Title: Distributed Microwave Oscillators: theory and design
1Distributed Microwave Oscillators theory and
design
Alessandro Acampora PhD Student Centre
Tecnologic de Telecomunicacions de Catalunya
(CTTC) Weekly Seminar Series, 10/06/2009
2Outline
- Motivation and Objectives
- Linear and Nonlinear Analysis of an Oscillator
- Numerical techniques for analyzing circuit at µW
frequencies - Time Domain
- Frequency Domain
- Parametric Analysis of nonlinear circuits
- Distributed Oscillators and VCO
- Examples
- Design Procedure of a reverse mode DVCO
- Simulation Results
- Implementation and test of a four stage R-DVCO
- Measurements Results
- Conclusions and Future Work
3Motivation Top level perspective
- A plethora of coexisting telco services overcrowd
the transmission bands (mostly UHF band ,0.3 to
3 GHz) - GSM, UMTS standard for mobile telephony
- Wi-Fi or IEEE 802.11(x) for wireless local area
networks - WIMAX for Broadband Regional wireless access to
the Internet - Wireless information exchange requires ever
increasing bit rates to support multimedia
services (translates in need for more bandwidth) - A flexible receiver architecture would be needed
to catch signals transmitted at different
frequencies in a certain range - Higher data rate services could be more easily
allocated to higher frequencies
4Device perspective
- Design and implement wide-band tunable receiver
sub- components - Oscillators (Voltage Controlled Oscillator)
- Mixers
- The aim will be pursued investigating appealing
solutions by means of - Theoretical Analysis
- Simulation, by means of EDA software
- Implementation Test
- Measurements
5SmallSignals (linear) Analysis of a FET Amplifier
- When signals have small magnitude, linear
approximation holds for the voltages and currents
of an active device.
Signals cause small variations around a quiescent
point.
6Block Diagram of a Positive Feedback Oscillator
- An oscillator consists of
- an Amplifier (made up of active devices
transistors) - a Frequency Selective Network (reactive elements,
crystal, resonant cavities) - a Feedback Path, providing regeneration of the
signal in order to build-up oscillations
Nyquist Criterion for the onset of oscillations
Energy is converted from DC to AC
7Linear Analysis of a Negative Resitance Oscillator
- Partition of the circuit in two parts
- Resonator
- Active part providing negative resistance to
ensure oscillation start-up and compensate for
losses -
8Determining the oscillation frequency
- The real part of the output admittance is
negative (delivering power) - The imaginary part presents positive slope
9Drawbacks of Linear Analysis
- Provides only an estimate of the oscillation
frequency - No information about the amplitude (power)
- neither about the harmonic content
- neither about possible spurious oscillations!
- Fact we are neglecting the nonlinear behaviour
of the FET! - Linear Analysis result could still be useful as a
first guess
10Numerical Analysis in Time Domain
- Expression of the fundamental circuit variables
by means of a system of differential equations
Forward Euler (one step explicit method)
Backward Euler (one step implicit method)
11Numerical Analysis in Frequency Domain Harmonic
Balance
- Time domain simulation could be computationally
expensive (computation time vs storage
requirements) - What is the right time step?
- A network containing distributed elements ( i.e.
microscrostrip lines) would give rise to
complicated differential delay equations - Assuming a periodic behaviour (1- tone
excitation), taking the Fourier transform at both
sides
12Harmonic Balance for Autonomous Circuits
- When the circuit to be analyzed is not externally
forced by input generators, frequency is an
unknown. - Convergence to DC steady state solution when
analyzed with the classical HB. - Auxiliary generators, mimic the behavior of the
real ones ? avoid the DC equilibrium point!
quéré , 1992
The probe is inserted in parallel between an
output node and a reference point (GND).
13Parametric Analysis
- Trace the solution curve versus a critical
parameter? - Correspond to a succession of HB problems!
- Difficulties arise when considering multi-valued
solution curve? a parameter switching algorithm
is invoked to overcome this difficulty
14Distributed Amplifier Basics
- Distributed Amplifier an old idea to overcome
the bandwidth-gain limitations Ginzton,1948 - Coupling N stages by means of k-filter sections
a linear increase in gain is observed, without
compromising the bandwidth Wong, 1993.
Currents from each output stage are combined in
an additive fashion while parasitic capacitances
are not accumulated
15Forward mode Distributed Oscillator
- Idea add a feedback path and a frequency
selective network to obtain an oscillator
Hajimiri, 2001 - The physical length of the path determines the
operating frequency (as it gets shorter, the
frequency gets higher) - Possibility of inserting a varactor diode in
order to provide limited tuning capabilities
Forward propagating waves are reinserted in the
input line through the feedback loop
16Reverse mode Distributed Voltage Controlled
Oscillator (1)
- Removing drain resistor, it is possible to
connect drain and gate line together in a
reverse manner to take advantage of backward
propagating waves Skvor, 1992 - Effective path length changes when we activate
one stage at a time, leaving the others switched
off - As a consequence, a discrete set of frequencies
will appear distributed in the pass-band of LC
filter sections
Higher Frequency correspond to the activation
of the first stage while lower frequency to the
activation of the last one
Transistor T2 is placed crosswise in order to
provide extra gain
17Reverse mode Distributed Voltage Controlled
Oscillator (2)
Impedance of a low pass constant-k filter section
Reverse Gain of the Distributed Amplifier Divina
and Skvor, 1998
The choice of the matching sections is critical
to avoid spurious oscillations!
18Reverse mode Distributed Voltage Controlled
Oscillator (3)
- Tuning capabilities biasing in a complementary
fashion, two adjacent stages at a time - A continuous set of frequencies are obtained,
between two discrete ones Divina and Skvor,
1995 - Intuitive explanation biasing simultaneously two
transistor changes the effective path length of
the feedback signal
19Nonlinear Analysis and Design of a reverse mode
DVCO
- Design Requirements
- Four stages
- Tunable in the range 1 to 3 GHz
- Flat output power (limited fluctuations around 5
dBm) - Step-by step approach.
- first schematic included only real transistor
models, and idealized lumped components and TL
sections - linear analysis is performed by checking
negative resistance zones ? estimate of the
oscillation frequencies (see next slide) - Parameters of the LC-sections are chosen
- HB simulation provides the power and harmonic
content for each discrete frequency
20Simulated Negative Conductance Zones
When each transistor is independently biased we
obtain a zone in which its output admittance
becomes negative. These zones partially overlap
assuring a continuous tuning range!!
21From Schematic to Layout.
The values for LC sections were tuned after
the introduction of layout elements
Vendor model for inductors and capacitors has
been used with appropriate values L3.3 nH and
C1.2 pF assure a (close to) 50 Ohm impedance and
5 GHz for the cut-off frequency of the k constant
filter section
22Nonlinear Analysis and Optimization of a reverse
mode DVCO (2)
Simulations performed by sweeping the frequency,
observing tuning voltages in pairs
23Optimized Output Power
- Heuristic procedure to demonstrate that some
different biasing scheme (using three stages)
could lead to a smooth output power - Tuning performance is nearly the same, while
output power variations are noticeably reduced
24Measurements Results (1) Tuning
Tuning range 750 MHz to 1.84 GHz
Frequency Gaps are present, the most critical is
between 1.425 and 1.650 GHz
Lab Experience Facing with Hysteresis Phenomena
and Instabilities.
Variation of the bias voltages is highly
symetrical (matching with simulation!!)
25Measurements results (2) Output Power, Current
Consumption
Output power curve and DC current show similar
trends.
26Lesson learned from DVCO implementation
- Accuracy of lumped inductors models was found to
be critical (shift in the frequency range) and
vendor -dependent - Comparison with vendors models and extracted
parameters (via calibration kit Vector Network
Analyzer) gave higher value for the measured
inductances - Redesign the TL sections in between the first and
third stage giving them smaller lengths would
probably reduce the gaps in frequency - Fine tuning of the lumped elements to reduce
spurious and unwanted oscillations
27Future Work
- Extend the DVCO tuning range (3 to 10 GHz)
- Study the phase noise in DVCO
- Identify the critical parameters to study
qualitative change in DVCO behaviour - Apply the mentioned techniques to study other
RF/MW sub-components (frequency dividers)
28References
- Wong, 1993 Thomas T.Y. Wong, Fundamentals of
Distributed Amplification, London, UK Artech
House 1993. - Ginzton,1948 E. L. Ginzton, W. R. Hewlett, J.
H. Jasberg, and J. D. Noe, Distributed
Amplification, Proc. IRE, pp. 956-69, August
1948. - Kundert, 1990 K.S. Kundert, J.K. White, A.
Sangiovanni-Vincentelli, Steady-state methods for
simulating analog and microwave circuits, Kluwer
Academic Publishers, 1990 - Wu and Hajimiri,2001 H. Wu, A. Hajimiri,
Silicon-Based Distributed Voltage-Controlled
Oscillators, IEEE Journal Of Solid-State
Circuits, VOL. 36, NO. 3, March 2001, pp 493-402 - Skvor, 1992 Z. Skvor, S. Saunders, and C. S.
Aitchitson, Novel decade electronically tuneable
microwave oscillator based on the distributed
amplifier Electron. Lett., vol. 28, no. 17, pp.
16471648, Aug. 1992. - Divina and Skvor, 1995 L. Divina and Z. kvor,
Experimental verification of a distributed
amplifier oscillator, in Proc. 25th EuMC 1995
Conf.. Kent, U.K. Nexus Media Limited, 1995, pp.
11631167. - Divina and Skvor, 1998 L. Divina, Z. Skvor,
The Distributed Oscillator at 4 GHz, IEEE
Transactions On Microwave Theory And Techniques,
VOL. 46, NO. 12, December 1998, pp 2240-2243.
29Thanks for your kind attention!
Alessandro Acampora Ph.D Student , CTTC Centre
Tecnologic de Telecomunicacions de
Catalunya alessandro.acampora_at_cttc.es