WaveForm Engineering

When considering the performance of RF Power Amplifiers (PA’s) or indeed other non-linear Devices, it is the terminal RF I-V waveforms that are the unifying theoretical link between transistor technology, circuit design, and system performance.

A logical expectation would be to find WaveForm Engineering driving both the measurement and mathematical analysis requirements of the RF PA design process at all stages—transistor optimization, circuit design, and system integration. However, due to limitations in measurement architectures the RF PA design approach has had to be based on other measurable properties, e.g., dc I-V characteristics, bias dependent S-parameters, load-pull contours, AM-AM and AM-PM behavior, scalar power, and output spectrum, to name but a few. This often led to a breakdown in the design chain where different departments employed their own, differing measurement techniques, this works until an issue arrises then the source of the problem can often be difficult to find.

The past 15 years have, however, seen the maturing of a number of RF characterization systems capable of measuring RF voltage and current waveforms. Coupling such systems with impedance control or emulation hardware also enables engineering of these RF waveforms during measurements. The result is a system that provides the PA designer engineer with a practical RF I-V waveform measurement and engineering solution. There is now a real opportunity to complement the present RF PA design approaches with waveform theory and experimental waveform information, finally allowing for robust linking of measured performance mathematically back to theoretical expectations.

Waveform engineering from Mesuro
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Designers have traditionally placed more reliance on the accurate measurement of key nonlinear performance parameters, including output power, gain, efficiency, linearity, etc. as a function of the fundamental load impedance, the key design parameter. This is typically referred to as load-pull. Basically, in this black-box approach, the aim is to find, by either trial and error or via a logical sequence of measurements, the harmonic load impedances with which the desired optimum performance is achieved. Theoretically, we know that optimum performance will occur when appropriately engineered current and voltage waveforms are achieved. During this experimental investigation, the PA designer is effectively modifying the current waveform by the choice of dc quiescent bias point and input drive level and the voltage waveform by the choice of fundamental load impedance, hence, is indirectly engaged in the process of WaveForm Engineering. However, there is no way of knowing whether the optimum solution has been found.
 
In addition, as PA designers are forced to meet increasingly demanding system requirements and thus consider more complex high-efficiency modes of operation, the black-box approach is becoming impractical. This is because it requires the systematic variation of many more parameters such as harmonic source and load impedances - clearly randomly optimizing so many variables becomes prohibitive. This experimentally based design process could be made more efficient if coupled with waveform measurements. This is a good example of where RF waveform measurement systems can contribute significantly to both RF power transistor characterization/development and RF PA design/optimization. The practical consequence of integrating RF waveform measurements with solutions that provide experimental control (engineering) of the terminal impedances is that the engineering of the RF current and voltage waveforms can be experimentally monitored. Systematic experimental PA design investigations can now become WaveForm Engineering driven; the design goal finally becoming the realization of the theoretically derived optimum waveforms. It is important to note that, even if the stimulus signal is confined to a single-tone (continuous wave) signal, the RF waveform generated by the nonlinear DUT is spectrally rich—it will also contain harmonic frequency components.

Waveform engineering from Mesuro
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Hence, RF WaveForm Engineering requires not only fundamental load-pull but also harmonic source- and load-pull. This highlights the fact that when investigating the behavior of a nonlinear DUT, impedance variation investigation should not be limited to the fundamental frequency alone. Note that, since there is a very strong interaction between all these parameters, it is essential during the process of WaveForm Engineering via fundamental and harmonic impedance variation that the waveforms are continuously observed. This ensures that the load-pull variations are appropriately directed and thus minimizes the need to do systematic sweeps. If the stimulus signal is a single-tone (continuous wave) signal, confining the problem to include spectral components up to the third harmonic would provide for a band-limited RF WaveForm Engineering compromise solution that is practical to realize.
In summary, from a nonlinear PA design perspective, the integration of RF WaveForm Engineering capability, whether passive or active, with RF waveform measurement capability is essential. With such systems, the practical design of PAs achieved by directly employing the theoretically based WaveForm Engineering approach is now experimentally possible.