The 2 new technologies that will radically change loudspeakers

[KSTR]

The model in the Klippel KCS is adaptive. While there has to be an initial model prototype in form of measurements of a "golden sample", that model is constantly updated in operation, by measuring voltage and current and deriving the parameters from the mapping. This means that the model parameters are inside a rather fast feedback loop so that the real-time feed-forward distortion compensation etc is always based on the current behavior of the driver.

 

[Waxx]

I'm not the technical expert here, but i just want to say that i've seen many variations of this tech being hyped as "the solution for speaker issues" but none got to a point that it was usable in real life. So i'm equal sceptical about this.

The "problem" with speakers is that the membrane that moves air has resonances that causes distortion, and eliminating or reducing them is key, that is true, but how to do it is still the big question and i don't thinnk this is the answer. What is the answer i don't know nieighter, but this reads as old wine in a new fancy bottle...

 

[KSTR]

Once everything else is sorted out, cone/dome resonances are the least of our problems. Once we can trade transducer motor linearity for efficiency (which is where KCS is a giant leap forward in that the motor nonlinearity can be compensated for), that addidional efficiency can be "wasted" on a heavier, more rigid and well-damped cone/dome.

If there ever was a company that really understands how electro-dynamic transducers work then it is Klippel. So, any technology that fights transducer issues which comes from this company will most likely be quite ahead of everything else in the industry. Klippel is in the know, definitely.

 

[Nugget_]

Regarding the KCS, let me add Powersoft IPAL for previously disclosed technologies with the same goal. They use a diffrential pressure control sensor drilled in the baffle, to measure sound/air pressure on both sides of the baffle, and electrically feeding the data back to the input, therefore reducing (large signal) unlinearities. But this is only for low frequencies.

Supposedly the user can even input T/S parameters thus creating a "synthesized driver" with parameters of his choice.

I don't recall if and to what extent the technology analyzes electrical signal behavior of the driver but i suppose it does, at least for voicecoil heating analysis.

 

[Audionaut]

Regarding the KCS, let me add Powersoft IPAL for previously disclosed technologies with the same goal.

Yes, I had looked into the IPAL system before, but it seemed too complicated to set up for home use.

IpalMod

www.powersoft.com

I was interested in this driver

18IPAL 2Ω - LF Drivers - B&C Speakers

www.bcspeakers.com

The system uses active feedback, which not only measures electrical parameters such as current and voltage, but also—via a differential pressure sensor (which is structurally coupled to a small, defined port on the front and back of the diaphragm)—the actual force acting on the diaphragm. A DSP and amplifier adjust the signal in real time so that the diaphragm movement corresponds as closely as possible to the setpoint.

 

[Nugget_]

They use a diffrential pressure control sensor drilled in the baffle, to measure sound/air pressure on both sides of the baffle

via a differential pressure sensor (which is structurally coupled to a small, defined port on the front and back of the diaphragm)

Where do you get the information, that the sensor is coupled to the diaphragm? As I wrote before, it is just drilled into the baffle. Powersoft has this drawing. I see what "port" you might mean (the hole below the sensor), although i don't quite understand why there would be any port/hole. Does seem to destroy the purpose of measuring both sides of the baffle independently...

but it seemed too complicated to set up for home use.

Why complicated? From my point of view, just expensive. And it may require a request to Powersoft to sell you the IPALMod, as it is not readily available on the market for private customers. But that shouldn't be a problem, AFAIK they give them out if you ask nicely and pay.

 

[kemmler3D]

What's different about them?

Noise cancellation is more or less linear, AFAIK it just takes a delay + gain + EQ operation to cancel out the incoming wave effectively.

Distortion cancellation is not linear, I think you need to generate a different set of harmonics for all inputs. The software that replicates the distortion profile of guitar amps (at least up until recently) required thousands of measurements of the original amp to essentially create a lookup table of distortion profiles that is referenced in real time. I think recently they've started using machine learning or "AI" to simplify this effect. To cancel out distortion you need to be able to generate the correct distortion for all inputs to the system with low latency, it's not trivial.

 

[tineared]

Noise cancellation is more or less linear, AFAIK it just takes a delay + gain + EQ operation to cancel out the incoming wave effectively.

Distortion cancellation is not linear, I think you need to generate a different set of harmonics for all inputs. The software that replicates the distortion profile of guitar amps (at least up until recently) required thousands of measurements of the original amp to essentially create a lookup table of distortion profiles that is referenced in real time. I think recently they've started using machine learning or "AI" to simplify this effect. To cancel out distortion you need to be able to generate the correct distortion for all inputs to the system with low latency, it's not trivial.

I understand that method of the input detection will be different, but since the distortion characteristics of the driver is a known quantity, it should be fairly straightforward to correct for it. Velodyne did it with their subs years ago. I believe they used some sort of accelerometer to accomplish it.

 

[Soniclife]

The Kef LS60 uses modelling and feed forward DSP to reduce distortion in the bass drivers, their white paper discusses it.

 

[Serval]

Hi,
Kartesian is pushing the state of the art in loudspeaker drivers to a new level with their VKI lineup.
They also have a YouTube channel where they go into the technical details
Découvrez _vKi | .Kartésien https://share.google/Dl712tCihdxrN2S84
Produits https://share.google/7kbgn1NCxRQtEcRoB

 

[Haskil]

That depends.

No, it's not new, because Klippel isn't the first to actively "control" loudspeakers. The basic idea of monitoring the movement of the diaphragm and correcting deviations has been around since the 1960s and 1970s. Particularly well-known are the Motional Feedback Systems (MFB) from Philips and the sensor-based controls from Backes & Müller. Both approaches work with a real mechanical sensor on the loudspeaker – usually an accelerometer or other motion sensor – that measures how the diaphragm actually moves. This signal is then compared with the input signal and corrected via a closed control loop. So if the diaphragm deviates from the desired movement pattern due to motor nonlinearities, the suspension, or the influence of the enclosure, the electronics intervene and counteract this. The goal is to track the diaphragm movement as precisely as possible, thereby significantly reducing distortion and improving control of the low-frequency range.

Yes, it is new, because while Philips and Backes & Müller use real sensors, Klippel attempts to calculate the membrane movement from a physical model of the loudspeaker. The system simply measures electrical variables such as voltage and current at the voice coil and combines these with detailed parameter models of the driver. This data is used to estimate in real time how the diaphragm is currently moving, and the drive signal is corrected accordingly. You could say that instead of measuring the movement directly, it is calculated virtually.

The real innovation lies not so much in the idea behind the control system itself—which has been around for a long time at Philips and Backes & Müller, for example—but in the way it is implemented. Klippel replaces the physical sensor with a mathematical model and uses the DSP computing power available today to apply this model in real time. This allows similar corrections to be made even with standard speaker chassis, without the need to integrate a special sensor into the driver. This is particularly interesting for series products because it reduces mechanical complexity.

Klippel uses a virtual "this is how this driver should behave in this situation" based on data measured in advance from a driver prototype and feeds this data into the correction program of the purely electronic circuit without mechanical sensors.

I find this approach highly interesting, even for very high-quality loudspeakers and not just for small Bluetooth devices.

And the French Cabasse in 1975 with a series of Monitor speakers used on public radio and also high-fidelity speakers that used a piezo sensor and a comparator that reinjected a signal in real time, correcting the incorrect movement of the speaker... These speakers were famous and are still sought after on the used market. Their production must have ceased around 2000. Afterwards, they greatly refined a 3-way concentric ring driver.

 

[Redacted]

Sounds like the mighty hand of marketing seizing a small design innovation to save the costs of using a larger cabinet or driver..
I'll not hold my breath until I see test results or hear it myself...

 

[ppierre]

Does it matter if it's cheaper while having same or better performance?

In then end, the volume and weight have a cost. From manufacturing to logistics. A technology solution who half this cost is nearly free lunch !

 

[Redacted]

Does it matter if it's cheaper while having same or better performance?

In then end, the volume and weight have a cost. From manufacturing to logistics. A technology solution who half this cost is nearly free lunch !

All fine by me. I'm just suspicious of smart packaging potentially being represented as something else that it may not be, in the audio sense. We'll see if one has legs...

 

[RetroStereo]

Thank you, @Keith_W, Yes, quite Fabulous all that is comeing, isn't it? Is it reasonable to suggest that all this began with the Bose 901 speakers?

I would say very unlikely.

I knew Amar Bose in the mid/late-'60's and attended a demonstration of his sound wall concept - an early precursor of the finished 901 product. It was a 7- panel array, each containing 5 identical 6x9" drivers. I don't know all the details, but essentially, the left-most and right-most panels got pure left and right stereo signals, the center panel got equal left+right signals, and the two additional panels between left & center and center & right got proportional parts of the two channels.

There were two major concepts incorporated in his system: 1) multiple small drivers meant that none had to work hard to deliver distortion-free high-SPL sound, and, 2) by spreading the two-channel sound over a wide sound stage, the impact was intended to be more realistic.

Obviously, a 7-panel array was an unrealistic product; however it led to the direct/reflected concept, including altering what information was reflected where with the active controller that became the wildly popular 901 system.

Edited to add: While intriguing, I never particularly liked either version. I was an AR3/AR3A advocate, then a Dahlquist DQ10/1W, and, finally, a MartinLogan Monolith fan.