The 220 ohm resistors were just insurance against op amp instability.
Yes, that is Bob Widlar! You gotta love that photo.
Cheers,
Bob
I want a simple tunable option, like the Tek 505, not very complex, just a dual track potentiometer.
I do not need an Audio Precision style oscillator with fast computer frequency control for production line tests, impressive as they are.
Well, yes that's why I wrote "Presumably".
I need to re-read exactly what he wrote. Do you have the link or a copy, not sure I can find it.
I understand there are effects from the finite bandwidth of the amp. I am not sure if this actually contradicts what GK wrote, maybe just another way to look at it, but perhaps I remember his view incorrectly.
This approximates Bode Maximum Feedback, which I have advocated over in the Solid State forum, so seems reasonable to me.
What is the Thomas reference itself?
Yes, I understand the effect on the relative quadrature outputs, and since I want to try an (enhanced) Sum of Squares leveler I want them equal.
AFAIK Bruce Hofer used gain simply to eliminate an op-amp.
That is not a major concern for my requirements, just wanted to check that there was not some other point I had missed.
I don't need to try that bit in Spice but I do realize that the macro-models are often pretty bad.
There was an excellent discussion by someone of attempts to model op-amps accurately that dissected many model problems, failures by the manufactures to address the problems, denial there was any problem, etc etc.
Wish I could remember where I saw it, sound familiar to anyone?
Best wishes
David
Last edited:
Another half remembered article was a measurement of the decay of an SVO with only the outer loop closed.
The author attributed essentially all the losses to ESR in the capacitors and calculated the putative value.
I was not convinced so I didn't bookmark it, now I'd like to re-examine it if anyone knows the reference.
Best wishes
David
There is one 68 and 180 ohm resistor mentioned on the table.
Are these for the integrator caps?
Dave you can close the K+ loop without an amplifier but you will have to add some damping somewhere and balance it with the K+ feedback. Otherwise the settling will be slow.
I eliminated one op amp by combining part of the multiplier with the HP section amplifier.
But I already had an inverted signal from the Mdac I/V convertor.
I eliminated one op amp by combining part of the multiplier with the HP section amplifier.
But I already had an inverted signal from the Mdac I/V convertor.
There are actually two 68 ohm resistors, the column header is (KM,LN) to indicate both the KM and the LN connections.
These are the ones in the oscillator, their effect on Q-enhancement is mentioned in p.57 second column.
Sorry, I don't follow the reference, what is the K+ ?
Best wishes
David
It been more than 25 years since I read though the article. You seem to have it fresh in your mind.
The k- and K+ are the loop gains. oscillation occurs when the gain is at unity: when
K- * K+ = 1.
Here is an analysis for Wien.
VK-1 Audio Oscillator
I have a copy from Bob's website, just checked that.
OK+, just needed to know your notation, I will think a bit more.
Also about your comment on the losses in the resistors, I have previously pondered this issue in feedback amplifiers.
I rechecked GK website and still can't find the comments about Gain-Bandwidth problems, probably deleted.
Best wishes
David
It is used in the Tek 505, thanks for the reminder.
Now I think of it, the optimal technique should be to use both Bruce's and Bob's tweaks.
Capacitor across the inverter feedback resistor in combination with resistors in series with the integrator capacitors.
My first intuition is that the time constants should match, to keep the loop "flat" (i.e. same level at every op-amp)
One extra capacitor seems a small price, I wonder how far the loop frequency can be pushed with, say, AD797s?
Best wishes
David
Last edited:
Keep in mind that input resistance of the op amp loads the integrator cap.
The LT1468 has a relatively high input resistance which is one of the reasons I choose it for my SVO.
The search is over:
http://www.janascard.cz/PDF/An ultra low distortion oscillator with THD below -140 dB.pdf
Jan
http://www.janascard.cz/PDF/An ultra low distortion oscillator with THD below -140 dB.pdf
Jan
The search is over:
http://www.janascard.cz/PDF/An ultra low distortion oscillator with THD below -140 dB.pdf
Jan
Thanks, Jan. Excellent article and superb performance.
I am especially thankful that the paid attention to, and addressed, oscillator noise, which I think would be a good area of additional focus on this thread.
However, note that it is a single-frequency oscillator with a tight agc trim. This greatly reduces the needed authority of the agc circuit, reducing both its noise and distortion contribution.
BTW, I heard that TI is dropping the LME IC line, or at least a very large part of it. Anyone know if there is any truth to that?
Cheers,
Bob
Yes. Most of the line is dropped with only a few remaining. TI said they would supply for one year and then poof.
I am reasonably sure this circuit was discussed, and dismissed, much earlier in the thread.
When I read the thread recently, in an effort to catch up, I found some trials of what I think was this circuit.
Unfortunately the link to the circuit was to a journal article and now dead, but seemed to be the same circuit as your link.
As Bob says, distortion can be reduced if you reduce control authority, and this circuit achieves it's low distortion with very low control authority.
The consensus of the discussion was that the control authority was sufficiently low as to make the circuit very "tweeky", very sensitive to component tolerances, mismatch etc.
This makes sense, it's a very conventional circuit, to make it perform much better than any other Wien B. will require heavy trade-offs.
Happy to be helpful, it's made me really think about this issue.
Now that I really think about it, I notice that GK's oscillator has the hi-frequency tweek capacitor across the inverter local feedback resistor.
This will produce exactly the opposite effect to what is done in the Tek 505 by Bruce.
Very odd, could this be the reason for GK's problems?
I checked, and both Bob and Bruce's tweeks do indeed have about the same time constant.
They both use the same variety of op-amp, so that reinforces my idea that the inverter and integrator TCs should match.
Also the Renardson SV oscillator has a 505 style capacitor and similar TC but the amps are different so not directly comparable
Best wishes
David
Last edited: