Settling time does not affect distortion measurements directly. Often, however, steps you take to improve distortion may negatively affect the settling behavior of the amplifier. For example, pole-zero pairs are often added to the open loop gain curve of an amplifier to boost loop gain at low frequency (which reduces distortion) and then to cut it at very high frequency (MHz range) to improve stability. This typically adds a small amount of over-shoot to the transient response.
You can see this in Figure 16 of the AD797 datasheet where a 2nd pole (after the dominant) is at 1MHz, and a zero around 5MHz to restore the phase before the loop intersects 0dB. Now refer to the pulse response plots which show a small amount of overshoot. This is quite common and not unique to op amps.
My best guess is its mainly down to output stage bias current. The discrete opamps I've seen tend to run much, much warmer and often feature significant region of classA operation. Plus that discretes just do better in the presence of RF which opamps seem to dislike with a passion. Is that related to junction isolation? Dielectrically isolated parts often seem to get relatively good rap for SQ.
Video opamps don't tend to care so much about noise so sometimes have degenerated LTPs - AD847 is one example. This gives better front-end linearity, especially handy to have in the presence of OOB noise.
I think you've put your finger on something vital - science has always been about coming up with testable hypotheses in an attempt to explain observations.
Whereas what we see here a lot is scientism - finding rationalisations (which must be non-falsifiable hence not testable by experiment) to explain away observations.
Just remember, Chris, many people enjoy playing "Who's line is it anyways?" science:
https://www.youtube.com/watch?v=9KAGwNtI26w
None of my effing business. The burden is on you to prove the extraordinary claims. Negatives can't be proven, asking others to prove your conclusion is wrong is a logical fallacy. At least in the rational world.
As such, if you expect others to prove that you don't hear "micro transients embedded in black noise background" then please do everybody a favour and hold your breath until.
Since I seem to be the only one here that actually reads these papers (iirc the correct term is 'the sucker'):
Dick Heysers design was a conceptual exercise for an amp where the output stage bias was provided by the signal. The idea being that the bias varies with the signal such that it always has the right bias level and gets rid of xover. In that sense it did what it was supposed to do.
It was not meant as the next best amp.
Jan
Tell me how a resistor or capacitor is made in an IC. It's constructed fundamentally different than the one that comes with two leads and is soldered on a PCB. No one has ever thought of this as a reason?
What about them? If the part can amplify and be clean 120dB below the signal who actually cares how things were implemented?
I try to concentrate on oscillators, serious wideband amplifiers, noise and
esp. phase noise.
That circuit avoids the crossover distortion in a more fundamental way. 🙂
So let the the negative attack time always be with it.
BTW Carl Battjes of Tektronix devised a better Darlington look-alike, beta is
only the sum of the transistors, it pushes ft and drops in a friendly way over
the octaves.
Those Tektronix guys were the real heroes. They had no better parts at
their disposal than the rest of the population, but built the amplifiers that
the rest could use to judge their circuits with.
Read the wideband amplifier books by Hollister and by Staric e.a.
Not exactly cheap, but worth every dime. Sort of concentrated TEK.
Maybe a little bit over the top for the audio frequency range.
regards, Gerhard
Last edited:
- Status
- Not open for further replies.