Every competent designer knows that a high C cable can interfere with a high inductance MM cartridge, by potentially creating resonances. 2m of PE or PTFE dielectric coax can have 150-200pF (depending on the size) which, with a high inductance (like 0.5H), can resonate in the audio band. Not so for MC cartridges.
The trouble starts when people are claiming that zip cords of the same RLC sound different (and generally worse) from boutique speaker cables. Not to mention the >1,000 squid power cable sound.
An other analogy for the need, or not, of very high maximal slew-rates :
is a car able to reach a speed of 300 km/h any better or useful than a standard car on a road where speed is limited either by law or because it is very winding ? The standard car may even have some advantages, among them probably more confortable and spacious.
You perfectly indicate the point : the sound of a cable alone can't have any meaning. A cable belongs to a whole circuit with a source (phono or power amplifier) and a receiver (preamplifier or loudspeaker) and they all interact. It is on the whole circuit that investigations have to be done.
For example, an almost unique study :
Capacitor Sounds, Speaker Cables and Crossover Inductors.
Just a reminder that ALL reliable Double Blind Listening Tests bla bla conducted last Millenium (including some of mine) AND this Millenium show clear preference for Bandwidth limited to 20kHz or lower.
This is for true golden pinnae who can reliably tell the difference. If you conduct these tests properly, you'll find loadsa Golden Pinnae who claim to be able to tell the difference but can't ... ie they are deaf.
_______________
A possible explanation is that small signal bandwidth is a measure of what is (demanded and an amp's slew rate is a measure of its capability of meeting this demand. ie if the closed loop bandwidth is 200kHz, this demands 10x the slew rate of one that has a bandwidth of 20kHz.
I don't actually subscribe to this view but would love to investigate it sometime.
But if you want to make something that is guaranteed to make your system sound better, make a linear sharp cut 20kHz filter to put before your amp input.
_______________
One of my Holy Grail quests in Power Amp design is to incorporate severe Bandwidth Limiting in the feedback design of the amp without compromising stability, HF THD & slew.
Maybe I should drop HF THD & slew and just have stability. 😱 Stability is another thing that is reliably preferred in DBLT bla bla 🙂
As I concluded,fewer gain stages, low global feedback, and flat thd for the whole audio range almost every time sounds great. If you see distortion rising over 1 or 5KHz as usual , get away from it... Bandwidth and slew rate should be reasonably high but no insane numbers get you great sound for it self.😉
This correlate my feelings... My enclosures cut at 16KHz, any attempt to add a tweeter were a failure.
I have carefully tuned a low pass filter on my amp with benefits...
Slew-rate is something else... or may-be have side effects...
While limited bandwidth mostly remove distortion products and help the sound to be more 'natural', it seems fast amps are better for 'details', just a feeling.
You know, like high sample rates are better for basses... Fast servo ?
I don't care for slew rate when no feedback...
If you insert, ahead of a very low distorsion amp, a circuit introducing the amount of distorsion required so that it buries the distorsion rising of the amp and so, get a flat THD across the audio range, will it sonically differ from an amp conceived so that it has a flat, but higher than the first one, THD ?
What should be a reasonably high maximal slew-rate is still my question to all the CFA promoters. Since the beginning of this thread, I did not get anything ressembling an answer else than "very high".
Lower than any of the CFA fans would like to admit. Slew rate is just another mechanism that creates distortions, here TIM distortions. Tell me how much TIM you can take and I'll tell you, more or less, the required slew rate. IMO, 1V/uS for each volt of output is already way under any audible threshold. The rest of arguments ("fist cycle distortions", "feedback chasing it's tail", confusion between phase shift and delay, group delay vs. phase delay, etc...) belong to the common, but audio specific, nonsense.
It's funny how some claim that 800V/uS has a SQ impact in a 100W/8ohm power amp, while praising the National/TI audio LME49710/LM4562 opamps for the "good sounding" at 20V/uS.
Pissst -- I already said- using same CMA circuit.... with thd+n level agreed upon is so low as to be Blameless -- and vary the SR (not SR limited) and listen.
[no I dont know how you will do that exactly in the circuit while keeping the HD at Blameless levels] Seems do-able though... this group, I'd bet a dollar, will figure it out.
So far as I can recall no one has done this... just made assumption using phono cartridge BW or 20KHz BW to guess what minimum SR might need to be. That is not the same thing as what is needed in terms of SR of the amplifier to make it also Blameless. Its just a marker, perhaps. Or not.
Thx-RNMarsh
[no I dont know how you will do that exactly in the circuit while keeping the HD at Blameless levels] Seems do-able though... this group, I'd bet a dollar, will figure it out.
So far as I can recall no one has done this... just made assumption using phono cartridge BW or 20KHz BW to guess what minimum SR might need to be. That is not the same thing as what is needed in terms of SR of the amplifier to make it also Blameless. Its just a marker, perhaps. Or not.
Thx-RNMarsh
Last edited:
Your LME49710/LM4562 opamps will be used at around 1V eff, output playing full power audio in an ordinary preamp. Means 2.8V peak to peak.
So, a 100 Watts 8 Ohms audio amplifier (28veff, 80V peak to peak) will ask 80/2.8*20 pp at full level for the same slope. This makes ~570V/µs.
An other way to figure out is to to multiply the preamp slew rate by the gain of the amplifier, usually 22.3.
20V/µs*20 = 445Vµs.
Surprise !
For obvious reasons, we will prefer each stage to be faster than the previous one. So: > 445Vµs.
This controversy is, once more, just boring. Asking what is the "reasonably high maximal slew-rate" is as clever as asking what is the *reasonably low maximal harmonic distortion*.
The only answer you can get is: "The best you can !" or "who knows ?".
Last edited:
20V/µs x 22.3 = ~445Vµs. Sorry for the typo
As everybody is free to design any amp he want with any characteristic or topology he prefers, can-we continue to explore what we like in peace without to be fired on our choices at each post ?
As everybody is free to design any amp he want with any characteristic or topology he prefers, can-we continue to explore what we like in peace without to be fired on our choices at each post ?
Last edited:
This is one of the most vulgar error about the mechanism of negative feedback and low-pass filters.
Esperado said:
Saying why will be welcome.
Esperado said:
A very low harmonic distortion means a preservation of the distribution of all the harmonic components of the input signal.
A very high slew-rate alone is not a guarantee of this.
However... to get very low distortion over all the audio bandwidth, it is well known that a reasonably high maximal slew-rate is necessary.
But... if it is the highest slew-rate which is desired, high currents in the circuit are required, the overall linearity suffers and compromises the aim of getting the lowest harmonic distortion. So the concept of excessive slew-rates emerges which, I am sorry to say, will probably sound very unpleasing to some.
I saw your circuit once, I remember it was very interesting but I can't find it any more. Where is it ?Bimo said:
Everybody is entitled to a choice, including yourself, but this doesn't make your calculations less absurd. For clues, you may want to look into the "full power bandwidth" concept.
Sorry for missing the obvious reasons, care to educate me as of why? And who would be "we", the CFA church members?
Last edited:
Oh, so sorry, talking about "absurd", and because you pontificated about slew-rate, i thought you knew the meaning of this measurement.
From Wikipedia: "slew rate is defined as the maximum rate of change of output voltage per unit of time".
Ask somebody to explain-you why, because time does not change, to reproduce the same way the same ideal square wave, the slew rate has to be proportional to the ...Output voltage.
From Wikipedia: "slew rate is defined as the maximum rate of change of output voltage per unit of time".
Ask somebody to explain-you why, because time does not change, to reproduce the same way the same ideal square wave, the slew rate has to be proportional to the ...Output voltage.
Last edited:
Have you really read and understood Mr. Cordell's otherwise excellent paper, very actual and relevant for the year of 1980 when it was published?
Now, after reading the above, care to explain:
1. Why CFA's with hundreds of V/uS slew rate are required for "good sound"?
2. What are the obvious reasons beyond "we will prefer each stage to be faster than the previous one"?
3. Who is "we" above?
Last edited:
I do not care so much about others opinions or calculations of what 'ought' to or 'should' be good enough for preventing TIM distortion. This is about what qualities the CMA can bring to the table. And, if higher than OK SR produces any other benefit - measureable or audible - direct (like SR) or as secondary/indirect. Or, the whole combination together.
Beyond that, there is the never ending issues of what choices to make when designing. Each can decide that for themselves without judgement from others about thier choice. Looking at it from a black-box point of view.
CFA with hundreds of v/us have not been stated as Required. thats a straw-man set up. If you read the fine print here - we want to accentuate that characteristics to see if it does matter and Then another threshold might appear other than minimum levels 12v/usec (or what ever) to prevent TIM distortion. If it becomes a point of preference to have a higher than 'OK' amount via listening - then that is interesting also. And, would like to know why if possible... otherwise, it becomes another 'marker'. But, I am sure not every single person on the planet will agree... they never do. No arguing about it is useful... just state your own position/facts and let others do their own thinking and drawing of conclusions.
Like my friend likes to say -- if you know all the answers, then there is nothing new for you to learn.
Thx-RNMarsh
Beyond that, there is the never ending issues of what choices to make when designing. Each can decide that for themselves without judgement from others about thier choice. Looking at it from a black-box point of view.
CFA with hundreds of v/us have not been stated as Required. thats a straw-man set up. If you read the fine print here - we want to accentuate that characteristics to see if it does matter and Then another threshold might appear other than minimum levels 12v/usec (or what ever) to prevent TIM distortion. If it becomes a point of preference to have a higher than 'OK' amount via listening - then that is interesting also. And, would like to know why if possible... otherwise, it becomes another 'marker'. But, I am sure not every single person on the planet will agree... they never do. No arguing about it is useful... just state your own position/facts and let others do their own thinking and drawing of conclusions.
Like my friend likes to say -- if you know all the answers, then there is nothing new for you to learn.
Thx-RNMarsh
Last edited:
One thing we have to remember, i think, is our amplifiers are not playing music with resistive loads.
Each electro-acoustic voltage/current to air pressure transducers have inertias, resonances,"force contre éléctromotrice"* and acts as microphones even with rooms resonances. Those will be added to the 'pure' output signal before to be subtracted with the input original signal, creating more complex signals.
Looking at the VAS input, it is easy to see the difference playing the same sample in the same amplifier loaded with a resistance or a speaker.
Not to forget the EMI/RFI returning by the speaker's cables (I use shielded ones).
This, may-be, can explain why some can feel a positive difference with more than the minimal theoretical requested SR = 2πF x Vpk.
But there is more than this. I remember how, at the time some mixing desk were full of 741, it was difficult to get mixages not too muddy, while each track was nice in solo.
As all this together is too complex to be modelized, i prefer to fear on my ears and experiences in this matter.
Nobody will be able to erase my decades of sweat and efforts to produce better sounding music with argues or numbers from school books, neither to sell-me snake oil... that i never sold on my side.
*What is the word in English ?
Each electro-acoustic voltage/current to air pressure transducers have inertias, resonances,"force contre éléctromotrice"* and acts as microphones even with rooms resonances. Those will be added to the 'pure' output signal before to be subtracted with the input original signal, creating more complex signals.
Looking at the VAS input, it is easy to see the difference playing the same sample in the same amplifier loaded with a resistance or a speaker.
Not to forget the EMI/RFI returning by the speaker's cables (I use shielded ones).
This, may-be, can explain why some can feel a positive difference with more than the minimal theoretical requested SR = 2πF x Vpk.
But there is more than this. I remember how, at the time some mixing desk were full of 741, it was difficult to get mixages not too muddy, while each track was nice in solo.
As all this together is too complex to be modelized, i prefer to fear on my ears and experiences in this matter.
Nobody will be able to erase my decades of sweat and efforts to produce better sounding music with argues or numbers from school books, neither to sell-me snake oil... that i never sold on my side.
*What is the word in English ?
Last edited: