Class E Power Amplifier

[sharethewell]

I have a question about excess inductance in class E Power Amplifier. It's said the LC tank in series are not tuned at fundamental frequency but actually off tuned a little bit to have an excess inductance Lx for higher efficiency. So my question is since LC tank is off tuned, then isn't the output signal (sine wave) distorted a little because of higher harmonics going through the filter?

What's normal inter stage impedance matching network used between driver stage and output stage of the class E power amplifier?

Will it distort the RF signal even when the output impedance of the driver stage is matched up to the input of the output stage?

Thanks a lot!

 

[vfone]

I don't know why you mention "LC tank is off tuned". The series LC tank IT IS tuned at fo (input frequency) and ideally behaving as an open circuit for all harmonics.
Impedance matching depends by the active components used, but generally the Class-E final stage use an input conjugate matching network .
Usually never use two Class-E stages one after the other. Generally the driver is in Class-AB and the final stage is Class-E.

 

[sharethewell]

I don't know why you mention "LC tank is off tuned". The series LC tank IT IS tuned at fo (input frequency) and ideally behaving as an open circuit for all harmonics.
Impedance matching depends by the active components used, but generally the Class-E final stage use an input conjugate matching network .
Usually never use two Class-E stages one after the other. Generally the driver is in Class-AB and the final stage is Class-E.

Well. The reason I said the LC tank series in the output stage is off tuned a little bit is for the higher frequency. If you tune the L and the C at 433MHz, the the drain voltage and current of output transister will have more overlap (power dissipation) compared to having a Lx (excess inductance). That's all I read from papers I found. So the LC series I use should have 0 impedance if it is tuned at fundamental frequency ideally, and higher impedance for other harmonics. That's whay I am a little confused that if the LC tank series is off tuned a little bit, then the output waveform is distorted a little. Since the operating frequency is 433MHz. The 2nd and 3rd harmonics are so close to 433MHz. It is difficult to use such an off-tuned LC filter series to filter out all harmonics perfectly.

For the driver stage, I am using class F topology.

By the way, regarding the inter-stage impedance matching network between driver stage and output stage, what kind of matching network is normally used?

 

[vfone]

There are Class-E PAs even for 13.56 MHz ISM band, and should be no problem at 433 MHz.

Class-F have one of the hardest to tune output matching network from all PA classes, so the interstage MN would be a challenge in your case.

You can try using a PI topology for interstage MN, which allows matching for wider range of impedances. Maybe you get a possibility to absorb the stray reactance of the shunt capacitor at the output of the Class-F driver into this PI network.

 

[jayce3390]

I am sorry but I also don't understand this concept of "off tuned LC tank", I was thinking LC tank resonates at the fundamental frequency.

For the second point, driver stage has a very low power consumption compared to final stage then if the idea is to improve efficiency I think a Class F driver stage is useless and time consuming. Do you have an idea of reached efficiency using classical Class AB or Class F driver?

 

[sharethewell]

I am sorry but I also don't understand this concept of "off tuned LC tank", I was thinking LC tank resonates at the fundamental frequency.

For the second point, driver stage has a very low power consumption compared to final stage then if the idea is to improve efficiency I think a Class F driver stage is useless and time consuming. Do you have an idea of reached efficiency using classical Class AB or Class F driver?

As far as excess inductance Lx in the output stage, that's what I understand about adding the Lx to the LC tank series.

Using Class F as driver stage is for creating a square wave for fast transision between on and off stage of the transistor in the output stage. That's main purpose. I haven't got to use Class AB yet.

---------- Post added at 02:00 ---------- Previous post was at 01:55 ----------

There are Class-E PAs even for 13.56 MHz ISM band, and should be no problem at 433 MHz.

Class-F have one of the hardest to tune output matching network from all PA classes, so the interstage MN would be a challenge in your case.

You can try using a PI topology for interstage MN, which allows matching for wider range of impedances. Maybe you get a possibility to absorb the stray reactance of the shunt capacitor at the output of the Class-F driver into this PI network.

Haven't tried PI network yet. I have been trying to use L matching network. Not very successful. By the way, when you measure input/output impedance of Class E PA, do you use ZM in SP analysis or ZM in PSP+PSS analysis?

 

[pancho_hideboo]

By the way, when you measure input/output impedance of Class E PA,
do you use ZM in SP analysis or ZM in PSP + PSS analysis?

ZM in PSS/PSP analysis.
Here you have to consider that operation points of circuit is affected
by not only drive power but also port impedance's value.

But you have to do both conventional SP and PSS/PSP Analyses.

Surely see my appends in the followings.
https://www.edaboard.com/threads/190138/
https://www.edaboard.com/threads/199753/

 

[Mamdos]

I got what you meant by "not tuned" to switching frequency. Try to analyze the circuit like this:

Instead of a series RLC load, consider an LC resonator and an R-L (RX) load. Now at the frequency of resonance the LC is shorted an the load is an R-L. Does this distort our signal? Of course! But the effect is normally negligible.
How? the excess L affects the Quality factor of our resonator (filter).
why is it normally negligible? Q is normally low itself ( around 10 ) the excess L is not going to change that a lot.

 

[sharethewell]

Thank you for sharing your thoughts. That makes sense to me.

I got what you meant by "not tuned" to switching frequency. Try to analyze the circuit like this:

Instead of a series RLC load, consider an LC resonator and an R-L (RX) load. Now at the frequency of resonance the LC is shorted an the load is an R-L. Does this distort our signal? Of course! But the effect is normally negligible.
How? the excess L affects the Quality factor of our resonator (filter).
why is it normally negligible? Q is normally low itself ( around 10 ) the excess L is not going to change that a lot.

 

[coolsaurabhsonu]

hiiiiiiiiiiii

i am also working on class E power amplifier at 403-405 MHz frequency and output power of (0 dBm ).
can you please help me designing the driver stage and output matching network.

thank you..