Doesnt that depend on your assumptions vs my goals?
----- ultra low distortion sources.... LPF and Notch filters?
To get lower distortion sources..... one way I suggested 4-5 years ago was a LPF and also a notch filter to eliminate the 2H and/or 3H. I have tried both ways. IMO-- That requires an auto-tune notch IMO. Ditto, for the Cordell Distortion magnifier for measuring.
If you followed the 339 mods, you saw me use a passive variable freq notch filter, modified with active Q enhancement for that purpose. but it was manually tuned also.
Remember this? In my mind, its all related... notch filters. manual vs auto-tune.... whether to improve measurements or to improve the source'
2016OK carry on.... what were you saying?
-Richard
Last edited:
Well, yes if you add cascoding... My point was you dont need cascoding if input source Z is very low in the parallel config. Their C's cancel.
THx-RNMarsh
Last edited:
Off top and rule breaking posts have been removed. You know better, and know what the consequences will be if you continue.
OK Chris,
Some here think i am high-jacking but maybe if they go back, i was very involved here. I was the one who found Victors gen deep on a back page of ebay and bought it and measured it and told DIYAudio. I was behind David doing a variable freq version...
I see no progress since then and that was YEARS ago. I see no progress here in other new gen designs. What are we doing with notch filters, anyway? That should be a different forum?
I am sorry to seem pushy but it has been YEARs you know. So, I went back to david's idea on how to improve his design and suggest we get back on track with his idea.
"The limiting factor in my SVO appears to be from loading on the op amps. Particularly the second integrator because it sees loading from both the tuning network, multiplier input and controller input. I was able to improve this some by placing a Jfet op amp as buffer between the leveling controller and oscillator stages. A composite including an high current output buffer I think would solve the problem and the oscillator could be operated at a much higher output. This could eliminate the need for an output buffer/amplifier. A buffered op amp would handle smaller resistors and larger caps in the integrators which would bring the noise some. An extended amplifier bandwidth would improve the loop stability because the Q enhancement effect would be less.
Can we make a composite amplifier with a high current buffer stage happy with functioning as an integrator?
CFAs can't be used as integrators but could they work in composite design? If so then the THS6012 might be good candidate."
I would like to see this gang of merry men go in this direction. Either to Davids design, or Victors or a new design with such composite amp as basis for generator/osc.
Test and measurement of it might be a separate subject of design? Its too bad that I had to disagree with SW and his reaction about the need for auto-null..... But, if you are going to measure such small levels, with high Q, auto-null is necessary.
THx-RNMarsh
Some here think i am high-jacking but maybe if they go back, i was very involved here. I was the one who found Victors gen deep on a back page of ebay and bought it and measured it and told DIYAudio. I was behind David doing a variable freq version...
I see no progress since then and that was YEARS ago. I see no progress here in other new gen designs. What are we doing with notch filters, anyway? That should be a different forum?
I am sorry to seem pushy but it has been YEARs you know. So, I went back to david's idea on how to improve his design and suggest we get back on track with his idea.
"The limiting factor in my SVO appears to be from loading on the op amps. Particularly the second integrator because it sees loading from both the tuning network, multiplier input and controller input. I was able to improve this some by placing a Jfet op amp as buffer between the leveling controller and oscillator stages. A composite including an high current output buffer I think would solve the problem and the oscillator could be operated at a much higher output. This could eliminate the need for an output buffer/amplifier. A buffered op amp would handle smaller resistors and larger caps in the integrators which would bring the noise some. An extended amplifier bandwidth would improve the loop stability because the Q enhancement effect would be less.
Can we make a composite amplifier with a high current buffer stage happy with functioning as an integrator?
CFAs can't be used as integrators but could they work in composite design? If so then the THS6012 might be good candidate."
I would like to see this gang of merry men go in this direction. Either to Davids design, or Victors or a new design with such composite amp as basis for generator/osc.
Test and measurement of it might be a separate subject of design? Its too bad that I had to disagree with SW and his reaction about the need for auto-null..... But, if you are going to measure such small levels, with high Q, auto-null is necessary.
THx-RNMarsh
Last edited:
A driven cascode on a unipolar input stage is ultimately simpler and does more for you. Parallel approaches depend on having the same capacitance nonlinearity and Early effect in both sexes of the paralleled input pairs to achieve the advertised distortion cancellation.
Cheers,
Bob
yes, it certainly does depend on the C matching on how much C is cancelled. And, it isnt getting any easier to find good compliments. But they do exist.
After finding good compliments, Here is what i do to get a more complete cancellation.... It usually isnt necessary for low distortion but to get the very very lowest... like down below AP2722..... I can trim the load R of each to affect the C or also change the PS voltage to one to affect the C. Or similar trims. However, in my case I start with FETs with very low C to begin with on the input. Cancelling leaves so little remainder. [2N5457-60]
Low PSRR can get handled in other ways.
THx-RNMarsh
Last edited:
Its just a different mind-set. Fewest possible devices. low complexity, very stable, quiet, wide BW and low feedback yet very low distortion. Does it matter - sound wise? Under some conditions, yes.
Maybe better discussed else where, though.
THx-RNMarsh
Last edited:
Hi Richard,
Your idea has merit with regard to improving an oscillator design with a composite op amp where there is higher loading. This is beyond my knowledge, but I would very much like to see David's, Victor's or a newer design. For personal reasons, I would very much like to see David's design reproduced and / or improved and produced. Either as boards and kits, a functioning unit with case would be great for those members who don't have the time for a kit or build from scratch. I think its more important that people have better equipment to use than forcing a build from files only to achieve the same end. It is simply beyond most of us to purchase state of the art type equipment. So this represents a chance for many of us to have improved equipment to work with and therefore learn more, or do a better job at design.
-Chris
Your idea has merit with regard to improving an oscillator design with a composite op amp where there is higher loading. This is beyond my knowledge, but I would very much like to see David's, Victor's or a newer design. For personal reasons, I would very much like to see David's design reproduced and / or improved and produced. Either as boards and kits, a functioning unit with case would be great for those members who don't have the time for a kit or build from scratch. I think its more important that people have better equipment to use than forcing a build from files only to achieve the same end. It is simply beyond most of us to purchase state of the art type equipment. So this represents a chance for many of us to have improved equipment to work with and therefore learn more, or do a better job at design.
-Chris
Hi Chris,
Great. Good managers leave the talented ones alone to do their thing. But, I just saw this going no-where as a better osc/gen forum. Get us back on track.
Here is a missing piece of David's design for you to keep.
View attachment dlbSVO V5 SCH Sheet 1 (1).pdf
THx-Richard
Do you recall what level of distortion David's oscillator was able to achieve at 1 kHz and 20 kHz?
Cheers,
Bob
Hi Rick,
My hands are more than full, and I don't have the experience necessary to successfully get this off the ground. David's widow would also like to see this running as it is a good design.
There is a working prototype and evidence of excellent performance. It has variable frequency and level from what I recall from talking with David and testing his software interface for it. So it isn't a fixed frequency, fixed level oscillator like the excellent one Victor has made (and I own).
-Chris
My hands are more than full, and I don't have the experience necessary to successfully get this off the ground. David's widow would also like to see this running as it is a good design.
There is a working prototype and evidence of excellent performance. It has variable frequency and level from what I recall from talking with David and testing his software interface for it. So it isn't a fixed frequency, fixed level oscillator like the excellent one Victor has made (and I own).
-Chris
Let's stop this bickering and everyone stay out of the sin bin. This is a really good thread and let's try to stay on track, with a reasonable amount of lattitude around the topic.
It is sad that we have lost some tremendously talented people over the years because of this sort of thing. I would never like to see anyone leave in disgust.
Bob
It is sad that we have lost some tremendously talented people over the years because of this sort of thing. I would never like to see anyone leave in disgust.
Bob