WA2WVL Sept 12, 2012 Email: wa2wvlfk@hotmail.com
Using Power Supply Mosfets in HF Power Amplifiers.
1. Input Circuits
Most power supply Mosfets have an input Q greater than 5 at 160 meters and should be driven with a voltage source such as squarewave driver chips or decrete npn/pnp circuits. This is because the input capacity of the Mosfet is the dominate load on the driver. An alternate method is to terminate the inputs in shunt resistance (with blocking capacitor) and drive them as resistive loads.
As the frequency increases to 40 meters the input Q decreases to well under 5 and a different approach should be used. It becomes very difficult to drive large input capacities from a voltage source so current drive (tuned transformer drive) should be used. This has been done as high as 6 meters with large Mosfets.
At 80 meters either method may be used.
Table 1 shows the input impedance and input Q of the IRFP32N50K Mosfet on some of the HF bands. At 40 meters the real part of the input impedance is about 1.13 ohms while the input reactance (in series) is 3.56 ohms. Q = 3.56/1.13 or 3.15. Going to 20 meters and higher the Q decreases further and transformer drive works even better and voltage drive has no chance.
2. Output Circuits
Much of the work on Class E amplifiers has been done with single ended designs. Single ended designs have a fundamental flaw in that power is generated only on the turnon part of the rf cycle but must be delivered to the load over the full rf cycle. Where does this power come from? The Mosfet must handle more than twice the load power (allowing for less than 50% duty) durning its conduction period. This is the reason that a minimum Q (as Rabb has calculated) is needed.
Push/Pull does not have this problem and is inherently more efficient than single ended. In Push/Pull the turnoff energy needed to turn off device #1 is mostly furnished by turnon of device #2 and need not be stored in the external circuit. I would welcome discussion on this point.
By the way, combining two single ended amplifiers, 180 degrees out of phase, is not the same as operating a push/pull amplifier since each amp still must store the turnoff energy and the bandwidth of the output transformer must be very wide since it is not handling a sine wave. In Push/Pull the output transformer is handling a sine wave and bandwidth is not a factor.
Efficiencies greater than 90% can be obtained in Push/Pull by strict attention to loss factors. Rds on should be low enough that it consumes less than 1% power. Transformer loss also can be designed to be less than 1%. Switching loss will be under 5% with proper output tuning (a critical adjustment). Lastly the output network loss can be held to 2-3% by using Loaded Q’s less than 5 and Unloaded coil Q’s as obtained with copper tubing coils. Don’t forget that the capacitors must also be high Q. Doorknob or surplus ceramic capacitors are NOT high Q. The best capacitors are porcelain (such as ATC-100), or multiple Mica’s or air dielectric. The best practice is to test the caps at the rf current expected in the circuit before designing them in.
Squarewave vs Sinewave Drive.
In a Push-Pull switchmode RF amplifier as the left device turns off and the right device turns on there may be some overlap since the turn off time is usually longer than the turn on time.
In the past I have used Sinewave drive to provide this small time (by delaying the drive by 10’s of ns) to prevent common mode conduction (both devices passing current). Where the drive can be adjusted, drive level can optimize efficiency. Also
Sinewave drive can provide the peak currents needed to turn the devices on (and off) due to the Q of the source output filter. The devices should be prebiased just below turnon so that they will be well saturated over the full half cycle.
I used this technique to build a Class E, 1500 watt amplifier for 40 meter RTTY with efficiency well above 90%.