# Direct FM generation using varactor diode

I am trying to generate a constant 1 MHz FM carrier wave. I have achieved it via LC parallel resonance circuit. Initially I let the switch B open and switch A closed, in order to charge the capacitor. When it is charged, I open the switch A and close switch B, to fulfill the oscillation phenomenon.

I am trying to generate the same carrier wave using varactor diode, but unable to do so.

At 1 V reverse biased voltage, datasheet of varactor diode tells that it has around 440 pF capacitance. Solving two capacitors in series gives 220 pF, which on solving with 0.115 mH inductor, gives 1 MHz frequency theoretically. Why am I unable to get the result on simulation?

• The frequency is vastly different (2 caps in series) and there is no current transient (its the current through L1 that is important) to kickoff oscillation. You also seem to think this will create a carrier ...it will not. The oscillation will decay. – Jack Creasey Jan 4 at 15:31
• No doubt, the model for Varactor diode has some internal resistive components that dissipate away any initial excitation of resonance. – glen_geek Jan 4 at 15:31
• Why don't you choose a classical series or parallel Oscillator design and insert Varicap with RC values for isolation? – Sunnyskyguy EE75 Jan 4 at 15:43
• @TonyEErocketscientist Please have a look again. I have just edited the question. Besides this is a classical parallel oscillator design with varicap. Is it not? – Muhammad Naufil Jan 4 at 15:48
• Those are only "oscillator" circuits in a world where coils, capacitors, and wires are lossless. In the real world you need a real oscillator. Seach on Colpitts, Hartley, "RF Oscillator", etc. As pointed out in the answer you can excite your varicap "oscillator" circuit with a pulse; you should see a damped sine wave that'll give you an idea of the Q of the circuit. – TimWescott Jan 4 at 17:09

This is not how to create a 1MHz FM generator but shows what works on a simulator using a 1kHz 1us pulse to energize the inductor with < 100mA with a Q of 2000 in theory. (10M/500)

In practise, you need actual RLC values for every part or use a CMOS inverter instead of a switch and clock.

• Thanks for responding. Why are you applying specifically 2.9 volts? – Muhammad Naufil Jan 4 at 17:21
• @Tony It's interesting that you don't need a continuous osc if you want to do simple FM chirp modulation instead of ASK. So a simple current pulse to produce symbols may be quite adequate. – Jack Creasey Jan 4 at 17:24
• And which simulator are you using? I liked it – Muhammad Naufil Jan 4 at 17:24
• Falstad web javascript tinyurl.com/yb2k2lkl It's awesome but time scale is under options > or mouse wheel and shift or control but no varactor model yet – Sunnyskyguy EE75 Jan 4 at 17:37
• @TonyEErocketscientist Why is it producing 30 to 40 Volts at output? – Muhammad Naufil Jan 5 at 9:13

Until a simulation was attempted (in LTSpice), this circuit seemed a simple one to excite.

Discovered some tricky interaction between simulation time-step and resonance that affected results. First, here's the circuit. It starts off at t=0 with DC voltage source at zero volts. The RC time constant (R=1MEG, C=1n) rises to +20 volts in about five milliseconds. This transient alone is enough to excite resonance. During this time, varactor capacitance decreases, which increases resonant frequency from about 52 kHz to 400 kHz.:

A transient simulation shows the resonant voltage across the inductor rising for a few milliseconds while resonant frequency changes, then decays slowly as resonant energy dissipates via R1's damping effect, and possibly by dissipative resistances in the Varactor diode D2 model. Note the fine timestep of 5 nanoseconds - this simulation seems reasonable:

The same circuit with only the timestep changed to 150 nanseconds yielded a different simulation result, showing no resonant decay, and a distorted envelope rise - this simulation seems to have run afoul:

These plots don't show resonant frequency change (one must zoom in very far to see cycle-by-cycle period). One would think that a 150ns step size should yield a decent simulation, but it seems that an upper limit of 5ns does a far better job. Estimates of resonant frequency benefit from small step sizes. Be sure to limit your simulator to a small time-step.