Global Behavioural Model Generation Using Coefficients Interpolation

This paper demonstrates a novel approach to reduce the density of load-pull measurements that are required to populate a high-density Cardiff behavioural model coefficient look-up table, versus input drive signal, covering a large dynamic range. The presented approach investigates the use of coefficient interpolation to achieve this objective using a recently published re-formulation of the Cardiff behavioural model. This approach provides global interpolation functions for the model coefficients with respect to input drive level (|a
₁₁
|). It is shown that this approach can provide for accurate interpolation of load-pull behaviour over a 2-4 dB power range. This knowledge can be used to reduce the number of measurements necessary, hence the time duration required, to populate the Cardiff behavioural model look-up table, without compromising accuracy when used in CAD simulations. The techniques are demonstrated on a 10W packaged Cree HFET.