Three transistor 400mW FM amplifier

Hi again Joy,
Thank you very much for the SPICE link.
I stopped using SPICE regularly in about 2000 when the version of EWB stopped running on newer computers.
Since then I have an old WinXP computer that I use only occasionally. I started using it to handle 3D files, so I started again with the old SPICE models.

But now I have got a nice link I will download and have a good look at it. With the old EWB some oscillators would not start unless you added a switch to give them a kick. But they were good for most things, especially filters and the like.

Best regards from Harry

Someone ran a spice simulation and I spend countless hours soldering.

Simulation shows 24dB (200X) power gain at 96MHz . My Xtal TX outputs around 3.6mW and I got around 670mW

= 186X power gain.

Simulation saves a lot of time and energy . I'll learn Spice

Deleted

Hi Dare,

I did run through your latest circuit on the simulator.
The gain is 24.9dB. Your power gain is from 6mW to 600mW, that is 100x power gain, in dB that is 20dB.
The simulated gain of 24.9dB, I would say it is quite close to it, meaning the simulation gives valuable info about the circuit!
The gain diagram shows V(ot) / V(in), that takes into consideration the 50 Ohm impedance of the voltage source.
The input impedance is about 41 Ohm.
The exact value of the output stage tuning capacitor, C9 in the simulation, barely made any difference. The change from 11.5pF, as you suggested, to 24pF, which gave the highest gain, was only about 0.4dB!
Again, the value of the C6 capacitor in the middle matching network was critical.

If you want to play with the simulation I can send you all the files, zipped.
The simulator, LTspice, can be downloaded from the Analog Devices web-site free, not even registration is required:
https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html#
The beauty of LTspice is that it is not "cripple-ware".

Hi Joy,
You need to position the coax cables at 50Ω points so as to avoid VSWR. This makes it a little more complicated, but the benefits are great, especially at higher powers.

BR Harry

Harmonic rejection with coaxial cable? I learned something new from you today.

Input impedance of the driver is 75ohm at 96MHz. Someone simulated this circuit on Spice but the output was not properly matched. I've sent them updated schematic. I'll let you know the results.

If my calculations are correct, then with 54ma of collector current , 300/96 * 26/54 = 1.4ohm is the input impedance of each final transistor therefore I designed the input matching network to feed 0.7ohm

But someone simulated it on Spice and input Z to the final stage is coming as 5ohm. But then a proper output matching network is also missing .

dare4444 wrote:For some reason a value of 68pf gave maximum gain.

Interesting that the reactance of 56pf has an impedance of about 33 ohms @ 90MHz. 68pf is not far off - still in that region.
It would be interesting to simulate that and see how it performs theoretically. Perhaps it could also depend on the method of construction?

I was interested how you used two paralleled PA transistors, and that you found, as I did, that GP audio/switching transistors work well at VHF. I have been using BC547/BC557 for years for just that reason. BTW you can put a plastic encapsulated BC547 into a bit of copper/brass brake pipe/tube, with some silicon heat paste, and get some heat sinking, screening and mechanical mounting. You can't do that with metal can transistors since the collector is connected to the can.

Another point is that you mentioned harmonic reduction. Using a low power illicit transmitter is often overlooked by authorities, unless you have a dirty transmitter. In Spain there are NO local radio station licenses at all. All the local radio stations are pirates. The authorities simply turn a blind eye as long as they operate sensibly. Coast FM Gold  are pushing 500-Watts into a 3 element beam, pointing down towards the city of Nerja from a mountain.

If you want to really work on harmonic reduction then you can use another technique:
1 - Make two simple (separate) PA stages - lets call them PA-A and PA-B
2 - Connect RF into PA-A and connect the output to the load through a short length of coaxial cable (coiled up)
3 - Connect RF into PA-B through a short length of coaxial cable (coiled up) and connect the output directly to the load

The bit of coaxial cable in each PA is identical length, so the phase delay is the same in both PA stages. BUT, any harmonics generated in the PA will have different phases due to the coaxial cable in only one PA output. Adjust the length of the coaxial cable to target the third (or second?) harmonics you want to cancel.

For example, the third harmonic of 100MHz is 300MHz. Wavelength is 1m. ½-wave is 500mm.
If cable VF = 0.67 then the coax length should be 333mm eg. 5 turns of 20mm Dia. looped cable, and that will fit on a PCB.

You can use a pot in the emitter of one PA to balance the levels a little so you can really notch out the 3rd harmonic. You should easily be able to get a 20dB reduction. If you are really careful with making the two PAs identical, then you can get 30dB or more reduction.

Just a little idea for you. There is nothing more impressive than a hobby-built bit of kit that can out-perform professional kit 

BR Harry

For some reason a value of 68pf gave maximum gain.

Decoupling the emitter resistor of the first transistor (33 ohms) with 68 pF ????

Deleted