A Caltech Library Service

Techniques for Mixed-Signal Linearization and Large Signal Handling in Radio-Frequency Receiver Circuits


Keehr, Edward Arthur (2011) Techniques for Mixed-Signal Linearization and Large Signal Handling in Radio-Frequency Receiver Circuits. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/WFGT-4823.


In this dissertation, two effective linearization schemes for radio-frequency receivers are introduced. The first of these comprises a mixed-signal feedforward path which regenerates third-order intermodulation (IM3) products at radio frequencies, downconverts these products, digitizes them, and then uses them to cancel corruptive IM3 products in the digital baseband portion of a nominally linear receiver path. The combined implemented receiver represents a SAW-less direct-conversion receiver for UMTS FDD Region 1 that achieves an uncorrected out-of-band IIP3 of -7.1dBm under worst-case blocking specifications. Under IM3 equalization, the receiver achieves an effective IIP3 of +5.3dBm and meets the UMTS BER sensitivity requirements with 3.7dB of margin. To enable this mixed-signal feedforward path, a multistage cubic term generator is introduced which uses cascaded nonlinear operations to generate reference IM3 products. The multistage nature of this circuit is considered in the context of the aforementioned linearization scheme and is shown to provide sufficient dynamic range for nearly complete IM3 cancellation while dissipating far less power than the original receiver front end. In particular, the effect of the group delay between stages is analyzed and shown to permit large IM3 cancellation ratios for interstage group delays less than 1ns.

Expanding upon the first effective linearization approach led to the development of a large signal handling receiver with an out-of-band 1-dB desensitization point of +12.5dBm. Enabling this large signal handling capability is a passive mixer downconverter preceded by a novel wide-swing LNTA. With a stacked push-pull class-AB common-gate architecture, the LNTA reduces the magnitude of input-referred distortion by up to 40dB beyond that predicted by an initial slope-of-3 characteristic while at the same time minimally impacting the effective small-signal gain of the receiver. To compensate for intermodulation distortion terms of order greater than 3, IM3 and IM2 products are processed down to digital baseband where they are successively multiplied to generate approximations to higher-order terms. In the case of a +12.4dBm QPSK-modulated signal and a -16.3dBm CW blocker, cancellation improves receiver input-referred error by over 24dB, resulting in an extrapolated IIP3 of +43.5dBm.

Item Type:Thesis (Dissertation (Ph.D.))
Subject Keywords:Adaptive equalization, adaptive signal processing, digital linearization, feedforward error cancellation, mixed-signal linearization, nonlinear circuits, RF receivers, system-level linearization, wireless communications.
Degree Grantor:California Institute of Technology
Division:Engineering and Applied Science
Major Option:Electrical Engineering
Awards:Charles and Ellen Wilts Prize, 2011
Thesis Availability:Public (worldwide access)
Research Advisor(s):
  • Hajimiri, Ali
Thesis Committee:
  • Hajimiri, Ali (chair)
  • Emami, Azita
  • Rutledge, David B.
  • Hassibi, Babak
  • D'Addario, Larry R.
Defense Date:1 June 2010
Record Number:CaltechTHESIS:08232010-071647727
Persistent URL:
Default Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:6005
Deposited By: Edward Keehr
Deposited On:28 Feb 2012 18:05
Last Modified:09 Oct 2019 17:05

Thesis Files

PDF - Final Version
See Usage Policy.


Repository Staff Only: item control page