Let's Design and Build a (simple) Analog Theremin!

Posted: 1/3/2019 10:57:13 PM

From: Madison, WI

Joined: 5/19/2016

You can use LTSpice to create wav files. Search help for .wave. Here are a few lines from the help file.

LTspice can write audio files.  These files can then be listened to or be used as the input of another simulation.

Syntax:  .wave   V(out) [V(out2) ...]

Example: .wave C:\output.wav 16 44.1K V(left) V(right)

I've used this many times with reasonable results.

Mark Allie

Posted: 1/20/2019 2:59:38 PM

From: Northern NJ, USA

Joined: 2/17/2012

Franklin Oscillator

Over on the "basic experiments" thread [LINK] Dominique mentioned the Franklin oscillator, which was news to me so I thought I'd investigate it a bit over here so as not to disturb that thread.

Here's an LTspice schematic:

And here's the LTspice file if you want to play along at home: [LINK]

It's just a non-inverting amplifier with very small drive and sense capacitors going to the LC tank.  It just barely oscillates in simulation so I presume it's worse in real life, and so I'm disinclined to take this investigation beyond the sim phase.  If you run the sim you'll see rather significant phase shift from the input to the output of the amplifier, which is a tank voltage swing killer, particularly with high Q tanks.  Amplifier input and output are connected together, which rarely works well.  And the larger the tank L the more trouble it has oscillating.  And literally all of the components have to be tweaked in order to get it to do anything at all.  I could be missing something, but I would avoid this circuit if I were me.

There doesn't seem to be any magic analog oscillator topology out there that works optimally with untapped inductors.  By "optimally" I mean that generates the full tank voltage swing that a high Q inductor is capable of.  A 3.3V supply is capable of +/-1.65V at the drive side, which can ideally produce +/-165V with a Q=100 coil.  Single transistor types rely on phase error to function, and two transistor types have inherently high transport delay, which usually introduces more phase error than single transistor types.  My digital quadrature LC DPLL topology is the only thing I've encountered that approaches full theoretical tank swing.  The EW with low impedance LC driving a high impedance EQ coil is close, but it is non-linear - worse, it has two resonances that you have to worry about, one of which is environmentally-based.

Posted: 1/20/2019 5:04:26 PM

From: Berlin Germany

Joined: 4/27/2016

Thank you, dewster, for estimation this oscillator type. The hint from Mark for making sound outputs in LTspice is super!

Posted: 11/10/2019 4:37:27 PM
Art Harrison

Joined: 3/17/2010

That is an interesting oscillator.

Posted: 11/12/2019 8:18:45 PM

From: Madison, WI

Joined: 5/19/2016

Art have you played around with this 2 transistor oscillator at 12V. I am still using your PNP differential oscillator in lab. Always works great. Gives roughly 6-8Vpp 400-500kHz or (300-350KHz)sine waves at the mixer inputs in our lab. I also added a resistor between the collector and the tank to control gain. I believe you mentioned doing this but decided not to bother.


Posted: 11/23/2019 4:03:28 PM

From: Northern NJ, USA

Joined: 2/17/2012


These two recent threads [LINK] [LINK] have got me wondering about the feasibility of some sort of RC LC oscillator "blend".  Have an RC oscillator drive a series EQ inductor L / antenna C, with that LC somehow "backdriving" the RC (or otherwise providing feedback).  High voltage and stable plus more pullable?  I understand that pullability isn't all that desirable in an analog Theremin, indeed it is often quashed via C padding, but more pullability would be welcome in digital and C sensor type applications.  I dunno, just a thought.  Might be easier to tune with a pot?  I suppose the LC resonance is rather invariant though, and the higher the Q the less you want to drive it off resonance.

You must be logged in to post a reply. Please log in or register for a new account.