Margin of Reliability and Monotonic Extensibility of the New MBF Memoryless Nonlinear Power Amplifier Behavioral Model

Mairtin O'Droma,Lei Yiming

DOI: https://doi.org/10.1109/ursi-at-rasc.2015.7302969

2015-01-01

Abstract:The new modified Bessel-Fourier (MBF) nonlinear RF power amplifier (PA) memoryless behavioural model will be presented. Like the BF envelope model, from which it emanates, it is directly extracted in the envelope domain of RF measurements but, unlike it, it exploits a previously unobserved potential for finding better models than are possible with the BF model. This potential is present in the inseparable underlying extraction of Fourier series (FS) approximations of hypothetical memoryless complex instantaneous RF voltage transfer characteristics (IVTCs) of the PA. The accuracy and behavioural prediction performance especially of its low third order model will be shown to be better than the best published low order models in use today. Besides inheriting all the key properties of the BF model, surpassing its low order accuracy and performance and matching its high-order accuracy and performance, the new model will be shown to come with two attractive attributes of ‘margin of reliability’ beyond the dynamic range set by the extraction measurements and model order ‘monotonic extensibility.’ These attributes are important additions to those found in the legacy BF model. The first attribute brings more assured stable usage by allowing PA input signals to drive the amplifier deeper into saturation than occurred in the model extraction measurements. This is something that is not permitted with most behavioural models, and normally would be deprecated. The benefit of the second attribute, besides overcoming problems of unpredictable model coefficient sensitivity, is its usefulness in contexts where controlled dynamic adaptive tradeoff is required between computation efficiency and