Re: Low distortion sinewave oscillator without big capacitor.

On 1/04/2025 2:09 pm, Edward Rawde wrote:

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vsdi3h$3nagd$2@dont-email.me...

On 31/03/2025 5:54 am, Edward Rawde wrote:

Not long ago JM posted a 1KHz sinewave oscillator with very low distortion.
It used a 470uF non polarized capacitor which in practice would probably be made from two 1000uF capacitors.
There's nothing wrong with that but I wanted to see whether I could make a working circuit without needing such a large
capacitor.
>
What I have so far is below.
>
Any comments?

>
It relies on the Analog Devices MAT-02 dual transistor, which is now obsolete
>
https://www.analog.com/media/en/technical-documentation/obsolete-data-sheets/mat02.pdf
>
The .asc file shows eight NPN transistors labelled MAT-02, presumably in four pairs of the part, but it isn't clear which of the
eight transistors should be paired up.

 BCM61B is available and very reasonably priced.
https://www.digikey.ca/en/products/detail/nexperia-usa-inc/BCM61B-215/2119400
It can be used for Q1 Q2 and Q3 Q4 in the circuit below.
 The remaining four transistors can use MAT14
https://www.digikey.be/en/products/detail/analog-devices-inc/MAT14ARZ-R7/2510588
which although a bit pricey has four independent matched transistors.
I can't find an LTSpice model for MAT14 so MAT02 is still shown in the circuit below.
 

>

What's the best way to control the output level?
Currently it's 5v pk-pk but I rather have half that.

>
The rectified currents from the four phased shifted versions of of the output waveform flow through R13, R14, R15 and R16 into R11
and through it into the virtual earth set up at the inverting input of U3, where the summed current is compared with a fixed
current drawn from the +15V rail through D1 and R10.
>
Doubling R10 from 330k to 680k would roughly halve the output amplitude.
One could be more precise, but it wouudl be hard to justify the extra effort.

 Ah yes, that takes care of the output level.

Not all that well. Using the positive rail as your voltage reference suck, and including the diode drops of the rectifying diodes is even worse.There are precision rectifiers that use op amps to take out the diode drop, and synchromous rectifiers built around transmission gates can be even more precise.

The revised circuit is below.
Line 459 will need to be unwrapped.

I did that, and the circuit does work, after a fashion.
It still uses eight transistors to do what John May did with three separate discrete transistors.
Because he didn't use a matched pair, he had to use two 250R emitter resistors to get the operating conditions he needed to make them act as parts of a three-transistor asymmetric Wilson current mirror.
https://en.wikipedia.org/wiki/Wilson_current_mirror
He then had to by-pass the resistors with a big capacitor to get the effect he needed. This introduced a phase shift, but John May's phase-shift oscillator offers four different phases from which he could pick off the right phase to get the correction signal he needed.
By summing current from two adjacent phases you can get exactly the phase shift you need, but he didn't to be all that exact.
Using a matched pair for the two transistors at the bottom of the mirror you can get rid of one of the emitter resistors and make the other one small enough not to matter (so you don't need the capacitor), but you do need to pick off a different phase to get the right feedback.
The MAT04 is totally unnecessary.
Or it least that's the way it strikes me. I've yet to get a simulation to work to illustrate the point - it's a complicated circuit, and once it hits saturation the subtle effects that stabilise it get swamped.
I need a better grasp of what's going on in the circuit, and a way to start it up that doesn't let it slide over into saturation before it stabilises.
--
Bill Sloman, Sydney