Wien bridge sine oscillator on 9V single supply

[nickcordle]

Hello all, I've cooked up something I think will be useful to me and thought I'd leave it here in case it's useful to anyone else, as I've gleaned a lot from these boards.

I want to measure various distortions so a clean sine is needed, but I don't want my recording interface fried by a mistake, and my 8-bit scope/AWG is the opposite of clean, so I started reading ...

It's a (roughly) 1kHz oscillator, as seen in TAOE, ESP audio pages, LTC AN43, etc.  Main thing I wanted to achieve was to run it on 9V single supply since I have plenty of that lying around, and not so much +/-15V like all the references seem to assume.

Attached spectrum shows the loudest harmonic in the neighborhood of -80dB below fundamental.  I've been able to trim the distortion down really low, but then amplitude bounces around ... seems to be a fundamental tradeoff with stabilization schemes like this.  If anyone has insights to share I'm all ears!

I'm gonna try wider rails, 9V battery/PS into TC1044SCPA for +/-9V, and any ideas for next steps ... what else should I try?  Rail-to-rail op amps maybe?  Not sure what this would do but I have noticed the NE5532s seem to get closer to one rail than the other when left to oscillate wildly.  So there's -some- asymmetrical behavior built in there, but I don't know how significant.

... And I'm gonna try a dual-gang pot to control the frequency and see how it behaves.

Would love to get the thoughts of the vastly experienced folk here!

(I know, I know, get a real AWG ... $$$)

Nick

[Scruffie]

Too tired to look at your current circuit but I will tell you a pair of LM13700 will make a nice clean sine wave (On +/- 9V) see the datasheet.

[Rob Strand]

... And I'm gonna try a dual-gang pot to control the frequency and see how it behaves.

If you make it variable you might find it misbehaves,  possibly motor-boating (or Squegging).  The R10+C3 network often needs to be made into a more elaborate RC network.    If your oscillator has a lot of distortion you can sometimes get away with the simple circuit.

You might want to look at oscillators by Jim Williams, but some of those are also fixed frequency.

One example is on p25 of this magazine.  You will see the complex network around IC5:
http://americanradiohistory.com/AUSTRALIA/Archive-Electronics-Australia/1986/EA-1986-12.pdf

------------

So there's -some- asymmetrical behavior built in there, but I don't know how significant.

Possibly due to the JFET.

[PRR]

-80dB, 0.01%, is astonishingly good performance for a diode detector and a JFET. I think you should be very chuffed.

The next steps (after perhaps JFET linearization) mean moving to much more complicated designs for every small decrement of THD. 3-phase oscillator and 6-phase detection, and on and on.

http://www.nanovolt.ch/ and read everything, follow all references. 

[Rob Strand]

This idea was pretty cool way to get low distortion with diode limiters. 
It originated in Wireless World Feb1982, which is referenced in the article:
page 84
http://americanradiohistory.com/AUSTRALIA/Archive-Electronics-Australia/1989/EA-1989-02.pdf
page 100
http://americanradiohistory.com/AUSTRALIA/Archive-Electronics-Australia/1989/EA-1989-03.pdf

The article was written by Phil Allison.   If you have been around the internet in the last 20 years I'm sure you will @$^*ing know him. 

The idea is explained in the article.  Basically the distortion products propagate down the chain differently to the fundamental and this allows you to cancel out the distortion.

My best effort using diodes was to use a state-variable structure like the nanovolt page.  The output at the second integrator naturally filters the distortion more then you subtract off what is left using the idea in the article.

[amz-fx]

Here is an article about low distortion oscillators, though not single supply:

https://www.edn.com/electronics-news/4389675/Low-distortion-oscillator-tests-measurement-circuits

I have a test oscillator made by the author of that article and it is excellent!

regards, Jack

[nickcordle]

I appreciate the links!  Pretty sure I haven't seen any of this stuff before.  I am a bit new at this I suppose.

I've glanced past state-variable filter designs in a few places but haven't gotten around to figuring out how to do them with a single supply, so maybe that's what I'll try next, in between reading binges.

To be continued ...

[imJonWain]

I don't know how little "minimum distortion" is but I found this link with a really simple single supply schematic.

http://www.techlib.com/electronics/audiooscillators.htm