New project proposal: voltage-mode Class D RF PA for HLx

[nekv...@gmail.com]

Based on the work I did with K1KP several years ago and subsequent simulation work, I believe it is possible to produce an affordable 100-W+ high-efficiency EER amplifier for Hermes Lite using GaN FETs manufactured by GaN Systems. I am writing to recruit a team to join me in this effort, as I don't have all of the necessary skills and important design decisions need to be made.

The two biggest decisions may be:

1. Do we need to move on to HL3, if only to replace the existing FPGA with a part having better availability?
2. Should we create an "HL2-EER" derivative, add EER-specific circuitry to a new HL3 or create a separate board to bolt-on the necessary circuitry to HL2?

Each of these top-level decisions carries with it a host of lower-level decisions. For example, an "EER derivative" could replace the post-AFE dual OpAmp (U9) with  comparators to produce the EER phase drive signals, while a separate board would apply the HL2's single-ended analog output to comparators. The derivative would need high-speed drivers to send the phase drive signals to amplifier board.

The duty factor of the phase drive signal needs to decrease with increasing carrier frequency. It seems to me that this is best accomplished using I2R-programmed DACs to control the voltages applied to the phase drive comparators, so hardware and software I2R integration knowledge is required.

Choice of the active device will depend on the power level to be achieved and the methods available for sinking the heat that is produced. These are very small devices, so dissipating even a few watts needs to be carefully monitored and exhausted. It seems likely that it would behoove us to instrument the design for input power, output power and device temperature, plus any circuitry/software to monitor and act on the data. Likewise with SWR; high SWR will degrade the efficiency, increasing device power dissipation, so SWR will have to be monitored and the data acted upon appropriately.

Luckily, while design decisions need to be made, the pulse-width modulator (PWM) is largely solved, at least conceptually. Here, again, system level integration may be necessary.

Anyone in the group interested in taking this on?

73 for the upcoming new year,

Brian K1LI