Driver Illusions



Drivers are a mystery to most, understood by some, and mastered by only a few.  The biased "advertising" words and frequency response curves are often disturbingly embellished/smoothed.  However, there is hope.  The earliest quote I remember was internalized in about 4th grade.  It is, "The truth is not apart from reality but lies within it."  I find this applicable to many things, and speaker drivers are certainly included.  There is some truth in what every manufacturer says about their drivers.  Some frequency response curves are accurate.

A waterfall plot can speak a thousand words.  A frequency response (FR) curve can speak only a few.  Unfortunately most manufacturers (except Raven, Jordan) don't publish their waterfall plots.  They only publish a FR curve.   The degree of truth varies from manufacturer to manufacturer.  SEAS FR curves tend to be very truthful, and this is evident with their metal cone drivers.  With stiff cone drivers, there is cone resonance.  My first project was a prime example of this.  The driver had an excellent reputation, and a very stiff cone. It was not a SEAS driver.  The manufacturer's FR curve seemed to indicate little/no resonance.  The designer (not me) applied a 2nd order crossover to this driver at 2200hz.  This all seemed quite logical, but it failed profoundly.  This is because the manufacturers FR curve was quite misleading.  

I didn't understand the issues concerning cone resonance initially.  I now understand that  uncontrolled resonance made the whole speaker sound really bad.  It was "nasally", "congested" and "bright".  I figured the resonance was very high, but even in the male vocals the speaker sounded bad.  Resonance effects the whole system.  The @$150 midwoofer (I will not disclose which one) and @$80 tweeter didn't sound any better than my B&W DM602.  I was disgruntled until Dennis Murphy came to the rescue.  He had the same problem with this driver and previously spent quality time designing a crossover for it.  He conquered!  Dennis has excellent crossover design skills.  He sent me the crossover plan he designed.  It was a 4th order parts count at 2100hz.  I implemented it and "viola" excellent sound happened.  The difference wasn't subtle.  It was amazing!   Stiff cone resonance requires a steep crossover to get "beneath it".  This entails great skill, and many crossover parts.  A good crossover is far more important than good drivers.

Most folks focus on the cone of the driver.  It is visible, and it sells speakers. The common "Joe" can readily grasp that a stiff cone will produce a flatter wave-front and a cleaner resultant sound.  A pretty cone seems to be equally important.  If it looks high tech then it sells well in hifi shops.  At least 50% of the cost in a commercial speaker goes toward marketing.  But there seems to be little marketing $ given to the motor structure of the driver.  IMO, this is more important than the cone material.  I would rather have my $ go into a good motor than a good cone.   

I read an article in speaker builder copy 7 of 1999 titled "Navigating Speaker Design: Sleuthing Driver Parameters" by Mark Wheeler. This article is very good at describing how subjectively "fast" sound will emanate from a given driver. Mr. Weeler suggests that one way to explain how fast a driver will be is to multiply the magnetic force (bl) by the radiating area (Sd) of the driver. Then divide this number by the weight of the cone (Mms). This coefficient will yield the audible "speed" of the driver. This concept is very similar to hotrod power (Bl) to weight ratio (Mms) with traction (Sd) used to figure out a ľ mile time. This analogy might be useful when trying to comprehend Mr. Wheelers assertions. While he is correct in his assertions, and his article is very informative, he leaves out a very important facet of what makes a driver audibly "fast".  He omits magnetic linearity.

Lambda acoustics does a nice job describing their motors and the issue of magnetic linearity.  I believe that magnetic linearity is very important too. I will make an attempt to describe magnetic linearity in crude fashion.  I will assume an overhung voice coil moving within the magnet to produce sound.  

The voice coil should operate within the confines of the pole piece.  Some distortion always occurs when the voice coil extends beyond the pole piece.  When the voice coil is within the confines of the pole piece there is some dilemma.  Most speaker manufacturers will call the region where the voice coil is within the confines of the pole piece the linear region, linear X-max, or something of the sort. But just how linear is it?? My crude theoretical compendium is that if the voice coil moves forward within the pole piece when responding to a bass note (50hz) it is off center for a moment and effects the field of the pole piece. Furthermore, the moving voice coil will induce current (eddy currents) within the pole piece.  While the motor's field is disturbed by the voice coil, the disturbed motor must tell the voice coil to respond to an upper octave piano note (@2000hz) the motor will, in varying degrees, smear this note. This is because there are other currents in the magnet - eddy currents.  The eddy currents interfere with the primary electrical field of the motor.  The midwoofer magnet needs a clear (uninterrupted) path for this 2000hz waveform to be most effective. 

Asking a motor to perform 50hz and 2000hz is a tall order. Focus on the issue of magnetic linearity (eddy currents) within the normal operating range of the voice coil becomes important when the difficulty of the drivers task is considered. Just because the voice coil is within the confines of the magnet doesnít mean that it is operating in totally "linear" fashion. In truth, all motors are effected by voice coil movement. Some are less effected than others. This fact isn't profoundly clear in factory advertised specifications.  The closest number for assessing linearity is inductance.  Better motors have lower inductance, and poorer motors have higher inductance.

Thus, the primary motor function is the voice coil inducing a varying electrical field upon the magnet/pole piece.  But there is a reverse effect too.  The pole piece induces an electrical effect from the eddy currents that messes up the magnetic field, and effects the voice coil.  Shorting rings result in a drastic reduction of this reverse effect, and their result can be measured in lower inductance and increased clarity.

A few common hifi driver companies have endeavored to conquer magnetic linearity issues. They are Focal (W series), Lambda, Scanspeak, Dynaudio, and SEAS (EXCEL line only), and the Peerless (HDS line only). These drivers are all on the expensive side (except for peerless). This isnít to say that all other drivers sound bad, but it is my opinion (however infantile) that magnetic linearity does make a difference. Thus, the "linear region" isnít the same from one company to another. And Mr. Wheelers theory isnít the end of the story.

Applying Mr. Wheelers raw power to weight ratio theory to Scanspeak drivers would lead you to believe that many Focal and Audax drivers would outperform the SS by a wide margin.  This doesn't appear to be true.  The Scanspeak drivers have a fairly poor power to weight ratio when compared to the Focal and Audax stuff.  The Vifa stuff would also seem far behind Focal and Audax with their comparatively poor (Mark Wheeler) numbers and their soft cones too. What appears true is that Vifa drivers are used in a disproportionately large number of hifi speakers for their seemingly poor (Mark Wheeler) numbers. Scanspeak drivers are used in some very high end systems despite their poor (Mark Wheeler) numbers. Audax drivers are pretty scarce despite their good numbers. Focal is represented outside of their own JM Labís line of speakers, but not to the degree Mr. Wheelers numbers would indicate. Thus, some examination of what drivers the commercial speaker manufacturers use seems to indicate that there is something else afoot.  There is more to this issue than Mr. Wheelers numbers would indicate. I believe the missing issues are voice coil inductance and magnetic linearity.

A very good place to start a study on drivers is at the LDSG webpage.  Upon entering this hobby, I spent about 100 hours internalizing the "LDSG".  It is a phenomenal web page where Bob Stout has spent a great deal of unselfish time and effort.