Aaron T. Radomski

New topology for Class E amplifiers

A Class E RF amplifier, which can operate into any load conditions without need for other additional circuitry to protect transistors, is introduced. This is provided by the new topology, which is called an inductive clamp. Our topology incorporates inductive clamp circuitry to the basic Class E amplifier circuit and it has all the benefits of Class E amplifiers. Additionally, it has inherent self-protection that comes with the inductive clamp circuit. A Class E amplifier with the new topology is designed, simulated and implemented. The experimental results are presented and found to be very close to the simulated results. The amplifier drain efficiency is measured around 88% at the rated power level and it is confirmed that the amplifier protected itself and was stable over entire VSWR and dynamic range within the bandwidth of operational frequency. Class E amplifiers with this topology can be used in applications where the load conditions are dynamic.

Systemic health evaluation of RF generators using Gaussian mixture models

We propose an application of specific machine learning techniques capable of evaluating systemic health of a Radio Frequency (RF) power generator. System signatures or fingerprints are collected from multivariate time-series data samples of sensor values under typical operational loads. These fingerprints are transformed into feature vectors using standard scaling/translation methods and the Fast Fourier Transform (FFT). The number of features per fingerprint are reduced by banding neighboring features and Principal Component Analysis (PCA). The reduced feature vectors are used with the Expectation Maximization (EM) algorithm to learn parameters for a Gaussian Mixture Model (GMM) to represent normal operation. One-class classification of normal fingerprints is achieved by thresholding the likelihood of a fingerprint feature vectors. Fingerprints were collected from normal operational conditions and seeded non-normal conditions. Preprocessing methods and algorithmic parameters have been selected using an iterative grid search. Average robust true positive rate achieved was 94.76% and best specificity reported is 86.56%.

In-Vivo Fault Analysis and Real-Time Fault Prediction for RF Generators Using State-of-the-Art Classifiers

In this paper we apply various machine learning techniques for fault detection of RF (Radio Frequency) Power Generators. Fast Fourier Transform features are used in our analysis for all experiments. Radial Basis Function Networks (RBF) is used to build a two class classifier to differentiate between normal and one fault condition. We apply three one class classifiers to model the normal operating conditions. The data is obtained from five different generators of the same model type.

Embedded one-class classification on RF generator using Mixture of Gaussians

In this paper we apply a specific machine learning technique for classification of normal and not-normal operation of RF (Radio Frequency) power generators. Pre-processing techniques using FFT and bandpower convert time-series system signatures into single feature vectors. These feature vectors are modeled using k-component Mixture of Gaussians (MoG) where components and corresponding parameters are learned using the Expectation Maximization (EM) algorithm. Data is obtained from three different generator models operating under normal and multiple different not-normal conditions. Exploration into algorithmic parameter effects is conducted and empirical evidence used to select sub-optimum parameters. Robust testing is reported to achieve a 3s classification accuracy of 95.91% for the targeted RF generator. Additionally, a custom C++ library is implemented to utilize the learned model for accurate classification of time-series data within an embedded environment such as a RF generator. The embedded implementation is reported to have a small storage footprint, reasonable memory consumption and overall fast execution time.

Power accuracy and source-pull effect for a high-power RF generator

RF high-power generators are extensively used for plasma etching technologies. In order to achieve high quality for the Silicon wafer process, power accuracy and stability become critical requirements for RF generators. Since a plasma chamber is regarded as a nonlinear active load, load-pull effect has been investigated thoroughly in recent years. However, power measurement is not just related to load situations. Source mismatch also plays an important role for power stability and accuracy. In this paper, power accuracy for a high-power RF generators is investigated through theoretical estimation and direct experiments. For low-reflection loads, the source-mismatch effect is dominant in power measurement error when a calibrated V-I probe is used for reflection and power measurement. In order to investigate this effect, a series of load-pull experiments have been made on a commercial RF generator with power feedback. It is shown that a given source mismatch can be greatly reduced through power feedback [14][15]. The remaining source mismatch effect becomes a comprehensive result related to three factors: the dynamics of nonlinear capacitance of the power transistors, static mismatch from the output filters and the load situation. Between the source mismatch and load reflection, there are some interesting relationships that can be used to correct the power error and thus improve system performance for the generator.

Tunable impedance matching networks based on phase-switched impedance modulation

The ability to provide accurate, rapid and dynamically-controlled impedance matching offers significant advantages to a wide range of present and emerging radio-frequency (RF) power applications. This work develops a new type of tunable impedance matching networks (TMN) that enables a combination of much faster and more accurate impedance matching than is available with conventional techniques. This implementation is based on a narrow-band technique, termed here phase-switched impedance modulation (PSIM), which entails the switching of passive elements at the RF operating frequency, effectively modulating their impedances. The proposed approach provides absorption of device parasitics and zero-voltage switching (ZVS) of the active devices, and we introduce control techniques that enable ZVS operation to be maintained across operating conditions. A prototype PSIM-based TMN is developed that provides a 50 Ohms match over a load impedance range suitable for inductively-coupled plasma processes. The prototype TMN operates at frequencies centered around 13.56 MHz at input RF power levels of up to 150 W.

Developing a linear model of RF power generators with pseudo random binary signals (PRBS)

In this paper, we will present an approach developing a linear model of a radio frequency (RF) power generator by using pseudo random binary signals (PRBS). We will compare two linear models obtained respectively by the PRBS approach and a traditional modeling approach. The result shows that both approaches achieve a very similar model of the RF power generator. Moreover, it can be shown that the PRBS approach is easily implemented in FPGA (Field-Programmable Gate Array) and can be adapted for the on-line system identification.

Improvement of Class E Amplifier with Inductive Clamp Circuit Topology and Its Applications

A class E amplifier with inductive clamp circuit topology is improved to have stability for all load conditions at any DC supply voltage, Vdc. Based on the new improved topology, the RF Amplifier is designed, simulated and built to be used as a driver for any class of RF amplifier operating within five percent of frequency bandwidth. It is shown that class E amplifiers with inductive clamp circuits using an improved topology give all the benefits of class E amplifiers with inductive clamp circuits and in addition have minimum spurious oscillations and harmonic levels. The improved topology and results that we introduce here can be employed in RF applications where the signal purity over entire VSWR and dynamic range is one of the amplifier requirements such as plasma excitation