Why does this implemented Hartley oscillator's amplitude vary significantly?

Question

I have implemented the below Hartley oscillator in LTspice for analysis before implementation. Later, it was implemented on a circuit board.

I see it oscillating in LTspice at the calculated frequency of where it should be, 1/2·π·sqrt(LC).

On a circuit board, I see it oscillating at about 40 kHz higher, I recall it being 578 kHz or so.

I am fine with its frequency.

When implemented and powered up, I watch the developed waveform on my oscilloscope. The waveform moves around and its amplitude varies more than in LTspice. The frequency seems constant, but the waveform, while looking sinusoidal, is moving up/down and varying in amplitude.

What is going wrong with this and how can I fix it?

I thought Hartley oscillators should have a constant amplitude?

I remember doing an experiment with a crystal oscillator and noting that the waveform was locked in and super steady.

The LC waveform from the below circuit, when implemented, almost seems unusable. It varies too much that makes triggering on a rising edge highlight its movement.

I saw an earlier question where someone mentioned that unless the inductors had very high q, it might not even oscillate. There was also mention of the collector output needing to be high-impedance and tapping the output might also kill oscillations.

My circuit oscillates, but not acceptably.

What could cause this and where can I learn more about the drivers of its variability?

I am hoping to see a constant/rock solid amplitude and frequency.

EDITED SCHEMATIC

Answer 1

What is going wrong with this and how can I fix it? I thought Hartley oscillators should have a constant amplitude?

There is nothing in the basic Hartley oscillator circuit that defines the output amplitude. Same story for Colpitts too. What will tend to keep the amplitude somewhat stable are non-linearities in the transistor leading to slight sinewave distortion at one end and clipping at the other extreme. Resistor component values can mitigate this but, it'll never be anything close to a true sinewave.

Your schematic also has R1 unconnected for some reason.

Answer 2

Generally, changing a "little" one-value component doesn't change the behavior of a classic oscillator. Just the frequency is "slipping".

Look at this: changing 100 nF to 300 nF (whatever is R3=500 or 1500 Ohm).

Answer 3

I have since figured out what was wrong. The emitter resistor needed to be lower. When I lowered it from 1500 to 100 Ohms, the output became perfect.

Answer 4

"I am hoping to see a constant/rock solid amplitude and frequency."

Each working oscillator needs a loop gain>1 which allows a safe start of oscillations (for small amplitudes at switch-on at t=0). Therefore, the amplitudes would increase until they are limited (clipped) at the power supply rail - unless the amplitudes are "softly" limited because of some non-linearities within the circuit.

Sometimes, such non-linearities are part of the design (diodes, for example) and sometimes the inherent non-linearities of the active device are exploited. But it is not possible to exactly determine the amplitudes which can reduce the loop gain to the value required by the oscillation condition (unity loop gain).