Theremin Oscillator Help

I'm an electrical engineering whos currently trying to design and construct their own theremin. As part of my coursework, I have worked a lot with amplifiers, but this is my first time building an oscillator of any kind. I recently designed this thing:

Which is based off of a negative impedance generator. From my calculations (and my very limited knowledge of oscillators), I managed to calculate that the resistance looking out of from the base of Q3 will be negative, which I think means that if I connected an LC tank to the same net as the base of Q3, then the circuit would refresh the energy lost by the LC tank will thus oscillate.

Unfortunately, after simulating the circuit with a simple LC tank connected to ground, this doesn't seem to be the case. The circuit will either flatline at a certain voltage, or there will be a period of oscillation that gets dampened out after the first 100us. 

I'm sure I'm missing something basic, but can someone help to point it out? Is my perspective of what makes a system oscillate wrong?

Try removing the R4.

Try removing the R4.

I tried that and it didn't work, but I managed to have some successes with different topologies over the course of the few days. I learned that my original oscillator didn't really work because it seems to be doing feedback around Q4 while the LC tank on the Q3 side. So I switched that connection and removed the current mirror because it was taking really long to simulate and the Caps also seemed to be adding a fair amount of noise into the oscillator loop.

After messing around with some of the values and topologies, I managed to get the following oscillator:

It works fine, but there are a few issues I feel like I should remedy before this thing is Theremin ready:

1. The resonance frequency seems to be located in the megahertz range. I'm not sure if this is bad or good, but most Theremins seem to have an oscillation frequency of around the 100Khz range, so ideally I would think I'd like the oscillation frequency of around there. Any attempts to change the LC tank oscillation will either result in a decaying or exploding oscillation.

2. The FFT plot generated by SPICE looks like this:

Which looks super gross because it seems like there will be a ton of noise when besides the resonance frequency, which is not ideal for a musical instrument.

Where to proceed from here is a little unclear for me because as of right now, my analysis process has been verifying the negative impedance looking in from the LC tank and then just messing around with values until it looks okay. But is clearly not. What should I do from here? Does anyone know any tips to reduce this noise?

Welcome bearimem!

It seems like you're building an Etherwave oscillator without the series inductor going to the antenna:

https://d-lev.com/support/Updated_Annotated-Etherwave-Schematic.pdf

Your emitter resistance is way too high, and your Q4 collector resistor should probably be eliminated.  You can bias Q3 via the inductor, this makes ground reference VCC, but it's effectively ground for all practical purposes.  The Harrison uses this topology with PNPs to do true ground reference, but PNPs are inferior devices:

https://theremin.us/145/145.html#Schematics

And I would recommend you breadboard these types of circuits along with Spicing them, a lot to learn from both approaches, some of it mutually exclusive.  You need to use a raw breadboard without a metal baseplate, and set it on top of a plastic box or something.

Welcome bearimem!It seems like you're building an Etherwave oscillator without the series inductor going to the antenna:https://d-lev.com/support/Updated_Annotated-Etherwave-Schematic.pdfYour emitter resistance is way too high, and your Q4 collector resistor should probably be eliminated.  You can bias Q3 via the inductor, this makes ground reference VCC, but it's effectively ground for all practical purposes.  The Harrison uses this topology with PNPs to do true ground reference, but PNPs are inferior devices:https://theremin.us/145/145.html#SchematicsAnd I would recommend you breadboard these types of circuits along with Spicing them, a lot to learn from both approaches, some of it mutually exclusive.  You need to use a raw breadboard without a metal baseplate, and set it on top of a plastic box or something.

Any tips on how to do the analysis for this kind of thing? Right now I'm just doing DC and AC, but I would like to know how to determine the criteria for whether a system will oscillate or not.

Also someone earlier mentioned earlier that the oscillators should not see each other and that I have to make 2 boards. Whats that all about?

"Any tips on how to do the analysis for this kind of thing? Right now I'm just doing DC and AC, but I would like to know how to determine the criteria for whether a system will oscillate or not."  - bearimem

AC transient analysis is of course the thing to really examine.  The oscillator may need a tiny kick to get going in simulation, either an initial current in the inductor or a tiny offset voltage (from rest) in a capacitor.  I use two analyses, one that records the whole thing for graphing and one that only graphs the final few cycles.  I've been fooled many times by the "kick" thinking it was true oscillation, when in fact it was just ringing down.

Many oscillators are simulated and benched in this thread: http://www.thereminworld.com/forums/T/30562?post=206703#206703

"Also someone earlier mentioned earlier that the oscillators should not see each other and that I have to make 2 boards. Whats that all about?"

I would suggest you breadboard stuff and follow your EE instincts.

bearimem,

In addition to Analog Theremin thread mentioned by dewster above, it makes sense to look at "Wallin oscillator" thread.
It's focused on current sensing oscillators.
I believe it may be interesting for your coursework.