fmfrequencysinevco
I am designing an FM modulator circuit using Colpitts topology. Unfortunately I encountered difficulties and do not know how to solve them as I am new in electronics.
Firstly I don't get a clean sine wave on ouput. Also, the FFT of the output differs from what is being expected – I get only one frequency for the whole input sine wave. So the circuit "shifts" the frequency but doesn't really modulate it.
I might be wrong but thanks for your help anyway.
Generate a sawtooth wave (so it has even harmonics) at 19kHz and use a 19kHz and 38kHz tank circuits to get the two frequencies. Or take the 19kHz sine and rectify it (full wave rectification doubles the frequency).
Here's a stereo decoder from my Leak Troughline Stereo tuner (only 3 transistors):
Q1 amplifies the pilot tone, which is then rectified (using D3 and D4), then Q2 generates the 38kHz signal that is phase locked to the pilot tone.
Though I'm sure it's possible to do this easier now compared to when the FM stereo was just introduced (and the tuner was designed).
I think a series clipper can do this job. The circuit of which is shown below.
Vd in schematic is the forward drop of diode D.
Best Answer
The MV2201 has a nominal capacitance of 6.8 pF at 4 volt DC and can span a range of 1.9 to one when the DC control voltage changes from 1 volt to 10 volt. So, the range might be from 8 pF (@1 volt) to about 4 pF (@10 volts).
Now ask yourself how much that is going to change the main frequency of the oscillator given that the varactor diode is, in effect, in parallel with C22 and C23 (series elements). C22 and C23 are both circa 5 nF hence, the net capacitance to L1 is about 2.5 nF and you are expecting to "observe" a frequency deviation due to that capacitance (2.5 nF) rising a couple of pico farads and falling a couple of pico farads.
My calculations tell me that a 2 pF change in 2.5 nF will result in a frequency change of 1.0004 so if your centre frequency is 1000 Hz then I would expect to see it rise to 1000.4 Hz and fall to 999.6 Hz. Is your spectrum analyser able to spot this change? It's very small.
If you made L1 ten times bigger and C22 and C23 ten times smaller, the effect might be visible.
Nobody expects it to have a clean frequency. The Colpitts oscillator is a phase shift oscillator and it doesn't run at the resonant frequency of the tank but to one side of it hence, the sinewave will always be distorted to some extent. The waveform I see in your picture is not unreasonable given your circuit.