Adaptive noise cancellation for IS95 transmitters

1. Adaptive noise cancellation for IS95 transmitters Tomás O’Sullivan and Peter Asbeck University of California, San Diego

2. Motivation Antenna • Modern modulation schemes can result in significant noise in the receive band • Degrades performance of FDD systems • Goal of this project is an electrically quiet receive band Noise Canceller Receiver LNA PA Duplexer

3. Presentation Outline • Feedforward architecture for noise cancellation • Design topology for low insertion loss and low power dissipation • System simulation, implementation and measurements • Techniques for adaptive operation • Conclusions

4. Feedforward for noise cancellation • Part of signal coupled off • Correct phase and amplitude adjustments are made • Signals merge to cause cancellation • Can achieve deep cancellation over a wide bandwidth Delay Gain and Phase adjust LNA

5. Improved Topology • System enhances performance of duplexer • Allows for low insertion loss and low power dissipation • Extra component, the TX band filter, is required ANT RX TX TXbandfilter LNA Gain and Phase adjust

6. Design Issues • Accurate gain and phase match required for good cancellation • 50dB cancellation requires 0.027dB gain match and 0.180 phase match 50dB

7. Delay mismatch issues • TX band filter introduces delay mismatch between the paths • 30dB cancellation over 25 MHz requires 400ps delay match 600ps 1ns 400ps

8. System Simulation • Entire system simulated in ADS • Expected narrowband operation was observed with an adjustable null • Optimization tool was used to achieve cancellation at different frequencies by adjusting phase and gain

9. Simulation Results • Performanceof duplexer is enhanced by feedforward action

10. Implemented System • Low noise, low power and linear class A amplifiers were used in the design, in conjunction with a voltage variable phase shifter and attenuator Epcos B4224 3rd order Notch filter Gain and Phase adjust Agilent AT-41511

11. Measurement Results • Measurement agrees well with simulation

12. Noise Canceller Characteristics • Added Attenuation = 20dB • Attenuation bandwidth = 2MHz • Insertion Loss = 0.27 dB • Power Dissipation = 9.75mW

13. Adaptation methodology • System provides cancellation for single channel only • System is calibrated and lookup table is populated with required gain and phase for each channel • Table can then be indexed to get optimum performance for particular channel • Periodic updating of table can combat drift and ageing of system

14. Algorithm Comparison Setup • Equipment centrally controlled by PC • Phase and gain varied by DC power supplies • Algorithms implemented in C++ and tested directly on system

15. Calibration Algorithms • Several optimization algorithms investigated to find best • Univariant search and pattern search techniques compared Variable 2 Variable 1

16. Performance of Algorithms • Univariant search outperforms pattern method in speed of convergence, otherwise they are similar

17. Conclusions • Feedforward techniques have been shown to be viable for noise cancellation in transmitters relaxing requirements on duplexer • Hardware for adaptive noise cancellation has been implemented • Optimization algorithms have been investigated to find best calibration routine