapped thru data to obtain probe data
Save probe data in .s2p file for later use in measuring DUTs
Repeat for other probe(s) if desired
ThruProbeProbe
2-port calibration planes
DUT De-embed swapped thru data from DUT
data to get probe dataProbe
1 2.s2p file .s2p file
2 1DUT
Page 102
安捷倫科技高頻元件量測研討會
Feb.23, 2006
Alternate Way to Get Probe + Thru DataPerform unknown thru cal in coax and with probeMeasure thru plus probeDe-embed swapped thru data to obtain probe dataSave probe data in .s2p file for later use in measuring DUTsRepeat for other probe(s) if desired
ThruProbeProbe
2-port calibration planes
Unknown thru and DUT De-embed swapped thru data from DUT
data to get probe dataProbe
1 2.s2p file .s2p file
2 1DUT
52
Page 103
安捷倫科技高頻元件量測研討會
Feb.23, 2006
How Do I Get My Fixture De-Embedding Data?
Perform an unknown thru cal using coax on one side, a probe on the other side, and the fixture itself for the unknown thru
Measure the fixture section and save data as .s2p file
Repeat for each section of fixture
Probe
2-port calibration planes
Unknown thru and DUT
www.agilent.com.tw
有關安捷倫科技電子量測產品、應用及服務的詳細資訊,可查詢我們的網站或來電洽詢。
台灣安捷倫科技股份有限公司
台北市104 復興南路一段2 號8 樓電話:(02) 8772-5888
桃園縣平鎮市324 高雙路20 號電話:(03) 492-9666
台中市408台中市文心路一段552號12C電話:(04) 2310-6915
高雄市802高雄市四維三路6號25樓之1電話:(07) 535-5035
本資料中的產品規格及說明如有修改,恕不另行通知。
©2004台灣安捷倫科技股份有限公司Issued date : 06/2005Printed in Taiwan 02/20065989-4862ZHA