Jason has been working on a portable software defined transceiver design for the past year. Every day this week he’ll discuss a different part of the hardware in a series of guest posts. You can chat with the designer in the forum. Today’s post is about the power amplifier.
Since this project is designed for portable (as in battery-powered) operation, it needs to have an efficient power amplifier that is capable of stable operation across the band and is as efficient as possible.
Lots more below.
There are a number of different types of RF amplifier in common use. The only ones with decent efficiency are class C, D, E, and F amplifiers. Each of these amplifiers uses a slightly different methodology in order to achieve very high levels of efficiency. With the exception of the class C amplifier, the theoretical efficiency of the remaining three types is 100%.
Class F amplifiers are quite difficult to tune, owing to the presence of several tuned circuits in the output stages which must be tuned to harmonics of the fundamental frequency. As a result, they don’t seem to be popular in hobbyist type projects.
Class E amplifiers achieve their efficiency by switching the transistor at a point in the cycle when the current flow through the switch (transistor) is zero. This eliminates power loss in the switch and results in high efficiency. The tuned circuits necessary to accomplish this are frequency specific, so the efficiency drops off rapidly as one tunes above and below a point of optimum efficiency. Furthermore, very high voltages are developed across the switch in this type of amplifier, which requires the use of transistors that have high breakdown voltages. Such transistors involve compromises which prevent the amplifier from working as well as it might with a transistor with a lower breakdown voltage.
Class D amplifiers are implemented as a push-pull design, but rather than having both transistors conduct current in the linear region (e.g. where they have finite resistance), they are again operated as switches which greatly improves their efficiency. The typical Class D amplifier is operated in voltage mode. In an ideal voltage mode class D amplifier, the output node sees a half-sinusoidal current waveform, and a square voltage waveform. This type of class D amplifier is popular at lower frequencies (widely used in audio amplifiers for example), but has not worked as well at RF frequencies for various reasons.
In 2001, an alternative type of class D amplifier was proposed by Kobayashi et al. and dubbed a current mode class D amplifier (CMCD for short). In this type of amplifier, the output node sees a half-sinusoidal voltage waveform and a square current waveform. This results in the same theoretical 100% efficiency as the voltage mode class D, but it is much easier to realize at RF frequencies because the transistor output capacitance is absorbed into the tank circuit.
The CMCD RF amplifier offers several advantages over the more popular class E RF amplifier. The high voltages present on the output node of class E amplifiers are not present on CMCD designs, allowing a better selection of transistors to be used as the switch. The fact that the output capacitance is absorbed into the tank circuit also allows the CMCD amplifier to operate at much higher frequencies than a similar class E design could.
In my previous experience with class E amplifiers, I have found that it is straightforward to obtain high efficiencies at low frequencies with common parts. But beginning at 20M, the efficiencies start to drop. I understand that more exotic RF transistors are available which can work at much higher frequencies, but these tend to be expensive and difficult to obtain. Furthermore, as they are generally not designed for class E mode operation, it is necessary to operate them at lower power levels than what they are capable of in order to avoid exceeding their breakdown voltage owing to the large voltage swings present on the output node.
After reading a bit about CMCD amplifiers, I decided to build one myself. With a bit of tweaking, I was able to reach 90% efficiency on the 15 M band using common FETs available for well under $1 each. This was significantly better than what I’ve managed to be able to do with class E, so I decided to use this type of amplifier for this project, figuring I would be able to reach higher power output levels as well as run at higher frequencies than I would with the same transistors running in class E mode.
Here is the schematic for the PA
Plus I have uploaded the matching PDF for anyone who wants to see the schematic more clearly. As usual, comments are suggestions are solicited. I still have a lot more to learn.
Via the forum.