A Symmetrical Transducer Topology

[smowry]

I have not designed a transducer in 10 years. I am retired in Phuket, Thailand. However, the work of Dr. Jack and his staff at KEF on their VECO feedback control revived my interest in transducer design. I realized that some of the alternative transducer topologies that I had independently investigated could have applications with control systems. I feel like there is some unfinished business; however, not to the extent that I would unretire and accept clients and/or commercial transducer development projects. Having said that, I am fascinated by the ASR forum phenomenon and the job that Amar et al. has done to bring together the AV community. Then my objective of this thread is to simulate and design a coaxial transducer array concept real time, posting simulation results for comments and discussion. In that way the ASR community becomes quasi-consultants to a, hopefully interesting and fun, transducer design exercise. Typically, transducer design is a secret agenda. Don't tell anyone what we are doing and therefore don't exchange ideas. Whereas, I agree with Dr. Jack, nothing is really new in transducers. Most transducers produced today still use the Rice-Kellogg topology with cantilever voice coils that were first used in 1924!

Here's a bit of history of the evolution of this concept dating back to 2005.

1. https://pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Mowry_Steve/Steallus_Motor_Design.pdf

2. https://pearl-hifi.com/06_Lit_Archi.../Mowry_Steve/Air_Core_Tweeter_Magnet_Assy.pdf

3. https://pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Mowry_Steve/DRAGONS.pdf

I plan to use Loudsoft's FINE software. I have contacted Peter and Dorit and I am waiting for their reply. I have invited them to visit this thread. Peter is perhaps the most experienced transducer engineer ever with the highest number of projects of all time.

Kindly find a transducer concept sketch below. None of the parts have been designed or documented except the tweeter. The parts in the sketch are essentially place holders. The concept has evolved quite a bit recently. Note the simplicity of the assembly.

The primary magnetic gap has linear permeability. The center coin is 304 Stainless, which has about the same electrical conductivity as that of the magnets and like the magnets has relative permeability ~1.0. The primary light alloy basket is the motor assembly's heat sink. A unique feature is that the high permeability secondary magnetic gaps contain passive sensor coils. With regards to the lumped parameter transducer model below, and knowing that if a AC Voltage is supplied to voice coil which is in a DC magnetic gap, then a velocity is created in the moving assembly but this implies that if we apply a velocity to a passive coil in a magnetic gap, then a voltage, is generated.

So now the transducer has additional terminals connected to the sensor coils. This concept requires a feedback controller and a current source amplifier for the low frequency transducer to reach its highest performance potential. The AMT tweeter has flat impedance and it is not know if a current source amp would improve performance. The coaxial array should be active with DSP so that the tweeter and midwoofer can be time aligned.

Typically, the first comment that I receive from colleagues is that it is isobaric. Well I don't see that, and it is misleading. The way I see the diaphragm topology is as a sandwich composite, a really thick and poor sandwich but no less a sandwich. There are faces and a core. The core happens to be air which is a fluid. The moving assembly is effectively one diaphragm with a front and a back side with air in between.

There are subtopics within this thread related to the concept development and implementation.

1. DSP
2. Feedback Control
3. Current Source
4. Almost Air Core Transducer
5. Spider-less or are the surrounds the spiders?
6. Full complementary symmetrical topology
7. Coaxial AMT

So please let the discussion begin. All serious comments, questions, and/or concerns are welcome and encouraged.

 

[olieb]

This looks like "the sandwich" could be an awesome basis to build a dipole/OB driver/speaker!
I suppose the foam part is thought to be open cell (to dampen cavity resonances in the sandwich)?

As I understand it, the tricky part of a coaxial - among other things - is the integration of the tweeter with the diaphragm and surround so that the FR/diffraction does not go awry. So the motor and assembly is only a (minor?) part of the problem.

I did not get through the lumped parameter transducer model but from what I understood in the text(s) from Rob Schmid about motion feedback a velocity sensor seems of somewhat limited benefit.

1. Proportional Velocity Feedback, which creates a force that counteracts movement. As this force is proportional to the velocity it acts like a mechanical damper, reducing the Q of the resonance (the yellow arrow). This is favorable, however hardly other beneficial effects on for instance distortion are noted as the effect is limited to the resonance frequency region.

 

[smowry]

Here's an entire thread on the topic of Motion Feedback Control. https://www.audiosciencereview.com/...-distortion–explained-by-its-inventors.64033/

 

[Curvature]

AMT

Why AMT, if you don't mind? I can understand the electrical advantages, but the vertical directivity is always restricted.

 

[smowry]

The Project objective is to minimize distortion, so the vertical directivity compromise was accepted. Furthermore, this is a less is more exercise, no spider and no iron around the gap. The typical dome tweeter's surround, surround to dome termination and voice coil to dome termination are eliminated along with a streamlined leadout from the AMT's film diaphragm.

 

[Curvature]

The Project objective is to minimize distortion, so the vertical directivity compromise was accepted. Furthermore, this is a less is more exercise, no spider and no iron around the gap. The typical dome tweeter's surround, surround to dome termination and voice coil to dome termination are eliminated along with a streamlined leadout from the AMT's film diaphragm.

I would be very interested to see the result.

 

[smowry]

My question: Why are there no vent holes within the motor assembly? The only function of the center hole is to provide a path for the AMT's leadouts.

My Answer: I despise small hole within the motor assembly along with the associated claims of convection cooling. Many years ago when he first introduced his distortion analyzer DA-1, Dr. Wolfgang published the results of a thermal data acquisition with his test system. He first ran the system with the DUT being a typical small woofer and made the acquisition. Then he ran the same test on the same DUT but this time he covered the "vented pole". Surprisingly, the voice coil in the woofer with the covered vent ran cooler. Although, it was not measured, I claim that there was some noise radiating from the uncovered vent.

Now did the designer/manufacturer perform CFD analysis to verify claims or did he assume something? And all for the sake of a stupid dust cap. Well this concept has no dust cap and no inherent whistles, less can be more. Jussi at GENELEC told once told me that without holes he can see it in the impedance curve. Well in this concept one one will see any of that; there is no dust cap to form the boundary of the pump diaphragm. This concept uses both convection and conduction cooling by utilizing the primary basket as a heat sink that is as close to the voice coil as possible and has huge surface area. The concept does not pump air through a spider nor through small holes. The front diaphragm has behavior, while the rear diaphragm has anti-behavior. The stiffness of the trapped air adds stiffness to the diaphragms; however, it also adds its mass.

 

[Curvature]

How will you be modeling these interactions?

 

[smowry]

The thermal model with FINEMotor Pro and FEMM, the diaphragms together, the sandwich, I am not sure yet, any suggestions? I could try FINECone but I am not sure if I can enter the more complex geometry and the question of the accuracy of the results. My plan is to begin with the motor assembly.

 

[Colin5619]

Very interesting

 

[kemmler3D]

Really interesting stuff, I wish I was technical enough to contribute something useful here!

 

[Curvature]

The thermal model with FINEMotor Pro and FEMM, the diaphragms together, the sandwich, I am not sure yet, any suggestions? I could try FINECone but I am not sure if I can enter the more complex geometry and the question of the accuracy of the results. My plan is to begin with the motor assembly.

What about COMSOL? More or less standard for FEA these days as far as I know (and I am only a very distant observer). @René - Acculution.com may have a suggestion.

 

[smowry]

Concept Assembly Update

 

[smowry]

COMSOL is too everything for me.
1. Too expensive
2. Too difficult to learn
3. Too much computing power required
4. Too labor intensive

I was a VF OPERA and HKS ABAQUS user for years. I'm done with that level of analysis. It's just no fun anymore. However, those tools are responsible for getting me to where I am with regards to this concept. And for example, the assembly is designed to facilitate 2D simulation. When we designed the racetrack surround for GENELEC's Ones. The mechanic in Finland had to let the COMSOL run all night to converge to a solution. When COMSOL was used for symmetrical geometries, 2D, like in their 3.5 inch, then the ABAQUS results were essentially equivalent to COMSOL's. The 2D ABAQUS simulations converged in minutes.

JP2018157592A - Loudspeaker having suspension member in which diaphragm of loudspeaker driver is attached to chassis of loudspeaker driver, driver, and driver - Google Patents

PROBLEM TO BE SOLVED: To provide a loudspeaker driver in which a high level distortion caused by a non-liner rigidity normally occurred in a driver using a progressive suspension member.SOLUTION: A suspension member 100 in which a diaphragm of a loudspeaker driver is attached to a chassis of the...

patents.google.com

 

[smowry]

Modular Gauged Assembly Example

 

[René - Acculution.com]

COMSOL is too everything for me.
1. Too expensive
2. Too difficult to learn
3. Too much computing power required
4. Too labor intensive

I was a VF OPERA and HKS ABAQUS user for years. I'm done with that level of analysis. It's just no fun anymore. However, those tools are responsible for getting me to where I am with regards to this concept. And for example, the assembly is designed to facilitate 2D simulation. When we designed the racetrack surround for GENELEC's Ones. The mechanic in Finland had to let the COMSOL run all night to converge to a solution. When COMSOL was used for symmetrical geometries, 2D, like in their 3.5 inch, then the ABAQUS results were essentially equivalent to COMSOL's. The 2D ABAQUS simulations converged in minutes.

JP2018157592A - Loudspeaker having suspension member in which diaphragm of loudspeaker driver is attached to chassis of loudspeaker driver, driver, and driver - Google Patents

PROBLEM TO BE SOLVED: To provide a loudspeaker driver in which a high level distortion caused by a non-liner rigidity normally occurred in a driver using a progressive suspension member.SOLUTION: A suspension member 100 in which a diaphragm of a loudspeaker driver is attached to a chassis of the...

patents.google.com

2D in COMSOL would probably be the same time as 2D in Abaqus, so not really a fair comparison here.

 

[smowry]

Hi Rene',


Nice to see that you stopped by. Let's see how thing go but once I select the cone/surround geometry I could send you a *.dxf file of the sandwich and invite you to simulate the vibroacoustics in 2D. What do you say?

 

[smowry]

Amir @amirm, I am sorry that I misspelled your name in the opening post. However, my comment was sincere.

 

[René - Acculution.com]

Hi Rene',


Nice to see that you stopped by. Let's see how thing go but once I select the cone/surround geometry I could send you a *.dxf file of the sandwich and invite you to simulate the vibroacoustics in 2D. What do you say?

I only take on paid projects, and I am fully booked for Q3. I follow what I can on the side line (forums and such) and contribute to the community with articles, but not with help in projects as such.

 

[smowry]

This is not a commercial endeavor. My goal is to generate interest in alternative transducer topologies and share ideas. I agree with Dr. Jack, "The trouble is that speaker engineering is an 80 year mature topic. Nothing is totally new...", but those claimed to be inventors continue to send Patent Application to the USPTO. Why? The more Patents a company has, the more money they can borrow. It's become a cottage industry for Patent Attorneys and the USPTO. Most Transducer Patents are a natural and straightforward progression of existing engineering design methods. So can we move beyond the Rice-Kellogg Cantilever Voice Coil Topology please?