The 2 new technologies that will radically change loudspeakers

[illusonic]

This is not the same as discussed here. The patent is about linear distortion. Feedback and feed forward error correction is mainly about nonlinear distortion although feedback methods also correct linear distortion (frequency response nonlinearities) by default. In some cases the latter is compensated to avoid different frequency response with the FB loop on/off. In feed forward implementations nonlinear signal processing is applied. That's not required in feedback implementations.

While I can ensure you that both technologies work well to reduce distortion, they can obviously not overcome mechanical excursion or thermal limits. That's why they are combined with software limiters to protect the speakers. How much additional clean output can be achieved depends on the range between onset of audible distortion and mechanical/thermal limits of the loudspeaker. For typical hifi applications mechanical limits are the predominant factor. Because distortion components are usually of (much) lower level than the desired signal, low additional power is required for their compensation.

 

[BobbyTimmons]

Nothing is going to "radically change loudspeakers". Speakers from the 1970s can still sound good if they're restored. There's been slow if any progress in speaker technology in the last few decades. It's a mature technology.

 

[Paul7777x]

Nothing is going to "radically change loudspeakers". Speakers from the 1970s can still sound good if they're restored. There's been slow if any progress in speaker technology in the last few decades. It's a mature technology.

Hahaha!!

 

[JeffS7444]

If I can actually use Brane Party Pro as both a subwoofer and a portable bluetooth speaker, I could see myself buying one.

 

[Cbdb2]

Combine mic feedback with a good limiter and you are effectively close to what KCS can do. In this scenario mic feedback has the advantage of correcting directly sensed acoustic error similar to a feedback loop in an amplifier (the reason class-D has become as good as it is today

How does it correct a signal that has already left the driver?

 

[Salt]

Hahaha!!

No joke.
Back in the days there existed excellent made transducers for the proposed use case, and some experts that made use of it.
Same as today.

 

[KSTR]

How does it correct a signal that has already left the driver?

In the same way any feedback-based amplifier corrects a signal "that has already left the output stage": It keeps the error from appearing in the first place, feedback happens in real time, not after the fact (well, in case of a mic that is not mounted to the cone there is obviously some dead time which may or may not restrict the usable frequency range because other restrictions may kick in first, like cone breakup).

This principle can both be used for motion control and for active sound traps (ANC).

 

[Salt]

In the same way any feedback-based amplifier corrects a signal "that has already left the output stage": It keeps the error from appearing in the first place, feedback happens in real time, not after the fact (well, in case of a mic that is not mounted to the cone there is obviously some dead time which may or may not restrict the usable frequency range because other restrictions may kick in first, like cone breakup).

This principle can both be used for motion control and for active sound traps (ANC).

You could grab the information of the cone's movement directly from a coil attached to it and interact via feedback directly without delay caused by the 'air-gap' (was discussed here in a thread with Steve, who left here a while ago).

 

[illusonic]

How does it correct a signal that has already left the driver?

A control loop will generally react to an error with the speed that the controlled system allows. You cannot make a tweeter from a woofer just by applying more power. If a control loop includes some dead time, like the time of flight of sound from the speaker membrane to the microphone or an AD/DA conversion in case of a digital controller, the reaction to the error will include some delay. If delays are sufficiently short, the control loop will catch the error while building up and avoid that it keeps increasing. This means the error cannot be avoided completely or at all when delays in the loop cause too much phase shift. That's why the mic is mounted close to the speaker membrane and the AD/DA conversion runs with high sampling rate (often 192- 384kHz in ANC). Typically a feedback loop reduces the error by the amount of gain that you see when you open the loop, insert a signal at the open end and measure the response. Today's ANC systems reduce noise and distortion by up to 40dB at some frequency where loop gain is at it's maximum. Over a wider band, typically about 20dB error reduction is possible. So 10% speaker distortion would be reduced to 1% or less in such an example.

 

[illusonic]

You could grab the information of the cone's movement directly from a coil attached to it and interact via feedback directly without delay caused by the 'air-gap' (was discussed here in a thread with Steve, who left here a while ago).

If the signal you can get from a sensing coil is accurate, this would be possible. If for example, the second voice coil of a dual coil woofer is used to sense the woofer position, the sensing signal will have an error similar to the error of the BL(x) curve (motor force over excursion). Unless this error is compensated in the sensing signal by some sort of nonlinear processing, it cannot be removed from the speaker.

 

[Salt]

ANC does not cover a 2 or 3 meters of usual listening distance in home environment.
With headphones it might work.

 

[illusonic]

ANC does not cover a 2 or 3 meters of usual listening distance in home environment.
With headphones it might work.

Of course the environmental noise is only reduced at the mic position. But noise and distortion from the speaker are also reduced at the hearing position.

 

[Cbdb2]

In the same way any feedback-based amplifier corrects a signal "that has already left the output stage": It keeps the error from appearing in the first place, feedback happens in real time, not after the fact (well, in case of a mic that is not mounted to the cone there is obviously some dead time which may or may not restrict the usable frequency range because other restrictions may kick in first, like cone breakup).

This principle can both be used for motion control and for active sound traps (ANC).

You have orders of magnitude more delay than in an amp (complicated DSP always takes some time). Would like to see the bode plots.

And also form https://community.sw.siemens.com/s/...ee-versus-diffuse-field-near-versus-far-field this:

"This transition from circulating to propagating continues in an unpredictable fashion until we reach the threshold distance of roughly a wavelength, or three times the largest dimension of the sound source, whichever is greater. This complex region is known as the acoustic "near field". This mix of circulating and propagating waves means that there is no fixed relationship between distance and sound pressure in the near field, and making measurements with a single microphone can be troublesome and unrepeatable. "

 

[illusonic]

The DSP is not complicated. It's only some IIR filters and sometimes a limiter. AD/DA and time of flight delays are bigger than a few cycles of a fast DSP. And you are aware that ANC headphones exist?

Merovinger subs are an example for mic-based feedback on larger speakers. They place the mic on the membrane, which is probably the best solution for high excursion speakers. Otherwise the distance speaker to mic would vary a lot in the nearfiled of the membrane. Other option would be beside the speaker membrane.

SUBWOOFER | MEROVINGER AUDIO

www.merovinger-audio.de

 

[Petrushka]

Thank you, although....

• The Bose 901 speaker system, introduced in 1968, was one of the first and most prominent consumer loudspeaker systems to utilize an active equalizer (often referred to as an "Active Equalizer Module" or "Active Equalizer Box") to achieve its intended sound. The term "parametric EQ" specifically refers to EQs with adjustable Q, frequency, and gain, the Bose 901 active equalizer was a specialized, pre-set active EQ designed to correct the frequency response of the 9 full-range drivers.

I have, for years, casually referred to any active equalizer as a Bose Box.

 

[RetroStereo]

You could grab the information of the cone's movement directly from a coil attached to it and interact via feedback directly without delay caused by the 'air-gap' (was discussed here in a thread with Steve, who left here a while ago).

That's a classic servo-controlled driver, e.g. the Rythmik F18 & similar. That technique goes back at least as far as the '60's and, possibly, a good bit further.

 

[KSTR]

Mic-based motion control of loudspeakers (cone-mounted or otherwise) is real and known as a well-working solution for decades, so no need for useless discussion whether is's working or not. Same goes for the many other ways to implement servo control.

 

[KSTR]

If the signal you can get from a sensing coil is accurate, this would be possible. If for example, the second voice coil of a dual coil woofer is used to sense the woofer position, the sensing signal will have an error similar to the error of the BL(x) curve (motor force over excursion). Unless this error is compensated in the sensing signal by some sort of nonlinear processing, it cannot be removed from the speaker.

MFB works best when the sensor really only picks up the desired quantity (displacement, velocity or acceleration). The second coil of a DVC woofer has a lot of crosstalk that needs to be factored out (not easy but can be done). Also, it picks up magnetic hum because it is not humbucking and that leads to correcting a signal that isn't there actually. Don't ask me how I know.

In general, the sensor should be more linear (and notably over a wider range of excursion) than the driver. But you always need proper limiters inside the control loop so that correction never tries to force the driver to do things it cannot do when approaching its physical limits. And more often than not the goal of expert MFB is not only general distortion reduction, rather it is to stabilize the driver and make it strictly follow the target signal in a 100% reproducible fashion, even with some distortion.

 

[Snarfie]

The combination of Klippel optimized speaker design and modern room correction (REW, Dirac, MathAudio). In theory it seems like a strong synergy i suppose. Improved linearity and directivity at the source, combined with in-room correction this “coupled approach” could become standard in speaker design, or even be integrated more deeply into future speaker systems (if applicable) rather than treated as separate components?

 

[BobbyTimmons]

Hahaha!!

Compare a computer from 1976 with one from 2026. Now compare a loudspeaker. Everyone would notice the difference between the computer from 1976 and 2026. The only thing about the loudspeaker most people would notice is a different external appearance. It's a mature technology that hasn't changed much for over fifty years.