Wien bridge slowly decaying oscillator

Question

I want to build a resonator that slowly (~3 seconds) decays without a stable oscillation. I've tried using a simple RLC circuit without luck - I've got ~1-2ms decay time at maximum; so, I've turned my attention to active oscillators, particularly to the Wien bridge oscillator shown below.

If I turn the R4 all the way to the right, I get a stable square-wave output; if I turn it to the left, the output shape becomes more and more sinusoidal and at some point it becomes unstable, decaying to silence after some time has passed after an induced by pressing the SW1 button disturbance. If I continue turning the R4 to the left, the decay time decreases; it seems logical, that if I want a decay time around several seconds, then I need to turn the R4 right just up to the point where the oscillation starts.

This solution, however, is impossible to embody in real parts, since even a small change in resistance (due to the temperature change, for example) will put the oscillator in the stable oscillation territory again, which I want to avoid at all costs.

So, my questions are:

1. How can the Wien bridge oscillator schematic be modified so that it will be slowly decaying without the risk of a stable oscillation?
2. Is there a better way to build a slowly decaying resonator?

I am looking, if possible, for a purely analog solution without inductors and without "cheating" by using a stable oscillator and a VCA to control the slowly decaying signal amplitude.

^{simulate this circuit – Schematic created using CircuitLab}

Answer 1

Just like it`s not possible to realize a linear circuit having exactly unity loop gain (steady-state oscillation) it is a problem to exactly meet a loop gain of 0.999 or so. I think, the "most critical path" is to meet exactly the gain requirements for the active part of the circuit.

Therefore, the following approach could solve the problem: Usage of an oscillator configuration based on a unity gain amplifier which does not need any resistors (with tolerances) in the feedback path. There are some circuit alternatives - start a search for "unity gain oscillators".

EDIT/UPDATE:

Perhaps a better and more reliable alternative:

Use a classical Wien-oscillator and multiply the output with a decaying gain factor (first-order RC-step response) - for example with an OTA. In this case, the decaying signal is used to drive the gain-determining current Iabc for the OTA.

Answer 2

As you've discovered, boosting the ridiculously-low Q of RC circuits up to high values (near oscillation) requires very stable components, and well-controlled gain. It is asking a great deal, and stability is a big problem.

Starting with an LC resonator would be much easier. Starting with a higher Q than RC networks means less gain is needed. But "L" you don't allow.

An alternative resonator might be a piezo element. No "L". Starting Q at resonance is high, requiring little extra gain to extend the decay time , or push into oscillation:

A one-transistor gain element is usually used to make an oscillator. Changing resistor values should allow gain to be reduced to the point where oscillations decay slowly. The only downside is that frequency is determined by piezo physical parameters:

^{simulate this circuit – Schematic created using CircuitLab}

Answer 3

@sx107 @TimWescott A bit delayed, but, without any info on what application your decaying oscillator will need to fit to, I like TimW's approach. So, why not buy a second-hand electric guitar. This mechanical oscillator is easily stepwise tunable, has a very high Q-factor (decaytimes 5 s or more) and can be electronically read out.