oscillator with constant amplitude

J-fet

I've done that. Distortion becomes more of an issue.

Apparently there are (or at least were) some studio-quality mixers that were implemented by shining variable amounts of light onto CdS cells for low-noise, low-distortion variable resistance controlled by a DC voltage.

Sure thing. I built circuits vaguely like that as a kid (audio companders for tape).

Jim Williams went on a long digression about HP200A oscillators and JFETs and such, and came up with something pretty good, if a bit involved.

Cheers

Phil Hobbs

Has many ways to do it, including CDS cells or jfet (.002%)

uses a cap to block DC across the jfet for an improvement in THD they only claim .01% though. With a jfet with interchangeable D-S it does better

Few jfets don't have interchangeable drain and source connections.

Winfield Hill's AOE 3 gives and example as figure 7.22 on page 438. He uses the drain-gate divider trick on the jfet to minimise distortion (which Jim Williams missed in AN-43 at Fig.43).

The National Semiconductor LME49710 might have been a better choice of op amp.

Probably not the best approach if you want to minimise current consumption. E-mail me - at snipped-for-privacy@ieee.org - for an alternative. Tell me your preferred frequency and I might be able to tailor my solution.

perhaps the word symmetrical would have been better

but is in the ap note no the less.

If you don't care too much about distortion the Wein bridge w/ diode limiting can keep the 3rd harmonic down to the 0.1 to 1% level. You can make it a bit better by taking the output (with another opamp as buffer) from the non-inverting input. (You get a little RC filtering "for free".)

(DDS has taken a lot of the fun out of oscillator circuits.)

George H.

Not if you want seriously low distortion - harmonic spurs more than 120dB below the fundamental.

I was going to ask what sort of "science" demands such low distortion. But then I thought of one... and I have data!

Above is a sweep through an optical pumping signal (I'd have to check the numbers.. but something like..) RF frequency = 100 kHz Sweep width (x-axis) is about 4 mG The signal is transmission at different RF amplitudes.

The peak on the right that grows linearly with amplitude is the 2nd harmonic distortion.. The generator (my Rigol) has some power at 2f. The signal on the left hand side that grows as ~RF amplitude squared (Again I'd have to look up the numbers.) is from two photon transitions.

The reason both signals don't sit right on top of each other is that there is a non-linear term in the Zeeman splitting of the ground state given (exactly) by the Breit-Rabi equation.

OK any other science or electronics that needs such low distortion?

George H.

(Oh you may see that the 2-photon feature is also somewhat broader.. that's also from the quadratic term in the Zeeman energy.)

ld

over

ower

er

act

e to

, if

d try

dB below the fundamental.

istortion%20oscillator%20with%20THD%20below%20-140%20dB.pdf

wanted it to test better-than-20-bit A/D converters.

My thought is that anybody who gets serious about sub-micro-degree temperat ure measurement is using an A/C (probably Blumlein) bridge, which can only be balanced at one frequency. Even low-level harmonic content in the bridge drive can swamp the desired signal and saturate the high-gain amplifiers t hat you need to pull out the unbalance signal at the desired frequency.

The short answer is that there's a small - but desperate - market out there for really low distortion sine wave sources.