The circuit is attached. After much calculation, experimenting, and getting help from members on the Forum, I designed this three transistor FM transmitter on a copper-clad board. It is voltage-tuned using a varactor diode, from 83 MHz to 113 MHz. Oscillator runs on a 12v battery, and varactor gets 21v for wide tuning range. The first stage is the Vackar oscillator. Second stage is a voltage follower used as a buffer. The third stage is a common emitter amplifier connected to a telescope antenna. It's very low power, maybe 10 mw. It does reach a few rooms in my house, but barely. It tunes beautifully and produces near perfect sine waves. The frequency stability is excellent with no drift, even after hours of running. I'm happy with it, except I think I made a few mistakes in my design which is causing a power loss. For example, I suspect the RFC and 1050 ohm resistor in the first stage might be backwards and need to be reversed. I haven't tried doing that yet. All three RFCs are small ferrite cylinders wound with 15 turns of #30 wire. The two air-core coils are 5 turns of #26 wire wound on a 7mm form. My other concern is the third stage. Do I even need the RFC in this stage? Maybe the 550 ohm resistor is all that's needed. I'm not getting much voltage gain, maybe 3x. Then again, that might be ok and I just need another stage of amplification. I tuned the two variable caps slowly until I got maximum output. Some measurements: First stage collector output is 4.5 volts. Second stage emitter output is 3.8 volts. This passes through a 47pf capacitor to the base of the third stage, for a base voltage of 2.2 volts. Output voltage at the antenna is 7 to 9 volts peak-to-peak over the tuning range. Maybe I should just take that 7 to 9 volts and pass it through a transformer to reduce the voltage and get more current to drive a final amplifier? Notes: I haven't included an audio source yet. I want to get this circuit right first. The oddball values on some resistors are actually measured values. In several cases I had to "bundle" them to get values close to my calculated values. I tried to provide all the details, but I'm sure I left something out. All suggestions are heartily welcomed!

 

After reviewing my mistakes in light of all the suggestions I got, I decided to rethink my design after the Vackar oscillator stage. The oscillator works great, but I did make one small change to the oscillator - I removed the RFC from the collector with no degradation to the signal. The following two stages are gonna be redone, even though the circuit as a whole does work fairly well. It's a learning process. I'll post the resulting circuit when I hopefully get there. Thanks to everyone who replied with help.

 

To get more power, you can remove resistor of 550 Ohm. Third stage

 

To get more power, you can remove resistor of 550 Ohm. Third stage

Thanks. I'll try it.

 

The bias on the first amplifier transistor is the collector voltage of the first stage. That may be the optimum bias point for that stage. A coupling cap can separate the voltages.

 

The bias on the first amplifier transistor is the collector voltage of the first stage. That may be the optimum bias point for that stage. A coupling cap can separate the voltages.

That's true. I experimented with various values for a coupling capacitor. I found that I got more output with no capacitor at all. But it affected the bias of that first amp because it put the collector resistor in parallel with the upper bias resistor. So I opted for a 47pf coupling capacitor and adjusted the bias resistor accordingly. I was actually surprised that it worked after my mistake was pointed out in a few posts. I think the change improved the stability. Anyway, I'm going drop the two stages after the oscillator and redesign it by connecting it to a cascode amplifier instead. We'll see how well it works. Thanks for your insight.