Yes I see this. It the relative improvement I'm looking for not so much accuracy of measurement. Even phase noise translates to amplitude change with a sampled servo agc.
The sample pedestal has to be in the exact position of the peak each sample.
There is a lot going on that can effect the amplitude. Sorting it all out is tricky.
If the servo is working right I should be able to blow on the lamps without seeing much disturbance.
It has come up several times in this thread already, you might want to look this up. IIRC there was also someone selling PCBs.
Of course ppb distortion range (I presume this is American counting, so 1 ppb = -180 dB) is a complete nonsense figure and simply reflects the ignorance of other distortion mechanisms.
Samuel
I think that number was only a claim that the op-amps don't contribute to THD. There was after all a distortion trim.
It appears twice in the text--the first time yes, just the opamps, but the second time is clearly related to the overall performance:
"The measurement of the harmonic distortion of this oscillator defies all of our resources, but appears to be well into the parts-per-billion range."
One might argue where the ppb range starts--if it is below -120 dB then I can agree.
Samuel
"The measurement of the harmonic distortion of this oscillator defies all of our resources, but appears to be well into the parts-per-billion range."
One might argue where the ppb range starts--if it is below -120 dB then I can agree.
Samuel
Hi samuel;
To argue the claims doesnt matter and is fruitless. The fact is the thd from the osc topology is extreamly low. Ive seen other dual opamp combo's before and think there is merit. Without a brand new approach (distortion cancellation, etc) this is the only practical way to get thd down -- esp at higher freqs. Let's discusss the merits and any weaknesses for a moment. A brief summery of what others have experienced with it/similar designs would also be helpful.
Thx-RNMarsh
To argue the claims doesnt matter and is fruitless. The fact is the thd from the osc topology is extreamly low. Ive seen other dual opamp combo's before and think there is merit. Without a brand new approach (distortion cancellation, etc) this is the only practical way to get thd down -- esp at higher freqs. Let's discusss the merits and any weaknesses for a moment. A brief summery of what others have experienced with it/similar designs would also be helpful.
Thx-RNMarsh
Last edited:
Of course, as I've noted earlier it is a very wise consideration for a very low distortion oscillator. You might want to check J. G. Graeme, Amplifier Applications of Op Amps. McGraw-Hill, 1999, which has an entire chapter on composite opamps.
Also I can share a special composite opamp based on two AD797, and using the distortion cancellation pin of one to sum the outputs:
I don't recall the exact distortion figures I've measured, but it surely was much better than any available single IC opamp in inverting mode, with reasonably low source Z (the e-mail I've sent to Scott about this some time back just says "Resulting distortion performance is stellar."). I'm not saying that's my final word on the topic, but it should give you something to start with.
It doesn't look that good actually--the composite opamp has about 6 nV/rtHz voltage noise density, and the operating level is relatively low.
Samuel
When I spoke with Jim Williams he laughed a little when I mentioned this. They AFAIK never actually went that low and IMHO the stability and startup problems would increase to the unbearable level approaching 10 ppb.
What would you do to make a dual opamp for lower thd but not be a royal pain with stability or startup issues?? Got a circuit we can try along these lines?
Thx-RNMarsh
Of course, as I've noted earlier it is a very wise consideration for a very low distortion oscillator. You might want to check J. G. Graeme, Amplifier Applications of Op Amps. McGraw-Hill, 1999, which has an entire chapter on composite opamps.
Also I can share a special composite opamp based on two AD797, and using the distortion cancellation pin of one to sum the outputs:
I don't recall the exact distortion figures I've measured, but it surely was much better than any available single IC opamp in inverting mode, with reasonably low source Z (the e-mail I've sent to Scott about this some time back just says "Resulting distortion performance is stellar."). I'm not saying that's my final word on the topic, but it should give you something to start with.
It doesn't look that good actually--the composite opamp has about 6 nV/rtHz voltage noise density, and the operating level is relatively low.
Samuel
I can see that with 1k Ohm et al it wouldnt be the lowest noise but better than most of the TT circuits from commercial sources... using high value R's and low value C's.
What is the approx noise of this dual opamp design you have put up here?
I have some older Burr-Brown books (1973) on Opamps which Graeme authored and coauthored to look at today. Thanks for the reminder.
Thx-RNMarsh
Last edited:
If you read through the trials and challenges of those who tried to make the 10KHz oscillator work its very challenging. Its a special case where the phase response of the "composite" amp was tuned to support the requirements at 10 KHz. With 180 dB of forward open loop gain its really tweaky to get running and very narrow band. They fiddled with the compensation to get usable phase response at the target frequency. I would love to try one with the CLT-1 to see how low the distortion really is. But not enough to build one.
Duplicating the filter I posted earlier could get to the same place with passive components. The inductors would be a challenge but not impossible.
Duplicating the filter I posted earlier could get to the same place with passive components. The inductors would be a challenge but not impossible.
My impression is that people just copy a lot of stuff.
The idea though is to increase the gain... but does it have to be THAT much. It would help just to get 20-30dB more gain (esp at 10KHz). And, that would not be so skidish to deal with.
Combinations of higher gain combined and/or with cancellation is the best approach so far as I can see. Baring something totally new and original.... which could also happen.
THx-RNMarsh
The idea though is to increase the gain... but does it have to be THAT much. It would help just to get 20-30dB more gain (esp at 10KHz). And, that would not be so skidish to deal with.
Combinations of higher gain combined and/or with cancellation is the best approach so far as I can see. Baring something totally new and original.... which could also happen.
THx-RNMarsh
Last edited:
Too bad that they seem not to have realized that o/l gain at the harmonics counts, not at the fundamental... As far as I understand from the comments in this forum, the difficulty is related to the authority of the multiplier being very low. This design would surely benefit from a better multiplier, so one could bring some authority back. But of course they had to use LT parts.
Samuel
Just adding a buffer on the output will increase the open loop gain as long as the buffer doesn't compromise phase margin. Same for adding an input stage. However dealing with common mode issues on the input can undo all of your efforts with an added input stage. Its all tradeoffs.
No. The open loop gain must be 1.000000000 at the oscillation frequency
with phase 0° or the oscillation would either die or grow to infinity, or the
frequency would run away until the phase condition is met.
There might be an inner loop, of course.
Any return to higher loop gain with rising frequency would invite
additional wild oscillation frequencies.
Gerhard
Last edited: