For a long time I have fussed and worried about the design of HF RF amplifiers. You see so many varieties on the web, but they seem to break down into two sorts: broadband, and tuned. The tuned ones are for dedicated operations like fixed HF transmitters and have tuned matching networks, where as the broadband ones are liked by radio amateurs to use across the HF bands. And one notable thing about these amateur designs is the use of the common IRF510 MOSFET clamping down the input with a very low value resistor (20-100R) intended to remove any resonance between the IRF510 input capacitance (180pF) and the inductance of the driver transformer secondary and try to flatten out the response.
Here's my methodology, right or wrong, for a simple amplifier design with a target of 5W output @ 7MHz using an IRF510 output stage and a 2 x 2N3904 drivers.
Note: the decoupling capacitors of 100nF have an impedance of 0.2R @ 7MHz, so two are used in parallel on the supply. A parallel 10-100uF may be needed for low frequencies and is normally recommended if you are using the amplifier for SSB.
A MOSFET is a voltage-in/current-out device, unlike a BJT. For the IRF510 the input capacitance Cin = 180pF, and the output capacitance Cout = 80pF. The enhancement mode bias needs to be about Vth = +4 to 4.5V to set the operating point. The transfer characteristic shows that a 0-1A output needed is achieved by a Vth ~ 1V signal.
On the data sheet the transconductance = 1.3amp/volt @ 4A, higher at lower IC values.
The load resistance for 5W output is
RL = Vcc^2/2*Pout or 144/10 = 14R
To match a 50R output impedance this means a transformer of
N = sqrt(50/14) = 1.9, or a turns ration of 9t:5t. 5 turns to the MOSFET drain side of course.
The transformer will be on two FT50-43 toroids, so
Lp = 11uH (5turns) or XL = 480R @ 7MHz, meeting the need for it to be x5 to x10 the matched impedance.
The secondary (9 turns), to match a 50R output load of the amplifier, is 36uH with an impedance of 1k5 @ 7MHz
The resonant frequency of the Cout = 80pF and the transformer L = 11uH is 3.5MHz with Q = 0.08.
The input impedance of the IRF510 with Cin = 180pF in parallel with a 68R resistor < 40R @ 7MHz.
Transconductance shows that an input voltage swing of 0-1V is needed to drive the MOSFET to 0-1A drain current. The quiescent current will be set to 100-200mA, empirically
The driver load resistance for a nominal 150mW output is
RL = (Vcc-Vee)^2/2*Pout = 121/0.3 = 400
So the interstage transformer is an FT50-43 and has a ratio of
N = sqrt(400/40) = 3, or a turns ratio of 12:4
The input to the driver will initially be capacitor coupled, later I will see if a transformer is required.
This is the circuit, I have implemented with parallel 2N3904s and a IRF510
The driver DC operating conditions were arrived at experimentally. At the start the emitter resistors were 10R, but when biased to 10- 20mA each the current drifted up a lot as the transistors got warm. So I have settled on 56R which is much more thermally stable. with Ic ~ 22mA each.
Here are the results using this setup,
Input from Si5351 SIgnal Generator via a 40m LPF,
In ~ 16dB, or 44mW
Output into the final stage impedance ~22dB (8db gain...)
For info on the SIGGEN go here. And for the RF METER go here. For the 30dB TAP go here. The LPF is a small kit from QRP Labs.
Just as an experiment I connected the DRIVER output to a small QRP PA from QRPver that I have here. This is rated at 10W out for 1W in. It uses a single stage MOSFET type RD16HHF and runs at 12-14V, its an excellent piece of kit.
With the driver stage input this produced an output of 3.3W in to a 50R load (10dB gain). Here's the schematic for interest. Notice the input design, the 50R input is transformed by the 4:1 transformer down to around 12R to drive the MOSFET... the output has a 1:3 transformer to match a 50R load.
As you can see it has RF triggered VOX and this worked fine with the 300mW input.
My power stage is a single IRF510, biased to around 150-200mA standing current (I used a normal pot for this, but it is very sensitive so a multi-turn pot will be better). The input of the IRF510 is very capacitive and the stage is prone to oscillations, so it's input is swamped by a 68R resistor, this makes the overall input impedance < 40R at 7MHz.
The overall gain of the two stages is 6dB (driver) and 7.5dB (output). The driver gain is very low, probably because the output stage input impedance is a lot less than 40R. This needs looking into. But first I need to add a pre-driver to get some more overall gain. At the moment the output is only 1.3W into 50R