FM Transmitter Power Amplifier and Oscillator

Where’s the frequency control on the oscillator? I understand that this isn’t the full circuit, but still the oscillator must have an input somewhere, right? Or is this just a placeholder oscillator? Also, I’m observing a weird instability in this oscillator that I’ve never seen before. Every once in a while the oscillations start decaying for no apparent reason and then are taken over by some 400-500MHz oscillations for a very short time and then the oscillator resumes normal operation for another while so that it essentially has periodic fits of instability with a frequency of maybe 1MHz (just a wild guess). The only change I made to the circuit was to add a disconnected 300MHz source so that I could simulate it at 10ns/s and still have an accurate simulation.

Now that I look at it a bit closer it looks like the whole oscillator bit on the left doesn’t actually resonate at 100MHz but actually 400-500MHz and the 100MHz resonance comes from the interaction of the 43nH inductor and 56pF capacitor. In that case maybe the damping amount of the 100MHz oscillations depends on the current through the lower 1μH inductor. In that case when the 100MHz oscillations are strong enough they cause current to build in the 1μH inductor and when enough current builds it kills the 100MHz oscillations allowing the 400-500MHz oscillations to manifest before the current through the inductor drops again fairly rapidly and allows the 100MHz oscillations to take over again. I’ll have to look at it closer and test this theory to be sure.

Hmm, it looks like the current through that inductor doesn’t have anything to do with it and only changes in response to the oscillations, not the other way around. In that case I have no idea what could be causing the oscillator to be doing what it’s doing.

The 50mΩ resistor and 110pF capacitor on the left seem completely useless as no current flows through them and when I force a temporary imbalance to make current flow through them it creates a 50MHz oscillation that is not in sync with the 100MHz one and that also decays over time.

In fact, the whole oscillator unit (everything to the left of the blue wire) doesn’t contribute at all to the 100MHz oscillation. Even when it’s removed the 100MHz oscillation will be amplified and sustained as long as it has enough initial amplitude to trigger the first transistor in the two transistor amplifier. This essentially reduces the “oscillator” to a weird kick-starter circuit that isn’t even necessary because simply connecting the power supply ought to deliver a large enough impulse to kick-start the 100MHz oscillations.

Ok, I finally found what causes the periodic instability. So, first the oscillator on the left starts oscillating at a few hundred MHz, then once the oscillations are large enough they trigger the transistor thereby kickstarting the 100MHz oscillation which then kills the fast oscillator by creating a charge across the 15pF capacitors in the high frequency oscillator which biases it’s transistors into a hard off. Once these capacitors eventually lose their charge thanks to the current from the 68kΩ resistor, the fast oscillator starts to damp the 100MHz oscillations and then resumes making fast oscillations which then puts a charge across the 15pF capacitor again thereby killing the fast oscillations again and so the cycle repeats.

I abandoned this design. The idea of the oscillator on the left is that the two halfs resonate at half the output frequency (thus less drift), but it's a bit unstable in my opinion, and would be a pain to tune throughout the whole FM range. In fact, I couldn't find a varicap that would have sufficient capacitive range to cover the whole band (keeping in mind Q and max current rating of the varicap)

Yeah, I figured that out, but in the end it ended up oscillating at 100MHz for a completely different reason and so the 50MHz oscillation would drift out of sync if simulated with enough time resolution. And also very unstable in my opinion. So yeah another design would be a lot better. Do you have it worked out well enough to put on EC yet?

Do you know why http://everycircuit.com/circuit/4648736044875776 is a VCO? Also, you might be able to use it for your PLL if you adjust the component values to oscillate at 100MHz, but the frequency variation isn’t super linear so you’ll need more microphone gain for one end of the FM spectrum than the other.

In reality, it would be much more stable, as the transistors which the design require, don't work well above 200MHz, plus all components will have active resistance to dampen oscillations. Especially the capacitors, varicaps will have enough resistance to dampen any unwanted oscillations.

I think I'll go with a regular colpitts oscillator. There's a PLL feedback after all, so any instabilities should get locked under a crystal oscillator reference, so it won't drift. In order to maximize linearity, I think I'll use one set of varicaps for the VCO control part, and a second set of varicaps for the audio modulation (fixed-biased at the most linear part of their characteristic).

But do you know why the linked circuit acts like a VCO? It doesn’t even have any variable capacitance.

I think the Miller's effect adds onto the overall capacitance in the circuit, by shifting the transistor to different gain levels. The collector-base capacitance is seemingly increased with increasing the gain of the transistor. Thus it forms a reactor modulator.

Actually, that can’t be because the base is held at a fixed voltage and the miller effect depends on the collector-base capacitance affecting the voltage at the base. I’ve determined that the current through the resistor determines the magnitude of the effect with low currents giving little frequency difference and large currents giving large frequency differences, but that hasn’t helped me pin down the cause of the VCO effect, only how to control the strength of the VCO effect.

BJTs gain is controlled by current not voltage. I've tested the Miller effect here in EC and the transistors exhibit this effect due to the very nature of the transistor. You see, the apparent increasing capacitance is actually a virtual effect due to the capacitive connection between input and output with common emitter amplifiers. This is usually given by the following equation Cmiller=C.(1+Av). The Av (voltage gain) value of common base amps can be affected here in EC, by choosing the bias current of the transistor. Remember that in common base amplifiers the gain is roughly equal to Ic/Ie, so by changing the base current, you also affect the gain. The only difference that's in our interest here is that in common base amplifiers, we use the emitter as the AC input, but use the base to set the DC gain, so it's still affected by the Miller gain, but just not in the same fashion as common emitter amps. With common emitter amps it affects the high frequency response, as the base is also the input for the AC signal, as well as the DC bias path. With common base amps it affects the oscillation frequency (when used as oscillator) due to varying apparent leakage of collector capacitance to ground, which is affecting the LC tank's resonance frequency.

I’ll be continuing this conversation over on my circuit since we don’t need it happening two places at once and I think my circuit is a little more relevant to the current conversation.