375-Watt Carrier Class-D AM Transmitter
For 160 and 75 Meters
Breadboard

While gathering parts for a Class E rf deck I got the urge to try out my recently assembled WA1QIX PDM generator into something other than a big resistor! Having several extra homebrew Class D rf deck carcases available from my 137 kHz transmitting setup it seemed like the perfect opportunity to give Class D a try at 75 meters. The first deck was designed for a 100-watt carrier using two 11N90s in push pull - each with it's own IXDD414 driver. It didn't take long to realize that the concepts that worked well at 137 kHz weren't going to work at 75 meters!

Long story short...switching from voltage mode Class D to current mode Class D and changing form a large torroidal output transformer to a 'binocular' style one had the project back on track. The transmitter then effortlessly cranked out 100 watts carrier at 92-93% efficiency and produced modulation peaks in excess of 150%. Linearity and spectrum checks showed the transmitter to be clean and suitable for on the air use and the transmitter worked well as expected. A schematic of the 100-watt deck can be found at 100-Watt Class D RF Deck. 100 watts does well during the day but tends to make you invisible at night - especially in the ghetto! The existing rf deck had room for two more FETs so moving up to a 200-watt carrier would presumably be a simple matter.

WA1QIX had indicated that IXDD414s are capable of driving two 11N90 gates on 75 meters so the 200-watt version would only required adding two more FETs plus an output transformer ratio change and some additional ferrite material. The 200-watt version worked as well as the 100 watt - clean rf spectrum and again 92-93% efficiency and was on the air in short order. A schematic of the 200-watt unit can be found at 200-Watt Class D RF Deck. One wouldn't think that a 3 dB difference in power would be all that impressive but it sure makes a difference - especially at night. Some have said that there is a 'power shelf' that one must get above at night to be heard and the 3 dB increase in power made a noticeable difference. If 200 watts was better, legal limit would be even better yet!

The logical progression was to maintain the same topology and just add more IXDD414s, 11N90s, adjust the output transformer ratio and add more core material and that was the approach that was followed. The legal limit transmitter uses four IXDD414s and eight 11N90s. The schematic for this version can be found at 375-Watt Class D RF Deck. Keep in mind that all of these transmitters are just in the breadboard stage and represent works in progress. The measurements below were made on the legal limit version although measurements made on the 100 and 200 watt units produced similar results. Efficiency for the 375-watt unit is also in the 92-93% range.

Looking at the schematic one will notice that these amplifiers have no tuning controls - unlike Class E amplifiers which have tuning and loading controls. This means that for the amplifier to operate properly its output must be presented with a load impedance close to 50 ohms. In my circumstance this doesn't present a problem as I always use antenna tuners on the low frequency antennas. In effect, the antenna tuner takes the place of the amplifier tuning and loading controls. The transmitter is well behaved into VSWRs up to 1.5:1. It may be fine into higher mismatches although no further measurments have been made. The transmitter has inadvertantly been operated into an open circuit and the wrong antenna several times and suffered no ill effects. No protection devices are currently included in the transmitter but may be added in the future. An overload/high VSWR shutdown system, such as offered by WA1QIX, is probably a good idea and will be added in the near future.

Test Setup:

Gate and drain waveforms:

The gate waveform as measured directly at the gates of the 11N90s with Vds removed. The waveform is somewhat more square-wave like at the output of the IXDD414. The twisted pair connecting the drivers and FETs is about 3 inches of #20. Moving the drivers closer to the FETs or increasing the wire size would no doubt square up the waveform but it seems to work okay as is. The drain waveform is classical Class D looking with the usual ringing at the top and bottom. Actually the waveform is somewhat cleaner (especially at the base line) than shown owing to extraneous pickup with a not so well grounded scope probe in close proximity to the 375 watt carrier!

Spectrum analyzer shots:

Spectral analysis of the 375 watt transmitter on 75 meters. The upper shot is a wideband view looking 0-30 Mhz and the top of the scale is +60 dBm (1 kW). Harmonics are greater than 54 dB down - this with the simple 3-element low-pass filter. Although legal as is, a future improvement will be to add two more elements to the low-pass filter to bring the harmonics even further down. The lower spectrum analysis is a close in view. The two 'pips' about 155 kHz on each side of the carrier are the PDM switching frequency components which are about 65 dB down from the carrier. This will be improved shortly with the addition of two more elements to the PDM output filter.

Triangle waveform test:

Results of the 100 Hz triangle test. Additional tests were run using sine, square and triangle waves throughout the transmitter audio passband with similar excellent results. Further testing will be conducted.

375 Watt RF Deck:

Click on image for larger view. Nothing special in the construction of the breadboard. #14 copper wire used for gate and drain buses. This will likely get replaced with flat stock. #20 twisted pair may get replaced with something bigger. Other than that it's pretty much good to go as is.

200 Watt RF Deck:

Click on image for larger view. Again, nothing special in the construction of the breadboard. This deck started out as the 100-watt version. The only differences being two less 11N90s, two less ferrite cores and a different transformer ratio. All three transmitters are very similar.

The rf decks shown here operate on both 160 and 75 meters with just a change of the output low-pass filter. A 200-watt 40-meter deck was constructed and used extensively during the summer of 2007. As expected it required individual IXDD414s for each 11N90. A bigger problem at 40 meters was the leakage inductance of the output transformer. A legal limit version is in the works and will be described here in the future. I'd be interested in hearing from anyone that duplicates these decks or has comments concerning the design.