Voicing an amplifier: general discussion

Status
Not open for further replies.
godfrey, marce
Welcome. All my dictionary editions have both spellings of timbrel and only those pretending to be obtuse would suggest either spelling could mean anything else. Yes, an ideal amp will perfectly reproduce what is recorded but will that really produce the best listening experience given what we know about recordings? I can tell you that my Ohm speakers image nicely with one mike recordings but get lost with recordings where more mikes are used. Nothing was changed in the audio system save the recordings. Change the speakers and the multi-mike recordings improve while the single mike recordings degrade. This was a strong indication to me that the same signal can have different aural results depending on how it is processed by the playback system. I chose to start with the amplifier but all the audio components bend, chip and smooth that signal in their own way. Is it so unreasonable to pursue a method that identifies what does what to the end result so we can learn to get what we want to hear? As for this not being a new concept, please provide a link predating this discussion where someone else dared to present a comprehensive framework attempting to reconcile a sonic result to an electrical characteristic.
 
I can tell you that my Ohm speakers image nicely with one mike recordings but get lost with recordings where more mikes are used. Nothing was changed in the audio system save the recordings. Change the speakers and the multi-mike recordings improve while the single mike recordings degrade. This was a strong indication to me that the same signal can have different aural results depending on how it is processed by the playback system.

Now you're coming close to making sense. Recordings, speakers, and rooms are important and demonstrable factors.
 
DF96
I don't think it's daft to try and reconcile sound reproduction attributes with electrical circuit attributes. In order to do this, we need to establish the sonic attributes first. The three broad categories from page one of this thread seem to describe these sonic attributes comprehensively and succinctly. No one here has offered an additional or alternative category that makes any sense. I am aware that this takes some mental gymnastics that are beyond the abilities and prejudices of some the members here but any new concept that disturbs knowledge hard won through years of effort will encounter this type of ridicule.

JMFahey
I believe timbral is an acceptable spelling of the word timbrel. If there is another meaning attached to it which would create confusion between the two, I apologize. I submit that the electrical signals do indeed represent the music and that to the extent that they can be manipulated, the playback experience of these sonic attributes can be either enhanced or muted to the taste of the listener. This is what I mean by voicing the amplifier. In a prior post you stated that the three postulates presented to describe the sonic output of an audio playback system 'were clearly not comprehensive' but offered no amendment. To me that seems like nay saying for the sake of nay saying.

DrDyna
Like JMFahey above, you also stated that the three postulates presented to describe the sonic output of an audio playback system 'were clearly not comprehensive' and you offered 'amplifier' as a way to describe sound. I may not be the only person who saw a problem with this :)
 
Last edited:
Pano said:
Is it daft to think that amplifiers behave differently at different power levels and with differing loads? If so, why is that a daft idea?
Some weak designs may do that, but what has that got to do with "PRAT" - which is set by the musicians and the recording engineer? To make an amplifier which changes the speed of the music you need a lot more than, say, a few thermal effects. Where in almost any amp circuit is the FIFO buffer needed to change timing?
 
yldouright said:
I don't think it's daft to try and reconcile sound reproduction attributes with electrical circuit attributes. In order to do this, we need to establish the sonic attributes first.
The sonic attributes which matter for sound reproduction are well known: sufficiently wide frequency response (but not too wide), sufficiently low distortion, sufficiently high output power, sufficiently low noise etc. We can argue about what is sufficient in each category, and how to measure it. Many of the threads on here do exactly that.

There are other possible aims for other audio fans, such as '"slam" or "timing" - but these have nothing to do with sound reproduction.

I am aware that this takes some mental gymnastics that are beyond the abilities and prejudices of some the members here but any new concept that disturbs knowledge hard won through years of effort will encounter this type of ridicule.
I admit that it is beyond my ability to imagine how a simple amp circuit, intended by its designer (and purchaser) to merely amplify a relatively undemanding low frequency signal, could change the timing of the music. Maybe I need a paradigm shift? Then I can throw out all the stuff I learnt about Maxwell, Fourier, Nyquist, Kirchoff etc.
 
I admit that it is beyond my ability to imagine how a simple amp circuit, intended by its designer (and purchaser) to merely amplify a relatively undemanding low frequency signal, could change the timing of the music. Maybe I need a paradigm shift? Then I can throw out all the stuff I learnt about Maxwell, Fourier, Nyquist, Kirchoff etc.

Other than the amp exhibiting a horrible amount of group delay, I can't imagine either.... but then to do so, the designer would specifically be attempting to distort the sound.

My rule of thumb is that an amp should be distortion free. The remainder of the equipment can be used to tailor the results. In my case, I want it all distortion free...

Why would I want to second guess how the mastering engineer intended it to sound?
 
DF96
The sonic attributes which matter for sound reproduction are well known: sufficiently wide frequency response (but not too wide), sufficiently low distortion, sufficiently high output power, sufficiently low noise etc. We can argue about what is sufficient in each category, and how to measure it. Many of the threads on here do exactly that.
These can all be viewed as voicing decisions that will either directly or indirectly effect the audible result. When we trade away bandwidth for stability we may also be removing artifacts in the higher frequencies which effect any of the three submitted audio postulates in ways which we have yet to fully understand. When we accept greater THD so we can reduce higher order harmonics, we are once again voicing the amp. Wouldn't it be advantageous to know how the postulates put forth herein are effected by these decisions?
There are other possible aims for other audio fans, such as '"slam" or "timing" - but these have nothing to do with sound reproduction.
And nothing to do with PRaT?
 
for audio amplifiers the high frequency low pass fliter group delay as I already mentioned can be < 1 microsecond, the differential group delay ~ 10s of nanoseconds - no "what ifing" required

Bob's MSOFET Power Amp article used custom built quadrature resolving IMD hardware to give PIM/TIM in Otala's terms of 10s of picoseconds

attachment.php


low frequency corner can be more significant but in many AC coupled audio power amps the corner frequency should be < 2 Hz
DC servo generally use film caps instead of Al Electrolytics if you are suspicious of them, and even 0.2 Hz corner is possible, even 2nd order - Cherry has suggested Bessel high pass if any were concerned by filter group delay in the room modal response region

but really for diy DC coupled with manual trim is quite good enough on offset, and "perfect" on group delay issues
 

Attachments

  • cordell_pim.PNG
    cordell_pim.PNG
    13.6 KB · Views: 346
Last edited:
What if the group delay were bandwidth dependent?

That would be even more broken than just plain ol' broken, such an amplifier having made it to production rather than have the schematics moved to the recycle bin is quite silly.

On a complex and busy piece of music...

Musical signals are not that complicated, comparatively. Like I said before, if a 12 cent integrated circuit can render all 2,073,600 pixels that make up my computer screen and do it flawlessly, for months or years without a single pixel being out of place or be the wrong color, why would we still expect that a basic gain stage, who's design intent was specifically NOT to do those things...still does those things..I guess I just don't get it.

I once had the same opinion that you do, I used to drool over someday having a pair of Krell FPB amplifiers...but you reach a point where you realize...why would I spend 10 grand on an amplifier, when all someone has to do is put a sheet up in front of it and I can't tell it apart from a sam's club AV receiver?
 
DrDyna
Pixels are either off or on, the wiggly lines of analog get abused when being controlled the same way.

jcx
For a few simple frequency tones, I absolutely agree but when the music gets complex and the tones get intertwined at various levels, I don't think you can use that same formula. This is where I believe the subjective nature of objective measuring becomes evident.
 
This discussion is interesting, yldouright, and I think you are on the right track.
With regard to amplifier design and 'voicing': For many decades I was not a believer in 'voicing', but about 20 years ago, I ran up against a real problem. I could make an amp with a good topology, all complementary push-pull, 100V/us, high current available, high voltage swing available, and Class A to the first 10W or so. YET, it failed the listening test, by two of my closest associates at the time. This does not include the earlier situation with another design, where I removed the input IC, to change the design and get it accepted by Stereophile.
What could it be? Could it be fixed? (more later)
 
For a few simple frequency tones, I absolutely agree but when the music gets complex and the tones get intertwined at various levels, I don't think you can use that same formula. This is where I believe the subjective nature of objective measuring becomes evident.
you're quite thoroughly wrong - your suspicions are not born out by measurements

and we do have the measurements of Blecher/noise fill/Fastest complex multitones

they are AP analyzer menu selections today


or you can do it at home - and they are in fact still quite boring for prosumer soundcards, their fairly indifferent op amps by audiophile standards and for low distortion audio power amplifiers designed to Cordell, Self "textbook" standards

DiffMaker if you have a favorite "difficult" music selection


there really is a deep literature, I've personally pointed to much of it on these forums

Cabot, Hofer, have good survey articles

but its hard to beat the depth of: Czerwinski “Multitone Testing of Sound System Components – Some Results and Conclusions, Part 1; History and Theory” JAES V49 #11 Nov 2001 p 1011-1047
 
Last edited:
Status
Not open for further replies.