Transcript Constructing and Benchmarking a Pulsed-RF, Pulsed

Improving the RF Active Circuit
Design Cycle Through
Innovations in Electrothermal
Modeling, Characterization, and
Design Techniques
Dr. Charles Baylis
Faculty Candidate
[email protected]
April 11, 2008
Overview
•
•
•
•
USF RACAM Research Group
The RF Active Circuit Design Cycle
Research Strategy/Overview
Nonlinear Modeling of Thermal and Trapping
Effects in GaN HEMTs
• Radar Power Amplifier Combining Techniques
for Sidelobe Reduction
• Prediction of Phase Noise in Amplifiers and
Frequency Multipliers
• Development of Microwave Measurement
Techniques for Cell Cultures
School of Engineering & Computer Science
University of South Florida
School of Engineering & Computer Science
RACAM at USF
• 4 graduate students, 2 undergraduate
students
School of Engineering & Computer Science
RF Active Circuit Design Cycle
Design
Measurements
Models
Simulation
Fabrication
Testing
SUCCESS!!!
Good models can
eliminate this path.
School of Engineering & Computer Science
Strategy for Obtaining Funds
• Obtain initial grass-roots funding from
industry contacts.
• From this foundation, apply for agency
funding.
School of Engineering & Computer Science
Research Overview
• Nonlinear Modeling for Power Amplifier Design
(Modelithics)
$25,000 + $25,000 State Program Match = $50,000
• Prediction of Phase Noise in Amplifiers and
Frequency Multipliers (Trak Microwave)
$12,000 + $6,000 State Program Match = $18,000
• Investigation of Combining Techniques for
Reduced Sidelobes in Radar Power Amplifiers
NRL Proposal – March 2008, $170,000 for 2 years
School of Engineering & Computer Science
Research Overview
• Accurate Bio-Impedance Measurements
NSF Proposal - February 2008, $361,381 for 3 years
• Research on a Wireless System for
Transportation Applications
Proposal to Sunovia Energy Systems, February 2008, 3
years
• Development of Model Scoring Metrics for RF
Circuit Design
Proposal to Raytheon RF Components (Andover,
Massachusetts), $49,500 for 1 year
School of Engineering & Computer Science
Electrothermal FET
Modeling
• Nonlinear RF CAD Models are extracted
from measurements of Field-Effect
Transistors (FETs):
– Current-Voltage (IV)
– S-Parameters
– Load Pull
• Pulsed IV measurements allow thermal
and trap effects to be accurately modeled.
School of Engineering & Computer Science
IV Curves
• The IV curves give the boundaries for the
large-signal performance of the FET:
ID
Maximum Current
Drain-gate breakdown
Knee Voltage
Load line for signal swing
VDS
School of Engineering & Computer Science
Model Equation Fitting to IV
Blue Dots = Measured
Red Lines = Simulated
0.20
Ids (A)
0.15
0.10
0.05
0.00
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Vds (V)
Ids  IPK 0  1  tanh   tanh(  Vds)(1  LAMBDA Vds  LSB0  exp(Vdg  VTR))
  ALPHAR  ALPHAS  (1  tanh(  ))
School of Engineering & Computer Science
Pulsed IV Measurements
• Static measurements
can be inaccurate for RF
models.
• The cause: Slow
Thermal and Trapping
Processes
• Solutions: Pulsed IV
and Measurements
• GaN HEMT Static (Dark
Lines) and Pulsed IV
(Light Lines)
School of Engineering & Computer Science
Pulsed IV Measurement
• Measurements are performed during brief
(~0.2 μs) excursions from a quiescent bias.
• The pulses are usually separated by at least 1
ms.
• Thermal and trap conditions during the
measurement are those of the quiescent bias,
as in high-frequency operation.
School of Engineering & Computer Science
Trapping Effects
• Trapping Effects in MESFETs (Charbonninud
et. al):
– Substrate Traps
– Surface Traps
• Electron Capture  Fast Process
• Electron Emission  Slow Process
S
G
Surface Traps
D
Electron Flow
Substrate Traps
School of Engineering & Computer Science
Summary of Trap Processes
ID
Surface Hole Capture (Slow)
SLOW PROCESSES
Substrate Electron Emission (Slow)
Surface Hole Capture (Slow)
Q
FAST PROCESSES 
Substrate Electron Capture (Fast)
Surface Hole Emission (Fast)
Surface Hole Emission (Fast)
School of Engineering & Computer Science
VDS
Bias-Dependent Trapping
Gate:
Drain
School of Engineering & Computer Science
Bias-Dependent FET Model
New Parameters
School of Engineering & Computer Science
Bias-Dependent FET Model
Model without
quiescent
dependence:
0.05
Model with
quiescent
dependence:
0.05
VGS=-1.000
0.04
VGS=-2.000
0.03
0.02
VGS=-3.000
VGS=-1.000
Ids (A)
Ids (A)
0.04
0.03
VGS=-2.000
0.02
VGS=-3.000
0.01
0.01
VGS=-4.000
VGS=-5.000
0.00
0
5
10
15
Vds (V)
20
25
VGS=-4.000
VGS=-5.000
0.00
0
5
10
15
20
25
Vds (V)
Quiescent-bias dependence allows flexibility
in predicting the IV curves.
School of Engineering & Computer Science
Proposed Upcoming Work
• Adapt USF bias-dependent approach for
use on other desired models.
• Add time-dependence of capture and
emission trapping as well as thermal
effects.
• Develop a straightforward characterization
scheme for the bias and time dependence
of thermal and trapping effects.
School of Engineering & Computer Science
Partial Jardel Circuit for Drain Traps*
To calculations of
backgating voltage
+
Vds(t)
_
+
Vcont(t)
_
*O. Jardel, F. DeGroote, C. Charbonniaud, T. Reveyrand, J. Teyssier, R. Quere, and D. Floriot,
“A Drain-Lag Model for AlGaN/GaN Power HEMTs,”
IEEE International Microwave Symposium, Honolulu, Hawaii, June 2007.
School of Engineering & Computer Science
Radar Power Amplifiers – NRL
Proposal
• Desire to reduce spectral
sidelobes transmitted by
shipboard radar systems.
• Chireix amplifier topology:
School of Engineering & Computer Science
Goals and Discussion
• Desired Isolation: 100 dB outside of 40
MHz bandwidth
Reprinted from J. de Graaf, H.
Faust, J. Alatishe, and S. Talapatra,
“Generation of Spectrally Confined
Transmitted Radar Waveforms,”
Proc. IEEE Conference on Radar,
2006, pp. 76-83
• The pulsed signal and nonlinear amplifier
use causes additional spectral
components  “spectral regrowth”.
School of Engineering & Computer Science
Chireix Amplifier Operation
• A method for operating amplifiers with high
linearity and efficiency.
• Based on a trigonometric identity:
cos( A  B)  cos( A  B)  2 cos A cos B
G
cos[ω(t)+arccos(M(t))]
PA
Phase
2GM(t)cos[ω(t)]
Modulated
Signals
G
cos[ω(t)-arccos(M(t))]
PA
School of Engineering & Computer Science
Combining Challenges
• A summer at microwave frequencies?
• Combining Techniques
– 180-degree coupler
– Chireix combiner
• A three-port network cannot be lossless, reciprocal, and
matched at all ports.
• Pros and Cons
– 180-degree coupler – matched, reciprocal, and lossy (3 dB)
– Chireix combiner – unmatched, reciprocal, and lossless
School of Engineering & Computer Science
24
180-Degree Coupler Simulations
School of Engineering & Computer Science
25
Radar PA Combining – Upcoming
Work
• Proposal to NRL
for $170,000 over 2 years
submitted March 2008
• Study different combining
techniques:
• Examine rejection for better
spectral masks.
• Continue simulation studies
and eventually implement in
hardware design
improvements.
• Meeting with NRL at USF
campus on April 14.
School of Engineering & Computer Science
Phase Noise Prediction
• Tremendous consequences for system-level
design considerations.
• Example: Phase Noise in 64-PSK Amplifier
• Transistor 1/f noise is a source of phase noise.
a
I
i 2  K 1 b f
f
• Project Goal: Predicting phase noise
accurately in circuits through accurate modeling
of 1/f noise.
School of Engineering & Computer Science
Demonstration Circuits
• Linear Amplifier
– Test phase noise prediction at multiple bias currents.
– Si BJT, SiGe HBT.
– Designed circuits presently in test phase.
• Frequency Multiplier
– Test phase noise prediction due to self-biasing in
large-signal operation.
– Si BJT, SiGe HBT.
– Circuits presently in design phase.
School of Engineering & Computer Science
Microwaves and Bio –
NSF Proposal
In collaboration with:
School of Engineering & Computer Science
Biological Motivation
• Tissue composition can
often be investigated
through permittivity
measurements:    ' "
• Relative Permittivity Shows
Three Dispersions:
Reprinted from H.
Schwan, “Electrical Characteristics
of Tissues,” Biophysik, 1963,
Vol. 1, No. 3, pp. 198-208
– Alpha Dispersion
– Beta Dispersion
– Gamma Dispersion
School of Engineering & Computer Science
Dispersions
• Alpha Dispersion
– In kHz Range
– Ionic diffusion (Foster and Schwan)
– Measurable with Low Frequency Impedance Analyzer
• Beta Dispersion
– Between 1 and 100 MHz
– Capacitive charging of the cell membrane (Tamura et al.)
– Can measure above this with microwave techniques
• Delta Dispersion
– Varying causes
– Between 0.1 and 3 GHz (Foster and Schwan)
– Measurable by microwave techniques
• Gamma Dispersion
– Dipole orientation in water molecules changes (Foster and
Schwan)
– 20 to 25 GHz (depending on temperature)
– Measurable by microwave techniques
School of Engineering & Computer Science
Measuring the Water Content of
Cells
• Above the Beta Dispersion, the cell
membrane (a capacitor) appears invisible
to the electrical signal.
• The measured impedance above the Beta
Dispersion is heavily dependent upon the
water content of the cells.
• Cancer cells often have a higher water
content than healthy cells (Foster and
Schwan).
School of Engineering & Computer Science
Cell Culture Measurements
• Cell cultures often have an impedance of
1 kΩ or higher in the microwave range.
• Measuring high impedances with high
precision using reflection coefficients is
difficult:
School of Engineering & Computer Science
Microwave Measurement
Technique for High Impedances
• New technique developed at Czech
Technical University:
Reprinted from M. Randus and K. Hoffman, “A Simple Method for Extreme
Impedances Measurement,” 70th Automatic RF Techniques Group (ARFTG)
Conference, Tempe, Arizona, November 2007.
School of Engineering & Computer Science
Future Research
• Presently constructing system to verify
method of Randus and Hoffman on large
impedances.
• Modify cell culture impedance
measurement setup of Prof. Shekhar
Bhansali to perform microwave
measurements.
• Examine applications (i.e. clinical
detection of cancer)
School of Engineering & Computer Science
Conclusions
• An ambitious first-year approach has allowed
expansion of USF’s modeling and
characterization program into design and
biological applications.
• A good industry base has been built, and agency
proposals are being generated.
• A significant number of publications, as well as
additional agency and industry proposals, are
expected to be generated from the present work.
School of Engineering & Computer Science
Acknowledgments
• Larry Cohen and Jean de Graaf, Naval
Research Laboratory
• Christopher Reul, USF
• Brent Seward, USF
• Nathaniel Varney, USF
• Dorielle Price, USF
• Dr. Shekhar Bhansali, USF
• Dr. Larry Dunleavy, USF and Modelithics, Inc.
• Martin Randus, Czech Technical Institute
• Dr. Karel Hoffman, Czech Technical Institute
School of Engineering & Computer Science
Thank you for your time
and attention!
School of Engineering & Computer Science