class F power amplifier

[Nixphe]

Hello,


i'm currently designing a 3rd harmonic class F amplifier at a few thens of MHz. In simulation I achieve 90% PAE with a reliable model for my active device, but i some things are unclear to me. I understand the principles for class F.

My active device is a BF199. First of all, it seems quite tricky to set biasing on cut-off. First i see the amplifier works as class B later, it get more harmonics i guess. However i thought the harmonics at would be gerenated by oversteering in the input rather than by correctly setting the bias voltage?

Normally, i would expect the collector current to be half a sine. Instead i find a very dirty signal, which is still quiet for little more than half a period. If i make my input signal smaller again, i find a nice half sine, but my effeciency drops. Is this normal?

Then one last question ... does my load impedance need to be matched to my transistor in any way? I would guess the transistor functions as a switch, and this is not necessary, but i'm not sure.

Thanks for your advice!

 

[leo_o2]

Smaller signal might lead to lower effeciency.

 

[WimRFP]

Hello,

When I understand you well, you are using third harmonic peaking to get a higher first harmonic component (given same supply). This means the active device sees a short for all, except for first and third harmonic.

If so, your transistor is operating in the linear region for most of the time. You cannot operate it as a switch as this will result in high current transients.

You need some current clipping to get the third harmonic current in the right (reverse) phase to get the correct third harmonic voltage with respect to the first harmonic voltage. The third harmonic current should be maximum negative when the first harmonic current reaches maximum positive value.

Using conduction angle < 180, requires more clipping (so more overdrive to get some voltage saturation) to get correct third harmonic starting phase.

Regarding current waveforms in real world, measuring is difficult (from my experience with a 50 W amplifier), and operating a BJT to get the good current shape is difficult. This is due to low impedance non-linear input characteristic and also there is feedback via Ccb.

Can you post a circuit diagram?

 

[Nixphe]

Smaller signal might lead to lower effeciency.

thanks for your reply. indeed, as a smaller input signal will eventually lead to class B as the transistor won't be switching anymore.

What i don't understand is that when i increase my input signal to the point that my transistor heavily saturates and my efficiency turns very high, my current waveform looks very nasty. It's all random spikes, looks like noise, but still half the the time (or little more) there is no current at all (this is when the voltage on the collector is high, resulting in high efficiency).

so when i am in class B situation, efficiency is low, current is half rectified sine. when i further increase input signal efficiency increases. at some point i clearly see my transistor saturates and i get lots op 3rd harmonic on the collector voltage, resulting in quite a square wave. i keep increasing my input signal, the efficiency keeps increasing, till at one point the current on the collector is no longer a half sine, but some dirty signal

---------- Post added at 22:08 ---------- Previous post was at 21:34 ----------

Hello,

When I understand you well, you are using third harmonic peaking to get a higher first harmonic component (given same supply). This means the active device sees a short for all, except for first and third harmonic.

Actually increasing the first harmonic is a nice side effect. The main purpose of keeping the third harmonic in the voltage waveform on the collector, though, is to make that waveform more square-wave-like. This means that during the half cycle where the transistor conducts, voltage is lower than when no 3rd harmonic is present and less energy is wasted in the active element.

If so, your transistor is operating in the linear region for most of the time. You cannot operate it as a switch as this will result in high current transients.

AFAIK Class F is all about operating the active element (almost) like a switch. And without driving the transistor beyond it's linear region

You need some current clipping to get the third harmonic current in the right (reverse) phase to get the correct third harmonic voltage with respect to the first harmonic voltage. The third harmonic current should be maximum negative when the first harmonic current reaches maximum positive value.

A third order harmonic current seems highly unwanted to me. It would lower the efficiency. By biasing the transistor right at threshold, half of the input sine is pinched off. This half rectified sine wave contains first harmonic and all even harmonics, but no odd harmonics. You're of course right about the phase for the voltages.

Using conduction angle < 180, requires more clipping (so more overdrive to get some voltage saturation) to get correct third harmonic starting phase.

setting conduction angle is done by tuning bias voltage only.

Regarding current waveforms in real world, measuring is difficult (from my experience with a 50 W amplifier), and operating a BJT to get the good current shape is difficult. This is due to low impedance non-linear input characteristic and also there is feedback via Ccb.

Can you post a circuit diagram?

 

[WimRFP]

Hello,

When you short all, except first and third harmonic voltage components, you can't have a switch as active component as the collector waveform isn't square wave. Only when introducing higher harmonics in the collector voltage waveform, the active device may be more or less operated as a switch.

You need some (very) small third harmonic component with right phase to excite the third harmonic resonance in your collector load. You are right, large third harmonic current reduces efficiency and is a sign of wrong collector loading (for the third harmonic).

When using 180 degrees conduction angle, the smallest current clipping is sufficient to provide the required third harmonic component in the collector current.

As your transistor has output capacitance, lead inductance, etc, you normally should retune the output network somewhat to provide the correct impedances at the transistor chip.

How do you measure actual RF transistor current?

 

[chuckey]

I am not sure that the collector current means much. If the collector circuit has a coupling to allow 3 rd harmonic to be coupled through, then it must have a hi Z to the fundemental, so the fundementals current must be lower then if you had a purely resistive load. I think the way I would proceed is to couple the input signal via a resistor in parallel with a RF bypass cap, so the more drive you put in, the more reverse bias there is on the transistor and its angle of conduction gets shorter.
Frank

 

[WimRFP]

Hello Frank,

If I understand his class F topology well, he only allows the first and third harmonic voltage component at the collector. The load receives the first harmonic only.

For conduction angle = 180 degrees, there is no third harmonic, but some saturation or non-linear IC-Vbe transfer curve will result in some third harmonic component in Ic with the correct phase.

For conduction angle < 180 degrees, the third harmonic current has the wrong phase. In order to get a low third harmonic current with correct phase to get the required collector voltage shape, you need to allow more (soft) clipping.

The risk of using automatic (negative) bias voltage is that the conduction angle reduces too much with increasing drive, so that you don't get a small third harmonic component with correct phase.

 

[royal12461]

F power amplifier classes has divided into two major groups: linear and non-linear. It has drawn more attention for easier implementation and better integrated to sub-micron CMOS technology.