Present system is the SI, from the French System Internationale.  The present terminology for flux is Tesla.  (the use of the word present instead of current is deliberate to avoid appearing punny)  

1 Tesla = 10,000 Gauss = 10,000 Oersteds (I may have messed up on the Oersteds conversion but the Tesla to Gauss is correct)

Therefore, 17000 Oersteds  = 17000 Gauss= 1.7 Teslas

  The fields of the 12 FS/AL and the RS/DD are both 17000 Gauss, whereas the W15 woofer is 13000 Gauss.  17 kilogauss is quite a lot for a woofer but the RS/DD is a full range

I remember reading in a paper from JBL that very high flux densities will overdamp a woofer. This almost seems to contradict the feelings of some who claim that neodymium magnet woofers lack bass. It is my contention that the perceived lack of bass may be due to the higher flux density in the gap resulting in better damping.  If done right, the diaphragm will respond closer to the incoming signal with less overshoot and less damped oscillation after the signal is removed.  This phenomenon may be referred to as runaway.  It's similar to a car wheel bouncing after hitting a bump due to a weak shock absorber.

    JBL's mid-range drivers and tweeters have flux densities of 22000Gauss.  Coupled with a 4 inch diaphragm and edgewound ribbon voice coil, these drivers have an amazing sensitivity of 118dB, 1w1m.  The 375, now the 2440 was used in the JBL Paragon 

Super 12 RS/DD voice coil magnet gap is 0.049"    This is derived by using small drills as gauges. A #55 doesn't fit but a #56 does.  Given that #55 is 0.052" and #56 is 0.046", the mean is 0.049".  In all probability, the original design was set at 0.050"

The measured sensitivity of the Super 12 RS/DD is about 100dB, 1w1m. above about 300hz.

The several dB peak between 2khz and 4khz can be seen in this published response curve below (Fig. 1) and also in Figs. 2 & 3    

The vertical scale of Fig.1 is 10dB/division.

On the left, are two Super 12 FS/AL units atop each other and a W12 FS on the right to facilitate comparison of magnet sizes.

The two Super 12 FS/AL units have Waldom cones and coils, rebuilt way back in the 80's at a friend's re-cone shop.  They are now woofers.

Photos 3, 4, 5, and 6 show the suspect and the cell (cabinet) into which it will be set.  It's use is primarily to protect the speaker for testing.

The effect of the cabinet is of no concern as the tests are at frequencies above 1khz.

The cabinet wasn't moved during this mod.  The front can be swung out from the top and lifted off the bottom.  After the mod, the assembled front was gently lowered onto the bottom of the cabinet and swung into position.

The mic had to be moved back 2 inches

RED: rear loaded;  GREY: front loaded

 The most significant difference appears in the band just under 2khz and to about 3khz.  This does seem to correlate to the frequencies around 2474hz whose wavelength is 11 inches, the diameter of the hole. The rise around 2khz is of the order of 4dB.  The band there is narrow but under certain conditions can add stridency to violins and some female vocalists.

Both of these curves are gated response curves.  The LMS is given the distance of the mic to the speaker and the length of the first (nearest) reflection.  For a mic and speaker center being midway (4ft) between floor and ceiling, this would be 104".  The speaker is pulsed and the mic is turned on before a time lapse of 2.9mS to avoid picking up ambient room noise.  It stays on for a little less than 4.7mS because the reflected wave is just about to arrive at the mic, 7.6mS later.  This allows the mic to hear only the wave from the speaker and not the reflection.  It's better than 90% anechoic simulated.  Any ambient room noise will be heard by the mic while it is on.  Typically, this is random noise of around 30dB; an anechoic chamber is so sound proofed that one with sensitive ears can actually hear their heartbeat. I've been in one at IBM in Poughkeepsie, N.Y.

The caveat of this system is system is the less than 4.7mS on time of the mic allows it to get a good sample above about 400hz.  I would assume that ambient noise can also affect that.  In the sample of Fig.3, the system failed at about 475hz. Moving the mic to a half meter from the speaker and applying 0.25W will allow the system to get a good sample to around 325hz - 350hz.   0.25W @ 0.5m is the same as 1w1m.  This proximity seems to approach near field and can have effects in the high frequency response, especially when measuring tweeters.  For woofers, it doesn't matter as the near field response is used for woofers below about 400hz.  The mic is placed as close as possible to the diaphragm of the speaker although half an inch is adequate as the pressure in the proximity of the cone is quite uniform.

This set of responses was run in FFT mode (Fast Fourier Transform).  It was preferred since is more closely resembles music.  A logarithmic sweep takes 341mS to complete on CLIO.  At any instant during the sweep, there is ONLY one frequency monitored, quite dissimilar to music.  The red FFT response shown here is without the notch filter.  Compare this red curve with the red one shown in FIG.2 which is a logarithmic sweep.  The advantage of FFT mode is that the sound is always on which allows the notch filter to be bypassed for instant audible and visual comparison.

These curves were run with the speaker front loaded.  This shows at 2khz where there is a rise.  Again, looking at the red trace of Fig.2, the 2khz rise is suppressed.

The filter comprises a capacitor, an inductor and a resistor wired in parallel.  The filter is then wired in series with the speaker.  They are not easy to design as Vance Dickason mentions in his book, The Loudspeaker Design Cookbook.  In that section, he gives a few equations to get one into the ball park, so to speak but then adds that from that point on, it's trial and error.  I.M.H.O., without a means to run a frequency response, designing such a filter can be quite challenging as well as frustrating.

The multiplicity of curves is explained below.

Schematic in Fig.8 close to the page bottom

From the top, these are the resistor values. RED=no filter; GREY=4W; VIO=8W; BLUE=16W; YEL=36W; ORN=50W

It can be seen that a 50W variable resistor of about 10W capacity will allow adjustability.  Actually, 5W may suffice. The slow dropping of the response can easily be corrected with a treble control since most such controls have little to no effect at 1khz and a maximum effect of +/-15dB at 20khz.

The slope difference above 1khz between that of Fig.2 and of Fig.4 (this one) is most likely due to the nature if pink noise which delivers equal power per octave, whereas white noise delivers equal power per frequency.  To me, based on that, white noise would most likely resemble a constant voltage sweep.

See Fig.6 below, comparison between pink & white noises

For ease of comparison, here's the unfiltered response (RED) and the filtered response (YEL) run in sweep mode, called LogChirp by CLIO.  It runs two successive sweeps logarithmically, each taking about a third of a second.  It actually sounds like a chirp.  Well, two of them.

Comparison of pink noise RED and white noise GREEN.

1w1m with notch filter C=2uf, L=1.5mh, R=36W

It appears that somewhere in between lies the listener's preference

The blue trace is near field

The RED trace is 1w1m and the GREEN trace is the red one smoothed 1/6^th octave

This shows that this room response dominates below 250hz.  This is true for most household rooms..

It is clear that the center frequency is 3khz

Photo 10 shows the filter

PHOTO 10

This shows the effect of the filter on speaker impedance.  RED is the speaker impedance and BLACK is the impedance of the speaker with the filter in series using a 16W resistor.

The photos of the cabinet into which the speaker is housed appears to be a sealed box.  However, these curves show otherwise, that of a vented box.  The reason:  the speaker basket is not sealed against the front baffle, hence the bump at 20hz.   This is of no importance as this discussion has nothing to do with bass response but all to do with the effect of the notch filter in the treble.

This was mentioned for clarification to anyone who would have noticed that. I may get around to it.

Using a 16W resistor, the impedance of the filter at 3khz is 15.5W (green curve Fig.9).  The impedance of the speaker at 3khz is 15W.  The sum is 30.5W yet the graph shows 35W at 3khz.  My S.W.A.G., (Scientific Wild Ass Guess) is the extra 4.5W may come from phase shift.