Electronics Forum / Design / August 2008

Depends on the enviroment, and a great deal on how you specify
distortion

20% in a supermarket
30% in a swmming pool
110% Parliament/congress

martin

On Jul 24, 7:17�pm, mrdarr...@gmail.com wrote:

I believe from my days at Nakamichi that the level is in the
neighborhood of 6% or so.
Obviously, it is measureable at levels much less than this, and quite
often, easily obtained.

Why do you ask?

On Jul 24, 3:17 pm, mrdarr...@gmail.com wrote:

Since you've been asking about biasing your amp, are you hearing a
'raspiness' at low levels? That could be classic crossover distortion.
The way I used to adjust on the bench (long ago) was to run 30kHz into
the amp at a low level of 1 or 2 V p-p into the dummy load. Watch the
output on the scope and advance the bias until the 0 cross notches go
away. You'll find that happens close to the 20 mA Phil Allison
mentioned earlier.

Many assume the distortion increases with signal but I haven't found
that to be the case. Of course at clipping all bets are off and
depending on program material may or may not be annoying. The other
classic annoyance is the crossover notch. Keep in mind it sounds a LOT
like a rubbing/sticking voice coil. Also, some old opamps have
crossover problems as well.

G²

It's very subjective and depends highly on the harmonic structure of the
distortion. Odd higher order harmonics tend to sound the worst. 2nd may
be virtually inaudible at several percent.

Graham

On Jul 24, 4:17 pm, mrdarr...@gmail.com wrote:

I will assume that you've read all the other replies and not repeat
the good points there.

Perceptible can be hard to define.  At small levels of distortion,
music sounds "different" but it is hard to put your finger on what is
different about it.  It can sound like it is louder than you would
like it to be or perhaps like something weird has happened to the high
end of the frequency response.  These things are very subjective and
vary a lot with the type of music etc.

At something like 5% many people will say "This is distorted".  It
starts to become obvious that it is distortion that is what is
different.

Below 5% people can hear the difference if you switch back and forth
between distorted and undistorted.  Depending on the details of the
type of distortion, most people can hear down below 0.1% distortion.

An output stage with cross over distortion and enclosed in a feedback
loop makes the sort of distortion that is most easy to hear.  When a
low frequency signal causes the output to cross through the dead band
of the cross over distortion the output does something like this:

Ascii art:

.....................**........
...........**.....**...........
...........*.*..**.............
...........*..**...............
...........*...................
...........*...................
...........*...................
...........*...................
...........*...................
......*****....................
....**.........................
..**...........................
**.............................

This produces a large amount of some of the higher harmonics of the
low frequency.

On Jul 24, 7:17 pm, mrdarr...@gmail.com wrote:

'Distortion' is a catchall term, so the simple measured number doesn't
correspond to a perception in a reliable way.

There was an old experiment (by Carver, reported in Stereo Review,
if I recall correctly) on crossover notch distortion, which is
a kind of harmonic distortion, in musical rendition (musical
instruments
all generate harmonics, so the distortion is 'hidden' in this case).
They found perception was reliable at 1.5% measured distortion.
They also found that listeners didn't object to this as hurting the
music, but DID have some response to 'clipping' distortion
(not sure how much).

The race to make lower distortion numbers is a marketing scam.  Below
0.1%, it's unlikely to matter to a music listener.   Bass  distortion
in
most integrated circuit amplifiers is VERY high, and few complaints
are heard.  Have a look at the LM380 data sheet.

On 25 jul, 01:17, mrdarr...@gmail.com wrote:

As others explained, it depends on the circumstances.
I do not see myself as 'general public' but there is a 'get used to
distortion'
factor too.
General public has been exposed to PWM class D cellphone songs....
give it 10% and they still like it.
I bought some cheap Chinese boxes (well 150W or so each), those
have those piezo tweeters in it...and when I switched those on, the
sound was horrible.
It is the 2 rear channels... so anyways I am used to it.
When I did put the Senheiser headphones on last week, it felt like a
relief not to hear those terrible tweeters.
So not only amp, but also speakers, and of course source material.
Way back, I think it was in the seventies, playing vinyl, Shure M44-7
element, my own designed pre-amp, huge boxes, listening for that
fine detail, elliptical stylus???
Digital came, 44.1 kHz sampling CD, and after that mp3 compression,
cellphones....
As others mentioned, cross-over is horrible, especially as it sticks
out on low volumes, clipping is a speaker killer, and somewhere in
between
zero and max undistorted sine waves, something should work.
Anyways, was just reading about some artist, Neil Young, making a case
for better audio....
It is the whole chain however, and few have the quiet living room or
workspace even to listen, for those the 80dB signal to noise is of no
use,
on the contrary, we need companders so we can listen in the car, or
next to the PC.
And with constant high volume, crossover is not so noticed, and
saves current ;-)
Conclusion?
None, let the super hifi audio freaks have their 96kHz sampling,
what not.
'General public' being converted to AAC, mp3, mp2, electronically
processed stuff in many ways.....
All depends what you want to do,  remember listening in a church to
some Quad
electrostats as monitor for some flute concert, got all excited,
was much younger then, could actually hear 18kHz, and all those
15625Hz TV whistles,
could even tell you if a TV was synced properly....without looking.
The average population gets older (grey), and those high frequencies
are now used
to scare away youth that hangs around, to make secret ring tones in
class that
the teachers cannot hear, and against insects and small mammals...
Maybe older ones get deaf too (has not happened with me yet), and you
need
more and more power.. if the distortion is low enough they won't hear
it ;-)

The largest contributor to measurable distortion in a playback system
is probably still the loudspeakers, which from memory can still run
consistently above 2% for a single harmonic.

RL

Back when I used to final test dispatch consoles for Motorola, I could
perceive about .8% distortion at 1000 Hz. This was communications
equipment and generally performed pretty well but if you over drove the
mike or line inputs, the compressor would introduce the distortion. I
never tried multi-tone distortion. I assume that the threshold would
have been lower.

Hearing perception is a funny and tricky area. The type of distortion
will factor into the amount you can stand or detect. I know that odd
harmonics are less obnoxious, often dsirablee compared to even mode
harmonics.. So, best have a look at the wave form and not just a % THD
meter for this one. Non harmonics related would even be worse and more
disrubing to human perception.

Best Regards,

Marc