Last night I did a rough calculation of the importance of the harmonic series using:
(N-1)!/2 as the operating variable.
I estimate that a ratio of 50 dB between 2nd and 7th harmonic distortion is about right.
Now, please look back at the LAMM.
My biggest problem, right now is reducing 7th harmonic distortion in my ST analyzer using AD797's and an AD744 in key positions. Any suggestions, Scott?
(N-1)!/2 as the operating variable.
I estimate that a ratio of 50 dB between 2nd and 7th harmonic distortion is about right.
Now, please look back at the LAMM.
My biggest problem, right now is reducing 7th harmonic distortion in my ST analyzer using AD797's and an AD744 in key positions. Any suggestions, Scott?
Hi Brian & Dimitri,
Also the VAS emitter resistor is omitted/ignored. According to E.M. Cherry essential for HF stability (JEAS, Vol. 30, May 1982, p.290)
Also the VAS emitter resistor is omitted/ignored. According to E.M. Cherry essential for HF stability (JEAS, Vol. 30, May 1982, p.290)
ratio
Hi John,
Why this ratio? I would think if the 7th harmonic is low enough, say -120dB, this ratio doesn't matter.
Cheers,
Edmond.
john curl said:
Hi John,
Why this ratio? I would think if the 7th harmonic is low enough, say -120dB, this ratio doesn't matter.
Cheers,
Edmond.
john curl said:
Could you enlighten us as to the origin of this new paradigm? So increasing seconds to -40dB so the 7ths can be -90dB is the goal. Your fully balanced topologies (that inherently cancel 2nds) are at a real disadvantage then?
I recommend a passive notch filter.
The ratio that I am measuring with my ST analyzer has perhaps a 25dB separation between 2'nd and 7th, at best.
The absolute magnitude may not be as important as usually presumed. Perhaps ratios are more important, because IF you use global negative feedback to reduce the measured amplitude distortion, you must be making a proportional amount of PM distortion, that may be even more audible in small quantities than AM distortion.
The absolute magnitude may not be as important as usually presumed. Perhaps ratios are more important, because IF you use global negative feedback to reduce the measured amplitude distortion, you must be making a proportional amount of PM distortion, that may be even more audible in small quantities than AM distortion.
Notch filter where, Scott? I already use a 6dB/octave rolloff on the oscillator and get its 7th harmonic out of the way.
Bob Stuart, of Meridian, once wrote a paper in 1973, for 'Wireless World' that denotes various multipliers for higher order distortion. One would have to get his paper to address his earlier references. Works for me.
OT
Sorry to be off topic, but I just discovered a nice toy:
http://websdr.ewi.utwente.nl:8901/
Have fun.
Sorry to be off topic, but I just discovered a nice toy:
http://websdr.ewi.utwente.nl:8901/
Have fun.
A 'classical' approach for eliminating e.g. crossover distortions in OPs was putting a single ended load (resistor or better a current source) from output to a supply rail (driving the output stage in class A). But I'm not sure if this is the source of the K7 you have and certainly K2 would rise significantly by doing so(!!), but the magic ratio would be maintained 😉
I think Bob Carver did such things with his pre amps a long time ago and OPs weren't as good as they are today ...
At which level is your K7, certainly less than 100nV ?!
Intelligent response, Jon. I designed this very thing into my new phono stage using an AD797. Now, to get a response from Scott about this. Have been trying for months.
Thank you John 🙂
I just tried to get some distortion out of the AD744 and AD797 PSPICE models, but they don't do K7 at all (certainly marketing inspired models).
The AD744 does some K2 and K3 and the AD797 only K2 and the rest is 180 dB (!!??) below 10V, 20kHz sine output and a gain of 40 dB.
As already said, these models seem to be only rough models.
Scott would you be able to provide more accurate simulation models ?
I just tried to get some distortion out of the AD744 and AD797 PSPICE models, but they don't do K7 at all (certainly marketing inspired models).
The AD744 does some K2 and K3 and the AD797 only K2 and the rest is 180 dB (!!??) below 10V, 20kHz sine output and a gain of 40 dB.
As already said, these models seem to be only rough models.
Scott would you be able to provide more accurate simulation models ?
Do you mean to tell me that SPICE as used by most of us, is not very accurate, when the going gets rough, or exact? OH NO! 
Jon Lord said:
Hi Jon,
Indeed, a classical approach. I did just that with a uA709, more than 40 years ago. It worked fine by the standards of that time. But for a power amp, like D. Self's class-XD for example, it's a terrible kludge (think of reactive loads!).
Cheers,
Edmond.
What reactive load on an op amp used to amplify signal proportional to it feedback resistor ratio?
Spice
Spice can be quite exact, even for calculating the distortion. Regrettably, most op-amp models are over simplified. If modeled at the device level the results may be within say +/- 20%.
john curl said:Do you mean to tell me that SPICE as used by most of us, is not very accurate, when the going gets rough, or exact? OH NO!
Spice can be quite exact, even for calculating the distortion. Regrettably, most op-amp models are over simplified. If modeled at the device level the results may be within say +/- 20%.
john curl said:Do you mean to tell me that SPICE as used by most of us, is not very accurate, when the going gets rough, or exact? OH NO!
Oh, a wiseguy, huh?
But seriously, if the op-amp model is not at the device level, one cannot get reasonable distortion simulations from it, nor should one expect to. OTOH, the device models for BJTs take into account many different types of errors that would be intractable to deal with using hand analysis. There's a great book by Massobrio and Antognetti that goes into lots of detail about the models and the equations behind them. This isn't a SPICE book per se, but rather goes into depth regarding the device physics and showing the relationship of the physics to the voltage-current and charge equations. Things like variation of beta and fT with collector current are covered in detail. Believe it or not, I think this book would be right up your alley - seriously.
If one has a device-level model of the op-amp (not just a bunch of ideal controlled sources as the macromodels use), and the device models are as good as what Scott has available (big if), then I'd expect pretty accurate distortion simulations.
JFET and vertical MOSFET models typically suck though.
- Status
- Not open for further replies.