Another source and the better one for us for small speakers was at a junk yard where they could be had for as little as a dime each; all one had to do was crawl under the dashboard, remove the radio and then the speaker, which was usually mounted above the radio and the speaker was always a 6 by 9.  Usually the yard owner would give us the speakers in trade for our work removing the radios for him.  Ahh, those were the good ol' days.

Sixteen 6 by 9 speakers has the diaphragm radiating area of 2.7 fifteens or 4 twelves, about 351 sq. in.  That could make the walls shake, even with 15 to 20 watts of vacuum tube amplification.  At about 10 watts power handling capacity for each speaker, the system could handle 160 watts but no one even imagined an amplifier with that kind of power.  Heck, back then, 50 watts was considered to be an arc welder.  A juke box had about 20 watts and they sure did sound good despite the boom box bass.

We knew absolutely nothing about speaker design except such things as Ohm's and Kirchhoff's laws; the messers Theile and Small were probably still in high school.  A cabinet with a large enough front to hold sixteen 6 b y 9s was about 38" by 38" and about a foot deep for a total internal volume of 9 cubic feet.  A hole was used for a vent or a slot by making the front a little smaller in one direction.  Yup, it was boomy but it sounded great, just like the juke boxes in our local hangouts, soda fountains and greasy spoons.

Enough nostalgia from this 76 year old buzzard.

The green curve is the impedance of the original system built in 2011

The purple curve is the impedance of the truncated system modified in 2018.  It was decided to wire the speakers for 16 ohms just because that was preferred back then over 4 ohms.

The following mess of curves was run after RTV was applied to the annuli.  Over the 7 years, the foam annuli had deteriorated a little and some split up to half an inch, circumferentially.  The RTV, an uncured rubber in a tube, although very elastic when cured would undoubtedly raise the fs but I was surprised at how little it was raised.

The heavy black curve is that of an unused spare run in 2018.  The thinner traces are those of the individual drivers after the application of the RTV.  The fs was raised by about 10hz, from 87hz to 98hz at various points  due to the amount of RTV applied.  It was either that or trash the speakers; they are no longer available, being a buyout from Parts Express.

This is a magnification of the hu7mp of the previous set of curves.  Only one unit exceeded 100hz at fs.

The incident distance is 1m and the reflection distance is 1.346m, a time difference of 1mS, which is 1khz.  below this frequency, the LMS can't get a clean sample as the microphone is switched off before the reflected wave arrives.  This simulates anechoic conditions.

The yellow curve is measured at a distance of 11' and a height of 62 inches, on axis and at ear height when standing.  The brown curve is measured at the same distance but at a height of 35", ear height when seated.

At 62" high, the mic is on axis at 11' but at 35" high, it's 13.6 degrees off axis, the result of which can be seen in the drop of the brown curve above 4khz as compared to the yellow curve.  Audibly, it very noticeable.

The blue curve was derived from the blue chair (bottom photo).  All curves are smoothes by 1/4 octave to make it easier to differentiate them. Wider smoothing can smooth the curves too much, rendering a better looking response, a somewhat deceptive practice.

The sound from this little project far exceeded my expectations.  the bass is somewhat weak and boosting was avoided as at 85-90 dB, the diaphragms were close to, if no at, Xmax, about 2mm.  A ten inch woofer would eliminate that deficiency.

Figure 4

The pink curve was run 8 years ago and the green, Nov. 2018.  The pink one was when the speakers were spaced 6.25" center to center.  According to theory, this would cause interferences at 2100hz and subsequent harmonics, 5200hz and 10,400hz.

In the curve, a dip can be seen just around 2400hz, another at 5200 and another just beyond 10khz.  The noticeable aberration between 3khz and 6khz.  With the speakers close together (green curve), that dip is gone.

If memory serves me well, I seem to recall the high frequency response to be dull, definitely not having the presence it now has.  The roll-off beyond 7khz (green) versus the rise there (pink) is less noticeable due to the lesser musical content.  the ear is more sensitive below 6khz than above.

According to the revised Fletcher-Munson curves, a `100hz tone would have to be 10dB louder than a 4khz tone to be perceived as being equally loud.

The truncated retrofit, prior to which the baffle was flat and the speakers spread too far apart, which was thought to cause serious peaks and dips in the high frequency response.   

The speaker in the dining room 11 feet from the mic which is in the living room.

Behind the Sweet 16 is a clone of the Wharfedale SFB3.

Now in the living room.  The mic can barely be seen on the left just above the blue chair against the background of the column supporting the purple speaker.  It's midway between top and bottom of the left side.