Motion controlled loudspeaker

[Terry Given]

Terry Given wrote:




Nice idea.

the trick to optimising efficiency is to shape the convex hull so that
it fits within the frequency response of the smps. the maths is ripped
straight out of image processing.

What do you do about cone breakup though ?

whats cone breakup? is that when oscillations take place across the
structure of the cone itself?

Graham

Cheers
Terry

 

[Scott Dorsey]

whats cone breakup? is that when oscillations take place across the
structure of the cone itself?

Right. It's the main nonlinearity issue with low-Xmax drivers. When the
Xmax starts to get very long, you run into linearity issues with the
whole driver as well, of course.

The problem is that the cone doesn't want to move as a piston. You can make
the cone stiffer, but that just moves the breakup modes higher and makes the
cone more massy and harder to move quickly.

One solution to the breakup problem for subwoofers is to use an acoustical
low-pass filter that rolls off the higher breakup modes, and a very stiff
cone that moves them up to frequencies above the filter point.
--scott

 

[Eeyore]

the trick to optimising efficiency is to shape the convex hull so that
it fits within the frequency response of the smps. the maths is ripped
straight out of image processing.



whats cone breakup? is that when oscillations take place across the
structure of the cone itself?

At higher frequencies ( accelerations ) the outide of the cone can't 'keep up'
with the driven part and it flexes. This is called cone break up. You can get
various modes of it. This is a fundamental flaw with cone loudspeakers so
there's a limit to what can be acheived.

Graham

 

[[email protected]]

No comment on position feedback, but John Popelish was wrong when
he said that a speaker is a position device below resonance. He
neglects the effect of back-EMF, which means that at a (close-to)
constant voltage, the coil is a constant *velocity* device. Try it
sometime with a broken woofer - remove the cone and apply a volt
across the coil. You'll see the coil rises at a fixed rate. I've
done this with 4" voice coils from an ancient "washing machine"
disk drive head mechanism, which makes the effect very visible.

No, John was correct. It's a velocity device AROUND resonance, where
the back-EMF isn't masked by either stiffness or mass (lower Q means
wider velocity region). Below that it's a displacement device.

 

[Terry Given]

Right. It's the main nonlinearity issue with low-Xmax drivers. When the
Xmax starts to get very long, you run into linearity issues with the
whole driver as well, of course.

The problem is that the cone doesn't want to move as a piston. You can make
the cone stiffer, but that just moves the breakup modes higher and makes the
cone more massy and harder to move quickly.

One solution to the breakup problem for subwoofers is to use an acoustical
low-pass filter that rolls off the higher breakup modes, and a very stiff
cone that moves them up to frequencies above the filter point.
--scott

thanks Scott and Graham. the mechanical dynamics of most things get
pretty complex when you look closely.

but it does make me wonder if its not possible to somehow damp out those
specific modes within the cone itself. I'm thinking some sort of crazy
multiple "voice" coil arrangement that allows 3-D control over the
actual fields produced - essentially little "compensating" coils. then
with a bit of computational horsepower, one should be able to generate a
suitable 3D flux pattern to result in the desired air displacement.
phased-array kind of stuff.

just random thoughts.

Cheers
Terry

 

[Eeyore]

thanks Scott and Graham. the mechanical dynamics of most things get
pretty complex when you look closely.

but it does make me wonder if its not possible to somehow damp out those
specific modes within the cone itself. I'm thinking some sort of crazy
multiple "voice" coil arrangement that allows 3-D control over the
actual fields produced - essentially little "compensating" coils. then
with a bit of computational horsepower, one should be able to generate a
suitable 3D flux pattern to result in the desired air displacement.
phased-array kind of stuff.

I know what you mean. If only the cone didn't have to be driven just from the
centre.

Graham

 

[Ban]

The same as I'd hope to gain in theory but with relaxed expectations
due to the non-ideal nature of reality...i.e to significantly reduce
the error in the position of or pressure created by the driver -
thereby reducing distortion.

You would need to invent a new device, something faster than light FTL. Your
driving signal is a couple of milliseconds ahead, because of the highpass
characteristic of the speaker. Whatever you do has to be done before. It
only works for the higher frequencies, at the resonance the phase shift is
already +90° leaving little stability margin and without differentiation no
way to pump in any real power. The amp will ad a significant phase shift at
these low frequencies, as will your sensor interface.
The speaker characteristic is of second order, whereas a PID can only
compensate a first order slope and give you max. 45° margin more.
But you could maybe develop a non-linear model for a DSP, glue an
accelerometer on the dustcap and thus determine the pulse response for
different input levels and then apply an inverse model to the input signal.
If you use adaptive filters, you could make up for thermal effects, there is
a lot to imagine.
And then the distortion is mostly because the cone is bending, you need
laser interferometry for that. And it is intrinsic to the speaker
construction.

I like this time-warp predistortion idea, and you do not need a contact
sensor, but can do it with a mike, soundcard and existing programs.

 

[Fred Bartoli]

Clifford Heath a écrit :

No comment on position feedback, but John Popelish was wrong when
he said that a speaker is a position device below resonance. He
neglects the effect of back-EMF, which means that at a (close-to)
constant voltage, the coil is a constant *velocity* device. Try it
sometime with a broken woofer - remove the cone and apply a volt
across the coil. You'll see the coil rises at a fixed rate. I've
done this with 4" voice coils from an ancient "washing machine"
disk drive head mechanism, which makes the effect very visible.

LOL!
Apply 1V or whatever you want on a speaker and look the cone pass
through the window, troposphere, asteroid belt,...

 

[Boris Mohar]

I know what you mean. If only the cone didn't have to be driven just from the
centre.

Graham

Like this?
http://amasci.com/esloud/eslhwto.html

 

[Robert Latest]

["Followup-To:" header set to sci.electronics.design.]
On Wed, 11 Oct 2006 20:56:38 GMT,

Probably not... PID control was invented to help people who sell PID
controllers to people who use PID controllers and it's become sort of self
replicating.

If memory serves me right, "PID control" was temporarily renamed
"Fuzzy Logic" in the 90's.

robert

 

[Eeyore]

Like this?
http://amasci.com/esloud/eslhwto.html

Interesting.

It's a shame electrostatic loudspeakers need to be so big to be effective.

Graham

 

[Arny Krueger]

Scott Dorsey wrote:

But can't the delay be shortened using a controller? That
is, if a steep slope is detected that the controller
knows the driver will struggle to accurately reproduce
due to it's sluggish response time, it can output a much
steeper slope to ensure the coil accelerates faster?

That's been tried. Just about everything has been tried.

The basic problem with direct radiating speakers is that from a cone motion
standpoint they are like an integrator starting just above the frequency
where cone mass provides most of the load on the voice coil. This "mass
loading" starts just above resonance, or around 30-80 Hz for most speakers.
It gets harder and harder to have a stable feedback system above this point.

Thing is, speakers are pretty linear in their mass-loaded region. So,
feedback isn't going to help much, anyway.

Servo control has thus been pretty much restricted to subwoofers because it
is feasible with them, and many of them can benefit from the linearizing
effects of feedback.

The long term trend over the past 30 years has been to simply build
loudspeaker drivers that are more linear at low frequencies and larger
excursions. KISS.

 

[David Grant]

That's been tried. Just about everything has been tried.

Is that a reason to stop trying? I don't ask these questions because I
think I've come up with a unique idea I ask them because it's valuable
to know why something doesn't work; I think you've explained the
pitfalls pretty succinctly, so thank you.

 

[Arny Krueger]

Is that a reason to stop trying?

Most would say that they don't want to try something that has already
already been tried by competent people who failed.

I don't ask these
questions because I think I've come up with a unique idea
I ask them because it's valuable to know why something
doesn't work; I think you've explained the pitfalls
pretty succinctly, so thank you.

One relevant situation relates to a friend of mine who is moderatly well
known in AES circles. He did a servo woofer for his undergraduate
engineering project and it was fairly sucessful in terms of performance. It
was based one of the most promising speakers of the day, the EV 30W. That
came and went long ago. Just lately he designed some subs for his personal
use. He went with some very long stroke drivers from the car audio world.
No feedback, just feedforward frequency response compensation.

 

[John Fields]

Interesting.

It's a shame electrostatic loudspeakers need to be so big to be effective.

 

[Predictor]

If memory serves me right, "PID control" was temporarily renamed
"Fuzzy Logic" in the 90's.

Fuzzy logic is used for more than mechanical control. See, for
instance:

http://eprint.uq.edu.au/archive/00000625/02/paper13.pdf
http://www.copernicus.org/EGU/hess/hessd/2/1107/hessd-2-1107_p.pdf
http://www3.sympatico.ca/alawnicz/PAGE0212.PDF
http://www.ensc.sfu.ca/~ljilja/ENSC833/Projects/chen/presentation.pdf
http://jds.fass.org/cgi/reprint/84/2/400.pdf
http://udel.edu/~ebenson/Journal_Articles/Benson_ASAE_2000_Adaptive_Edge_Detection.pdf


-Will

 

[[email protected]]

Philips ( of all people ! ) tried motional feedback on loudspeakers
decades ago.

Clearly the results weren't good enough for it to become widely used
and Philips dropped it too .

Yes, but for more money you got a smallish, for then, bookshelf pair
that claimed the bass of JBLs or B&Ws but sounded a lot nastier. And
you had to get power to them. They did lead the way though, AIR active
sub-woofers with MFB came along a few years later.

--
Paul Repacholi 1 Crescent Rd.,
+61 (08) 9257-1001 Kalamunda.
West Australia 6076
comp.os.vms,- The Older, Grumpier Slashdot
Raw, Cooked or Well-done, it's all half baked.
EPIC, The Architecture of the future, always has been, always will be.

 

[jasen]

David Grant <[email protected]> wrote:

You can make feedback systems work pretty well at 20 Hz... Velodyne has done
it on subwoofers since the 1980s, and the modern accelerometers sold for
automotive air bags make it easier and cheaper than ever before. You can't
make them work at 1 KHz because the delay is too severe.

if you want an accelerometer that'll operate a high frequencies use a
voice-coil and integrate the output.


Bye.
Jasen

 

[Chris Malcolm]

Terry Given wrote:

Nice idea.

What do you do about cone breakup though ?

Electrostatic drive. Solves the other problems as well

 

[Scott Dorsey]

if you want an accelerometer that'll operate a high frequencies use a
voice-coil and integrate the output.

Some folks have tried this; there was even an article in a recent AudioXpress
about using a dual-coil subwoofer off the shelf with one coil as a feedback
winding.

The problem with this is that it's hard to avoid coupling between the two
coils.
--scott