Our concerns about numbers, harmonics, distortion, square waves, THD, measurements

[myhrrhleine]

Re: Our concerns about numbers, harmonics, distortion, square waves, THD, measurements

Post your ideas about.


Lumba is especially invited.

ear plus brain cannot process that information.... the thing works slowly because bio-chemical....

test your theory...
DAC with 10uS clock jitter

that's 100khz
inaudible, right?

 

[Pingrs]

very few successful audio designers have a duel degree in psychoacoustics

I seem to remember reading that Doug Self is one 🙂

Brian.

 

[Michael Chua]

Originally posted by Bigun

back to photography for a second - people rate lenses on a subjective quality called 'bokeh' which deals with how an image looks in the out of focus areas. Sure it's possible to try and equate this with measurements of spherical aberration but by-and-by people prefer to describe it in words using terms like 'smooth' and 'harsh'. Even a 'bad' lens with a pleasing bokeh will be deemed a good lens by those who 'know'.

I know its off-topic but is that the perception of the general public nowadays? Whatever happened to resolution, contrast, hues, aberrations etc.

 

[Bob Cordell]

What would you say are the limits of these measurements, given the above?



And again how much can you tell from this?

I'm not being contentious here, I'm just interested in the perceived limits of these methods. What important aspects of amplifier behaviour would you say is beyond the capability of these methods to measure?

These are all very fair questions. There are no black/white answers here. And just because I assert that THD is useful, I would never say that it alone can separate reliably how amplifiers will sound. Is this a reason to say THD-20 is worthless and throw it out? No.

I could easily give you two amplifiers both with 20 kHz THD of 0.1% at full power, yet one would sound great and the other might sound awful. The first might have very soft, low-order distortions, the second might have its 0.1% distortion as a result of crossover distortion, one of the most insidious distortions an amplifier can make.

It would be far harder for me to give you two amplifiers each with 0.001% THD-20 and have one sound great and one sound bad, assuming all other variables, like component quality and frequency-response-affecting things were equal.

An amplifier with very low single-number THD-20 has a lot less wiggle room to sound bad because of the nature of its non-linear distortion. Nevertheless, low THD-20 is neither necessary nor sufficient for a given amplifier to sound superb; it just increases the probabilty of good sound if all else is equal.

I fear that many people completely do not understand that THD is not what we hear, per se, but rather a measurable symptom of nonlinearity that may do bad things to the signal that we can hear. This may not be unlike the medical reality that a doctor often diagnoses a person's disease by the presence of certain antibodies.

We often hear people say that THD-20 is meaningless because we presumably can't even hear the second harmonic. The inevitable response will be someone else saying that we indeed can "sense" these higher frequencies. Both are probably wrong, and both miss the point. When we hear distortion, it is often not the harmonics that damage the sound but rather intermodulation among many different frequencies. Some people don't realize that the same nonlinearity that causes THD will cause intermodulation. Both distortion measurements are just indicators of underlying nonlinearity.

There are many aspects of amplifier misbehavior that simple THD measurements will not capture, and there are probably some effects we don't even know how to measure. A simple example of one thing that is not usually captured is effects of power supply rails fluctuating with program material. Another would be output stages going wildly out of proper bias because of program-induced dynamic temperature swings in the output stage. There are many others, but still no reason to throw out THD. The doctor still uses his stethescope, even though it does not tell all.

Cheers,
Bob

 

[ostripper]

perception of the general public nowadays

The general "pubic" 😀 😀 ,now that is a term to reckon with.
1. CHEAP .. If it is on sale at walmart and says "sony" , they
will buy it . (even if they have to buy it again every year
because it is junk).

2. "SHINY".. curvy lines, technical hype (2000W PMPO) ,if it shakes
their flatscreen and they feel air coming out of the port , its
good to go. What it sounds like playing music... they can
live with that when it's home..

3. LOUD.. this is the general public definition of "Hi - Fi"
if it is loud it must be good.. like the bass units in the "gangsta"
cars. BOOM .... Bzzzzt, BOOM.. Bzzzzt (I can hear the 30%
45hz distortion out the window now 😡 )

What a sad testament to the state of consumer ignorance
in the last days of the empire..
OS

 

[jkeny]

Bob,
I'm not suggesting throwing out THD but refining it (or another measure) to get to more of a direct correlation between the measurements & the sound. I know we have spectral waterfalls,etc but apart from a weakish agreement that a decreasing cascade from 2nd through to higher order harmonics is a positive thing, I don't think we have advanced much further than 20 years ago in our correlation system? Have we?

That really was my question - have we got any further agreement on correlating measurements with sonics?

Given this impasse, then it seems to make sense to try approach the subject from a different angle and my take on that angle is psychoacoustics which will hopefully lead us back to a better measurement system with correlation to what we hear.

 

[Rafael L]

A simple example of one thing that is not usually captured is effects of power supply rails fluctuating with program material.

I believe that talking about linear distortion, I am sure?
Low regulation of the transformer, decreases the bass amplitude, the audio seems compressed equal to a radio of batteries...

 

[scott wurcer]

Re: Re: Our concerns about numbers, harmonics, distortion, square waves, THD, measurements

test your theory...
DAC with 10uS clock jitter

that's 100khz
inaudible, right?

No actually, on both points. The peak deviation tells nothing of the jitter spectrum, you have your domains mixed up.

 

[Bigun]

I know its off-topic but is that the perception of the general public nowadays? Whatever happened to resolution, contrast, hues, aberrations etc.

I'm afraid the general public are only interested in long zoom range and auto-focus 🙁

 

[CBS240]

...... When we hear distortion, it is often not the harmonics that damage the sound but rather intermodulation among many different frequencies. Some people don't realize that the same nonlinearity that causes THD will cause intermodulation. Both distortion measurements are just indicators of underlying nonlinearity.

There are many aspects of amplifier misbehavior that simple THD measurements will not capture, and there are probably some effects we don't even know how to measure. A simple example of one thing that is not usually captured is effects of power supply rails fluctuating with program material. Another would be output stages going wildly out of proper bias because of program-induced dynamic temperature swings in the output stage. There are many others, but still no reason to throw out THD. The doctor still uses his stethescope, even though it does not tell all.

Cheers,
Bob

I think this 'technical' argument may help to support the 'subjective' campers' view that some amps with similar THD figures sound better.

3. LOUD.. this is the general public definition of "Hi - Fi"
if it is loud it must be good.. like the bass units in the "gangsta"
cars. BOOM .... Bzzzzt, BOOM.. Bzzzzt (I can hear the 30%
45hz distortion out the window now 😡 )

What a sad testament to the state of consumer ignorance
in the last days of the empire..
OS

If you can make money selling such cr@p to fools, and it makes them happy, what's the problem?..........other than the nuisance to the rest of us who they seem to think need to listen to that garbage.😡 But I don't blame the people who make and sell the stuff.

It's amazing to me a lot of these people build these immense current sucking stereos in thier cars yet only have a stock 40-50A alternator to power it. Way back in high school, (before class D amps were popular) it took a friend of mine 2 alternator failures to realize the common sense I tried to explain to him

 

[Salas]

We often hear people say that THD-20 is meaningless because we presumably can't even hear the second harmonic. The inevitable response will be someone else saying that we indeed can "sense" these higher frequencies. Both are probably wrong, and both miss the point. When we hear distortion, it is often not the harmonics that damage the sound but rather intermodulation among many different frequencies. Some people don't realize that the same nonlinearity that causes THD will cause intermodulation. Both distortion measurements are just indicators of underlying nonlinearity.

Cheers,
Bob

Your explanation makes sense. We could not tell an amp from another no matter their THD difference if it was only for the amplitude of its 2nd or 3rd etc. Simply because we listen to them through speakers that produce much more.

 

[ostripper]

By cbs240-If you can make money selling such **** to fools, and it makes them happy, what's the problem?..........other than the nuisance to the rest of us who they seem to think need to listen to that garbage. But I don't blame the people who make and sell the stuff.

I actually do make money from them (and shake my head afterwards as I count the $$$) 🙂

I fill their trunks with extra batteries (no more burnt alternators)
and 1 farad "boom caps" ha ha ... but they come in and
hear my HT sub and wonder why the autoexpress stuff
does not sound the same..

There are many aspects of amplifier misbehavior that simple THD measurements will not capture, and there are probably some effects we don't even know how to measure. A simple example of one thing that is not usually captured is effects of power supply rails fluctuating with program material. Another would be output stages going wildly out of proper bias because of program-induced dynamic temperature swings in the output stage. There are many others, but still no reason to throw out THD. The doctor still uses his stethescope, even though it does not tell all.

The supply aspect was one that came with "hairpulling" experience. Simulation did not foretell this, had to build it
and Decoupling , rail traces on PCB/ground returns,
all make a world of difference. I wonder under what conditions
the OEM's rate distortion ?...
with the amp cold ,warmed up and thermally
stabilized , or in a state of extreme thermal stress. (party duty)
I suppose they just go for the "ideal" to get the best numbers. 🙁
OS

 

[mosfets]

Back in the 80's

...I purchased an amp once and it had a great THD spec of 0.001 at full rated output and it was the worst sounding piece of junk I (and all of my friends) ever heard. The amp had a strong market value based on its name and model number so I quickly recovered most of my money.

Shortly there after, I purchased an amp with 0.1% THD, and then I discovered that, published specs mean nothing unless you know allot about the people that publish them. Who are these people? What and why do they want to list these "specifications" about their equipment? Very Important.

Great topic by the way!

For ME...all of this is semantics. Now I feel listening to music is more important than absolute detail and some times fine details are so enjoyable but more often than not it is the passion or energy in the music that I continue to enjoy year after year.

I love listening to great sonic recordings that reveal every weakness & strength in my Hi-Fi.BUT I most always enjoy listening to old scratchy vinyl records that contain emotion that can make me cry or whatever….you know what I mean?

 

[destroyer X]

dear Bear....we left to the acoustic guys the problem since long time ago

we have entered the measurement systems and we left them free to do nothing about... speakers are awfull... air pump...ridiculous and having the same technology we had 100 years ago.

Italian Suonno magazine, down the eigthies, have published pictures from speaker diafragm in motion (high speed film or photography showing speaker reproducing music... a scandall)...my God!... a tragedy..of course this was producing noises together sound, as the cones were so strange twisted the coil should touch the magnetic iron core!.... and i said to myself swetting a lot, so nervous i was.... oh dear!... we are searching for troubles into amplifiers...the wrong thing to worry about!

"os diafragmas estavam absurdamente deformados, entortados de uma forma alarmante!"

As we turn satisfied with our measurement system and we have entered deeply into those things, ignoring how the speaker will disturb, them they have survived (speaker factories, acoustic guys, acoustic engineers) doing almost the same all those years...we have never made some pressure on them...we never have asked them to make better work, to research for better transducers to produce sound..all they made was to change diafragm material from paper to plastic, or aluminium or silk and paper.... they have increased the magnet, also some of them reduced the coil weigth using square section of aluminium and copper wires...well...they did some but was a make up from the old air pump we had in 1900... the main difference is the permanent magnet when in the past you had a transformer to magnetize constantly a big piece of iron.

But we used to stay inside our own world, developing measurements and fashion moments arrived...linearity, and them intermodulation distortion, and them harmonic distortion..and transient.... and some analisis were created...and we were reducing and reducing numbers..while speakers were there hucking the sound... we were cleaning the sound sent to speaker and they were destroying the quality using those ridiculous air pumps (speakers)

This is the problem dear Bear.... while we left acoustic to the acoustic enginneers..while we were dreaming with a perfect speaker they remain the same...no pressure on them..we were very satisfied with our delirant graphics and numbers....all that fooling us...because speaker was there..remaining the same.

Some guys tried better "conversation" between amplifier and speaker...they had deep concerns about EMF and power transference, dumping, Vas efficiency and output stage quality...but even doing that... we made much better work than the speaker factories..speaker drivers and enclosures are still from stone age.

Not only the understanding about what those numbers means to human brain/perception...but also the perception were the problem is and to stop our delirant behavior to fix something is already good while ignoring the real souce of problems we have.

Graham Maynard made a great job about that...his GEM amplifier has better "conversation" with speakers.

Carlos

 

[destroyer X]

I have not words to write what i have perceived into speakers in english language

them i have tried the google translator...and this was the result.

Sorry...seems google translator is so bad as i am.

"Diaphragms, cones of speakers were deformed,
bent, and bend so that the assumed
misalignment of the coil in the mobile core
magnetico .. when that happens the distance between the
mobile coil and magnetic core reduces or tends to
zero ... this is the coil scratch, scratch, touch the core
magnetic sound."

 

[PMA]

It would be far harder for me to give you two amplifiers each with 0.001% THD-20 and have one sound great and one sound bad, assuming all other variables, like component quality and frequency-response-affecting things were equal.

Bob,

I have made a record of my standard power amplifier output level during my usual room listening conditions. My speakers have 94dB/2.83Vrms/m. It is a record of 1minute of Mozart’s No. 41 Symphony – Jupiter, covering loud passages as well. As we can see, power output is no more than 250mW/4ohm. The quiet passage voltage is in tens of mV. Where can we see 1mW – 1W spectra in Stereophile measurements, where can we see SFDR at these levels? The only plot is always the meaningless THD+N. Now, when you investigate amplifiers in this area of output power, you will find huge differences. They will never be shown in magazine measurements, as they are not „politically correct“. Then, we read that the measurements are meaningless.

The levels shown in my image are exactly those where most of amplifiers are in a transition zone of their output stages.

Regards,
Pavel

 

[jkeny]

Aha, very good point Pavel - this is the sort of thing I'm interested in i.e. is there an area we can focus in on that reveals more about an amplifier than the very gross measure of THD-20.

So could this be put to the test - can we get 1mW - 1W spectra for a number of different amps and see how these correlate to the sound? Is this also got to do with an amp being able to reproduce the very low level detail in a recording which gives it a realism?

 

[lumanauw]

amplifiers shouldn't change the sound. unless you're a guitar player and you WANT the amplifier to change the sound

A member posted a link about power amp, intersting first line
http://www.dartzeel.com/index.php?option=com_content&task=view&id=47&Itemid=69

We consider our products to be musical instruments.

What does this mean? Tailor the spectrum response?

 

[AKSA]

I am now using 125mW into 8R//100nF as a good indication of what is happening at these very low, highly significant levels. Spectral distortion at low power into reactive loads is crucial.

 

[PMA]

So could this be put to the test - can we get 1mW - 1W spectra for a number of different amps and see how these correlate to the sound? Is this also got to do with an amp being able to reproduce the very low level detail in a recording which gives it a realism?

As you may guess, I would not write it if I had not results. We were collecting data of THD (spectrum), SFDR and noise floor of several amplifiers of different topologies in range of 2mW - 1W. However, I would like to see the same data from well known designs, as JC-1, Halcro dm68, Pass XA160, Ayre MX-R etc. Just a plot of THD of 4 'no named' amplifiers:

(someone might try to guess what are the topologies).