Re: How 3-capacitor sine generator works really?

On 3/05/2024 4:14 pm, RodionGork wrote:

Hi Friends!
 Schematic / simulation in "Falstad online simulator":
https://tinyurl.com/23hcg8np
 This is probably very old and widely known schematic of single-transistor
generator which requires no inductance, but instead uses three capacitors - actually it seems to be a chain
of high-pass single-stage filters with transistor serving as feedback from output to input.
 One can find it, for example, in classic stylophone schematic (the part creating low-frequency
oscillations for "vibratto" effect).
 I teach it to my pupils for years probably and I always thought I less or
more understood what is happening inside - each filter stage gives shift in phase and hence when
amplifying feedback is added you get harmonic oscillations.
 However on the schematic given above I added 4 scopes over the length of
the filter (potentials
at the points A, B, C, D according to labels - here A and B are potentials
at points between capacitors, C is at the base and D at collector) - I
suddenly found that intermediate voltages are
not pretty harmonic! They could be distorted by the current drawn into
transistor base though. And
I'm not sure the output is exactly sine now. Though probably it is a matter
of adding some resistor  to improve input impedance of transistor cascade?
 Regretfully I can't find any thorough explanation of the schematic
(probably due to keywords being too general and I don't know if this design
has fancy proper name). So I would be grateful either
for links or for verbal clarifications.

It's a phase shift oscillator - one of many.
https://en.wikipedia.org/wiki/Phase-shift_oscillator
The amplitude is limited by clipping in the single transistor amplifier. If you model it with LTSpice, you can get the program to produce a Fourier transform of the output waveform and it is going to have all the harmonics out to the cut-off frequency of the transistor.
You can do better, but it takes more components.
Here's a solution I came up with back in 1986, developed as a retrofit to excite a linear variable differential transformer used to measure the progressively increasing mass of a single crystal of gallium arsenide (GaAs) being grown in the Metals Research GaAs Liquid-Encapsulated Czochralski (LEC) crystal puller. The circuit it replaced had been developed a decade earlier and used components that had become obsolete in 1986. The new circuit replaced it in new machines and was retrofitted to some older machines.
Only about 50% of the power fed into the oscillator ends up in the load, rather than the better than 90% transfer you can get with a classic Class-D oscillator – but it’s quite a lot more efficient than any of the low distortion oscillators I know about, and it lends itself to very precise control of the output amplitude. I’ve generated quite a few Spice models of various implementations of the idea, but I’ve yet to get around to building a current version of the real circuit – the 1986 version worked fine, but at that time I wasn’t aware how good the circuit could be and didn’t have any reason to check out its performance in detail.
Here's a proof-of-principle simulation - which doesn't have anything in common with the 1986 circuit.
http://sophia-electronica.com/BillsBaxandall.html
--
Bill Sloman, Sydney