I have read the datasheet of the AD844 - is that pin 5 ONLY to add feedforward compensation?
Seems like I and Jan (to name only two) use it for other things as well. And if such a thing is important enough to do it for the 844, why didn't ADI (and their competitors) bring out that connection for all their other fast CFB amps?
I'm trying to get into the mind of the chip designer here, not get a "Read the ******* Manual" answer...
Regards, Allen
Seems like I and Jan (to name only two) use it for other things as well. And if such a thing is important enough to do it for the 844, why didn't ADI (and their competitors) bring out that connection for all their other fast CFB amps?
I'm trying to get into the mind of the chip designer here, not get a "Read the ******* Manual" answer...
Regards, Allen
john curl said:
It controlled only common mode DC, with 10k or 5k resistors. Outputs of OAs saturated.
It controls both common mode and differential mode DC, with 2k output resistors. Courages hint is correct. Output DC depends on OA offset, and can be as low as microvolts.
More than 2 minutes needed 😉
PMA said:
OK, I see now exactly what is happening.
The mid-band gain of your circuit is slightly higher because you are altering the amount of local degeneration of the input stage.
The low-frequency gain of my circuit is changed because the "servo" is not just sending back DC, but also low-frequency AC. There is more AC gain at the gates of the cascodes than at the sources of the input stage. If you want to restore the low-frequency response and still use my circuit, either increase the time constant of the "servo" amp or decrease the gain of the "servo" amp (to compensate for the increased gain of that input node).
Allen Wright said:
I think Bob Pease used to call it "compensectomy" when you had to actually cut off the comp pin to get rid of unwanted capacitance. This pin 5 connection is only a 'feature' to a small sub-group of designers that actually know what to do with it. We have more recent amps with this feature AD829 and a few others. With the recent CF amps pushing 1-3GHz the pF's on this pin just get in the way not to mention the unavoidable .5 or so nH of inductance. The AD811 was designed for high output current but topolgically it is as simple as a text book CF amp gets.
The AD744 has the comp pin because it was used in a hybrid circuit with an external pnp, it's from the bad old lateral pnp days. It still sells rather well relatively considering a number of more recent parts have been obsoleted.
G.Kleinschmidt said:
Grey, You also lost me on this example. Absolute phase is easily measurable, not really an example of hearing things that aren't. I'm sure there are people who think +-.25 dB RIAA accuracy is not audible, but even the diehard objectivists of the BAS (the same ones that said 44/16 is indistinguishable from SACD) insist on +-.1dB equalization.
I have read that the neuro-physiology of sound transduction supports the possibility of absolute phase sensitivity but Dr Geddes who claims to spend lots of effort on perceptual testing of loudspeakers seems to still be unconvinced absolute phase perception is important in sound reproduction
I believe he is saying that most instances of "obvious" absolute phase perception are a function of the system nonlinearities - particularly dynamic driver loudspeakers
Given the physiologic basis for its perception and the technically "trivial" solution there appears to be no reason to ignore absolute polarity - but its relative low importance in the perception of reproduced sound would explain current practice
I believe he is saying that most instances of "obvious" absolute phase perception are a function of the system nonlinearities - particularly dynamic driver loudspeakers
Given the physiologic basis for its perception and the technically "trivial" solution there appears to be no reason to ignore absolute polarity - but its relative low importance in the perception of reproduced sound would explain current practice
"Audible" and "important" are two different things.
IME, it is very much a function of speakers- it was MUCH harder to hear on my full-range, phase-coherent ESLs than on the dynamic speakers they replaced. I'd also guess that this has to do with driver nonlinearity, specifically even-order.
IME, it is very much a function of speakers- it was MUCH harder to hear on my full-range, phase-coherent ESLs than on the dynamic speakers they replaced. I'd also guess that this has to do with driver nonlinearity, specifically even-order.
it could also mean that SE triode users have have an additional "leg up" on poor old Doug Self with his blameless amps when it comes to detecting absolute polarity by "just listening"
janneman said:
If I may be so bold as to qoute my own post 😉 :
Any opinions on this?
Jan Didden
jcx said:
If absolute phase is audible (I suspect it is, based on a lot of AES work done in the 70's, but am not sure about the degree of importance), then that brings up a big issue for loudspeakers with certain kinds of crossovers (I realize this is slightly OT). The most obvious is those loudspeakers in which one or more of the drivers is connected in reverse polarity to make the crossover work properly. Sometimes this is the mid-range.
Any thoughts on this?
Cheers,
Bob
PMA, regarding your 'final' schematic: I would simplify the servos and make sure that the effective bandwidth was VERY low, or else what bothers Charles about servos will come true.
Bob Cordell said:
Not to pursue the off-topic too far, but Larry Klein, audio writer for Popular Electronics, was very insistent on the undesirability of out-of-phase crossover designs. He pointed out that woodwinds, among other instruments, have asymmetrical waveforms. Below is the acoustic impedance of a clarinet, taken from this website. As you can see, the centerline rises with frequency. If the midrange is out-of-phase with the woofer and the tweeter, the reproduced waveform will be very affected.
Unfortunately, the second order crossover used from the early days of audio involves these out-of-phase hookups.
Attachments
Scott,
I have no earthly idea what the limits of human hearing are, but I know that I have managed to set the level controls on an active crossover to within .1dB of previous settings using only my ears. This measured with a Fluke 8060A after getting done. It's not something that I can do quickly; it takes days or weeks of fiddling to achieve that sort of thing. The first day you get it somewhere in the ballpark. The second day you get fairly close. By the end of the first week it's pretty damned good and would serve perfectly well as a 'forever' setting...but then you sit down to listen one evening the next week or the week after and you get this little itch that you can do better still, so you spend thirty minutes or an hour doing that last, little bit of adjustment.
I didn't set out to do this as a stunt. No one else was involved at all. It was just me and my stereo and a wide array of classical, jazz, and rock records. I had pulled the old crossover and put in a new one, so the whole balancing of the system had to be done from scratch. I knew where the levels ended up in the past (sine wave input, measured at the speaker terminals), and managed to find those notes (not always an easy task), but chose not to set the levels 'the easy way' by using test signals.
Having done this on at least three or four occasions, I'm fairly sure the results are not a fluke (no pun intended).
Now, as to absolute phase being easily measured--of course it is. Well, not measured, per se, but you can tell in the binary on/off sense whether it's 0 or 180 degrees. For that matter, you don't even have to measure anything at all, you can just run your finger across the schematic and tell what you're dealing with.
You and I don't know each other personally, but I gather from your posts here that you are at least passably familiar with high end audio, though not necessarily deeply involved in it. As such, you may not be aware of the long-standing loathing of people like dear Kleinschmidt and others of his ilk for things that they can't explain. It's not that (in this case) absolute phase can't be explained, it's that no one had any idea how or why it should matter to the human ear. People who live and die by sine wave bench tests correctly point out that the human ear cannot discriminate between a sine wave and a cosine wave, the difference being a mere few degrees of phase rotation. From this they incorrectly deduce that the human ear cannot tell absolute phase. The fly in the ointment being, of course, the difference between our reactions to steady state and impulse sounds.
But why, dammit, why?
I've even seen long-winded arguments based on evolution that since there's no evolutionary advantage to such a thing then obviously it can't exist. All this does is show the ignorance of the one making the argument, as evolution doesn't necessarily favor blue eyes over green for instance, but as long as the mutation isn't deleterious it can persist in the gene pool forever. I do, however, wonder if there might be some people possessed of an 'absolute phase gene' and others who lack it. If so, it would go a long ways towards explaining things.
In the same vein, I happen to favor wide bandwidth. Really wide bandwidth. No matter how good my hearing might be (and it's pretty good), I will personally guarantee you that I can't hear discrete tones at 250kHz. And yet I can hear a difference in the sound of a wide bandwidth circuit. Why? My working hypothesis is that it's a question, not of discrete tones being audible, but of waveform. Take something like a square wave and start bandwidth limiting the circuit. You can see visible changes on an oscilloscope when the bandwidth is still multiple octaves away from the fundamental. No real surprise there, it's just basic Fourier mathematics. But, like absolute phase, you can measure it and see it on the scope, but according to classic, die hard, Kleinschmidt-esque thinking, 20-20kHz bandwidth is all you need for audio reproduction (Red Book CD, anyone?), and yet...and yet...it takes only a bare minimum of experimentation to demonstrate that a wider bandwidth circuit sounds better. The market seems to have settled at around 100kHz as a reasonable compromise, but I argue that there's more to be had beyond that, although I recognize that it may not be feasible in a commercial product due to RF problems in some markets.
Regardless of whether it can be measured (e.g. absolute phase, bandwidth) or not (cable direction, differences in resistors) people like Self and Kleinschmidt think that everything is known that needs to be known and that anyone who claims to hear audible improvement or degradation of anything outside their narrow view of the universe is deluded and/or lying. It's all the same to them. To the most hard core, THD is the only meaningful specification and nothing else matters. If absolute phase does not worsen THD, then it's irrelevant in their eyes. Anyone who claims otherwise is in league with the devil. To stray even a single step from the straight and narrow is analogous to the old fears about marijuana: Take a toke on a joint and tomorrow you'll be snorting coke, the day after you'll be mainlining heroin. To open the door to things that can't be explained in audio is the same way: Absolute phase, differences in speaker cables, sticky dots on speakers and who knows what else? Best not to take that first step.
I'm not saying whether sticky dots on speakers have merit or not. I haven't heard them and have no opinion. On the face of it, I don't see how they would accomplish anything, but don't misconstrue that as meaning "they don't have any effect," because I meant what I said in the most literal fashion. I don't see an obvious mechanism, but that doesn't mean that there isn't one. Ditto for any number of other audio-related ideas for improvement. Does suspending your cable above the floor improve the sound? I don't know. I haven't tried it. Do the Bybee gizmos work? No idea, haven't heard them. Etc. Lacking a gazillion dollar influx from the government to improve audio, we're left on our own to try things that may--or may not--work. Just because some of them don't seem to make sense at first (why should absolute phase matter?) doesn't mean they're not audible. And some things that are measurable (like entire percentiles of lower order harmonic distortion remaining undetected under certain conditions) don't seem to matter when they theoretically should. To argue pedantically that something...anything...can or cannot be heard is foolish. Whether it can be measured or not.
Grey
N.B.: I need to go tend to the kids, so this post may read awkwardly. Didn't have time to fine tune things.
I have no earthly idea what the limits of human hearing are, but I know that I have managed to set the level controls on an active crossover to within .1dB of previous settings using only my ears. This measured with a Fluke 8060A after getting done. It's not something that I can do quickly; it takes days or weeks of fiddling to achieve that sort of thing. The first day you get it somewhere in the ballpark. The second day you get fairly close. By the end of the first week it's pretty damned good and would serve perfectly well as a 'forever' setting...but then you sit down to listen one evening the next week or the week after and you get this little itch that you can do better still, so you spend thirty minutes or an hour doing that last, little bit of adjustment.
I didn't set out to do this as a stunt. No one else was involved at all. It was just me and my stereo and a wide array of classical, jazz, and rock records. I had pulled the old crossover and put in a new one, so the whole balancing of the system had to be done from scratch. I knew where the levels ended up in the past (sine wave input, measured at the speaker terminals), and managed to find those notes (not always an easy task), but chose not to set the levels 'the easy way' by using test signals.
Having done this on at least three or four occasions, I'm fairly sure the results are not a fluke (no pun intended).
Now, as to absolute phase being easily measured--of course it is. Well, not measured, per se, but you can tell in the binary on/off sense whether it's 0 or 180 degrees. For that matter, you don't even have to measure anything at all, you can just run your finger across the schematic and tell what you're dealing with.
You and I don't know each other personally, but I gather from your posts here that you are at least passably familiar with high end audio, though not necessarily deeply involved in it. As such, you may not be aware of the long-standing loathing of people like dear Kleinschmidt and others of his ilk for things that they can't explain. It's not that (in this case) absolute phase can't be explained, it's that no one had any idea how or why it should matter to the human ear. People who live and die by sine wave bench tests correctly point out that the human ear cannot discriminate between a sine wave and a cosine wave, the difference being a mere few degrees of phase rotation. From this they incorrectly deduce that the human ear cannot tell absolute phase. The fly in the ointment being, of course, the difference between our reactions to steady state and impulse sounds.
But why, dammit, why?
I've even seen long-winded arguments based on evolution that since there's no evolutionary advantage to such a thing then obviously it can't exist. All this does is show the ignorance of the one making the argument, as evolution doesn't necessarily favor blue eyes over green for instance, but as long as the mutation isn't deleterious it can persist in the gene pool forever. I do, however, wonder if there might be some people possessed of an 'absolute phase gene' and others who lack it. If so, it would go a long ways towards explaining things.
In the same vein, I happen to favor wide bandwidth. Really wide bandwidth. No matter how good my hearing might be (and it's pretty good), I will personally guarantee you that I can't hear discrete tones at 250kHz. And yet I can hear a difference in the sound of a wide bandwidth circuit. Why? My working hypothesis is that it's a question, not of discrete tones being audible, but of waveform. Take something like a square wave and start bandwidth limiting the circuit. You can see visible changes on an oscilloscope when the bandwidth is still multiple octaves away from the fundamental. No real surprise there, it's just basic Fourier mathematics. But, like absolute phase, you can measure it and see it on the scope, but according to classic, die hard, Kleinschmidt-esque thinking, 20-20kHz bandwidth is all you need for audio reproduction (Red Book CD, anyone?), and yet...and yet...it takes only a bare minimum of experimentation to demonstrate that a wider bandwidth circuit sounds better. The market seems to have settled at around 100kHz as a reasonable compromise, but I argue that there's more to be had beyond that, although I recognize that it may not be feasible in a commercial product due to RF problems in some markets.
Regardless of whether it can be measured (e.g. absolute phase, bandwidth) or not (cable direction, differences in resistors) people like Self and Kleinschmidt think that everything is known that needs to be known and that anyone who claims to hear audible improvement or degradation of anything outside their narrow view of the universe is deluded and/or lying. It's all the same to them. To the most hard core, THD is the only meaningful specification and nothing else matters. If absolute phase does not worsen THD, then it's irrelevant in their eyes. Anyone who claims otherwise is in league with the devil. To stray even a single step from the straight and narrow is analogous to the old fears about marijuana: Take a toke on a joint and tomorrow you'll be snorting coke, the day after you'll be mainlining heroin. To open the door to things that can't be explained in audio is the same way: Absolute phase, differences in speaker cables, sticky dots on speakers and who knows what else? Best not to take that first step.
I'm not saying whether sticky dots on speakers have merit or not. I haven't heard them and have no opinion. On the face of it, I don't see how they would accomplish anything, but don't misconstrue that as meaning "they don't have any effect," because I meant what I said in the most literal fashion. I don't see an obvious mechanism, but that doesn't mean that there isn't one. Ditto for any number of other audio-related ideas for improvement. Does suspending your cable above the floor improve the sound? I don't know. I haven't tried it. Do the Bybee gizmos work? No idea, haven't heard them. Etc. Lacking a gazillion dollar influx from the government to improve audio, we're left on our own to try things that may--or may not--work. Just because some of them don't seem to make sense at first (why should absolute phase matter?) doesn't mean they're not audible. And some things that are measurable (like entire percentiles of lower order harmonic distortion remaining undetected under certain conditions) don't seem to matter when they theoretically should. To argue pedantically that something...anything...can or cannot be heard is foolish. Whether it can be measured or not.
Grey
N.B.: I need to go tend to the kids, so this post may read awkwardly. Didn't have time to fine tune things.
We have studied 'absolute polarity' for the last 35 years. The best paper on the subject, in general, is the 1975 article describing some Bell Labs (remember them) research published by the 'Proceedings of the IEEE' in a tutorial by Manfred Schroeder. I have added a polarity switch to my best preamps since 1981. Some find it very important, some find it less important.
Of course, the Lipshitz criterion for ABX testing will make 'proving' absolute polarity very difficult, and nearly impossible, but you can just listen for yourself. Personally, I find it less important, but it does depend on loudspeakers, xovers, and personal sensitivity to it.
Of course, the Lipshitz criterion for ABX testing will make 'proving' absolute polarity very difficult, and nearly impossible, but you can just listen for yourself. Personally, I find it less important, but it does depend on loudspeakers, xovers, and personal sensitivity to it.
Quite the opposite, John, Lipshitz had no trouble demonstrating audibility with DBT. He also trained himself to detect midrange phase shift of the sort mentioned by kelticwizard.
I know for a fact, SY, that he had trouble with ABX'ing it. He said so, himself, decades ago.
Now, if he could only now again ABX the worst TANTALUM caps that Walt Jung could find, and hear them in an ABX test. Last time he tried, he couldn't, so I guess that removing tantalum caps from mike preamps (Scott, please note this) is unnecessary. That's the way with ABX testing.
Now, if he could only now again ABX the worst TANTALUM caps that Walt Jung could find, and hear them in an ABX test. Last time he tried, he couldn't, so I guess that removing tantalum caps from mike preamps (Scott, please note this) is unnecessary. That's the way with ABX testing.
Lipshitz, S. P., Pocock, M., and Vanderkooy, J., "On the Audibility of Midrange Phase Distortion in Audio Systems," J. Audio Eng. Soc., vol. 30,
pp. 580-595 (1982). No problems at all hearing it, assuming the source signal was appropriately selected. Also there's a MS thesis of Daisuke, but I don't have that reference to hand.
Linkwitz on the same subject:
http://www.linkwitzlab.com/frontiers.htm#F
pp. 580-595 (1982). No problems at all hearing it, assuming the source signal was appropriately selected. Also there's a MS thesis of Daisuke, but I don't have that reference to hand.
Linkwitz on the same subject:
http://www.linkwitzlab.com/frontiers.htm#F
- Status
- Not open for further replies.