Speaker Designer Roundtable

Blog
Categories:
Floorstanding,
Stand-mount
Speaker Designer Roundtable

As part of our focus on affordable speakers, we asked four of the Industry’s leading designers to describe their approach to creating the most musical speakers for the price.


Dean Hartley
Technical Director, Monitor Audio

Dean Hartley joined Monitor Audio twenty years ago as head of product development. Over this period, he has been responsible for introducing new technologies, innovations, and numerous patents for the company as well as continuing to serve as the company’s “golden ears.” Today, Dean directs a team of twenty acoustic, electronic, and mechanical designers at Monitor Audio’s research and development facility in England.

Dean is a trained electronic engineer with a career spanning 30 years in loudspeaker design, working for well-known hi-fi and professional audio brands. As a musician, audio enthusiast, and technologist, he continues to be excited by the challenge of re-invention.

Is designing an inexpensive speaker easier or more challenging than creating a larger and more elaborate model? Why?
The goals for a new speaker design may be focused in many ways. With a lower-priced speaker, the common denominator always comes down to cost. This involves considering much finer details earlier in the design phase, where we would be looking for greater gains in a more expensive speaker, as cost is less of a concern. So, the design window of a more expensive speaker can be broader and more flexible.

The designer of an inexpensive speaker obviously has to make compromises and trade-offs. What musical qualities are you unwilling to give up, and which qualities can be sacrificed for the most satisfying overall performance?
We never consider compromises in musicality, or in the key ingredients that determine musical enjoyment. Any trade-offs are usually related to structural integrity of cabinet construction, driver magnetics, and crossover components. With a high-end loudspeaker our aim is to create a faithful (often referred to as neutral) sound balance, while retaining the desired musical qualities. We always aim to ensure our speakers are enjoyable to hear and to not “doctor” the balance to make the speaker sound “enhanced” in certain areas. This is perhaps the trickiest part of designing a lower-cost model, as the cabinet and other components can impart their effects on the speaker’s character.

We also need to consider the typical quality of equipment being partnered with the speaker. An example would be with a two-way stand-mount speaker (like our entry-level Bronze 2), where the design must yield a more efficient output and easier load to get the best from an amplifier that may not have high current delivery. With a high-end design (like our Platinum II PL100), we can consider lower efficiency and resulting lower LF response.

How much better are today’s speakers of a given price than those of ten years ago? Why?
I can only comment on developments we have made in our design with the investment in advanced FEA, simulation, and 3-D prototyping over the last ten years. I know people will have heard these words so many times now that they have lost meaning or ability to create a glimmer of interest. But let’s look at how this has made speakers better. Ten years ago (after the design of our first Platinum Series), we only had basic simulation tools for magnetics. Today, we can model a complete loudspeaker using multi-physics FEA. This means we know the effects of small changes in components or material parameters “on-the-fly.” We get closer to our target the first time, with subsequent prototyping feeding actual measured data back into the model to close the loop. Let’s use an example: We encounter an undesirable break-up mode in a new speaker cone design. Ten years ago, we would have made practical prototypes, measured them, and found this problem. At that point we would not know what part of the geometry or material issues to deal with to correct it. So, guesswork would lead us to perform many iterations of prototypes. We may stumble upon one that works, but still not be sure why. This has been the case for most driver designs in the past. Today, we can get closer to eradicating these issues long before any physical prototyping begins. The advancements in speaker design over the last ten years are primarily down to applying our trade using new technology, with smaller gains in the new materials available to us.

Do you think that designing very inexpensive speakers confers advantages to the designer when he’s creating upper- end models?
I am not sure there is a great correlation between high-end and price-driven designs, as the thought process and focus are quite different. We use the words “trickle down” a lot, and they’ve perhaps been overused when a component from a higher-level range finds its way into the range below. Making lower-cost speakers usually means higher production volumes, and to ensure efficiency the design must be close tolerance with (close to) zero rejections. Applying these volume-production engineering techniques and processes to the high end is something that we are very keen on. So, you could say that in some ways high-end designs benefit from the application, knowledge, and experience of designing lower-cost speakers. While this may not be evident to the customer, we may, for example, have found a new assembly technique that resulted in the closer tweeter tolerances required for a low-cost/high-efficiency design. So rather than affecting the fundamental design parameters of drivers and system, the benefits are more related to gains in production engineering and assembly. 

How much of the design is done in a computer and how much in real-world experimentation and listening? Describe the process.
The design is an iterative process with gated decision points, like most logical developments. From a concept we will determine what is required for a new design; how much of the design is being re-used and improved, or completely re-thought. We will commence simulation of the parts in question, which may be suspension, cones, magnetics, and the like. Once we have the individual parts to an acceptable level, we will simulate a whole driver and adjust parameters to match our target design. From there we will make a first prototype. If we are confident, then we may tool actual production-spec parts, and in some cases the requirements of tooling mean we must do this anyway. Once we have a driver, then we will run a series of measurements (LSI and LPM) to determine if it meets our target and simulation. If it does, then we will put this together in the system and start the listening/initial-voicing process to give us a starting point. If not, then we will closely analyze the physical prototype and simulation to try and find out why they fell short. This may result in a second, third, fourth prototype, as required. I have just referred to the driver design, as this is the heart of any speaker. However, in parallel we may be developing system and crossover design ideas. Or DSP and filter blocks, if it’s an active design. Our voicing/listening process can vary in time and number of sessions, depending on how happy we are with the results. A typical single speaker design will involve around 50 hours of listening and refining throughout the design stages. In our case the design stages are Prototype, Engineering Stage 1, Engineering Stage 2, Pre-Production, and Mass Production. My acoustic team (including myself) maintain design responsibility, listen, and provide approvals all the way through to Mass Production. We don’t hand this part over to a production team as we want to ensure complete consistency and integrity in all our designs.


Dr. Jack Oclee-Brown
Head of Acoustics, KEF

Dr. Jack Oclee-Brown received an MEng degree in acoustical engineering from the Institute of Sound and Vibration Research (ISVR), University of Southampton, UK, in 2004. Since then he has been with KEF Audio, where he now holds the position of Head of Acoustics. In 2012, he received his PhD from the ISVR for a thesis on the acoustic design of compression driver phase plugs. He is interested in all aspects of loudspeaker engineering and design. His work for KEF is currently focused on loudspeaker modeling, the development of software tools to aid loudspeaker design, and transducer design. 

Is designing an inexpensive speaker easier or more challenging than creating a larger and more elaborate model? Why? 
There is certainly less room for error when designing an inexpensive loudspeaker, since you’re often asking more from each aspect of the loudspeaker. If one area falls shorter than your expectations, this can really let down the overall performance.

The designer of an inexpensive speaker obviously has to make compromises and tradeoffs. What musical qualities are you unwilling to give up, and which qualities can be sacrificed for the most satisfying overall performance?
We always try and deliver both measured accuracy and musicality, but for an inexpensive or compact product it sometimes is not possible to do both. A classic example would be choosing between a flatter frequency response at crossover, or a higher crossover frequency that gives the tweeter an easier ride. We’d always aim to be able to do the former, but if in the listening this sounds strained or fatiguing we’d compromise the measured performance a little to give better musicality. Our best products are those where everything comes together nicely and there are fewer tough compromises in the listening room.

How much better are today’s speakers of a given price than those of ten years ago? Why?
Loudspeaker technology is very mature, and the fundamental behavior has been understood for more than fifty years. Consequently, the speaker has not changed whatsoever in some aspects, so a small bookshelf from ten years ago will likely have more or less the same sensitivity and bass extension as one you could buy today. However, computer simulation, materials, and manufacturing have improved significantly in this time. The products available today are much further refined than those of the past, and this is most obvious in the transparency and clarity of the midrange and treble.

Do you think that designing very inexpensive speakers confers advantages to the designer when he’s creating upper-end models? 
There are areas where the engineering techniques for an inexpensive and expensive product are quite dissimilar. However, in general the experience of designing inexpensive products can really help with upper-end models. For example, getting cabinet vibration under control in an inexpensive speaker is very challenging and means you have to really study and understand the behavior so you can have the biggest impact possible. This knowledge can be very usefully applied when you come to a higher-end design, too. The most useful aspect is doing a variety of work on many speakers of different shapes and sizes, as this gives you new challenges, keeps your skills sharp, and gets you familiar with the compromises of different approaches.

How much of the design is done in a computer and how much in real-world experimentation and listening? Describe the process.
A great deal of computer simulation is done in the early stages of the product design, particularly to develop the drivers for a loudspeaker. Using the computer it is possible to create hundreds of “virtual prototypes” and fine-tune the performance very intensively. This means that when we get first driver prototypes we already have a very mature design that performs far better than we could achieve by prototyping alone. It also means that we can normally put the first driver prototypes directly into prototype cabinets and start working on the crossover and listening to the product. In general, computer simulation is a huge advantage because it means you have a very clear idea of how each component should perform before you prototype it, and this means that you have fewer surprises and can focus on getting a speaker that performs really well in both measurement and listening. 


Andrew Jones
VP of Engineering, ELAC

Andrew Jones’ near lifelong passion for audio began in his early teens and quickly became focused on loudspeakers. This led to him to study physics with a special interest in acoustics, followed by six years of postgraduate research in the fields of crossover design and active noise cancellation. Subsequently he joined KEF, becoming Chief Engineer, followed by a move to the U.S. to work for Infinity, Pioneer, and TAD. He is currently VP of Engineering at Elac.

Is designing an inexpensive speaker easier or more challenging than creating a larger and more elaborate model? Why?
Overall, I believe it is more difficult. The cost constraints are so tight, with no allowance to increase the final selling price, that one has to be very clear on component part design and selection in order to meet those constraints. Knowing the cost of each part is critical so that one can make the needed trade-offs, and the control of design of all those parts is key to helping with cost optimization. The design also has to appeal to a much wider audience. Additionally, the speaker is going to be used with lesser quality amplification and source, so the likely sound capability of these ancillary components has to be taken into account, along with the sound quality of the recording.

The designer of an inexpensive speaker obviously has to make compromises and tradeoffs. What musical qualities are you unwilling to give up, and which qualities can be sacrificed for the most satisfying overall performance?
One of the compromises I always go for is to trade extended low-frequency response for higher sensitivity/maximum dynamic output from the speaker. Going for higher sensitivity gains you the ability to play louder for any given amount of input power, but how often is this required compared to how often will you miss hearing the bass in the music? I’m not, of course, advocating for hyped-up bass, but if we are to attract listeners that are more used to hearing music over headphones or in cars, where extended powerful bass is much easier to generate, then the budget speaker should be capable of offering some semblance of this bass performance.

How much better are today’s speakers of a given price than those of ten years ago? Why?
At first glance, today’s affordable speakers seem to be very little changed from those of yesteryear: typically a two-way system with most commonly a soft-dome tweeter and a plastic/woven/metal cone woofer in a closed or vented box. But they are definitely better. I would describe these differences as a greater degree of refinement. The reason is a mixture of materials science and manufacturing costs. Advanced materials for cones and domes have become available at much lower prices, with a wider range of sources and choices of variants. This allows for much greater ability to experiment and find the best material for the particular design in-hand. At the same time, as onshore manufacturing costs have risen and the manufacturing base has shrunk, offshore costs remain much lower yet quality of manufacturing has improved dramatically. Consistency of performance on even entry-level drive units is now at a point that even high-end manufacturers struggle to match. This has changed the price/performance ratio drastically.

Do you think that designing very inexpensive speakers confers advantages to the designer when he’s creating upper-end models?
For me it’s the other way around. Designing upper-end speakers and systems gives me the knowledge of what really good sound can be; it serves as a reference point. The goal in the affordable speaker is to distill the essence of that sound in such a way that the listener is still drawn into the enjoyment of the music, and not distracted from it by what the speaker is doing wrong. It comes down to the expectation of performance: seeking a higher performance than one would expect from a particular price point, and not being persuaded by arguments of “it’s good enough; what do you expect at this price point?”

How much of the design is done in a computer and how much in real-world experimentation and listening? Describe the process.
A lot of the design work is done on the computer these days. Modeling and CAD and measurement are so much more cost effective and widely available nowadays that much of the early design process is done this way. But the questions become what to model, and how accurate is the modeling. Widespread availability means that such techniques are used without a full understanding of their applicability or accuracy in the design under consideration, especially when it comes to measuring. It seems that there are more inaccurate measurements made than accurate ones! Therefore, real-world experimentation and real-world experience are musts for guiding one to the best uses of computer modeling.

The process, therefore, becomes model, build, test, listen, then cycle back to the beginning to adjust the model. The great thing about designing initially with computer modeling is that a greater number of iterations can be performed prior to building, thus either saving development time or allowing greater refinement within the allowed development time. Of course, most engineers are never going to finish early, but will keep on refining until the end!


Vince Bruzzese
Founder, Totem Acoustic

Vince Bruzzese spent six years at McGill University where he earned a degree in science. After a wonderful and fruitful 19-year career teaching science and math in both English and French, he founded Totem Acoustic in Montreal in 1987. From humble beginnings with a single model (the Model One), Totem has grown into a creator of a wide range of high-performance loudspeakers, with more than 50 models for every type of application. The company has specialized in small-volume enclosures that deliver large imaging and dynamics. Vince continues the principal design work for Totem Acoustic. His acoustic ethos has led to the development of Torrent technology and the creation of flexible real-world products that ring true on all axes.

Is designing an inexpensive speaker easier or more challenging than creating a larger and more elaborate model? Why?
The challenge is exacting for both expensive and modest designs. But it is actually harder to extract the absolute maximum in performance from less expensive models. Many of our speakers may look fairly simple from the outside, but they take many years of R&D to develop. For example, it took over two years of continuous development to finalize the 5¼" driver used in the Sky, which has a 3" voice coil. And although it’s not intuitive, not using a crossover in the woofer path is much more difficult than using a conventional crossover. The final tuning with the tweeter and its associated first-order crossover in a specialized cabinet requires hundreds if not thousands of hours of listening in choosing the absolutely few and correct parts necessary. Every driver is purpose-built for the particular application no matter what the price level. When Totem introduces a new model we want it to have a lifespan measured in decades rather than in years.

The designer of an inexpensive speaker obviously has to make compromises and tradeoffs. What musical qualities are you unwilling to give up, and which qualities can be sacrificed for the most satisfying overall performance?
Our objective is to create speakers that are “balanced” in performance. By this we mean that the overall presentation is “complete” so that the listening experience is stimulating, engaging, and rewarding over a very long time frame. The deepest “measurable” bass extension can be sacrificed in less costly models, but not in a way that detracts from the overall complete picture the speaker presents.

Our history has been one of continuous development. We want the bass, timbre, speed, pace, impact, and accuracy to be as true to life as possible across a wide listening area, and to portray the emotion of a recording. To realize these objectives we developed our own mathematical tables, not just the Thiele/Small parameters, but our own approach that we have fine-tuned over time as technology has evolved.

How much better are today’s speakers of a given price than those of ten years ago? Why?
I can’t really comment on other speakers on the market but I can expand on what we are doing. Flexibility and ease of integration are key. The altar of sound that existed in the past (a stereo system properly set up in the living room) does not really apply in today’s homes. Multi-functional rooms have replaced to a great extent the single-seat audiophile setup of a decade ago. This gives us an opportunity to create products that fit into people’s lifestyles. We want to include listeners who may have only experienced music through thin TVs, or earbuds, or smartphones.

Ten years go we envisioned a thin and compact full-range speaker that would provide off-axis intelligibility with no phase shift, and that would also provide true bass extension, in a unit that would fit on a wall. That product, the Tribe series, was made possible by our “Torrent” technology. Torrent is a driver design that is magnetically and electrically engineered to allow the woofer to be operated with no crossover in its signal path. The chassis and internal parts are all machined to high precision from high-tech alloys. The chassis provides a totally controlled foundation for the rest of the advanced components to perform optimally. Torrent provides remarkable on- and off-axis response along with phase correctness that results in a more genuine musical experience no matter where the speakers are positioned or where the listener is located. Before Torrent we couldn’t have achieved what we have with the Tribe series. People buy it for its looks and ease of integration but end up falling in love with the sound. We think it’s an important innovation in our industry. We also think that cost or placement limitations shouldn’t lower expectations for intrinsic performance.

Do you think that designing very inexpensive speakers confers advantages to the designer when he’s creating upper-end models?
No other aspect of audio design marries science, art, and emotion as much as speaker design. The speaker is an inanimate object that makes air molecules vibrate, but its function is felt on many dimensions of our sensory systems. Our hearing has evolved over millions of years, and that exquisite stimulation of our senses shouldn’t be limited to the exclusive few who can afford more costly speakers. Totem has a quest to extract the maximum performance, and to deliver the greatest emotional impact and connection, from all our designs. We’ve pursued technologies (such as Torrent) that have allowed us to do this.

How much of the design is done in a computer and how much in real-world experimentation and listening? Describe the process.
In Totem’s 30-year history we’ve seen an evolution in computer design that has certainly helped with modeling and simulations, as well as in the design work for our custom-installation and architectural products. However, that work on the computer is a small fraction of the time spent developing a new model’s function, aesthetic design, and capabilities. We approach speaker design as one would approach creating a musical instrument. It’s as much artisanal as technical. For example, our Sky and Signature One models each have more than 2750 hours of experimentation and listening, but only a few hours of computer modeling. This isn’t a statement on the efficiency of computer-aided design, but rather on our real-world, hands-on approach. We have accumulated perhaps the largest array of crossover parts anywhere in the world. We listen to countless permutations of these parts until we achieve the result we’re looking for. Computer simulations and measurement can show us only a very small glimpse of our goal of making the most musical speaker possible. We’re fortunate to have Lucy Lentini, our VP, who has a fine artistic and aesthetic perspective. She constantly challenges and inspires us to realize the highest ideals for our products.

Featured Articles

Lists