What goes into every Electromotive Laboratories loudspeaker system?

As with any technologically advanced system, the final product is a result of the hard work and learning of everyone who made each component for the system. Choice of components is critical to the end product performance and the better fit the individual component the better the end result will be. As with many systems the components react with each other in complex ways, one cannot simply choose a component in isolation based on written specifications of questionable validity that are intended to sell the product to the masses.

Careful selection processes must be followed.

While specifications allow us to narrow down component choices there is a lot of information missing that can only be obtained when the components are inserted into a system and the component is allowed to inter-react with others. It is a long and painstaking process where the exchange of one item for another can have far reaching consequences in the final result.

To arrive at a final choice of components for a specialised complex system can take years of not only auditioning components, but learning how far the interdependent reactions of certain components can extend into other components of a system.

When we are dealing with electromotive transducers they not only have to do what the drive system asks of them, but their physical and electrical properties can severely affect the performance of the systems that drive them. Even a seemingly benign inanimate object such as a cable or connection can behave in a different manner depending on where it is and how it is connected.

When we look deeper into how things are made to work as a single seamless unit, there are immense complexities.

The Electromotive Laboratories systems are all intricately engineered to work as seamlessly together as possible.

Here is a little of what went into the development of our studio monitor range.

The enclosure. It is not just another simple box.

We start with the cabinet. In order to produce any degree of high output full bandwidth sound we really must isolate the forward radiation of the loudspeaker cone from the rearward radiation that is in inverse polarity to that of the forward radiation. In most sensible cases we require a closed volume of air, free from as many resonances and distortions as possible and this requires a robust sealed chamber of adequate dimension.

The lower in frequency we wish to go the larger this enclosure of air needs to be. This air mass ideally should not resonate at all. Any resonance is a distortion of the time response of the loudspeaker and is superimposed upon the signal that is to be reproduced. An idealised enclosed non-resonant volume of air of wide bandwidth is a very hard thing indeed to achieve, a “holy grail” so to speak.

We are unfortunately, always limited by practical size constraints and these bring us great challenges. In the majority of cases a fully sealed well controlled enclosure is somewhat lacking in good low frequency output, it would require the inconvenient summation of more devices in the lower ranges to really adequately extend a practically sized loudspeaker to low enough frequencies to be said to be full bandwidth. Very large sealed systems can go adequately low, but they are often of impractical size. One further serious compromise of a small to medium volume sealed loudspeaker cabinet is that it is not possible to drive the low frequencies harder to achieve more output and maintain a distortion free signal at anything other than low to medium levels.

An inconvenient property of the enclosed air is that it resists compression of the enclosed air far more than it resists expansion of that air. To explain; if we had a 1 litre enclosure we would need infinite energy to move a piston back 1 litre to compress the air to absolutely nothing, yet if we moved a piston 1 litre forwards we have only reduced the internal pressure by half, a relatively low energy process in comparison. As a result of this, any long excursion drive unit in a small to medium volume sealed enclosure will have an asymmetric air load on the cone, there is more resistance to inward movement than outward movement, so even if the drive unit has adequate excursion, equalisation of the input signal to extend low frequency response will result in increased distortion, and also a asymmetric back electromotive force upon the amplifier causing further difficulties with the drive signal.

A sensible way to extend the response below the usual roll-off of a sealed enclosure it has been found that a vented cabinet in which a mass of air is trapped in an enclosed path (open at either end) between the inside and outside of an enclosure can extend the low frequency output by creating a second acoustic source. This is a common way (tuned vented box) of solving to some degree the issues of sealed boxes in the lower frequencies. This is however not a magic bullet. The system is inherently resonant, and if the designer is not careful it can lead to things such as “one note bass” or boomy sounding loudspeakers, both of which cannot be classed as a reference standard product. It is common for loudspeaker system manufacturers to build a box to a predetermined size for sales reasons, and thus find themselves limited in bass response due to dimensional constraints. As no seriously band restricted loudspeaker could be said to be a reference standard monitor most manufacturers try and push the systems as low as possible through extended bass tuning, unfortunately this leads to a very resonant bass when the tuning is in the bass range between 50Hz and 75Hz.

At Electromotive Laboratories our systems are all ported systems in order to avoid the non-linear properties of sealed enclosures, but we tune all of our systems at the lowest possible frequencies where the wavelengths are so long and slow that port resonance is not a significant factor in audible time domain interference. We couple this with adequate port area to avoid high velocity port artefacts, and a highly complex enclosure internal geometry to prevent any enclosure acoustic resonances and internal standing wave issues.

There is a vast amount of knowledge and experience that goes into our enclosure designs.

It is all about how things are mechanically put together.

In addition to the enclosure air volume tuning we pay great attention to potential enclosure distortion issues, At high outputs, enclosures will flex and mounted components will be subject to quite severe shocks and vibrations. If an enclosure is not well constructed and designed, it will tend to cause severe distortion at high output levels. There are many ways in which an enclosure can cause such problems.

One disadvantage of using a horn (or waveguide) HF device is that it places a significant weight a distance behind the baffle, this can form a lever, that when excited by the low frequency energy can amplify the stresses on both the horn mountings and the baffle itself. These stresses can be seen in seemingly unexplained resonances in the low frequency band and also manifest themselves in higher frequency mechanical vibrations of mounting hardware.

Simple internal resiliently attached supplementary support braces for the motor structure do not resolve this issue, a kind of circular hole that a drive unit slides into as the horn is inserted does not firmly grasp the motor structure.

In the Electromotive Laboratories series of loudspeakers we have eliminated all these issues by major design changes. All of our waveguide structures are incorporated into a one piece baffle structure, there is no fixing to suffer stresses and vibrations. All motor devices are solidly mounted on a very substantial mounting frame that bypasses the waveguide mechanical mounting, the waveguide supports nothing except itself, and finally the waveguide is heavily acoustically damped by irregular latex based beads bonded to the rear of the waveguide.

All waveguide mouths are blended seamlessly into the baffle structure to avoid mouth termination reflections.

To avoid any baffle resonances all of our baffles are dual layer bonded structures of at least 4cm thickness. It is not a cosmetic reason that we place our drive units on the smallest possible side of the enclosure. As the front baffle has the least integrity of any side of an enclosure, due to all of the essential holes for the sound to come out of, and lack of easy strengthening bracing positions it is prudent to have as little of it as possible and what is there to be as supported as much as possible by all the other cabinet side walls. Our enclosures are engineered to have as strong a baffle structure as possible with as few mechanical fixings as possible.

Our drive unit chassis connection to the enclosure is similarly robust. Nothing is resiliently mounted. When a loudspeaker unit is mounted on a resilient sealing gasket (such as rubber or cork) it can often tend to bounce at high power outputs. There are sometimes hundreds of watts of mechanical force acting upon the drive unit which the mounting hardware has to cope with, force that also vibrates things severely. If a drive unit is fixed on a resilient mount these forces can be adequate to cause unwanted noises, especially at the lowest of frequencies these noises can often seem inaudible, but can often be very problematic. At Electromotive Laboratories we choose to use sealing materials that will compress as flat to the cabinet as possible when the loudspeaker is fixed in position, all bolts press down directly on the chassis and hold that chassis tight to the cabinet. Every loudspeaker we make is carefully stress tested with a full power 20Hz signal before delivery to ensure nothing can make any unwanted noise.

Then it is about how we get all the parts to work together as one.

The ideal loudspeaker is always a single point source, the more sources a loudspeaker has the more problems the designer faces, unfortunately nobody as yet has invented a single high-power drive unit that can work from below 20 Hz all the way to 20,000 Hz. This means we need to make two (or more) devices work as one. For this we need to separate the signal into bands that each device is capable of reproducing. This is usually done by an electronic crossover filter.

Previously, before we had the understanding we have today this was done by a passive crossover network of capacitors, inductors, and resistors. This is now comparatively stone age technology.

These types of filters introduced as many problems as they solved, and often more problems. The components were incredibly sensitive to circuit resonances, and the loudspeaker drivers can change electrical properties as they operate causing both unpredictable constant frequency and phase shifts.

It was not possible to time align the systems and different amplifiers could produce vastly different source loads to the components in a resonant circuit affecting the tuning and causing vastly different behaviour from the loudspeakers.

Additionally, a passive crossover would all but eliminate the damping of the drive unit piston by the amplifier causing excessive resonances and poor transient response.

As a loudspeaker is driven harder we see the voice coil heat considerably, as this happens the electrical resistance increases and as crossover frequency is a function of this we see the crossover frequency change. It is very common to see a passive crossover develop a hole in the frequency response between drivers as they get louder, this is especially so in systems that use dome tweeters where coil heating is a bigger problem.

No professional loudspeaker should have a passive crossover in this day and age, the technology is seriously flawed.

A very early four way actibe studio monitor loudspeaker layout designed by us in 1984 before we better understood the unresolvable problems with such systems and such poor layout and excessive crossover bands.

The electronic crossover was always a far better option, but significantly more expensive. It required sometimes up to four times the amplifier count and often the crossover itself cost a thousand or more (Dollars, Pounds, Euros).

The electronic crossover allowed each driver to be perfectly driven from a low impedance source and driver property variation shifts only affected output level in that band (still a problem, but far less than a passive system)

Driver resonance was better controlled and the amplifier had a simpler load to deal with, thus, could perform better.

An analogue electronic crossover usually just provided a simple filter slope of fixed steepness (Q) and frequency, although some provided switchable options. More complex units provided basic phase and limiting options. In general a loudspeaker designer had to choose from a basic range of slopes and frequencies to get close to the properties of the drive units selected.

It was normally the case that drive units had to be selected to be nominally flat in the frequency domain within the band they needed to work in, and as most drive units are not flat for very long in the frequency domain it often leads to many ways being used to keep optimum flatness.

This issue led to four or even five-way systems being common place if one wanted to provide a flat amplitude response.

In most cases these systems were catastrophically bad in the time domain, leading to a fuzzy, soft, ill-defined sound that some found less obtrusive, unfortunately there was no accuracy in this type of system.

This was very much like looking at a picture or a beautiful portrait with a soft focus blur, many see this lack of detailed image as having a delicate beauty to it, but therein lies the hidden trap, for almost any old ugly face can be made beautiful with a little soft focus blur. When we are the ones creating the beauty we are the ones that need to see all the flaws that have to be treated, not the ones that should appreciate our work being made to look artificially nice before we have finished it.

Parts of the spectrum within -10 dB range where inter driver signals will highly likely conflict, these can be increased if the drivers have peaks beyond crossover that are adequate to extend this zone.

A common 4 way 18 dB/Oct crossover.

A crossover of shallow to medium slope can have two separate acoustic sources working at the same time for well over one octave in a way that both are audible.

In a four or five-way loudspeaker this can mean that the entire useful audio bandwidth is produced from multiple dislocated sources effectively causing the system to be impossible to time align.

As most analogue active electronic crossovers did not have time alignment facilities and such things were not closely considered this led to some systems having vast time domain problems, it could take a few milliseconds for all the different bits of the waveform to arrive at the listener. This can ruin positional and transient response performance and make it very difficult to accurately mix on such loudspeakers.

We now know far better than to do this. The advent of very powerful easily accessible analysis systems and DSP processing has enabled us to fix most of the issues that led to such terrible compromises. We are now able to extend the range of loudspeakers that were considered to not be flat through proper inverse modelling of the response anomalies, we are able to correct time domain (phase) issues to match units perfectly, and we are able to create super high rate filters to avoid band overlap as much as possible.

Parts of the spectrum within -10 dB range where inter driver signals will highly likely conflict..

A modern high order two way crossover as used in Electromotive Laboratories systems

On all Electromotive Laboratories systems, we only run two-way systems from around 20 Hz right up beyond 20,000 Hz. We cross over at the most optimum frequency for both amplitude response and directivity control. And we only have both drivers working at the same time for a fraction of one octave where they are correctly time aligned to cover  a wide vertical listening area.

We have also been able to develop crossover slopes that are designed in the acoustic domain, not simply the electronic domain. In this case the actual acoustic output of the drive unit is what follows the exact cut-off slope, not the electrical input signal. Many systems have poorly controlled responses within the driver slopes that can create strange anomalies in the output that are often not easy to resolve, our acoustic domain crossover slopes provide exceptional control through crossover.

Acoustic output through crossover producing correct slopes (LEFT)

Actual resulting response through Crossover

(RIGHT)