DAC1 did, yes. They claimed it was transparent at the time.
Looking at the DAC1 reminds me of how long ago it was...
Too bad there is no successor to AD1955 (DAC1 used '1853, though) with some minor updates and a 3.3V VDD supply.
Last edited:
+1.
It is easy to record any analog source we want, play-it back and compare them, listening if any major difference. With my AK4490, i cannot notice any obvious artifact.
While I can with any analog gear difference, including op-amps.
Not to talk about tape recorders or vinyl copy of my masters: I can design the copy (blind) with 100% of accuracy.
This quest for ultimate performances are unrealistic.
Personally, for home use, I consider anything that is under -90dB and 0.005% of distortion as technically satisfying and do not waste our listening pleasure. No one of our speakers does not meet these specifications. That is where i believe we have to concentrate our efforts. And, of course, our listening rooms acoustic.
That said, there are still some mysteries to clarify for me. Why can I hear micro-dynamic differences between analog stages whose measurements are much better than that ?
My solution is to 'overkill' power supplies and current capacity of my output stages.
Last edited:
Perhaps you'd be kind enough to help me understand this. Baxandall / Putzeys F-word seems (to my small understanding) to mean that Black 1928 is an approximation. Black is a linear equation that can't generate new terms, so can only be correct if the "second pass" through the amplifier is perfectly linear. Baxandall / Putzeys shows that new terms appear, so, to my limited understanding, Black must be somehow flawed.
To my naive view, Black seems to work as if distortions occur on the "first pass through" but not on the second and subsequent "passes through". That whole thought path is obviously wrong, but the contradictions between Black and "F-word" cause trouble for slower moving creatures like me.
If there really isn't a contradiction, or if my math skills are simply too primitive, I'd really appreciate any pointers you, or anyone else, could give me.
Always grateful, not yet dead,
Chris
Well, AK4490 sort of has op-amps internal (SCF). Not disagreeing with what you said though.
We agree completely. 😕
I wasn't aware that Black treated a non-linear transfer function in detail at all, in that sense one might say "approximation". I would call that a somewhat poor choice of words.
The problem has a solution, http://www.its.caltech.edu/~musiclab/feedback-paper-acrobat.pdf
Unless there is a significant genuine time delay in the signal going through the amplifier then feedback is not recursive. Such a time delay is extremely unlikely for an audio amplifier, although it can be a problem for people designing linear RF systems.
The new terms (re-entrant distortion) do not appear because the signal has gone round the loop again. They appear as an unavoidable matter of algebra and so can appear even in the ideal case of instantaneous signal propagation. The feedback is trying to adjust the input seen by the amplifier (the 'error' signal in servo terms) so that the output follows the signal input; anything in the output which is not in the input is attenuated. Essentially the feedback is doing an inversion.
To give a simple example, assume the amp output is 1+x - where x is something not in the input. To correct this the feedback has to generate 1/(1+x).
1/(1+x) = 1 -x +x^2 -x^3 +x^4 etc.
Can you see how simple algebra generates re-entrant distortion even when everything happens simultaneously?
The new terms (re-entrant distortion) do not appear because the signal has gone round the loop again. They appear as an unavoidable matter of algebra and so can appear even in the ideal case of instantaneous signal propagation. The feedback is trying to adjust the input seen by the amplifier (the 'error' signal in servo terms) so that the output follows the signal input; anything in the output which is not in the input is attenuated. Essentially the feedback is doing an inversion.
To give a simple example, assume the amp output is 1+x - where x is something not in the input. To correct this the feedback has to generate 1/(1+x).
1/(1+x) = 1 -x +x^2 -x^3 +x^4 etc.
Can you see how simple algebra generates re-entrant distortion even when everything happens simultaneously?
I read the three Summaries and the Discussion.
Scott, I think you just gave Mr. Curl a relative fresh paper to be used as supporting evidence against N.F. application. Why? 🙂
George
Not sure that would be the right take away reading all the examples. BTW Jim and Jerry believe in >100kHz for audio. Jerry is beyond smart (jn would appreciate that he is also a clockworks expert).
Don't-you think it is a good idea to try to get, as an electronic target, a flat phase curve up to , say, 20KHz in audio ?
Last edited:
If that is a goal, why do we not come close to achieving it? Speakers have excessive amounts of phase shift.
It isn't phase shift (time domain waveform distortion) per se that matters so much to humans (except maybe at LF and for the punch of percussive sounds which tend to be heard more according to their time domain waveforms), as much as it is rate of change of phase with respect to frequency, aka Group Delay, IMHO.
Last edited:
+1
That why I said "for electronic".
As, usually, we are not in a hurry, listening to music, delay is not an issue ;-)
Good paper, thanks Scott. Yes, some of us actually believe our ears, and extended bandwidth is one criterion that seems necessary for best audio reproduction.
How people “read” differently the same script🙂
IMO, the paper puts the psychoacoustics into the discussion (as to what - other than "numbers" of measurements- is of importance when listening to music)
George
Right. Except I would say that a particular type Group Delay occurs if a constant time delay for all frequencies (not necessarily an issue for music playback), but it can also refer to fast changing rate of change of phase vs frequency associated with resonances such as the phasey sound of wah-wah pedal phase shift (which can sound objectionable in overall music reproduction). That phasey sound can occur with some filters, including dac interpolation filters, and is more likely with very steep transition band frequency response. Might occur with some speaker resonances too, I guess. As usual, all the foregoing is IMHO.
Last edited:
😱 I'm shock to see myself agreeing with T.
How was the listening comparison performed? It's a critical aspect especially when small change in listening position can alter the soundwave arriving to our ears.
- Status
- Not open for further replies.