1kHz 130dB distortion sinewave oscillator.

I'd probably ask you what you thought the best way of measuring it would be.

I would too.

I wasn't expecting to get below -120dB in a simulation but the current circuit is approaching -140dB in a simulation. I have definitely found it fun to do and also to improve it using suggestions from people here. And I have learned a lot about how to use LTSpice which makes it much quicker and easier to simulate anything I might get paid for.

From a sinewave oscillator point of view the original objective was to find out whether it's really necessary (at least at one fixed frequency) to use lamps, thermistors or opto devices to get the lowest possible distortion. That question seems to have been answered.

That's the Stoic philosophy, and I'm actually pretty good at following it in real life.

You should. He was taking flying saucer stories seriously, and the evidence has been looked ito for years, and there's nothing in it.

Religion is big on that sort of nonsense. Spinoza pointed out the problem a couple of hundred years ago, and there are still people who take it seriously - including a friend of mine who is a professor of theology.

One of the local fringe religions has had a group of eleven people charged with murder. They were convinced that an eight-year-old type-1 diabetic really didn't need insulin, and she died.

There are some levels of irresponsibility that really can't be tolerated. John Larkin, Cursitor Doom and Donald Trump think that anthroppgenic global warming is a matter of opinion, and that is already killing people, and will probably kill a whole lot more.

Of course it's nice to minimise the parts count, but that's not the only thing you should pay attention to. Most 82-year-olds are unemployable - some of them because they are dead or close to it. I used to build quite a lot of what I simulated, but it would be more of a performance these days.

Not exactly. It certainly didn't include you or Cursitor Doom.

So could pretty much everybody who is interested in the problem. Buy a good audio A/D converter, digitise a couple seconds of output and Fourier transformer the that data to pull out the harmonics.

The fact that you couldn't be bothered to post that explicity is telling.

Simulations have to simulate a simplified version of reality. Sometimes quantisation error can make simulation perform worse than reality - it happened with out ultra-sound scanner back in 1977 - but it isn't common.

You'd probably wouldn't bother if you had never built a circuit.

It's more of a long term investment.

It can be a long term investment.

But it's the wrong question. The real questions are "can you stabilse the output of an oscillator without using a non-linear device", to which the answer is that you can't, and "which non-linear device gives you least distortion in the output" to which the answer is "it depends on exactly what you are trying to do".

This illustrates another advantage of the svo. The three stages each attenuate the harmonics coming from the FET.

I've downloaded John May's .asc code and modified it by pushing R12 up to 39k, which degrades the harmonic content of the output, but the second harmonic in the output is still 148dB below the fundamental while the third harmonic is 143dB down.

Looking at V(n003) - closer to the FET the second harmonic is only 140dB below the fundamental and the third harmonic 124dB down. V(n002) give

Unsurprisingly the effect is more marked for the third harmonic which is further away from the fundamental.

I've swapped in Edward Rawle's LT1678 and the circuit does much the same with more nearly real opp amps, and simulates just as fast.

OPA1655/56