This is an example of a high efficiency QRP transmitter designed to work at very low supply voltages (3v-5v). It can produce 2 watts a 4 volt supply @ 70% efficiency. It uses small, inexpensive switching mosfets. The primary requirement for these mosfets is low output capacitance, a VDS of >20V, a logic level VGS and a drain current rating of a couple amps. There are many devices that will work. Unlike a Class E amplifier, this design requires no special alignment, providing for multiband operation easily. Only the output filter consisting of a L Network and Pi network in series need to be changed for a given band. It is tolerant of all kind of load conditions including infinite Z and maintains efficiency when poorly matched. While this circuit utilizes a microcontroller, display and clk generator, the logic buffer can operate from any oscillator source so the amplifier can be adapted to simpler designs.
Circuit Salad 
Interesting PA, I have been looking for a circuit that could give a few watts out on low voltage supplies so I was pleasantly surprised by the mail notification on this one.
I guess you do PWM for the bias to the FETs?
What current do think the bias current is if I wanted to test out the circuit as-is without the MCU?
Thank you for sharing another novel circuit!
//Harry
No current just a bias voltage from the MCU’s DAC module. With most logic level FETS you don’t need it. That particular FET has a threshold V low enough that it works without any bias. I used the DAC to produce about 1 volt shift up(shifts the gate zero voltage above ground) to slam them a little harder. Any voltage source like a LED forward V..through a divider or POT would work fine if desired. You would adjust to just at or below turn on.
OK, got it. Thank you!
Cool design. Is it working in Class-D? In my experience HF amplifiers claiming high efficiency often work at least partially in Class-E mode. Did you measure currents / voltages on the PA transistors?
Does the keying circuit shape the dits?
I am in a process of building a FOX transmitter for my kids from a junk box, your design would fit the bill.
Its not running class E, it’s not exactly class D either..Its a traditional push pull class C amplifier but driven with a switching waveform, resulting in class D like behavior and high efficiency. My efficiency calculation is simple: power out/power into the final. There is very slight wave shaping..when I have used this buffer drive circuit in the past I have had little key click effect. If you use it at 20 meters or below you can use the buffer drive chip as a 1 watt amplifier by itself..that is on my blog as well.
Interesting circuit. My LTSpice simulation suggests the output impedance is lower, around 5 ohms.
Decoupling the coils on T1 appears to improve efficiency and power output.
With the correct matching, at 3.7 V, the output power is simulated to be around 3.3 W and the MOSFET’s dissipate close to what they can safely handle.
Be interesting to build the circuit and see what happens in reality.
So the output transformer should perform better I would think as even order harmonics are canceled and the core size can be smaller as the currents cancel each other reducing net current across the transformer. I briefly experimented with separate inductors(trying to minimize specialized parts) early on and found the performance poor. I interpret this topology roughly as a switched push pull class C amplifier . From this premise, I did back of the napkin calculations for the design, so I wanted to get 1-2 watts from a 5v supply. This gives us a single ended Z of (5^2)/2*Pwr -> 6.25 ohms. we can then doubled this because of the push pull topology…this yields about 13 ohm @ 5V and 2 watts out. With the actual design I get 2 watts out at just over 4 volts…so my filter network may be off a little or my use of a class C model may be inaccurate. Regardless, it seems to work well.
Somehow, when looking for inspiration and insight I’ve landed on your pages. I’ve been developing a barebones stm32 sdr tranceiver for some time now and just ordered the first PCBs. In many ways your 4V transmitter post set the design philosophy of the transceiver. I did not want to use a regulator and was tired of difficult to match rf pa filters, your had a solution with real results. I have given circuit salad credit in the project readme, let me know if that is fine and I hope you find my project interesting. https://gitlab.com/m0rph/pocket-rocket-sdr
I’m glad you like the circuit..I think you will find it is easy to get running