Why Is High Feedback Considered Bad In Audio? In Simple Terms

[Bjarne Bäckström]

Sorry, I didn't see the original post.

[...]
Google for "transient intermodulation distortion" and "Matti Otala".
--

 

[Phil Allison]

""Bjarne Bäckström""
"abby"

Google for "transient intermodulation distortion" and "Matti Otala".

** Might as well Google up " Lewis Carroll " and "Alice in Wonderland"
while you are at it.

"The Mad Hatter" makes more sense than that blatant fake Otala ever
did.




....... Phil

 

[Jasen Betts]

Some classes of higher-order filters ring, some don't at all. More
math.

for real?
how do they manage that?

 

[Adrian Tuddenham]

There is actually a simple solution to that, one I thought of
in the seventies, and was actually used in commercial designs.
Put a simple transistor stage before the power amp (in the power amp),
and make sure that transistor stage limits just before the power amp does.
The transistor stage needs no feedback, and the power amp always works
in the linear range.
When clipping happens, in the transistor stage, the output looks nice.

Several of my own P.A. designs (all transistor) have soft clipping in
the driver stages. On the occasions where I have had to run them
overloaded[1], they sounded much louder than their numerical wattage
would have suggested and there were no complaints about distortion.


[1] When the client says it's only a small job and it turns out to be a
much bigger one, for which I really should have brought more kit.

 

[Adrian Tuddenham]

If one were going to go after purveyors of that kind of gear, then for
consistency, one would also have to go after a large part of the entire
cosmetic and beauty industry. In common with the impugned audio
equipment vendors, they too are at best selling people the means to
delude themselves.

Which is not to say that would be a bad thing, but it illustrates the
true scale of the problem.

Except that people aren't poisoned and don't have have life-threatening
allergies triggered by gold-plated CDs. The cosmetics and perfume
industies are literally getting away with murder and (in Europe at
least) are virtually unregulated.

 

[Vladimir Vassilevsky]

Several of my own P.A. designs (all transistor) have soft clipping in
the driver stages. On the occasions where I have had to run them
overloaded[1], they sounded much louder than their numerical wattage
would have suggested and there were no complaints about distortion.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
A typical audiofool ignorance.


The amount of nonlinear distortion = deviation of the transfer curve
from the straight line. Soft clipping = more distortion if compared to
hard clipping to the same ceiling.

If the amp has to enter the area where it clips the signal, one has to
limit the windup of the integrators in the feedback path. So the
recovery from the saturation will be quick.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[Vladimir Vassilevsky]

Gaussian and Bessel (linear-phase) filters don't ring, no mater what
the order.

Wrong.

Gaussian and Bessel filters do ring, although not nearly as much as the
other filter types.

You can make a sharp-cutoff (delay equalized) analog filter that
doesn't ring, but it's a pain.
True.

Since ears aren't phase sensitive at high frequencies (especially
above 20 KHz!) a little ringing doesn't matter anyhow.

BTW, there is no connection between the "ringing" and the linearity of
the phase in the filter.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[Adrian Tuddenham]

Several of my own P.A. designs (all transistor) have soft clipping in
the driver stages. On the occasions where I have had to run them
overloaded[1], they sounded much louder than their numerical wattage
would have suggested and there were no complaints about distortion.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
A typical audiofool ignorance.

I am just as scathing of audiophoolery as you are; this comment was
based on 40 years of practical P.A. experience as a designer and as an
operator.

The amount of nonlinear distortion = deviation of the transfer curve
from the straight line. Soft clipping = more distortion if compared to
hard clipping to the same ceiling.

Soft clipping gives less audible distortion on a low fidelity P.A.
loudspeaker system. If an amplifier has to wound up to too high a level
in order to deal with an awkward situation 'in the field', it is
infinitely preferable to hear soft clipping in the top 6dB of the signal
than to endure the harsh distortion from flat-topping.

If the amp has to enter the area where it clips the signal, one has to
limit the windup of the integrators in the feedback path. So the
recovery from the saturation will be quick.

With soft pre-clipping, the power stages are less likely to saturate at
all.

There are other reasons for using as little feedback as you can get away
with on a P.A. amplifier:

Most designs need an output transformer to feed the floating 100v line,
so a tertiary winding, close coupled to the secondary, is the only way
to give meaningful feedback. There may be considerable phase shifts due
to resonance between the the self-capacitance and inductance of the 100
volt winding.

The load is unpredictable and the amplifier must remain stable under all
sorts of conditions from an open overhead line several hundred yards
long (during set-up) to a short circuit (when some idiot drives a nail
through the cable). If there is significant leakage inductance between
primary and secondary, it won't take much feedback from the secondary to
give instability under some conditions of loading.



My comments about soft clipping applied to difficult underpowered P.A.
situations where turning the volume down isn't an option and you are
forced to choose the least of several evils. I am not advocating soft
clipping for high quality sound reproduction (where adequate amplifier
power is the obvious first step).

 

[krw]

If one were going to go after purveyors of that kind of gear, then for
consistency, one would also have to go after a large part of the entire
cosmetic and beauty industry. In common with the impugned audio
equipment vendors, they too are at best selling people the means to
delude themselves.

No, the users of cosmetics delude others, which is exactly what the
manufacturers advertise.

Which is not to say that would be a bad thing, but it illustrates the
true scale of the problem.

That people are vain? Maybe. OTOH, I'd rather see a
dressed-to-the-nines woman walking down the street than one with
unbrushed teeth, just out of bed, and in an old torn housecoat.

 

[krw]

Sorry, I didn't see the original post.

[...]
Google for "transient intermodulation distortion" and "Matti Otala".

You should be able to answer a question that's been bugging me. Which
do audiophools spend more money for, speaker cables or HDMI cables?

 

[Vladimir Vassilevsky]

I haven't rolled my own power amplifiers for 30+ years now.

Commercial audio amps is one of the things that I design as a business.
In our days, that's mainly Class D.

In the past, though, it was my standard procedure to have way more
than adequate _real_ audio power... like 100W per channel in my living
room ;-)

Loud sounds, big guns, shiny cars and such attributes are often
associated with weak personalities

But, I also utilized soft clipping in the driver stages to ensure I
didn't rail the output stages.

Limiting the signal upfront the amplifier can be useful. However I
disagree about limiting in the driver stage.

I always used RF-grade BJT's in my designs. If I were to build
something again I'd still use BJT's... since I have, in my secret
pocket, a way to A-B bias reliably ;-)

Using the RF BJTs in the linear amp would be nice from the purist point
of view. It would accomplish the unconditional stability with the deeper
feedback. However the audio applications are severely constrained by the
cost rather then by the performance. So the main difficulty is not to
get to the ultimate perfection, but to make it reasonably good while
minimizing the cost.

BTW, the main push towards class D is because of the material savings
due to the reduced size of the heat sinks.



Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[qrk]

Some classes of higher-order filters ring, some don't at all. More
math.

Ringing can be a linear phenom, in which case it's just peaking in the
frequency domain. And if the ringing is above 20 KHz, there's nothing
to peak and nobody can hear it anyhow.

John

I'll have to add "grittiness" to my list of audio metrology terms...

5N, 6N, 7N (as in 99.99999 pure copper)
acceleration
agressive
air
articulation
bass transient response
blackness
bloom
bright
burnin (cables)
clarity
color
confused
congested
continuousness
crisp
delineation of inner detail
dynamics
effortlessness
etched
extension
fast bass
free-flowing ease
grainy
granularity
hangover
hard
harmonic completeness
hashy
holographic
honesty
impact
jump
layering
liquid
liquidity
liveliness
lush
macrodynamics
microdynamics
musicality
openness
pace
quantum purifier
reference
refinement
relaxed
roundness
reveal
single-crystal copper
slam
slow bass
smear
soundstaging
space
sparkle effect
spatial resolution
speed
stunning (everything is stunning)
thin
tight
tight (bass)
timbre
tipped-up
transparency
truth
unfussiness
wire direction

"Toasting". A process of processing a cable which improves the
fidelity of such cable. Cables that are toasted are directional.
Unfortuantely, I can't find the original reference to toasting cables
which was most amusing to read. The testimonials proved the worth of
toasting.

http://www.noteworthyaudio.co.uk/Services.html#4
http://www.ultraaudio.com/opinion/20030801.htm

Terms in the second link...
vividness
tonal accuracy

 

[Bjarne Bäckström]

Sorry, I didn't see the original post.

[...]

Hi,

Can someone explain, in simple terms, why high feedback is considered
inferior to low feedback in audio circuits? To me, linear is linear
and
a high feedback op-amp circuit is linear. Apparently, that isn't
entirely
true.

Thanks,
Gary

Google for "transient intermodulation distortion" and "Matti Otala".

You should be able to answer a question that's been bugging me. Which
do audiophools spend more money for, speaker cables or HDMI cables?

Well, I left this discipline behind me about 30 years ago, so I don't
know what's "in" these days. But I'd say, since people don't seem to
change much, in my experience they favor things that escape scientific
explanations.
--

 

[krw]

On Fri, 23 Jan 2009 17:08:00 -0800, John Larkin
[snip]

stunning (everything is stunning)
thin
tight
tight (bass)
timbre
tipped-up
transparency
truth
unfussiness
wire direction

"Toasting". A process of processing a cable which improves the
fidelity of such cable. Cables that are toasted are directional.
Unfortuantely, I can't find the original reference to toasting cables
which was most amusing to read. The testimonials proved the worth of
toasting.

http://www.noteworthyaudio.co.uk/Services.html#4
http://www.ultraaudio.com/opinion/20030801.htm

Terms in the second link...
vividness
tonal accuracy

It's all coming back to me now...

Anyone, besides me, remember when cactus needles were all the rage for
stylus use ?

I know some who used certain cactus to, um, "enhance" their
appreciation.

 

[MooseFET]

In your opinion, would a filter that has no overshoot in the time domain be
capable of ringing?  

Consider this response to a step:


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

Ringing but no overshoot.

 

[Sylvia Else]

No, the users of cosmetics delude others, which is exactly what the
manufacturers advertise.

Some products have that stated purpose, it's true, but a great many make
claims (usually expressed alongside weasel words) as to benefits that
are not supportable.

Sylvia.

 

[[email protected]]

Gaussian and Bessel (linear-phase) filters don't ring, no mater what
the order. Sharp-cutoff (Butterworth, Chebychev, elliptical) filters
do.

You can make a sharp-cutoff (delay equalized) analog filter that
doesn't ring, but it's a pain. Digitally, it's easier.

Since ears aren't phase sensitive at high frequencies (especially
above 20 KHz!) a little ringing doesn't matter anyhow.

John

Higher order Bessel filters ring. You can easily convince yourself of
this by looking at the impulse response. The time domain response is
the convolution of the signal and impulse response. If the impulse
responses goes negative, it must ring. [The situation is the same in
DSP for FIR filters, i.e. if a tap is negative, the filter must ring.
Well, except for all negative taps, which of course is just a polarity
change.]]

The Gaussian never rings.

The Bessel is designed for maximally flat group delay. Now by
coincidence, some of the lower order functions don't ring.

 

[[email protected]]

Several of my own P.A. designs (all transistor) have soft clipping in
the driver stages.  On the occasions where I have had to run them
overloaded[1], they sounded much louder than their numerical wattage
would have suggested and there were no complaints about distortion.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
A typical audiofool ignorance.

The amount of nonlinear distortion = deviation of the transfer curve
from the straight line. Soft clipping = more distortion if compared to
hard clipping to the same ceiling.

If the amp has to enter the area where it clips the signal, one has to
limit the windup of the integrators in the feedback path. So the
recovery from the saturation will be quick.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultanthttp://www.abvolt.com

Yes, saturation is the "sticking" seen on a scope. You see some
photodiode amplifier circuits that do clamping to avoid sticking since
the effect on recovery time is better for a designed clamp versus
"organic" sticking.

 

[JosephKK]

On Fri, 23 Jan 2009 18:27:12 -0700, Jim Thompson




I'll add it. Is that measured with a thermocouple?

John

Only if you want inferior results. For proper measurement you must
Platinum RTDs.

 

[Vladimir Vassilevsky]

On Sat, 24 Jan 2009 16:58:14 -0800 (PST), "[email protected]"


I don't get that. A simple CRCR network can have an impulse response
that swings negative, but it doesn't ring.

Highpass filters ring by definition. The step response goes up and then
down.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com