[sdiy] Dynamics and speakers, was:Advice

[Thomas Dunker]

On Tue, 3 Feb 2004, Richard Wentk wrote:

> The real point of compression is that given the limitations of most
> speakers, there's no point trying to accurately reproduce the peaks anyway.
> Studio monitors can reproduce peaks accurately (more or less) but nothing
> else in the world can, so that headroom is effectively wasted.


 Hey, couldn't resist commenting on this since the reproduction of
uh..."life-like" dynamics has been my quest as a DIY audio nut for the
past 10 years I have spent a great deal of time trying to figure out what
it is the audio industry actually wants us to believe. Some of the most
able speaker technology in terms of dynamics (or dynamic linearity to be
more precise) just happens to be the oldest, and in domestic hi-fi terms
also the "most obsolete" - namely horn speakers. Wherever horns are used
in the pro audio industry today (SR rigs, stadiums, movie theatres etc.)
it's precisely because they have vastly better dynamic linearity
(distortion vs. output SPL) than most alternatives. Horns are also used a
lot specifically for other primary reasons, like dispersion control, but
they got me interested because of the dynamic potential.

 Anyway, the sad story goes something like this: Ever since the direct
radiator speaker came about (circa 1928, by Rice and Kellogg in the US who
beat P.G.A.H. Voigt to the patent office - Voigt also had working
prototypes at the time), an ongoing trend has conspired to effectively
remove "lifelike dynamics" from most people's criteria for "high fidelity"
reproduction. The first theater sound systems had speakers with something
like 50% efficiency (modern hi-fi speakers have something typically around
0.1% efficiency, which probably explains the fact that efficiency is
represented by a decibel figure at 1 watt input, yielding a more
impressive number) and were powered with sub-10W triode amps. This would
fill an entire movie theater with sound of unprecedented fidelity for the
time (1930s).

 Movie sound was developed into the forefront of audio technology during
the 1930s, and advanced to give birth to multi-track studio recording
techniques and fantastic analog reproduction systems by the end of the
thirties. This is exemplified by the Fantasound system developed for Walt
Disney's picture Fantasia by Hollywood sound engineers like Howard
Tremaine, along with speakers and amps by RCA and Altec, and the
three-track optical stereo soundtrack developed by Bell Labs. This optical
sound track utilized a proprietary system called TOGAD (tone operated
gain-adjusting device) which formed the basis of what must have been one
of the very first "companding" schemes for recording. Separate gain
control voltages for each of the three main channels were modulated onto
different frequency carriers that were mixed and recorded as a separate
track that ran along the three audio channels. These four tracks were
recorded onto a dedicated 35 mm film that was syncronized with the picture
film. During playback, the gain control voltages were recovered from the
TOGAD track and fed to VCAs (!) in conjunction with the playback preamps.

 Believe it or not, but this optical recording system enabled the Fantasia
score to be recorded and reproduced with about 90dB of dynamic range!

THAT'S ANALOG RECORDING STATE OF THE ART IN THE YEAR 1940

 During the Fantasia recording sessions, multitrack recording technology
as we know it was invented, except they mixed it down to three discrete
channels (left, center, right). Orchestral "sound effects" were also
worked into the mix and triggered by the TOGAD system to be played by
various speakers placed around the theater, so even surround sound is a
60+ years old technology.

 The speaker technology used reflected the demands of the
recording/reproducing chain. The specs were insane for the time, something
like 120dB peak in the rear of the theater, with 6dB of headroom. Huge,
balanced class A triode amps powered bass horns and HF horns. A few
theaters were equipped with insanely expensive Fantasound systems, but
there was also a portable system complete with diesel generators,
projectors, cables and everything.

Read more here:

http://www.widescreenmuseum.com/sound/Fantasound1.htm

(I got a lot of first hand dope on Fantasound by one of the aging
engineers who worked on the system in the 1940s, a guy named Jack Strayer)

 OK, this is megalomaniac triode-age movie theater HI-FI madness, but
demonstrates how we should probably be a little careful using phrases like
"the state of the art" today, realizing that the actual state of the art
TODAY is very far from reaching into most people's experience with
reproduced audio.

 And that's because of how amplifiers and speakers have developed since
the days of triodes and horns. Triodes were replaced by pentodes, making
amp power cheaper, causing some idiot to think hey, what do we need this
kind of speaker efficiency for, and along come smaller, less efficient
speakers. Pentodes were eventually replaced by transistors and by the
1970s typical "hi-fi" speaker efficiency had settled somewhere around
87dB/1W/1m - or somewhere below 0.1% relative efficiency. (Or 99.9% of
input power directly converted into HEAT and other losses, rather than
sound).

 Specifying the efficiency in decibels is useful some times. The most
inefficient speakers, like LS3/5As, Magnepans and such, are around
85dB/1W/1m. The most efficient speakers (big/horn based systems) can
easily be around 105db/1W/1m. Most people don't realize how much a 20dB
difference is in terms of amplifier power. If the speaker is 20dB LESS
efficient, it needs 20dB or A HUNDRED TIMES MORE POWER. Think about it:

 If you drive the 105db speaker with a humble 10W amp, comfortably
below the speaker's max power rating, with 115dB peaks, you would -
theoretically - have to have a 1000W amp driving the 85 dB speaker to get
to the same max SPL or useful dynamic range - EXCEPT the 85dB speaker's
voice coils would begin to self destruct at a mere couple hundred watts of
input. No matter how much you up the amp power, the bottleneck will be in
the speaker's power rating - BECAUSE OF ITS ULTRA LOW EFFICIENCY. It is
therefore impossible to entirely compensate for inefficient speakers by
increasing amp power as far as dynamics are concerned.

 Next time you see a monster amp driving tiny speakers, don't be
impressed, be sad.

 Okay, so speakers can be more or less efficient, but it's not as simple
as that. I brought up how 99% or more of the input power is converted into
heat loss in the speaker. This has brought us fabulous innovations like
high temperature cements and coil former materials allowing the
low-efficiency insanity to escalate. An often overlooked fact is that when
a piece of copper wire gets HOT its resistance increases. In a speaker it
can typically imply that the impedance of the speaker may DOUBLE at
"highish" input power, still well below max power rating. Since the
industry has also decided that speakers - inherently current controlled
devices - MUST ONLY be driven by low output impedance (undistorted
voltage) amplifiers, another no-brain industry standard - this gives rise
to what we call "POWER COMPRESSION", well known in the pro sound industry,
but obviously a taboo in the "hi-fi" branch.

 Quantitatively, power compression can very easily amount to something
like 6dB loss of dynamics, give or take some. This compression will also
be subject to unpredictable effects as the copper coils in different drivers
will have different thermal time constants. Consider passive
crossover/filter networks and you have the textbook definition of "A
MESS". Not even the Q parameters or the actual efficiency of the drivers
can be constant if the voice coil resistance isn't. A matter of speaker
effciency and/or applied power.

 I have spent years reading about all kinds of  distortion effects in
speakers and the amp-speaker interface, so it's a little hard for me to
get off the soapbox when I'm at it. But, the single most important thing
to understand is that the harder a speaker must work (the more the cones
move, the more power it tries to get rid of etc.) the more distortion
(relative to the input signal) it makes. It's not cool that the woofer
cones move an inch, it tells you you're probably listening to 20-30% THD
and intermodulation distortion and horribly distorted transients too.

 The subjective perception of music dynamics has to do with more than just
SPL ratios and dynamic distortion. It also has something to do with the
reproduction of transients, which typically has to do not just with
amplitude levels, but also with time domain stuff. Like, how well are the
system resonances damped? A bass reflex speaker is incapable of
reproducing transients without some degree of periodic "overshoot",
because the whole idea is that the resonance should only be damped to a
certain degree so we get a little extra "bass" for cheap out of speakers
that are too small. This is another "standardized" type of distortion
typical in speakers.

 Summed up, "state of the art" speakers and the amps that drive them are
not suited for really big dynamics. Even most studio monitors except maybe
Westlakes, UREIs, TADs and nice big old JBLs. Not the undersized stuff
they tend to use today.  It should be quite possible - with the right speakers -
to hear more of the recorded dynamic range than the recording engineer
did when listening to the final mix.

 It's not the recording industry's fault that the consumer audio people
turn out so much misengineered junk, but it's sad how it's defined
lifelike dynamics out of the "hi-fi" world. It's like they try to bring it
back with amazing new digital media - WHY, IF THE SIGNAL WILL BE
COMPRESSED DOWN TO 30-40dB DYNAMIC RANGE ANYWAY. What do we need all this
dynamic resolution for then? You *can* get more dynamic range recorded
onto 33 1/3 RPM vinyl than is common in most CDs out there.
 How cool is it to have 24 bit resolution and 120dB S/N ratio when the
recorded signal ends up being compressed to the dynamic range of worn
cassette tape - or worse. Why are we supposed to get all excited about
THAT? What a joke!

 Still, the most dynamically capable speakers can make even the most
awfully compressed recordings come alive, I know from experience, that
only proves how much compression the average speaker ADDS. I had no idea
until I could hear it for myself. Never would have happened at the hi-fi
store.

Pardon the ramble and off-topicness...


Thomas Dunker

Trondheim, Norway