Identifying Legitimately High Fidelity Loudspeakers: Speaker Diaphragm Material

Now back to what we can see when we go shopping.  One of the most salient features of any loudspeaker is its diaphragm.  There is a lot of science and pseudo science surrounding the choice of a diaphragm material and what effects it really has on performance.  One of those debates, which had been raging for decades, is about metal vs. paper or cloth diaphragms.  Each has its merits, and everyone it seems, has their preferences.  Let's discuss the basic function of the loudspeaker, and how the diaphragm is supposed to function, and what is actually happening in the “real world”.

An ideal loudspeaker diaphragm would never flex, thus it would exhibit perfect pistonic motion.  It would always be infinitely rigid, well damped, and inherently in that characteristic, the speed of sound through the material would be infinite, thus the mass would ideally be zero and the break up mode frequency would be infinity.  Loudspeaker engineers have a word for this ideal cone material.  Unobtainium (apologies to James Cameron from Avatar). 

Treated Paper Woofer (courtesy of Morel)

Paper and Plastic Diaphragms

The best thing about plastic (“polypropylene”) cones, or whatever mysterious formulation or name the Marketing Department comes up with that month) is that they perform reasonably well and their manufacturing uniformity is very high, with pretty low cost. Unit-to-unit consistency with low cost is the ideal for any manufacturing concern, not just loudspeakers. Unfortunately, the need for constant incoming quality control is actually a downside to polypropylene cones.  Poly cones have a major benefit in that they are highly damped and lossy.  This means they can have a controlled break-up, which results in a smoother high frequency roll-off.  Controlled breakup also can lead to a gradual reduction in Sd (projected area of the driver diaphragm) which reduces beaming compared to a rigid diaphragm. Some people argue that plastic cones tend to not sound as snappy or lively as stiff cone counterparts.  This can be attributed to to an almost capacitive storage effect inherent in soft cone type drivers.  However, if well engineered, good, well damped plastic diaphragms can excel at midrange frequencies and at a fraction of the cost of producing a quality stiff cone driver.

Paper formulations have all the good points spoken about above, but (depending on how/whether it’s sealed or not) the paper cone can actually absorb moisture from the air, changing its mass and damping characteristics. Cost, manufacturing complexity-uniformity issues, performance considerations, etc. make diaphragm materials selection a very difficult aspect of loudspeaker design.

Paper cones, made in part from felt and wool, for all their lack of sex appeal, have a combination of stiffness, low mass, and loss inherent in the material that often make the best sounding cones.  If the cone can be made small enough, and stiff enough so that the problem area is above the region (frequency range) where the cone is excited, then light and stiff is the best way to go.  If, on the other hand, you expect to cross over an 8 inch speaker with a 6 inch cone to a 1 inch dome tweeter at 2000 Hz (typical), then it is unlikely you will ever find anything affordable to do the job better than paper.  Calling loudspeaker cones paper is a bit like calling everything that grows and converts CO2 into oxygen “vegetables”.  Formulations are mixed and processed and doped with chemicals in a way where art truly meets science.  All paper cones are NOT created equal.

Aluminum Cone Woofer (courtesy of SEAS)

Metal Diaphragms

On the surface, it seems like Diamond or Beryllium would be the ideal obtainable material for a loudspeaker cone.  Then, there is the cost to consider.  So, how about a compromise?  Aluminum can be made hard with alloys and heat treating (almost none of which is either ever done or done right) in the consumer marketplace.  Aluminum is abundant, and relatively inexpensive.  It is relatively easy to form and does an excellent job of dissipating heat.  Since we actually want to sell this speaker, let's pick aluminum as a starting point.  Now, the theory tells us that the speaker should never flex.  It also tells us every time we double the weight, we end up making the speaker 6db less efficient. (Meaning we lost ¾ of our prior efficiency.)  If the aluminum is too thin, then it won't be stiff enough.  Too thick, then it won't be sensitive enough.  So once again, our engineering choice is not the best of all worlds scenario, it is in fact another compromise.  While it is true that making a diaphragm from metal has this advantage, it brings with it a disadvantage as well.  Any standing waves on the cone will end up creating a great deal of ringing and flexing because what makes this material so strong (inability to flex) is also its downfall.  What happens is that unless the diaphragm is indeed very small relative to the wavelength radiated, it will eventually flex.  Unfortunately, when it finally does so, it will do it dramatically, and it will involve much of the cone in the process, which results in a rather significant peak or dip in the response and ringing.  Ringing is bad for system linearity and sound quality.  Minimizing this is a paramount concern for making a good sounding, and well behaved driver.

We can mitigate this problem by attempting to dope the cone with a chemical designed to absorb the sound waves, but then we are troubled again by the mass issue.  It seems that the very thing we were trying to avoid at the start (compliance of flexure of the cone material) turns out to be a useful solution to the problem.  While it will mean the diaphragm ceases to behave in a pistonic (acts as a uniform whole) fashion at a lower frequency, the compliance (accompanied by damping or loss) within the material tends to localize the disturbance of the standing wave to a much smaller area of the cone.  Because less of the cone is involved in the “break up”, the impact on the frequency response is less pronounced. The peaks and dips are much less dramatic when they occur, and the point at which the cone stops making high frequencies and just dies (secondary resonance) is reduced.

The best sounding metallic cone speakers we’ve ever heard have never ignored the damping question, and most frequently combine metallic domes with lossy surrounds or an alternative material to impart damping to the cone.  Back in the late 1980's, M&K Sound purchased raw drive units from both Eminence and Peerless (Denmark).  One of the Peerless engineers named Knud Thorborg was asked by management to make a metal dome tweeter.   He grumbled, and then finally caved in with a caveat.  He would make it work, but it had to be done right, which meant essentially his way.  Well, some people at Peerless were pretty upset with Knud.  His way was 1 layer of aluminum, and 6 layers of “other”.  Finally after beating him up in a dark alley (speaking metaphorically of course) they settled on 3 layers as a compromise.  Now, while damping gets a lot of press, what is also true is that quite often the changes attributed to damping are often the result of differences in mass, stiffness, or geometry.  All play a critical role in determining a loudspeaker’s tonal characteristics above the range where it is acting like a piston.  Remember, if you cross over your 8 inch woofer to your 1 inch tweeter at 2000 Hz, the voice coil in the woofer has likely changed directions twice before the wave propagates up and to the end of the surround.  When you use a speaker above its piston range (where it is basically an air pump) it behaves more like a transmission line than a pump.  It is at this point in design, where tools like Finite Element Analysis (FEA) can be useful, as the simple Thiele-Small model falls apart.  Though FEA is still a maturing technology and not ideal for cone development.  Many loudspeaker driver engineers rely more on Klippel’s scanning vibrometer.  This tool actually measures the displacement and velocity of the device under test. Essentially, anyone doing FEA without also using the scanning vibrometer to measure the results is relying too much on theory and not enough of real world.

It is challenging to make a driver good for both truly low and high frequency performance simultaneously.  For one, any driver becomes more directional with increasing diaphragm diameter.  This causes the overall sound to increase on axis, while it drops off axis at the upper range of the woofer, before it hands off to the tweeter.  Another issue is woofer breakup, or “cone cry”, whereby the woofer cone itself resonates at a “favorite” frequency, typically located just above the cross point.

The designers will often make compromises in order to achieve a cost effective and pleasing result.  A good rule of thumb is to cross the woofer over no higher than the wavelength of the cone diameter.  So in the case of an 8” woofer with an actual 6.5” cone diameter, you’d want to cross it over at around (13,524 inch-per-second / 6.5”) = 2,080Hz.  For a 10” woofer with a cone diameter of 8.5”, the highest crossover point should be around 1,600Hz.  The problem with using such a large woofer in a two-way system is it forces the designer to use a higher cross point than the woofer would effectively be able to produce or a lower crossover point than the tweeter could handle without being overly strained (by power handling or distortion).   For this reason, you don’t see many two-way speakers employing an 8” or larger woofer because the larger cone doesn’t effectively bridge that last 1/3rd octave gap to the tweeter.  No matter how it is designed, no woofer can effectively reproduce more than 5-6 octaves without tangible performance issues.  This is why multiple drivers are often used with specific design properties to play within their intended bandwidth and collectively work as one unit to produce full-range sound.

Tabulated below is a comparison of various metals utilized in speaker diaphragms.  As we stated earlier, the ideal driver would be infinitely rigid and the speed of sound through the diaphragm would be infinite, while the material itself would have little or no mass.  While no such material exists, we can get pretty close to ideal by having a material with a very low density and high propagation velocity.  

Materials Comparison (courtesy of Harman, Inc)

A property that affects diaphragm break up is the velocity of sound that propagates through it. The higher velocity means an increased wavelength for the sound wave in the structure, or a higher resonant frequency for the structure, thus reducing the chance for destructive resonance under random excitation. Although the speed of sound is second to diamond, the first evidence of break up in beryllium is still above 50 kHz. 

As you can see, pure Beryllium has the lowest density and hence the best Poisson Ratio (the measure of elasticity).  Density is directly related to weight, and thus arguably a critical metric in determining the tweeter’s sound quality. When reduced power is required for a given output, the coil heats up less and does not alter the impedance of the tweeter, which minimizes dynamic compression and lowers distortion. Also power handling is improved. This is a major advantage for beryllium, as the density is by far the lowest of the bunch and almost half that of diamond!  Theoretically such a material would be ideal for maximizing extension and reducing distortion and compression. 

Pure Beryllium cones are expensive and out of the reach of most budget designs and also toxic if burned, hence why Aluminum often makes for a great alternative.  For those that claim diaphragm material doesn’t make a difference, we’d argue that science disagrees.  

The Infinity team developed a laminated Ceramic Metal Matrix Diaphragm (CMMD) material that exhibits Low mass, high velocity and high damping.  It is deep anodized aluminum, where the anodizing (ceramic) penetrates up to 1/3 of the cone thickness.  Check “technology” at www.infinitysystems.com.  Lower cost products got a less aggressive anodizing, In the Infinity Prelude system for example, there were no breakup modes within the audible bandwidth.  The material is now used in models from different Harman brands and in car audio systems.

Soft vs Hard Dome Tweeters
There has always been an ongoing debate about the sonic characteristics of tweeter dome materials. Do metal dome tweeters really sound "harsh," as some claim? Do silk domes sound, well, "silky"?

Different materials do have different physical properties as it relates to diaphragm performance and these can and do influence the audible behavior of the driver. With full knowlege that many well-respected designers and engineers have opposing views and opinions on the subject, there are a few basic generalizations that can be made about different tweeter dome materials.

Metal dome tweeters are generally constructed of alloys of either berylium, aluminum or titanium. (All of which are chosen for their high stiffness to mass ratio.) They tend to have larger amplitude high frequency break-up modes than soft domes. Softer domes tend to have more break up modes which occur at lower frequencies than metal domes. When a metal dome breaks up, it does so much more dramatically than do soft dome tweeters which have the advantage of having more compliance and damping distributed throughout the dome. The more severe break up modes involve a much higher percentage of the tweeters surface area than a soft dome which tends to do a better job of localizing the disturbance, and isolating the other areas of the surface from the break up. The more dramatic metal dome break ups often will show up as a large dip and/or a narrow spike in the upper end frequency-response. There may be one or several peaks and dips which may be minimized by octave or fractional octave SPL averaging common to PC based measurement programs. Silk-dome tweeters tend to break up more often, but more gently. Some audiophiles feel that makes them more pleasing to the ear even when the break up mode is lower in the audio band than modes commonly seen in metal dome diaphragms.

The metal dome breakup manifests itself in the frequency response curve as a sharp peak, or a sharp dip in its usable frequency range. A quick glance at any credible published frequency response curve—such as those in Stereophile magazine-shows this behavior quite clearly. Interestingly, the ultrasonic peaks seen between 20-26 kHz are often accompanied by a depression in a metal dome tweeter’s response between 10-20 kHz. The May 2012 edition of Stereophile’s test report of the excellent Paradigm Studio 20 v5 speaker shows this classic metal dome FR behavior in Fig 4 on P. 103, as does the June 2012 issue, P. 117, Fig 5 of the Monitor RX6 with a "Ceramic-Aluminum/Magnesium" hard-dome tweeter. Perfect illustration, again.

So the "harsh" character often attributed to metal domes is not necessarily because of elevated frequency response levels in the highest audible octave (10-20 kHz). Since metal by its nature is not well-damped, (meaning the energy is not absorbed in the material) if a metal tweeter sounds harsh, it’s likely due to ringing or resonances in the dome that manifest themselves as audible artifacts, often times at frequencies lower than the curves would suggest.

Silk, treated cloth, polyamid, Kortec, Sonatex, are inherently better damped than single layered metal diaphragms. (The best metal dome designs get damping from their surrounds and absorption from acoustical treatments placed under the domes). Cloth and soft materials do not ring the way metal does, and the frequency response of a cloth tweeter is generally smoother. Soft domes with a lower and less resonant response roll-off at lower frequencies at the top of their usuable bandwidth.

View Stereophile’s published frequency response curves and you’ll see this pretty consistently. In that same Stereophile June 2012 issue, P. 105 Fig 4 of the Joseph Audio Pulsar, the soft dome’s response is very flat—no suckout between 10-20 kHz like a hard metal dome—but you can see it beginning its rolloff at 18 kHz. (Not atypical for a cloth or silk dome of high quality). Not the highest audible frequency, but still very close and good enough for reproducing 98% of all musical content. It must be added that the voice coil mass and motor force and attachment are as critical for determining the top end of the tweeters response as is the dome material and mass. Heavy voice coils will not work properly to bring the top end of a tweeter up to a very high frequency limit.

Regardless of the dome material chosen, a lot comes down to the implementation of the technology and, overall, how the speaker has been designed. Another thing to consider is that ringing at lower frequencies in a tweeters bandwidth may result from the tweeter's motor being under damped. If a higher order high pass filter is being used on the tweeter it may also induce ringing unless it is critically damped. (Which is not what is typically used in filters. What is typically used are Butterworth filters which all ring at any order above the first). Ringing of this nature, indeed any low frequency ringing has nothing to do with the dome material but rather its mass, mechanical resistance, and its motor force.

There are excellent-sounding speakers using all kinds of tweeters. There are also hybrid dome tweeters that use a metal dome overlay over a soft material underneath in an attempt to get the best of both worlds.

In the end, the overall sonic character of the speaker has as much—if not more—to do with the designer’s voicing decisions and other design choices as it does with any single arbitrary tweeter dome material. What one hears in the highest frequency ranges covered by the tweeter is a complex combination of materials, geometry of the parts and the tweeter frontplate, the baffle, the grille, and the crossover network, even if we eliminate the clear contribution of the amplifier and source material. This can vary significantly from one angle to another. Often moving 30 degrees off axis will radically affect the response you hear in the top octave. There is NO guarantee that a resonance will produce a peak more than a dip. This is a 50/50 split. There is NO guarantee of the Q or amplitude of that resonance. These things will often confound confuse and surprise even designers that have done it many times before. The amount of small issues that can go wrong and create real audible measurable artifacts is large enough to warrant a chapter on its own.

Bottom Line on Loudspeaker Diaphragm Material:
Stiff cone drivers are typically better than flexible cone drivers (all things being equal).  However, no magic cone material alone will determine the quality of the sound or performance of the driver.  Cone geometry and proper dampening of the cone material to better manage its behavior at and above its break-up mode plays a vital role in how the cone will sound.  It's critical to avoid operating the drivers in their break up mode so a clear understanding by the designer through a properly executed crossover network is key.