How to Better the Best
The Development of Meyer Sound's High Drivers
Photos courtesy Tara Meyer
All of the low frequency drivers used throughout our product range are designed in-house and manufactured specifically for us on an OEM basis by external suppliers. This arrangement allows us to fine-tune the designs over time, optimizing performance and improving integration with the cabinet designs and our in-house power amplifier electronics.
However, for many years we purchased high frequency compression drivers from manufacturers such as Yamaha (for three-inch units) and JBL (four-inch drivers) as standard customers. These companies were not interested in OEM arrangements and, as relatively small volume customers; we could not influence them to modify their products to our specific requirements. As a consequence, we have had to re-engineer these expensive bought-in units in order to extend their performance to the high levels or performance, reliability and consistency we demand.
The high driver is really the heart of the product and is critical to the final sound so the obvious answer was to develop our own range of compression drivers. However, it is a big step from taking and modifying someone else's product to building your own from scratch. Not many people make high drivers at this quality level - there's only really TAD and JBL - and we would be going into something unknown to us, but in competition with these well established products.
The benefits were significant though. Manufacturing our own high drivers would ensure security and longevity of supply. We would have complete control over all design parameters, allowing us to optimize the relationship between the driver, the electronics and the horns. Working directly with the electronics designer we can work around the loop much faster to optimize both the amplifier and the driver to work ideally with each other. We would also be able to refine the product in the light of experience or new technologies, improve their reliability and guarantee consistency between units.
However, we didn't want the pressure of having to develop a state-of-the-art high driver in a hurry so we set up a complete new facility with new people and ran it independently to avoid any kind of confusion with the driver rebuilding work which continued in parallel.
Our first high driver design was a 3-inch compression unit - the 1401M - simply because the technical issues involved were easier to address in a unit of this size. It is not under as much stress or pressure as a four-inch unit because there is less mass involved in the smaller mechanism. The 1401M replaced a modified Yamaha driver and has been in production now for almost three years. Only when we were confident about that design did we go ahead with the new four-inch unit, the 2010. Our goal was to match the performance of our modified JBL units, but with a greatly extended life. Every aspect of the design was evaluated and no detail was too small for careful consideration.
For example, we had many choices of material to use for the diaphragm. Titanium has good high frequency response because it is very stiff but it also has relatively high distortion and tends to fatigue quickly. Pure aluminum has a much longer life cycle but its high frequency response is not so good. Beryllium has a good blend of qualities in terms of efficiency, longevity and frequency response, but has significant drawbacks that make it less than ideal for all but the most demanding applications. Our research eventually lead to an aluminum alloy that provided a satisfactory frequency response with a good life expectancy.
Our three-inch driver uses a ceramic magnet but the larger diameter of the voice coil in the 2010 requires a much stronger magnetic field to achieve the necessary flux density. To do that with a ceramic magnet would have produced a driver weighing about 30 pounds that would break the horn it was mounted on! To be practical the driver had to weigh no more than about 10 pounds and therefore neodymium was the obvious choice.
The strength of the magnetic field is critical in a high driver because
it limits the high frequency response. For example, an 18,000 gauss field
allows the generation of higher frequencies than a 16,000 gauss field.
With a bass driver you may not need more than 14,000 or 16000 gauss because
you are not trying to drive it at 20kHz, and in that case a ceramic magnet
works perfectly well. However, an HF driver needs a really strong magnetic
field to get a decent high frequency performance and that implies the
use of more costly either alnico or neodymium magnets.
After the diaphragm material and the strength of the magnetic flux, it is the phasing plug that next determines high frequency performance, so we spent a long time developing this element of the design. In the 2010 the phasing plug is composed of five separate rings, each with edges of different angles. It is manufactured from a lightweight plastic that is resistant to environmental changes such as temperature and humidity. When assembled, the rings produce a series of slits through the plug that channel the sound from the diaphragm into the throat of the horn. This is a very complex part of the design because the diaphragm flexes differently at different frequencies, and so the sound moves through the channels of the phasing plug in different ways.
One of the limitations of the modified JBL unit was a reduction of output energy around 8kHz caused, amongst other things, by internal cancellations in the phasing plug. In our design we were keen to increase the amount of energy in this region by around 3 or 4dB and we also wanted to improve the overall efficiency by 1dB or so. However, we had to ensure that our design was sufficiently similar to the JBL unit to allow it to be retrofitted it into existing and past products without major modifications. After a great deal of experimentation with different phasing plug designs and various slot geometry's we eventually found the performance gains we were looking for. We also decided to keep using ferrofluid to match the acoustic characteristics of the JBL.
Linearity in a high driver is fairly easy to achieve because the excursions
are so small that the voice coil is always maintained within the magnetic
field. Instead the main distortion mechanism arises from incorrect matching
of the domed diaphragm profile to the phasing plug - the shape has to
be very precise to get the pressure loading to work properly. If it is
slightly out of true unbalanced loading causes all sorts of mechanical
problems and creates distortion.
Pure copper is relatively heavy and if used to make a voice coil the high mass requires a stronger field to achieve a satisfactory frequency response. An alternative is to use lighter aluminum wire for the voice coil but that has to be welded to the connecting leads because you cannot solder it. We investigated various voice coil materials and based on manufacturability and reliability we decided to use copper-clad aluminum wire in the voice coil which is reasonably light but can also be soldered.
The coil connecting leads are critical because they transfer all the power and have to flex continually so the leads are radiused to improve their flexibility and are supported to reduce fatigue to extend their life as far as possible.
A major goal of the driver design was to improve longevity. A standard JBL unit would typically last about one to two years, depending on how hard is was driven and the environment in which it was used - under test, in particularly hard conditions, we found drivers sometimes failed after only 8 months! Even though our modified units achieved much greater longevity and lower failure rates, we really wanted to double the life of the driver, giving up to five years of reliable operation. The main failure mechanism of the JBL was fatigue of the diaphragm suspension - not through excessive power, because our electronics can control that very accurately - but because the metal 'diamond' suspension work-hardens as it flexes and bends. We designed a flexible but robust plastic suspension system that is bonded to the aluminum diaphragm and gives a much longer working life.
The fundamental design of the prototype took about 18 months and created
a driver that achieved the sound quality, specifications and performance
we required. However, at that stage its construction didn't provide the
long-term reliability we needed and we spent a further year refining the
way the driver was assembled and the adhesives we used in its construction
- everything in the driver is held in position by a glue of one kind or
another. We went through a lot of subtle changes during this process,
each improving the design in small incremental ways until we arrived at
a very reliable and consistent production unit.
After all the laboratory testing, with extended high power music and
noise signals we beta tested the production prototypes alongside our original
modified-JBL units. It is important to run tests in real working environments
because this is the way the units will be used and there are conditions
here that cannot be replicated in the lab. For example, the horns attached
to the driver act in both directions - not only do they transform and
radiate sound from the driver to the outside world, but they also focus
external sounds on to the diaphragm, adding significantly to its stresses.
Our beta sites were mainly chosen from the venues that were giving us
a warranty headache - those places that were breaking the JBL units the
fastest! These were obviously the sites that were pushing the drivers
the hardest and so provided the best testing grounds.
Both the three- and four-inch high drivers are produced in a dedicated facility within our Berkeley factory, manufactured in small batches, hand-built by a team of highly skilled technicians. This process is both a science and an art! Although we work with a kind of assembly line idea, everyone understands the entire product construction process and can perform any part of it. This means that they appreciate the importance of every stage and the effects poor construction has on the test results! We have also worked with our suppliers so that they understand the importance of component tolerances. For example the top plate and pole piece have to be machined to very high tolerances because any fluctuation in the width of the magnetic gap has a direct impact on the frequency response.
Having perfected the tooling and manufacturing techniques we achieve
a very high degree of consistency with production failures almost non-existent.
Our average internal rejection rate so far has been about 2% but we often
run for weeks with no failures at all. An important aspect in achieving
such low failure rates is the careful inspection of component parts before
assembly. All of the critical components go through 100% inspection when
they arrive here from the suppliers so that we can maintain high levels
of consistency and accuracy - for example, voice coils are checked for
resistance, inside diameter, outside diameter, flatness and so on.
To protect the diaphragm from dirt ingress through the horn - sand or water, for example - we install a fine stainless steel mesh. This doesn't affect the frequency response at all because the airflow is so small at that point in the throat of the horn.
Although both the 3- and 4-inch drivers are in production now, their development is not complete. In the three years of production of the 1401M several improvements have been made to enhance reliability, for example, the most recent change being a new coating process to increase the life of the diaphragm by preventing corrosion. Most of these improvements have come through exhaustive experiments and testing. For example frequency response improvements resulted from studying the interaction of the phasing plug, the suspension stiffness, the voice coil resistance, and the optimum gap width. By adjusting these parameters we were able to find ways of smoothing out the peaks and bumps to provide a very uniform response.
Compared to the first pre-production units the current 1401M has, through incremental improvements, better repeatability from driver to driver, a flatter frequency response, and longer life of the diaphragm. Similar developmental improvements will probably be applied to the 2010 over time, as experience of the product grows and new technologies become available.
The 1401M and 2010 high drivers are now being used in virtually all our products, gradually replacing the modified Yamaha and JBL drivers in each product line. We test the new drivers in each system carefully before approving the replacement of the original units because every design is different and we want to ensure that performance is maintained to the highest standards. For example, the MSL6 crosses over at a lower frequency than some other products so we had to test the 2010 in this role very carefully to make absolutely sure that the new driver can handle these demands and that the extra excursion required doesn't cause any problems. Although this testing and approval process is slow, it has worked to our advantage, as we have been able to introduce the new drivers to products at a rate that matched the numbers of units we were producing. We weren't in any real hurry to phase in the new drivers and since they form such an important part of the loudspeaker systems we wanted to get it absolutely right.