View Single Post
  #4   Report Post  
Posted to rec.audio.high-end
[email protected] dpierce.cartchunk.org@gmail.com is offline
external usenet poster
 
Posts: 334
Default Motional feedback in speakers

On Wednesday, November 6, 2019 at 11:12:22 AM UTC-5, Peter Wieck wrote:
No surprise at all, back in the day when tube-based amplifier
damping factors were somewhere between 10 and 15 on a good day
with a tailwind, and acoustic suspension was uncommon. The 'cure'
for tubby bass would be some choice between a stiffer spider,
stiffer surround, smaller magnet, shorter excursion, any or all.


Uhm, no, not anywhere near a "cure", in fact, demonstrably feeds the
disease. Let's look at it piece by piece. Assume everything else
remains the same, and by "tubby bass", you mean excessive Q at resonance:

1. Stiffer spider, stiffer surround, either or both will raise the
resonant frequency of the driver, thus a higher Q at resonance.
Result: more "tubby bass".

2. Smaller magnet: presumably what youare in effect saying is reduce
the BL product of the motor system. Well, since the Q at resonance
is an inverse function of the BL product, that raises the Q of the
system. Result: more "tubby bass."

3. Shorter excursion: that, by itself will have little effect on
"tubby bass" per what it will do is cause more distortion at
lower levels. Result: more distorted "tubby bass."

Now, if the issue is speakers with "tubby bass" and low damping factor
amplifiers, motional feedback IS NOT THE CURE AT ALL. The entire point
of motional feedback was to attempt to linearize the speaker system,
i.e. reduce distortion. It does NOTHING to reduce Q, reduce "tubby bass."

One of the big problems with drivers is that the suspension stiffness
and the magnet's BL product is, for the most part, highly non-linear:
the stiffness increases with excursion, the BL product decreases
with excursion. And the greater the excursion, the greater the
deviation of both. And the greatest excursion occurs at low frequencies
(all other things being equal) and, ultimately, at resonance.

Now, IF the difference between sensor (the sense coil, in at least on
case) says where the woofer is vs where the amplifier thinks the
woofer should be, the feedback SHOULD provide a complementary non-linear
change in its output voltage to compensate.

There are several different ways of doing this: a sense coil
will provide an output that's a function of velocity, while
things like the piezo sensor in things like the Velodyne output
a signal that's proportional to acceleration. In the former
case, you integrate once to get position, in the later, you
integrate twice to get position.

But here's the rub: there IS NOT a simple functional relationship
between either the position, velocity or acceleration of the cone
and its acoustic output. In other words: the sensor providing the
feedback signal CANNOT be used to correct the frequency response
of the system, i.e., it's not fixing "tubby bass", it's trying to fix
"distorted bass".

Let me give one example that shows why this is true: look at ANY
bass reflex system: let's assume, for the moment, that it is
"optimally" tuned, i.e., that its frequency response is that of a
"perfect" lossless B4 alignment: it's dead flat down to cutoff,
and it rolls off at 24 dB/octave below that. Now, look at what
the cone is doing as it goes down in frequency through the region
of cutoff:

Acceleration: Constant with frequency as you go lower until
you approach the enclosure tuning frequency, at which point
is approaches 0, ad blow which it increases back up to
the same constant as above tuning.
Velocity: Increases as inverse of frequency until you approach
the enclosure tuning frequency, then goes to 0, goes back up for
a bit then starts decreasing with decreasing frequency.
Position: Increases as the inverse square of frequency until
you approach the enclosure tuning, at which point it approaches
0, then below it starts increasing with decreasing frequency.

And, oh by the way, where all the mechanical stuff goes to 0 at
the enclosure tuning frequency, the output of the port is increasing
and the bulk of the speaker's acoustic output comes NOT from the driver,
but from the port.

In other words, as far as frequency response is concerned, neither the
acceleration, nor the velocity, nor the position of the cone is a
useful predictor of the acoustic output in such a system.

Seems overly complex - and I can see where it did not last long,
or penetrate past a very few makers. By the way, Philips speakers
were "active" speakers, needing only a pre-amp input.


Well, I "was there" when the the phillips was being marketed, and at
the same time there was an actually reasonably well-implemented B6
system from EV, designed by, I believe D. B. Keele. Neither speaker was
"outstanding", but both were quite reasonable systems, performance-wise.

Both died in the market for a variety of reasons, but, especially with
the Philips, customers resisted it because they wanted to use it with
THEIR amplifier. Often times, they already had a system and wanted to
upgrade speakers, and looked at their existing amplifier as now being
a "waste". And, with the EV, that little EQ box, no matter how technically
sound the approach was, was just an obstacle for most people. To many,
it reminded them of the Bose 901, and for many, that was just too much.

It had NOTHING to do with performance, it was all about perception,
rational or otherwise. Regardless, systems like this withered on the
vine having little to do with their technical merit.