Linearity and efficiency are both critical parameters for radio frequency transmitter applications. In theory, a Doherty amplifier is a linear amplifier that is significantly more efficient than comparable conventional linear amplifiers. It comprises two amplifiers, connected at their outputs by a quarter-wave transformer. The main amplifier is always on, while the peaking amplifier is off during low power levels. Load modulation of the main amplifier occurs when the peaking amplifier is on due to the quarter-wave transformer, ensuring the main amplifier never enters saturation. This results in an efficiency characteristic that increases with respect to input power at twice the normal rate at low power levels, and plateaus to a high value at high power levels. However, in much of the research that has been done to-date, less-than-ideal results have been achieved (although efficiency was better than a conventional amplifier). It was decided to investigate the cause of the discrepancy between theoretical and practical results, and devise a method to counteract the problem. It was discovered that the main cause of the discrepancy was non-ideal transistor gate-voltage to drain-current characteristics. The implementation of a gate bias control scheme based upon measured transistor transfer characteristics, and the desired main and peaking amplifier output currents, resulted in a robust method to ensure near-ideal results. A prototype amplifier was constructed to test the control scheme, and theoretical, simulated and measured results were well matched. The amplifier had a region of high efficiency in the high power levels (over 34% for the last 6 dB of input power), and the gain was nearly constant with respect to input power (between 4 and 5 dB over the dynamic range). Furthermore, it was decided to investigate the role harmonics play within the Doherty amplifier. A classical implementation shunts unwanted harmonics to ground within the main and peaking amplifiers. However, odd harmonics generated by the peaking amplifier can be used to operate the main amplifier like a class F amplifier. This means its supply voltage can be lowered, without the amplifier entering saturation, and the efficiency of the Doherty amplifier can be increased without a detrimental effect on the its linearity. A prototype amplifier was constructed to test this theory, and gave good results, with better efficiency than that of a conventional amplifier, and a constant gain with respect to input power (between 6.4 dB and 6.5 dB over the dynamic range).

La technologie ultra large bande impulsionnelle améliore la précision des systèmes de localisation en intérieur, en limitant les perturbations potentielles liées aux multi-trajets. Une nouvelle norme, destinée aux situations d’urgence, relève le niveau d’émission de puissance de 20 dB, afin de faciliter la localisation au travers d’obstacles épais et l’intervention des secours. Cette thèse s’est attelée à la conception d’un amplificateur conforme à cette norme. Après un état de l’art des amplificateurs de puissance ultra large bande existants, plusieurs circuits sont conçus : une cellule de puissance pour atteindre les niveaux de puissance requis en sortie, et une cellule pré amplificatrice qui fournit un apport de gain. Enfin, un système d’amplification de puissance à polarisation contrôlée exploitant la nature impulsionnelle du signal est proposé. La limitation de l’activation de l’amplificateur à la seule manifestation d’une impulsion permet de réduire drastiquement l’impact énergétique du PA sur un système de localisation. / The use of impulse radio (IR) ultra-wideband (UWB) technology for indoor localization systems both improves signal accuracy and reduces disturbances from multipath. A new standard for emergency situations requires the design of a higher output power UWB power amplifier (PA) as the usual maximum output power has been raised by 20 dB in order to allow the signal to pass through thick obstacles, thus helping rescue operations. First, a UWB PA state of the art is presented, and the design of both a power cell and a driver cell are detailed, to achieve the required output power level and additional gain, respectively. Finally, a bias controlled amplification system is described, taking advantage of the IR profile of the UWB signal. The PA is activated only for pulse occurrences, reducing drastically its power consumption.