"Normal" non linearity (such as harmonic distortion) is one thing - where you can have less ad less, but never reach zero. Things like TIM that only happen if a certain condition is met is different - just like oscillation. There are analog amplifiers that oscillate in some conditions, and other that don't oscillate at all. As in *zero*. Not 0.0001 or anything, but not at all. Same with slew rate distortions. As long as the correct boundary conditions are observed, they don't happen - at all.
if I recall correctly, it's you who said that audio is not a question of belief.
do we?
I feel I need to remind everyone that the published measurements of the NCORE display a 0.0003% THD @20kHz, @20W power, not 0.01%. that is 3 ppm (parts per million). whereas some "high end" amps are nearing 1% @20kHz. at which point we need to remember what THD is. Total Harmonic Distortion is the ratio of the RMS voltage of all harmonics divided by the RMS voltage of the fundamental. one can easily infer from that the amplitude of the error waveform. the Hypex datasheets show the distortion spectrum and the fact that THD is pretty much invariant with frequency suggests one other important thing: that the spectrum does not vary much over it, meaning the distortion is of simple nature.
so, we have a few parts per million of error waveform with a full-scale 20kHz sine, the fastest slewing signal present at the input (which is a few times faster, if not orders of magnitude, compared to the slew rate of a kick-drum attack).
and... how come (cough cough) many "fast" amps display rising distortion with frequency, whereas others don't? what does that SR do, exactly? how does it help?
time to rephrase the issue: where is that TIM distortion?
maybe it's time to answer with facts (your measurements or that AES article would be a start) instead of cherry-picking and straw men.
Last edited:
I've not heard such partnering but would love to, and imagine it's a great match, really wake up those planar panels!
Hum...
As long as you design a closed loop amplifier, you design an oscillator. 'More or less' damped. A stable amp, according to Nyquist, is an oscillator with not enough Q at its resonance frequency while an unstable amplifier (according to Murphy) has too much
Same thing with this kind of distortion due to the delay between error and its correction, in presence of fast transition, that some call TIM. It is an inter-modulation distortion that will never be 0.
I disagree. As long as the feedback loop of the amp can handle the fastest input signal, there won't be *any* slew rate related distortion (of which TIM is a small special case).
From personal experience, this is not necessary - eliminating all the weaknesses that inject disturbing low level distortion, and making sure that the power amp can cleanly amplify to the transient peak SPLs is all that's required.
If the system is sufficently capable in this sense, then all recordings acquit themselves very nicely indeed, subjectively conveying the same impact as the "real thing" would ...
Julf, please, think twice about servos.
As long as your closed loop amplifier will have a transmission delay (and they all have) and the overall gain is a function of the feedback, you will have distortion.
Attached, bandwidth curves of the same amplifier took inside the loop (input of the second stage). First image in a voltage feedback configuration (222V/µs, 1Mhz) second in current feedback (1200V/µs, 5Mhz).
As you can see, we are far to be flat in the audio bandwitch, even in the second case.
My religion is just the product of studies and experiences.
Last edited:
I do believe you, but you seem to be talking about "some" amps, not the NCORE. whereas I'm trying to say that with 0.0003 % THD at 20kHz, 20W, I find it very difficult to believe that awful things happen with real music signals and I would need to see some hard data in order to change my mind. you mentioned the AES study and some personal measurements of yours.
Julf, i don't see your point. Level of feedback is just a change in degree of magnitude.
Some remarks: Harmonic distortion is not so important, as it only change (a little) the tonal texture of reproduced instruments. The Evil is IM, witch add uncorrelated signals (that you can consider as additional noises) that blur the image. The evil is instant dynamic gain errors that kill presences.
Static distortion is not so important, as our brain is able to filter them. Dynamic distortion is.
I do not want to argue endless on the same subjects. I just wanted to share a little of my experience, bringing some elements to your reflexions.
Now, when you design amps, you live in real world. Where all components are evil. Conductance of active devices varies with temperature, witch is a function of the level, both global and instant. Even the shortest printed board tracks is an inductance and add delays. Caps are slow to deliver their currents... Even resistances are not linear with temperatures and currents.
You spend a lot of time to try and listen, fighting against known and unknown evils.
Amounts of global of feedback ? You will try to manage the best compromise between locals and global feedback in a given situation. And each different solution will give-you audible different behavior: Chose your poison.
Global feedback evils ? it is not a question of religion (again) but price and industry. You can design devices with no global feedbacks, with expensive and sorted components. this is not the best track for an industrial product where it is more accurate to work on schematics with average components.
Both can give good results with clever design.
What about Bruno ? He is brilliant in pole calculations and achieved the nice idea to include the output lowpass filter in the feedback loop of its class D amps, in order to cancel more their distortions. Did he pretend to own any definitive conclusions ? I don't think so. I don't fully agree with the paper you refer. He is true in its analyzes from the point of view he consider, witch is only part of the total landscape. (Nobody mastered the whole landscape)
The subject was some were complaining about lack of definition in Class D amps. Real or imaginary, class D amps have their own limitation due to the speed of the power devices (switching frequency).
I just wanted to clarify some amplifiers behaviors.
There is nothing wrong with class D (on the contrary). They don't have anything you can refer as: "Class D amps sound that way". It was the same kind of wrong rumors printed at the beginning of digital.
Some remarks: Harmonic distortion is not so important, as it only change (a little) the tonal texture of reproduced instruments. The Evil is IM, witch add uncorrelated signals (that you can consider as additional noises) that blur the image. The evil is instant dynamic gain errors that kill presences.
Static distortion is not so important, as our brain is able to filter them. Dynamic distortion is.
I do not want to argue endless on the same subjects. I just wanted to share a little of my experience, bringing some elements to your reflexions.
Now, when you design amps, you live in real world. Where all components are evil. Conductance of active devices varies with temperature, witch is a function of the level, both global and instant. Even the shortest printed board tracks is an inductance and add delays. Caps are slow to deliver their currents... Even resistances are not linear with temperatures and currents.
You spend a lot of time to try and listen, fighting against known and unknown evils.
Amounts of global of feedback ? You will try to manage the best compromise between locals and global feedback in a given situation. And each different solution will give-you audible different behavior: Chose your poison.
Global feedback evils ? it is not a question of religion (again) but price and industry. You can design devices with no global feedbacks, with expensive and sorted components. this is not the best track for an industrial product where it is more accurate to work on schematics with average components.
Both can give good results with clever design.
What about Bruno ? He is brilliant in pole calculations and achieved the nice idea to include the output lowpass filter in the feedback loop of its class D amps, in order to cancel more their distortions. Did he pretend to own any definitive conclusions ? I don't think so. I don't fully agree with the paper you refer. He is true in its analyzes from the point of view he consider, witch is only part of the total landscape. (Nobody mastered the whole landscape)
The subject was some were complaining about lack of definition in Class D amps. Real or imaginary, class D amps have their own limitation due to the speed of the power devices (switching frequency).
I just wanted to clarify some amplifiers behaviors.
There is nothing wrong with class D (on the contrary). They don't have anything you can refer as: "Class D amps sound that way". It was the same kind of wrong rumors printed at the beginning of digital.
Esperado, I do see your point, in a way.
but try to view it from this perspective. in the 70s and 80s there have been quite a few studies along the lines of "better distortion measurements which correlate with subjective perception". for instace there's the Belcher test which consists of measuring distortion with two added MLS signals of harmonically unrelated base frequencies (the spectrum of a MLS signal is a "comb"). it is said that such a signal mimics many of the characteristics of music signals. they say the intermodulation products give a better picture of what's going on when you listen to music, as distortion producs are hard(er) to measure with a music signal.
but what baffles me with all these studies is that they fail to address a major point: any amplifier with low distortion will get "higher grades" at these tests. it's as if miraculously, for instance the NCORE will have higher distortion in, say, the test described above, while, a "high end" amp of high distortion will pass it with flying colors. which is not the case. why all this wasted energy to devise these tests when the overall target is stated very simply: higher linearity.
I still can't understand why you refrain from giving us some hard data, like TIM measurements done under realistic conditions.
but try to view it from this perspective. in the 70s and 80s there have been quite a few studies along the lines of "better distortion measurements which correlate with subjective perception". for instace there's the Belcher test which consists of measuring distortion with two added MLS signals of harmonically unrelated base frequencies (the spectrum of a MLS signal is a "comb"). it is said that such a signal mimics many of the characteristics of music signals. they say the intermodulation products give a better picture of what's going on when you listen to music, as distortion producs are hard(er) to measure with a music signal.
but what baffles me with all these studies is that they fail to address a major point: any amplifier with low distortion will get "higher grades" at these tests. it's as if miraculously, for instance the NCORE will have higher distortion in, say, the test described above, while, a "high end" amp of high distortion will pass it with flying colors. which is not the case. why all this wasted energy to devise these tests when the overall target is stated very simply: higher linearity.
I still can't understand why you refrain from giving us some hard data, like TIM measurements done under realistic conditions.
Last edited:
There is several reasons:
1- I'm lazy and not paid for that.
2- I'm retired and no expensive measurement instruments i used in my working places were offered as a farewell gift. I just work now with my ears, a meter and an oscilloscope.
3- If we used measurements instruments, sometimes, to verify our work progresses the right direction, we mostly used them to provide marvelous numbers about a specific product to the marketting office. It is an other work to write white papers. We used our ears for the real thing, like all of you, i presume. They are always plugged in and often more accurate than measurement instruments.
5- I tried several time on this forum to provide simulations graphs (as i've done under) to demonstrate some theory: each time, i have controversial reactions, like "your models are wrong", "this does not work" etc... i'm very tired of those negative and useless reactions.
6- Up to you to make your own simulations to verify BY YOURSELF any audio aspect you are interested in.
When i was a sound engineer and some colleague told-me enthusiastic comments about any hype new equipment, i was not believing-it: just i asked for a demo or test period in my studios to make my own opinion. A good way to stay away from stupid things like Aphex, or so many poor sounding mixing desks considered at a time like a must have ;-)
About measurement instruments, it is so easy, nowadays, to digitally record and compare music samples (input/output from an audio device).
Last edited:
You are just following an other typical pattern that i will not describe to not be as disagreeable as you are.
I provided original data, that you'll not see everywhere: Where are your comments ? Did -you understood what those two pictures means ? It is sims, so, just calculations: What best to demonstrate a theory ?
Where is YOUR DATAs to demonstrate the contrary of my so calling 'asserts' ?
What Issue is under the carpet ? Like this guy pretending in an other thread that one of my amps, yet build by several other people and that i was listening while reading its post, cannot be stable ? What do you expect i can do, but smile ?
I'm indeed not too much interested by 'words' when they just goes against 50 years of verified and correlated work and experiences in audio design and recording, because laws of physic tend to reproduce themselves. I'm just interested by new ideas or technos i can learn or discover before i die. Schematics.
Anyway, i agree with you on is the interest of this discussion ? Any new idea from you i can discover or any original and clever schematic i can enjoy to look-at ?
I was reading this thread, expecting entries from Bruno, as i'm exited (and not experienced) in class D design. He cleverly stay away from those sterile discussions.
By habit, we agree on most of the things, between audio designers with similar experience. And i agree with Bruno's work, as well as good deigners on this forum, like Lazy Cat, sharing its work and experiences.
Last edited:
To resume my 'asserts' in this discussion, because it seems they were not clear to everybody:
Ncore are class D amps using analog global feedback. Rules of servo systems applies as well as in a pure analog amplifier.
An amp will not be perfect on all the audio bandwidth if the delays in the feedback loop are not low enough to provide a flat response inside the loop (flat open loop bandwidth).
Even fastest analog amps are not fast enough to fulfill this requisite, and with the same power devices, a class D amp will present a /2 cutting frequency comparing to an analog equivalent.
This mean an analog amp *can* be slightly better in the upper frequencies. But more expensive, for sure.
On an subjective point of view, Ncores are among the best class D amps i had listened to, with one other using a combination of class D and analog class A.
And do not suffer from any special evil.
Harmonic distortion measurements do not tell much about how sound an amplifier, so i do not wanted to comment those numbers.
What the hell ?
Ncore are class D amps using analog global feedback. Rules of servo systems applies as well as in a pure analog amplifier.
An amp will not be perfect on all the audio bandwidth if the delays in the feedback loop are not low enough to provide a flat response inside the loop (flat open loop bandwidth).
Even fastest analog amps are not fast enough to fulfill this requisite, and with the same power devices, a class D amp will present a /2 cutting frequency comparing to an analog equivalent.
This mean an analog amp *can* be slightly better in the upper frequencies. But more expensive, for sure.
On an subjective point of view, Ncores are among the best class D amps i had listened to, with one other using a combination of class D and analog class A.
And do not suffer from any special evil.
Harmonic distortion measurements do not tell much about how sound an amplifier, so i do not wanted to comment those numbers.
What the hell ?
Does this apply to NCore frequency response drop of 0.5dB at 20kHz, or is it something else? (I'm not an electric engineer)
I'm not an expert, so I'm not able to answer you. But if you are interested, you can check the some explanation of the KV2Audio philosophy from the George Krampera himself ...
Video 11 - musikmesse & prolight+sound 2011 - Am Stand von KV2 Audio - YouTube
- Status
- Not open for further replies.