|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
76.24.78.116
This is a vexing topic, and one that gets debated hotly. The designers and engineers measure the performance for their given product (amp, speaker, CD/DVD player, turntable setup), but those measurements alone really does not (completely) tell one “how the component” actually sounds, or moreover, how component “x” sounds as part of the overall system.
Each component in a system can be measured, but do the measurements alone tell one enough about what one hears while using it? While measurements tell a LOT about how a piece of gear performs, it does not always easily translate into how it sounds. Examples of this abound, but a generic example is amp topology, where many folks (subjectively) will prefer the sound from a given amp that may not measure as well in a number of areas (take your pick, noise, THD, IMD, etc.) as another amp that measures better overall, but “just does not sound as good”, due to issues such as crossover distortion, excessive amp feedback to obtain a lower distortion measurement, filter networks in Class D amp output stages, output transformers with both tube and solid state amps, etc.
What are the most important aspects of getting your sound reproduction right for your system, and what aspects of playback do you listen for do you think are not addressed by measurements that are readily available?
"What this country needs is a good 5 watt amplifier!" (Paul Klipsch)
Follow Ups:
Our hearing is pretty decent - test tones and pink noise signals give us an accuity of analysis that is easily surpased by measurement equipment - these days. On the other hand - when we play music - live music that is a whole new part of the brain activates - the cortex and "pleasure centers" that provide us an emotional response. And when we play reproduced music - through our systems - even more of the prain activates - allowing us to imagine all kinds of things that are not measureable from those two somewhat correlated left and right signals.
Sure we can measure if some part or all of the signals are out of phase -or delays and level differences that might hint to the placement (or lack there of) in our imagined "space". But by looking at the signal and measuring a portion of it we could not likely distinguish Prokofiev from Pink Floyd - whick our brains can do in less than a portion of a second!
Three most important things in Audio reproduction: Keep the noise levels low, the power high and the room diffuse.
Oh, and read Richard Heyser "A View Through Different Windows". Audio Magazine
(1) We can measure stuff is many different ways
(2) The measurement requires we select a frame of reference from a selection of alternative
(3) Different frames of reference contains the same information but the observations will vary.
Key Quotation: "The prime audio problem ("how can we measure what we hear?" arises when the Flatlander [the tester who has selected a frame of reference] tries to convey measures of fidelity who sees things through a higher dimensional window"
Most ignored quotation, "Perhaps we cannot measure the illusion of sound, but we might be able one day to grasp some of its structural properties, as a perceived higher dimentional experience."
Three most important things in Audio reproduction: Keep the noise levels low, the power high and the room diffuse.
Good insight. Thanks for the thoughts.
There are some objective audio folks who subscribe to what I’ll call the “church of the audio critic”, think a few measurements tell the whole story, and resort to name calling if you do not agree to their point of view. It’s very hard to have a dialogue when there is no room for additional information to be presented.
Obviously, measurements tell you quite a bit, but as someone else pointed out earlier, not all the measurements are listed when reviews are conducted. So, there is only so much one can get out of a review. Additionally, let’s face it: When is the last time you read a negative review that basically stated “don’t bother” listening to a piece of gear.
I think there are measurements that could be made that would capture why a 60 WPC Class A amp sounds better (subjectively) than say a 200 watt Class D amp.
"What this country needs is a good 5 watt amplifier!" (Paul Klipsch)
Try to keep frequency response (in room, at the listening seat) fairly flat and extended, try to retrieve all of the information that is on a recording, etc... But go with what sounds right. And if what sounds right is not what the measurements tell you is "right"? Well, at least you can come back here and say that you tried to be objective.
"He was one of those men who live in poverty so that their lines of questioning may continue." - John Steinbeck
Edits: 05/13/12
Scientific method is based on repeated observation, and observation by more than one observer in more than one setting, just to control bias and situational error. Measurement by instrumentation is only one form of observation, but it becomes the preferred method, because it's quantifiable and more standardized. That makes it more impressive; it doesn't make it "right".
Our "hearing" is really a highly complex combination of body excitation translated by the brain. It's perception, not numerical perfection. Hence, under the best of circumstances each of us "hears" a little differently in exactly the same situation in real life. It turns out that our hearing can distinguish upwards of 10,000 different tones changing in a few femtoseconds - an exquisite combination that exceeds many measuring devices' capacities.
That is our receiving end. That doesn't even begin to deal with the transmitting end of musical communication nor with the intervening acoustical environment, only the former being what is purported to being measured in those graphs and numbers you see published.
All that lined up together defines the individual experience of listening to music, to which you must add the emotional and social context of the individual listener's background expectations.
To put it another way, recently having to learn how to walk again after hip replacement, my physical therapist told me that he could measure the action of individual muscles very accurately, but it didn't measure up to simply watching me walk, which is a highly complex interaction of many muscles all at once.
What is measured in only part of the story, and that's why.
might believe nanoseconds...
Everest's (sp?) "Handbook of Acoustics". Ain't that fast enough? LOL
Thus far in my audio journey, in many instances I seem to prefer components or technologies which measure worse.
My favourite amps tend to be single-ended triodes - to me they sound the most life-like yet they tend to measure much worse than high power solid state amps. The caveat being that of course they do have to be mated to an appropriate speaker.
All of the hardline digital crowd will talk endlessly about how digital has a lower noise floor, higher signal-to-noise-ratio, greater dynamic range etc,etc...yet to my ears the sound of high quality analogue still has the edge over high quality digital in precisely these areas - more effortless, palpable, natural, 3-dimensional, real. Yes digital has improved by leaps and bounds over the past 30 years but perfect sound forever it definitely ain't.
Bottom line is IMO current measurements can tell us if something is way off - for example, ridiculous amounts of distortion, impedance mismatches, grossly skewed frequency response etc...however beyond that I don't really think they mean that much. If listening is what we use the equipment for, the only measuring tools which REALLY matter are those cartilaginous ones on either side of our heads.
I think we have to remember that the measurements we see in magazines and from magazines are typically only a subset of the relevant ones.
Also, that more is known about the interpretation of measurements than most people realize -- for example, there's some good research on the relative audibility of harmonic distortion, variations in on-axis and room response, the audibility of resonances as a function of Q, etc., that most of us are either unfamiliar with, or have no way to apply in practice.
It's very easy to forget too that what looks bad in a graph doesn't necessarily sound bad, and vice-versa.
Even then, the best measurements can't be correlated entirely with subjective experience.
Still, I find that you can predict an a surprising amount about the sound if you're familiar with the technology and the implementation, and have a reasonable suite of measurements. Over the years, patterns emerge, e.g., I've learned to correlate a sharp dropoff in a waterfall plot with the prized detail of electrostatics and other highly revealing transducers.
As to what I prize, I'd have to say neutrality above all else -- the ability to reproduce something that sounds sort of like a live acoustical instrument. But I'd add the caveat that that isn't just a matter of on-axis and polar response, distortion, macrodynamics, and the like. Spatial accuracy is an important part of realism as well. It can be correlated with measurements and design, but often in not-very-obvious ways.
In my own listening, I find that I prefer all-around performers to "one trick ponies" that do some things very well at the expense of others. Compromises must be made, but I think the most convincing results are obtained when they're spread widely, with the usual caveats, e.g., that midrange accuracy is disproportionately important, etc.
To me, acoustics and the limitations of 2 channel stereo are the final frontier, because until we solve the limitations of stereo and reproducing one acoustic inside another, usually smaller space, reproduced sound will always sound canned. But I think most of us can do better than we do with the technology that we have, and agree with Jim that setup and acoustics can make more of a difference to the sound than just about anything.
Sadly, I think there is one thing that does even more damage to sound than bad acoustics -- the recording industry. As long as we get close- and multimiked, equalized, compressed and limited crap, we won't be able to realize the potential of our systems. A kid with a stereo mic can make a better recording than most of what you can buy in a store. And I'm talking about classical music here. I don't even want to think about Auto-Tune and the loudness wars.
No degree of high quality equipment will fix a poor recording.
I think the song is the thing.
~~~Our lunacies have been infringed..
I designed amplifiers for about 20 years. Not audio amps, but RF amplifiers - which share more with tube amps than solid state.
There are a lot of simple measurements and more sophisticated ones. If you knew what to measure, I am convinced you could characterize such a device fully enough to predict the "sound" (presentation, colorations, etc). Where the "debate" falls down - is that in lots of reviews a set of basic measurements are made, but rarely does it get to the level of detail and nuance that could fully measure and characterize why, say, a Pass amplifier sounds "different" than a Ayre amplifier.
Where I part ways with so-called "objectivists" is many feel that the simple measurements are enough. And having built and designed amplifiers for several different applications (from 2W to 5kW, from 27MHz to 2GHz, linear, CW, and some feed forward and feed back systems) - that the basic measurements "on offer" are a good start but don't tell the full story.
Many times, we'd have to develop a measurement to characterize some anomoly or other we found and had to understand. I beleive that many companies for Audio electronics do this as well - and like we did - never widely publicized it.
"Knowledge is knowing that a tomato is a fruit. Wisdom is knowing not to put it in a fruit salad"
"Many times, we'd have to develop a measurement to characterize some anomoly or other we found and had to understand. I beleive that many companies for Audio electronics do this as well - and like we did - never widely publicized it."
I bet you're right. It's really tough to make progress without correlated measurements to help reveal and quantify the problems and hopefully your progress. But why tell your competitors? I actually designed consumer electronics for a while, (not home audio!) and once marketing wanted a "white paper" so I did one, but they said it was too technical. So I inquired of a customer relations guy what it was the customer really wanted to know and was told that he wanted to know that the product would work well and last a long time!
That actually made a lot of sense after the shock wore off... The rest is merely buzz words for most folks and any will do. And that explains why my preamp sounds better than yours, mine has millimeter technology!
Rick
Good post. Thanks for your insight.
One issue that I wonder about is Class A amp sound vs. any other topology. Class A designs (using Nelson Pass as a example) sounds subjectively better than Class AB. I took a Threshold S300 (150 wpc AB), re-configured to a SA/3 (50 wpc Class A), and it sounds subjectively better. Not sure what measurement captures this.
"What this country needs is a good 5 watt amplifier!" (Paul Klipsch)
Class A does not have the crossover distortion of AB. Designers use feedback (local or global) to eliminate that form of distortion - but since feedback takes some time in order to come into effect - you get some "open loop" crossover distortion in a transient that a simple measurement won't appear - but may be very apparent in the time domain. The sound will be DE-sweetened vs a class A. That is a rather large effect.
There are other effects surrounding things like device parameters (input capacitances, diode effects, etc. depending upon the device used) that behave much better with class A than AB.
ALso there are a couple of ways of handling the bias of class A. If the bias is "open loop" then on peaks it can break into AB (though the gain can drop a bit unless you wrap feedback around it and play games with bias level) which can affect the sound. You can also force a particular bias on a transistor "come Hell or high water" which I personally prefer - but it has its own problems, too - but this will affect the sound of Class A and whcy some class A stuff doens't sound like others.
Another BIG factor is how the power supply is bypassed. This is a dirtly little secret on how electronics manufacturers get their "house sound" - the amount and value of capacitors on the power supply lines will have a huge impact on the sound.
Nelson Pass does it one way, David Belles does it another way. When you measure using static tones you will see they all measure the same, but in the time domain you will see very small differences at different frequencies that give a particular amp a "warm" or "cold" or even "detailed" vs "musical" ... all good stuff!
"Knowledge is knowing that a tomato is a fruit. Wisdom is knowing not to put it in a fruit salad"
" but since feedback takes some time in order to come into effect - you get some "open loop" crossover distortion in a transient that a simple measurement won't appear - but may be very apparent in the time domain."
So exactly how much time does it take feedback to come into effect?
Our understanding of feedback and control theory and application is obviously quite different.
Alternatively, one might posit that the crossover distortion is not well served by a single feedback signal derived from unmatched paralleled output devices. Too much variance. Obviously, a real class A amp doesn't have to address crossover distortion, you sidestep the problem. Class AB crossover distortion can also be minimized ala Self's 'Blameless' implementation Alternatively, one can 'simply' match each rail's output devices by Vbe and hFE. It gets exponentially more difficult to match these parameter's with an arbitrary range, say 2% as the number of output transistors increase. Anecdotally, my 2 favorite amps are the AKSA Lifeforce and Bel 1001 (yes, the Bel is AB, regardless of the marketing literature), both with closely matched output devices. I'm simply positing, not asserting, but at least I'm not asserting that feedback theory is wrong.
FWIW
Great feedback, thanks Bromo33333. Over the years, the only amps I have kept are Nelson Pass Class units. I still have a Threshold 400A, three SA/3 for the HT, and a XA 30.5 for the two channel.
"What this country needs is a good 5 watt amplifier!" (Paul Klipsch)
The problem is that we are trying to make objective measurements for what is basically a subjective experience. We want an emotional connection to the music. How can you measure that. We do not even fully understand what is going on in our brain that makes us for example sad when we hear a certain piece of music. You must listen to equipment to determine what it sounds like to you and measurements cannot predict that.
Alan
Have you ever read Meyer's Emotion and Meaning in Music? It's been years and I confess I don't remember all the details, but at the time I read it, it was the best take I'd seen on why music can have an emotional effect on us. (His later paper on music and information theory is also a classic.)
What are the most important aspects of getting your sound reproduction right for your system, and what aspects of playback do you listen for do you think are not addressed by measurements that are readily available?In my totally biased - although experienced - opinion, worrying about how electrons move through circuits and wires is totally inconsequential when compared to getting the acoustic wave-launch into your room right and your reception of it intact at your seat.
Once that has been accomplished, adressing how electrons flow can be productive - in fact, infinitely more so, since you will be able to fully appreciate any differences.
Best regards,
Jim Smith
Edits: 05/12/12
I think there is much truth in what you say, Jim.
"What this country needs is a good 5 watt amplifier!" (Paul Klipsch)
I can only hear what I hear.
Observe, don't think
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: