I've been thinking quite a bit about how to characterize the sound capability of an amplifier and or audio playback system and I've come up with three main categories:
1. soundstaging
2. timbral accuracy
3. PRAT
Now, before any of you jump in with any of the other numerous adjectives that can be used to define the elements of sound, I will tell you why I think these will fall under the three main descriptions above and more importantly, how they correspond to electrical design decisions. Let's set aside the discussion of the interrelationship of all three for the time being and see if the treatise presented below is valid or useful in any way to anyone trying to voice their audio to their taste.
Soundstaging is a product of level accuracy and dynamics. Voices and instruments are placed in the soundscape by their levels and the persistence of this image is mostly a product of dynamic accuracy. I considered that the coherence of the soundstage is also dependent on the PRAT but as I said before, I am trying to start with a simpler taxonomy as the basis for something more comprehensive later.
Timbral accuracy was the easiest in my mind to identify. It is the frequency response reproduced with the most accurate tonal spectrum. Easy to conceptualize, less easy to produce in practice.
PRAT (pace, rhythm and timing) is mostly a function of the slew or speed of the amp. In the CFA vs. VFA rumble thread, I suggested that the greatest slew rate does not necessarily provide the most accurate PRAT so going balls out for amp speed may result in unintended sound artifacts which may or may not be pleasant.
Okay, I'm now ready for the slings and arrows. Have I nailed it or is my simplified characterization only useful to simpletons?
1. soundstaging
2. timbral accuracy
3. PRAT
Now, before any of you jump in with any of the other numerous adjectives that can be used to define the elements of sound, I will tell you why I think these will fall under the three main descriptions above and more importantly, how they correspond to electrical design decisions. Let's set aside the discussion of the interrelationship of all three for the time being and see if the treatise presented below is valid or useful in any way to anyone trying to voice their audio to their taste.
Soundstaging is a product of level accuracy and dynamics. Voices and instruments are placed in the soundscape by their levels and the persistence of this image is mostly a product of dynamic accuracy. I considered that the coherence of the soundstage is also dependent on the PRAT but as I said before, I am trying to start with a simpler taxonomy as the basis for something more comprehensive later.
Timbral accuracy was the easiest in my mind to identify. It is the frequency response reproduced with the most accurate tonal spectrum. Easy to conceptualize, less easy to produce in practice.
PRAT (pace, rhythm and timing) is mostly a function of the slew or speed of the amp. In the CFA vs. VFA rumble thread, I suggested that the greatest slew rate does not necessarily provide the most accurate PRAT so going balls out for amp speed may result in unintended sound artifacts which may or may not be pleasant.
Okay, I'm now ready for the slings and arrows. Have I nailed it or is my simplified characterization only useful to simpletons?
I can tell you how I do it, others may disagree. Electronics are a no-brainer. If the electronic system frequency response is flat, say -3 dB at 5 and 30 kHz, the distortion is low, say under 0.01% over the same range, and the amp has enough current to do the job with whatever speakers are in use, the job is done. You can do better than those numbers, but they're in the ballpark. Anything meeting those criteria will be near indistinguishable. If the load is really screwy you might need to look at stability. If you're doing vinyl the preamp performance needs to be up to snuff as well.
The rest is transducers. What's the input source, how is the speaker designed and voiced and how is the room arranged and treated. That's where the real bang for the buck is.
The rest is transducers. What's the input source, how is the speaker designed and voiced and how is the room arranged and treated. That's where the real bang for the buck is.
Thanks all for the enthusiastic opening of this thread. Ideally, I'd like the discussion to center around amp design decisions that emphasize the three sonic qualities described above but only if we can agree those three qualities are comprehensive descriptions of what we are trying to achieve in sound reproduction. In other words, is there any other sonic attribute that I didn't include or that doesn't fall under one of those three categories?
Conrad Hoffman
When you make your amps, assign the percentage of emphasis you place on the sonic result using those categories such as:
soundstage: 40%
timbral accuracy: 40%
PRAT: 20%
Your build description gives the impression that you aren't afraid of applying negative feedback and the resulting higher harmonics as long as you can get the THD within the range you want. Have I assessed you properly?
hongm
You're right, it is an audio system and all components interact but I'd like the focus on the amp alone for now.
Speedskater
I guess that couldn't hurt. Let's use two popular diy amps as an example. I have a myRefC and a battery powered gainclone which I use with Vandersteen model 3 speakers directly with the LAME digital content on my Toshiba player. Their sonic character to me:
Gainclone
soundstage 28%
timbral accuracy 31%
PRAT 19%
myRefC
soundstage 33%
timbral accuracy 38%
PRAT 17%
I will rate an old big block Threshhold by comparison:
soundstage 14%
timbral accuracy 38%
PRAT 18%
Conrad Hoffman
When you make your amps, assign the percentage of emphasis you place on the sonic result using those categories such as:
soundstage: 40%
timbral accuracy: 40%
PRAT: 20%
Your build description gives the impression that you aren't afraid of applying negative feedback and the resulting higher harmonics as long as you can get the THD within the range you want. Have I assessed you properly?
hongm
You're right, it is an audio system and all components interact but I'd like the focus on the amp alone for now.
Speedskater
I guess that couldn't hurt. Let's use two popular diy amps as an example. I have a myRefC and a battery powered gainclone which I use with Vandersteen model 3 speakers directly with the LAME digital content on my Toshiba player. Their sonic character to me:
Gainclone
soundstage 28%
timbral accuracy 31%
PRAT 19%
myRefC
soundstage 33%
timbral accuracy 38%
PRAT 17%
I will rate an old big block Threshhold by comparison:
soundstage 14%
timbral accuracy 38%
PRAT 18%
Last edited:
Also, you assigned these percentages purely based on subjectivity, and that is going to be a problem. It's not like any of these ratings are calculated using a ratio of some sort. You have too many variables such as the quality of the recording, your playback source, the quality of your DAC, and the accuracy of your speakers. As you can see from the forum threads, any of these factors can trigger a huge debate.
hongrn
Okay, let's assume a distortion analyzer/FFT can show us timbral accuracy and slew shows us PRAT. I don't necessarily agree that they automatically do but I accept they point us closer to the goal. What measures soundstage? Is it even a real attribute if it can't be measured? I hope you see why you can't completely remove subjectivity from preferences but that shouldn't stop us from exploring the practices that can achieve those preferred attributes.
Okay, let's assume a distortion analyzer/FFT can show us timbral accuracy and slew shows us PRAT. I don't necessarily agree that they automatically do but I accept they point us closer to the goal. What measures soundstage? Is it even a real attribute if it can't be measured? I hope you see why you can't completely remove subjectivity from preferences but that shouldn't stop us from exploring the practices that can achieve those preferred attributes.
Well, my amps don't have sonic properties when compared to other well designed amplifiers. I'm not afraid of negative feedback, in fact I think Harold Black was a pretty smart dude. What I've found is that if an amplifier sounds different from others, be it extra good or just bad, there's something wrong with it. Sometimes it's something obvious, found with the usual measurements on the test bench. Sometimes it's something subtle like RF getting in or a slight oscillation at certain amplitudes or with certain loads. Eliminate the design blunders and you eliminate the audiophile tonal descriptions.
It is difficult to have any meaningful discussion before terms are defined and agreed.
Timbral accuracy? WTF is that? Do you mean frequency response? Precious!
PRAT? I could say a lot more, but I will refrain.
Soundstage? An illusion depending on where things are panned by the mixing desk. Great fun but no more.
Voicing! Pretentious nonsense, unless you are dealing with a physical acoustic device like a cello or a loudspeaker.
Last edited:
cliffforest
Welcome to my thread and please, don't hold back
I believe I explained timbral accuracy and it is more than just frequency response, it is the ability to reproduce a waterfall plot of mulitiple simultaneous frequencies. When you hear a trumpet play a note, it is not playing only that frequency. A soundstage very well may be an illusion but it is curious that many people in a room can identify the same location of a sound when an amp is capable of creating one. Not being able to measure it does not mean we give up identifying what creates one.
Conrad Hoffman
There are numerous examples of amps with poor or mediocre THD numbers that sound really good and have some of those three attributes. One has to look no further than Class A amps to verify this. I'm sure you're not suggesting that amps that introduce any of the three qualities above into your audio system are faulty by their excellence but it is curious that you wouldn't strive to achieve those qualities in your designs. What parts do you typically use to build your amps?
Welcome to my thread and please, don't hold back
I believe I explained timbral accuracy and it is more than just frequency response, it is the ability to reproduce a waterfall plot of mulitiple simultaneous frequencies. When you hear a trumpet play a note, it is not playing only that frequency. A soundstage very well may be an illusion but it is curious that many people in a room can identify the same location of a sound when an amp is capable of creating one. Not being able to measure it does not mean we give up identifying what creates one.
Conrad Hoffman
There are numerous examples of amps with poor or mediocre THD numbers that sound really good and have some of those three attributes. One has to look no further than Class A amps to verify this. I'm sure you're not suggesting that amps that introduce any of the three qualities above into your audio system are faulty by their excellence but it is curious that you wouldn't strive to achieve those qualities in your designs. What parts do you typically use to build your amps?
Last edited:
1st would be establishing if your proposed subjective evaluation criteria can be heard, at what levels
http://www.madebydelta.com/imported...l_T4_Perceptual_Audio_Evaluation_Tutorial.pdf
then compare with audio amplifier parameters, design criteria, decide if you are making quality music reproduction system component
or are building an effects box
I have designed scientific and industrial electronics professionally for decades, really like signals and systems, control theory, learning parts types quirks, best application
my audio electronics interest started in uni reading through back issues, eagerly awaiting the next JAES, all of the stereo equipment magazines in the library back when they had amplifier schematics in them
I have occasionally attended BAS, JAES talks, demos over much of that time period too
Own and have read a number of psychoacoustics textbooks, "popularized" audio perception/hearing books, many shelf feet of analog EE design books
my "amp theory" subfolder in my "audio" documents has 1/2 G of pdf of IEEE, JAES papers, manufacturer white papers, app notes, magazine articles...
built my 1st audio power of my own design 35 years ago
and despite all of that I can't see that amplifiers can be designed for your categories of "voices"
in all that time I haven't seen any psychoacoustics literature that would give any credence to your categories being "free variables" in quality audio music reproduction amplifier electronics design
the amplifier's contributions would be in phase and frequency response flatness, matching between channels - which can easily be better than any psychoacoustic thresholds by orders of magnitude
group delay in amplifiers as low pass filters may approach 1 microsecond, differential group delay in audio bandwidth though can be 10s of nanoseconds
those numbers are equivalent to ~ 0.010" air path delay, 1/10 of a thou delta
while speakers, rooms "soundstage effects" are orders of magnitude above psychoacoustic thresholds - time/phase errors in milliseconds, dynamic modulation by 10-100s of microseconds, frequency response in 10s of dB, again modulating by single digit dB with signal for many dynamic multiway loudspeakers
generally considered to be the balance of harmonics amplitude and phase relative to fundamental of notes
again amplifier contributions to this in playback would be phase, frequency response flatness, stability, lack of distortions which is easily orders of magnitude better than human hearing thresholds for competent design electronics
IMD, particularly "TIM" is a popular audiophile meme - with no explanatory power - no psychoacoustic evidence for it being heard any differently than any other IMD (which it is a subset of and is included in "conventional" IMD measurements) and is by numbers a solved problem from Cordell's 1983 MOSFET Amplifier with Error Correction article, TIM analysis 30+ years ago
from lossy codec design to many psychoacoustics textbook documented effects we seem to need ~ 1% SPL change in any critical band for human JND and can't hear the effect of smaller changes
again loudspeakers, room effects, even variability with signal are way bigger
don't even see anyone I disbelieve having a coherent theory/recipe that comes within miles of basic psychoacoustic knowledge
Pace Rhythm and Timing are simply beyond amplifier's "memory" time scale
10s to 100 milliseconds timing errors are needed for your buddies to boot you from a really bad high school garage band as lacking any timing
world class musicians obviously do orders of magnitude better - but still large fractions of milliseconds
not tiny fractions of microseconds
you need DSP and memory to change the literal PRAT of a recording - or a turntable with your thumb on the platter
http://www.madebydelta.com/imported...l_T4_Perceptual_Audio_Evaluation_Tutorial.pdf
then compare with audio amplifier parameters, design criteria, decide if you are making quality music reproduction system component
or are building an effects box
I have designed scientific and industrial electronics professionally for decades, really like signals and systems, control theory, learning parts types quirks, best application
my audio electronics interest started in uni reading through back issues, eagerly awaiting the next JAES, all of the stereo equipment magazines in the library back when they had amplifier schematics in them
I have occasionally attended BAS, JAES talks, demos over much of that time period too
Own and have read a number of psychoacoustics textbooks, "popularized" audio perception/hearing books, many shelf feet of analog EE design books
my "amp theory" subfolder in my "audio" documents has 1/2 G of pdf of IEEE, JAES papers, manufacturer white papers, app notes, magazine articles...
built my 1st audio power of my own design 35 years ago
and despite all of that I can't see that amplifiers can be designed for your categories of "voices"
in all that time I haven't seen any psychoacoustics literature that would give any credence to your categories being "free variables" in quality audio music reproduction amplifier electronics design
the amplifier's contributions would be in phase and frequency response flatness, matching between channels - which can easily be better than any psychoacoustic thresholds by orders of magnitude
group delay in amplifiers as low pass filters may approach 1 microsecond, differential group delay in audio bandwidth though can be 10s of nanoseconds
those numbers are equivalent to ~ 0.010" air path delay, 1/10 of a thou delta
while speakers, rooms "soundstage effects" are orders of magnitude above psychoacoustic thresholds - time/phase errors in milliseconds, dynamic modulation by 10-100s of microseconds, frequency response in 10s of dB, again modulating by single digit dB with signal for many dynamic multiway loudspeakers
generally considered to be the balance of harmonics amplitude and phase relative to fundamental of notes
again amplifier contributions to this in playback would be phase, frequency response flatness, stability, lack of distortions which is easily orders of magnitude better than human hearing thresholds for competent design electronics
IMD, particularly "TIM" is a popular audiophile meme - with no explanatory power - no psychoacoustic evidence for it being heard any differently than any other IMD (which it is a subset of and is included in "conventional" IMD measurements) and is by numbers a solved problem from Cordell's 1983 MOSFET Amplifier with Error Correction article, TIM analysis 30+ years ago
from lossy codec design to many psychoacoustics textbook documented effects we seem to need ~ 1% SPL change in any critical band for human JND and can't hear the effect of smaller changes
again loudspeakers, room effects, even variability with signal are way bigger
don't even see anyone I disbelieve having a coherent theory/recipe that comes within miles of basic psychoacoustic knowledge
Pace Rhythm and Timing are simply beyond amplifier's "memory" time scale
10s to 100 milliseconds timing errors are needed for your buddies to boot you from a really bad high school garage band as lacking any timing
world class musicians obviously do orders of magnitude better - but still large fractions of milliseconds
not tiny fractions of microseconds
you need DSP and memory to change the literal PRAT of a recording - or a turntable with your thumb on the platter
Last edited:
Yes! That is exactly what I'm saying. If it sounds bad, it's bad. If it sounds good, it's bad. The qualities you're after should not be coming from the amplification system.
Once you get on that merry-go-round, there are no standards to design to, and you end up chasing an undefined target pretty much forever, using flawed listening tests.
Yldouright, you won't get the answers with this group of 'critics'. However, let me say that a successful amp design requires a lot of attention to a lot of details, including AC leakage, cap quality, as well as a smooth transfer function. I have made amps that measure almost perfectly and sound bad, then taking them apart and making them sound wonderful. Why, is difficult to measure.
Yes, certainly. I think the more rational reviewers understand this.
When a quality is deservedly lauded, the logical implication is that other components are inferior in that respect,
not that this unit "improves" the sound. Rather, it degrades the sound less.
Last edited:
You're just arguing for the sake of arguing. Why don't you pick a recording (one that the rest of us can download or purchase), a DAC, a pair of speakers (not DIY), a commercial amp, and start defining the criteria you mentioned (PRAT, timbre, etc...). Present your data in an objective manner and the reasons why you assigned the percentages that you did. All you've done so far is to throw out a bunch of terms that make no sense to me. Makes sense?
M Gregg
Thanks for the link and welcome.
hongrm
No I'm not.
rayma
It's hardly civil to call someone irrational just because they include their ears in the arsenal of test equipment.
john curl
Welcome. I have to say I'm probably like many who have spent too much time on the blowtorch thread. Do mosfets deserve their reputation as poor imagers or is it more function of their individual quirkiness and the difficulty in making them behave?
Conrad Hoffman
Please don't get the impression that I don't believe in measuring but too many people are unaware of the realities behind the numbers they read. One has to look no further than the commercial rubbish produced in the 1980's that had remarkable numbers and equally unlistenable sound.
Thanks for the link and welcome.
hongrm
No I'm not.
rayma
It's hardly civil to call someone irrational just because they include their ears in the arsenal of test equipment.
john curl
Welcome. I have to say I'm probably like many who have spent too much time on the blowtorch thread. Do mosfets deserve their reputation as poor imagers or is it more function of their individual quirkiness and the difficulty in making them behave?
Conrad Hoffman
Please don't get the impression that I don't believe in measuring but too many people are unaware of the realities behind the numbers they read. One has to look no further than the commercial rubbish produced in the 1980's that had remarkable numbers and equally unlistenable sound.
You've built a heck of a lot more amps than I have, and so far as the "usual suspects" of measurement go, just response and THD, I'd have to agree, but I've never found a sonic difference that didn't come down to something measurable by one obscure technique or another. Identical signals must sound identical. Non-identical signals might or might not.
- Status
- Not open for further replies.