Low-distortion Audio-range Oscillator

[scott wurcer]

Originally Posted by davada

Do you mean this one?

The Effect of u-Circuit Non-Linearity
on the Amplitude Stability of RC Oscillators

http://www.hpl.hp.com/hpjournal/pdfs...Fs/1960-04.pdf

Yes, the math here is very difficult remember this predates computers. Jim Williams sent me this with the comment that he thought this was the best analysis of a basic circuit concept that he had ever seen.

[dirkwright]

Originally Posted by dirkwright

It's a very nice looking board:
[3rd round] AN67 Ultra-low THD 10kHz sine oscillator PCB Group buy. - Discussion Page 10 - diyAudio

He did a great job. It's designed to fit into a specific chassis. It's certainly the highest performance Wein bridge in the known world.


Dirk

However, builders are having problems with it, as mentioned in the discussions section. This doesn't look like an easy oscillator to make stable. They tried to use it for different frequencies also with not so good results. I imagine anything with 180dB of gain would be an oscillating nightmare.

[richiem]

@dirkwright -- the gain and phase responses were specifically tailored for operation at 10kHz -- doing other frequencies would be a significant engineering project, I think.

[dirkwright]

OK, Glen is getting a group buy together for his generator boards, which are part of his THD analyzer project I've posted on here before.
Who would like an ultra-low distortion audio oscillator PCB? - Page 1
He is in OZ.

[dirkwright]

Originally Posted by richiem

@dirkwright -- the gain and phase responses were specifically tailored for operation at 10kHz -- doing other frequencies would be a significant engineering project, I think.

Yeah, you are right. I don't know what they did with it to try to make it produce different frequencies. There's not enough detail in the thread that I can see.

[dirkwright]

"The notch filter & distortion amplifier PCB is the board in the middle. The board to the left is the state variable oscillator board and the "dead bug" rats nest to the right is the prototype auto-tune board."

From Glen's website. The generator board looks nice and compact, perfect for a rack mount box.

[davada]

Originally Posted by scott wurcer

Yes, the math here is very difficult remember this predates computers. Jim Williams sent me this with the comment that he thought this was the best analysis of a basic circuit concept that he had ever seen.

HI Scott,

Dick Moore and I were discussing this article recently. Both Dick and myself have worked with lamp stabilized oscillators using op amps. I have done it successfully with a state variable oscillator with remarkable results. Distortion levels as good or better than any other SVO.

Although Dick has witnessed the effect in an original Hp oscillator, neither one us have observed the effect with amplifiers having near pure spectral power.

Why is this?

Cheers,

[scott wurcer]

Originally Posted by davada

HI Scott,

Dick Moore and I were discussing this article recently. Both Dick and myself have worked with lamp stabilized oscillators using op amps. I have done it successfully with a state variable oscillator with remarkable results. Distortion levels as good or better than any other SVO.

Although Dick has witnessed the effect in an original Hp oscillator, neither one us have observed the effect with amplifiers having near pure spectral power.

Why is this?

Cheers,

The lamp has two non-linearities, the thermal one and a voltage coefficient of resistance. Measure the lamp's distortion at 10kHz by comparing it to a resistor. The voltage coefficient of non-linearity is what actually stabalizes the oscillation with the thermal one setting the amplitude.

Try a sim with a perfect amplifier, Wein network, and a very small third order non-linearity in the amplifier. Kick it with a current pulse and the oscillation builds up to an amplitude determined by the non-linearity alone. The oscillator has a complex pair of poles that "walk" back and forth on the axis according to the instantaneous gain (gain of exactly 3 is on the axis) . The instantaneous gain is perturbed by the small nonlinearity and the amplitude is stable when the integrated gain over one cycle is exactly 3. Now keep decreasing the non-linearity and the amplitude keeps rising. A low frequency amplitude control is also needed.

I have tried to make a Wein bridge oscillator with no nonlinearity (<-140dB) in the circuit, indeed it never stabalizes except for a few seconds at a time.

[Samuel Groner]

Interesting--do you knwo an explanation why the lamp has significant voltage coefficient? So far I've suspected that this is mostly a well behaved metal wire.

Samuel

[KSTR]

Hi,

I had a chance to hook Viktors' 1kHz Oscillator on the AP and compare it against AP's own generator (direkt GENMON routing) and the Tek SG505. Test level was 2.7Vrms (Full scale on Victors')

No shielding of the Osc, and fed by lab supply.

The results speak for themselves, or in one word : INCREDIBLE!

Only with massive synchronuous averaging (4096 runs, checked for no/little loss of harmonics magnitudes) it was possible to see the distortion of this oscillator and it fares better than AP's by 10dB...15dB. The SG505 (with only 1024 blocks averaged) is far behind and plaged by hum which also seemed to disturb the autotracking of the notch filter of the AP.