I'm not sure what you are saying here. 0.1% of low order distortion would probably be enough to satisfy most listeners seeking high fidelity sound reproduction. Others will want 0.001% just because they can. Some prefer 2% as it adds some 'warmth' and they prefer recorded sound to live sound.Discopete said:
How transparent as defined by how truthful to non reproduced sound. I can only speak for myself. I cited .1% because most of the competition was sporting numbers like .05, .007 et al. The point being impressive THD numbers didn't mean a whole lot when the highest ones represented the sota at the time. SOTA is always defined by consensus in terms of sq. This was the era of the NF wars. Those amps sounded horrible to say the least.
I'm guessing it was the artful application of NF + serious attention to design parameters as is being discussed here that resulted in those great sounding old amps.
I never said that phase shift did not matter, merely that a delay can be considered in terms of its equivalent phase angle at a specific frequency.
Given the context of traderbam’s reply of “Except for Class D?” in post #52, I thought that context might include a discrete time system which the feedback signal could in fact have a delay, and not just phase shift as in a continuous time feedback system.
So, I'll update my statement - we're not all saying the same thing. I said what I wrote 🙂
Delay and phase shift are two quite different things??? They imply the same thing in the contemporary doctrine. Shifting the phase means delaying an entity, moving it back or forward in time, which is a deeply unphysical concept but anyway...
Making a distinction between the confusedly and arbitrarily used terms Phase and Polarity could be regarded as a good starting point.
Making a distinction between the confusedly and arbitrarily used terms Phase and Polarity could be regarded as a good starting point.
In a class D or switch mode amplifier, I believe the average output “update” delay is the switching period. In the order of micro-seconds.
Time delays are treated differently, mathematically, from phase shifts. Phase shifts are like a sinusoidal representation of electrical inertia. Inertia has no time delay.
Delays are the nemeses of control systems. The feedback unity loop gain f needs to be well below the sampling f. This is why class D uses extremely steep roll-off filters in the forward path to reduce the loop gain. It’s a hack. I’m not sure such a circuit has lots of control when really it is impotent above a few tens of kHz.
Time delays are treated differently, mathematically, from phase shifts. Phase shifts are like a sinusoidal representation of electrical inertia. Inertia has no time delay.
Delays are the nemeses of control systems. The feedback unity loop gain f needs to be well below the sampling f. This is why class D uses extremely steep roll-off filters in the forward path to reduce the loop gain. It’s a hack. I’m not sure such a circuit has lots of control when really it is impotent above a few tens of kHz.
Yes. Delay is one cause of phase shift. You can have phase shift without delay, therefore they cannot be the same thing. Delay is measured in seconds. Phase shift is measured in degrees (or radians). They don't even have the same dimensions.N101N said:
It is phase shift which affects feedback loop stability. Where the phase shft comes from (delay or something else) is a separate issue. Talk of delay in the feedback of an audio amplifier just feeds the popular myth that feedback corrects the previous error instead of the current error.
Not to beat the dead horse here, but 'phase shift' is only a result, a measurement. There is no 'phase shift' mechanism that produces 'phase shift' exclusively. Rather, a number of circuits and systems have behavior that, when examined in a certain way, exhibit 'phase shift'.
It's like frequency - it doesn't actually exist, it is a characterization of reality that is helpful when examining how actual circuits and systems behave. You can have a frequency of paying rent - it's not exclusive to an LTI circuit. Same thing with phase shift - my rent can be late 30 degrees.
I'll shut up now.
It's like frequency - it doesn't actually exist, it is a characterization of reality that is helpful when examining how actual circuits and systems behave. You can have a frequency of paying rent - it's not exclusive to an LTI circuit. Same thing with phase shift - my rent can be late 30 degrees.
I'll shut up now.
We talk electronics here (ie physics), not abstruse philosophy! Frequency and phase exist and are talked about all the time. You might as well say blue doesn't exist.
Apart from an all-pass filter?
Frequency is just as real as time. Phase shift is just as real as the number 2. Bar room metaphysics does not help newbies understand feedback.
Let's keep the insults to a dull roar. My ONLY point is that frequency, phase shift, amplitude, and a host of other quantities are properties of circuits or systems, not actual devices or circuits themselves.
So, the property of phase shift can indeed be exhibited by any system that makes a signal late, including, gasp, a delay. Just because this behavior does not happen in the few circuits you're currently thinking about does not mean that a delay cannot bring about behavior that will cause a feedback system to have instability, much like a phase shifted feedback signal. Imagine for example someone closing a feedback loop with an Arduino, sampling the output and computing a feedback signal. The delay caused by the sample period will in fact behave like a phase shift.
I apologize for muddying the waters for newbies, and I can assure you this will not happen again.
So, the property of phase shift can indeed be exhibited by any system that makes a signal late, including, gasp, a delay. Just because this behavior does not happen in the few circuits you're currently thinking about does not mean that a delay cannot bring about behavior that will cause a feedback system to have instability, much like a phase shifted feedback signal. Imagine for example someone closing a feedback loop with an Arduino, sampling the output and computing a feedback signal. The delay caused by the sample period will in fact behave like a phase shift.
I apologize for muddying the waters for newbies, and I can assure you this will not happen again.
Phase Shift or Phase Delay, Phase Difference, Phase Offset, whatever, is a property of periodic wave expressed by a mathematical function in Mathematical Analysis, invented to meet mathematical tractability. The phase shift can only be 90 degrees due to systematic incapability. Periodic functions are independent of amplitude and phase, also magnitude is independent of time.
On the other hand, the term Phase Angle is used in chemistry, optics, biology, physics, it has a different meaning and cannot be solved analytically. The syntax rules in Mathematical Analysis and the definitions in Physical Theory are not the same. Mathematical Analysis is not a representation of reality, but some frivolous fancy nonsense.
On the other hand, the term Phase Angle is used in chemistry, optics, biology, physics, it has a different meaning and cannot be solved analytically. The syntax rules in Mathematical Analysis and the definitions in Physical Theory are not the same. Mathematical Analysis is not a representation of reality, but some frivolous fancy nonsense.
It is difficult to work out exactly what you mean by this sentence.Monty McGuire said:
I was not saying that delay cannot cause phase shift. I was saying that loop stability in audio amplifiers arises from phase shift (as a function of frequency). Phase shift can arise from delay, but this will normally be negligible in an audio amplifier. Why am I repeating myself?
No. No. ?N101N said:
No idea what you are talking about.
The whole point of stating this was in response to traderbam's post #52:
traderbam said:
I, probably mistakenly, thought that he was considering the effect on phase margin of a delay applied to the feedback signal in a class D amplifier. On hindsight, aside from possibly a one sample delay, that shouldn't amount to more than a wart on that toad, so it's probably best ignored except for pathological cases.
There is no need to repeat yourself. I'd be happiest if we could move past this dead carcass.
Why are you obsessed with requiring infinite periodicity? No need to, you can generalize the maths very easily to finite segments of a periodic function, it ought to be very obvious how this works by extrapolation. After all these are mathematical models, we build them, we can choose what we do to get the most useful and practical results.
And the earth is flat. With out this "frivolous fancy nonsense" most of the technology you use everyday would not exist.
I don't believe many people understand the complexity of music wave form fidelity as it is infinitely more complex than what most tests use. The phase, amplitude of the composite wave form is extremely rapid in changing when seen though an oscilloscope. The old engineers knew this and produced some of the most outstanding components to reproduce music ever. This has been absolutely lost. Just as the cluelessness of the flywheel effect of the weight and balance of a turntable platter shows without a doubt, and like the industry in power generators know once up to speed the weight on the outside is critical and the interior must be stiff but not unnecessarily heavy. The old Denons, Technics, Kenwood L-07D and others understood this and were balanced and the weight was primarily on the outer part of the platter.
The total cluelessness is profound when it is so clear when listening to some of these and the likes of Spectral and the best OpAmps are the fastest, far greater than the garbage engineered by looks and no science or math. The high dollar shiny stuff is garbage. The "cogging" is by visual and not measurable when I have seen tested in analog the tape output of the finest measuring systems such as B&K for turntables, the same works for amplification. It really gets ugly when a Sony 880 can produce better wave form fidelity in their recordings of the metal tapes from subsonic frequencies out to 40khz than the output of some of the most expensive, thrown together half assed amplifiers on the planet. The idiocy of the turntables, arms and heavyweight cartridges of no compliance with arms that are at best 5-20x the friction of the EPA-100 is frustrating.
IMHO, it's a suckers market and the worst is the most expensive. The Stereohedron, Fritz Geiger and Microridge cartridges of old simply make the expensive tragically overpriced and under engineered 99% sound like the old 45rpm player in a box. Preamplifiers lack the engineering integrity for the older vertical manufacturing which is not even possible now as VFETs are gone and Nelson Pass has stated he has heard nothing even close to them sonically. Innovation is dead and most every category is going backward except for some versions of class D which I did get to hear the Infinity pair. Now much improved, these have some of the best components since Purifi entered into competition and now, those companies which have been on their butts have had to improve or die.
As far as speakers go, the best are made by DIY, not the kit stuff like so many that use the kit formulas of SEAS, Radian, SB Acoustics, Wavecor and several others and still, the crossovers are the slowest cheapest bunch of garbage even the high end Japanese. Fast capacitors are NOT ELECTROLYTIC and the best resistors are not SAND CAST and there are other solutions from a large number of suppliers such as Jenzsen, Audio Note, Audyn, Mundorf,... but rarely used and active crossovers are nowhere to be seen in high end audio except some studio PMC, ATC, Focal which also use far better components to their mass market over priced sucker speakers.
Rise time and Slew Rate ARE important. It's not for those who prefer lazy and easy on the ears. There is no competive market in any category and the reviewers are incompetent 99% of the time. Listen to a Spectral System and some of the better Sound Lab electrostatics or what they can do for the better Planar technologies, although Magnepan has the cheapest bunch of components for crossovers I've seen, just like SVS and other pretenders. Wilson is unlistenable to me and the more components the worse it gets. Keep with the best of the KISS principle and use quality components. There is a reason why the Grand Integra 508, 510, Yamaha B1 B2 and beyond, Kenwood L-07II and L-08, JVC VFets and Sansui's finest are moving into the 5 digits for units in excellent shape. The specifications people brag about for Benchmark AB amplifiers are no better than the best 30-40 years ago.
The total cluelessness is profound when it is so clear when listening to some of these and the likes of Spectral and the best OpAmps are the fastest, far greater than the garbage engineered by looks and no science or math. The high dollar shiny stuff is garbage. The "cogging" is by visual and not measurable when I have seen tested in analog the tape output of the finest measuring systems such as B&K for turntables, the same works for amplification. It really gets ugly when a Sony 880 can produce better wave form fidelity in their recordings of the metal tapes from subsonic frequencies out to 40khz than the output of some of the most expensive, thrown together half assed amplifiers on the planet. The idiocy of the turntables, arms and heavyweight cartridges of no compliance with arms that are at best 5-20x the friction of the EPA-100 is frustrating.
IMHO, it's a suckers market and the worst is the most expensive. The Stereohedron, Fritz Geiger and Microridge cartridges of old simply make the expensive tragically overpriced and under engineered 99% sound like the old 45rpm player in a box. Preamplifiers lack the engineering integrity for the older vertical manufacturing which is not even possible now as VFETs are gone and Nelson Pass has stated he has heard nothing even close to them sonically. Innovation is dead and most every category is going backward except for some versions of class D which I did get to hear the Infinity pair. Now much improved, these have some of the best components since Purifi entered into competition and now, those companies which have been on their butts have had to improve or die.
As far as speakers go, the best are made by DIY, not the kit stuff like so many that use the kit formulas of SEAS, Radian, SB Acoustics, Wavecor and several others and still, the crossovers are the slowest cheapest bunch of garbage even the high end Japanese. Fast capacitors are NOT ELECTROLYTIC and the best resistors are not SAND CAST and there are other solutions from a large number of suppliers such as Jenzsen, Audio Note, Audyn, Mundorf,... but rarely used and active crossovers are nowhere to be seen in high end audio except some studio PMC, ATC, Focal which also use far better components to their mass market over priced sucker speakers.
Rise time and Slew Rate ARE important. It's not for those who prefer lazy and easy on the ears. There is no competive market in any category and the reviewers are incompetent 99% of the time. Listen to a Spectral System and some of the better Sound Lab electrostatics or what they can do for the better Planar technologies, although Magnepan has the cheapest bunch of components for crossovers I've seen, just like SVS and other pretenders. Wilson is unlistenable to me and the more components the worse it gets. Keep with the best of the KISS principle and use quality components. There is a reason why the Grand Integra 508, 510, Yamaha B1 B2 and beyond, Kenwood L-07II and L-08, JVC VFets and Sansui's finest are moving into the 5 digits for units in excellent shape. The specifications people brag about for Benchmark AB amplifiers are no better than the best 30-40 years ago.
This is incorrect. Audio is very simple, but many are mislead by what they see on a scope. OMG look at all those wriggles, how complex!
Yet it isn't, it's just a bunch of sinewaves coming and going, the complexity is all inside our heads.
It also is not extremely rapid changing, a 20kHz signal takes a leisure 25us to change from one peak to another. Ages for electronic circuits; why, even a mechanical tweeter can follow that!
Another issue that hasn't been mentioned I believe is that rise time is a small signal spec, while slew rate is a large signal spec. An opamp can have a rise time of 1us yet be unable to faithfully reproduce a 20kHz wave. Because rise time is measured at 1V amplitude or even less, while slew rate comes into play at max amplitude. If you have a large slew rate you probably also have a fast rise time, but the opposite is not necessarily true.
Jan
I think you are projecting your own limitations. Slew limiting is a form of clipping. Typically the power amplifier has a LTP stage between the input and a slew limiting VAS with a compensation capacitor. The slew rate is the current output of the LTP divided by that compensation capacitance. When the limit of the LTP current is reached, ie clips, the maximum slew happens in the VAS. If the input to the amplifier is band limited and amplitude limited, it is impossible to cause slew limiting in the amplifier assuming the potential slew rate is greater than that of a sine wave at the maximum frequency and amplitude. No combination of lower frequencies that do not cause clipping has a higher slew requirement because slew rates sums the same as amplitude.
When I was young, the issue of slew limiting was discovered because people tried to use 741 op-amps for audio. This was because the 741 and similar had a very poor slew rate. The same problem happened in power amps that has too much gain and therefore required draconian amounts of compensation. Massive amounts of loop gain was assumed to reduce THD without limitations, but that ignored the slew rate issue, and often marginal stability. Then when the slew rate issue was "discovered", it became the base of marketing hype. Someone here said that "transparency" is about low THD. That's not true. If fact supernatural "transparency" is often the result of carefully tailored even harmonic distortion. The human hearing system can be hacked. It is not a precision instrument.