Set units for LTspice signals

Sure. I could have a computer monitor the output level and control a motor to adjust R1.

Of course Headmaster.

Congratulations. I've put stawberry jam on chicken and found it worked well.

Not that you can detail how.

DACs aren't that good, which is the kind of detail you ought to know. Or at least DACs anybody here could afford to use.

The point about the low distortion sine wave oscillators is that they are cheap. I've got a version built around a pair of AD734 four-quadrant multipliers, but they cost almost $60 each, so I'm not offering it as an option.

A twin-T notch filters can block a pure sine wave very effectively (though you do have to keep re-tuning it) to the point where you can amplify the residual signal to the point where the first four harmonics are detectable.

You can lock a sixty time higher frequency VC0 to the pure sine wave you want to look at - a 4046 will do it - and generate the in-phase and quadrature square waves that you'd need to detect the fundamental and the first four harmonics.

The twin-T filter will mess up the amplitude and phase of what you measure, but to a stable and predictable, and thus corrigible, degree.

Perhaps. It doesn't sound like a cheap option.

As if you had the capacity to learn.

I've never done it with a real circuit, but it ought to work. The parts aren't that expensive.

I may have missed the point here but why are you trying to generate and measure infra-low distortion sine waves?

Any harmonics120dB or more below the signal will be lost in the noise floor and can only be recovered by narrow-bandwidth techniques, so which practical applications require distortion levels below those obtainable from common existing oscillators?

When I tried it, the chicken broke its neck trying to peck the jam off - after that it didn't work at all.

None that I know of. It's a fairly silly game.

Jim Williams put a low distortion sine wave oscillator into one of his application notes, claiming to have got a harmonic content more that

AD5791 is a nice fast 1 PPM DAC. It can make a pretty good sine wave.

Add a 16-bit DAC off to the side to tweak out the residual distortion. IF you can measure it somehow. One could do the same trick using cheaper DACs.

That's the issue: how to measure sub-PPM distortion. If you can measure it, you can easily tweak out the harmonics.

Just the passive parts will have PPMs of distortion.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

And it costs about $100. That's an extravagant solution

Pull the other leg. You have to map the distortion before you can correct it, and you are skeptical about measuring the distortion in the first place.

Do show us how.

You measure it by blocking out the sine wave with a notch filter and look at what's left over. Twin-T filters make pretty good notches, but you do have to tune them carefully and keep tuning them as the temperature drifts. I suppose you could automate that but it would take work.

But it won't matter in this application. You want to pull out just the harmonics, and if the square waves repeat at the right intervals a little uncertainty on when the switching occurs won't introduce cumulative errors.

Really? Great excuse for not trying.

Sum a couple of 16-bit DACs, coarse and fine. Make a sine wave with the big DAC and fine-tune out the harmonics with the fine one.

Easy. The problem becomes measuring the harmonics.

You just admitted that "It's a fairly silly game."

But I'm designing a PPM-class programmable voltage DC source, so things like resistor voltage coefficients matter.

Not so much cap voltage coefficients, for DC. Dielectric absorption might matter.

DAC1220 is a real 20-bit DAC, only $11, but it's too slow for audio. There are some audio delta-sigma DACs that could do low distortion.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

As I'd mentioned in the paragraphs below, which you didn't bother to read before re-iterating your half-baked solution.

They mattered to me back in the 1980's when we were polishing up the Cambridge Instruments electron beam microfabricator, which included a two stage 18-bit DAC - where the lower 12-bits were 10MHz fast and the top six bits had a 1msec settling time. I bought some very tightly specified resistors from Vishay at the time, and heat pipe assembly to cool them (which probably hadn't been pumped out hard enough before it got back-filled with water vapour and turned out not to help).

Dielectric absorbtion won't matter for single frequency AC, though it can be right pain in lower frequency applications.

But you don't know how.

Now that blueworldhosting has recovered after a three day outage I can reply.

You are missing a vital part needed for system stability. Although it's not shown with a chicken, the necessary part is shown here:

Somebody likes a bit of ruff.