Any open back will have similar acoustical properties versus a sealed or ported enclosure, the main deficiency of the open back is the lack of lower bass. As with any speaker system, the location of the system in any particular room will have quite a substantial effect on the sound.  The thing about open back systems is their open sound.  If you've heard a Magnepan, you'll know what I mean.  They, too, suffer from lack of low bass.

Having a collection of vintage Wharfedale speakers, those with the cool looking red magnets, it was decided per a suggestion from a friend in New York to build one, just for schitts 'n giggles. To say the least, I was quite astounded by the sound and that was from hearing only one, mono, not stereo.  A link to a technical overview of that system is here.  https://www.ln271828.net/SFB3.html

The review posted by Inner magazines (top) praises the midrange but that may have been by listening at moderate levels around 80dB and I'll agree with that.  However, the type of music being played as well as who's doing the listening can dictate how loud to play.  In my case, Rock 'n Roll, 60's and 70's rock, Rockabilly, Country and the like usually find their way above 90dB and sometimes close to 100dB.  A consistent diet of SPLs above 95dB can, and will, do permanent damage to the ears and such damage is irreversible.  Although small, it is cumulative.

At these high levels, female vocals can be close to ear shattering.  That's due in part to the response of the twelve and the ten inch units,   See Figs. 6, 7 & 8 here.   https://www.ln271828.net/SFB3.html

Measures were taken as described on that page so there's no point in reiterating them here.

The next thing was to search for modern drivers to use in the SFB.  Again, just about any OB will suffice but the style of the SFB seems appealing.  The problem with its dimensions is that it provides no room to front fire the Super 3 tweeter, the response of which falls like a rock after 8khz if upward pointed.  A more modern tweeter with a sensitivity around or above 98dB, 1w1m, would do much better pointing towards the ceiling.

The search began with the woofer, then the so called midrange and then, the tweeter. (and then Along Came Jones)

The Woofer

Although a very satisfactory bass output was achieved with the W12FS, a slightly higher output can be realised with a W15FS or CS.  The W15 units have an effective cone area (Sd) of 99 sq.in.  The CS (cloth annulus) is slightly larger due to the less porous felt offering some radiation whereas the foam offers little to none due to it open cell nature.  Either one is considerably smaller in area than a standard 15 incher with its 133 sq.in. area.  The W12FS has an Sd of 71 sq.in.  The effect of the ten inch unit being wired in parallel to either the W12 or W15 would increase the output a little.

Having Bass Box Pro (BBP) surely simplified the choice of woofer although still very time consuming as the latest Thiele-Small parameters had to be entered manually from published data files.  To simulate the open back which has no actual volume, an enclosure size of 20 cubic feet was used in BBP.  This was found to be a practical limit allowing for baffle and floor reflections.  There was a low amplitude bump around 10hz which vanished if the box volume entered approached 100 cubic feet.  That would essentially simulate the speaker in free air, no baffle.

The search considered high frequency output of the fifteen so as to negate the need of a low pass filter.  The normalised and custom responses of BBP were also considered.  Power handling capacity was considered at a practical 200 watts but some had power ratings as high as 500 watts.  Cost was kept at $100 or lower, per unit.  Only one driver out of over a dozen seemed to give me that for which I was looking, a duplicate to the W15.  Both the W15 and the others were limited to 50 watts as that will affect the Custom Amplitude Responses.  The W12 used was never taken above 45 watts (rms) using a scope and that was assuming a 15 ohms load.  Actually, at those low frequencies, the impedance is considerably higher than nominal so the power transferred was actually much less.  The W12 cone excursion was around 4mm, never bottoming or making sounds as if the cone were buckling, usually heard as a snap or cack, a sharper sound that that made by the coil form striking the rear plate.  Of the W15 units of which I have several, two are as close to a perfect match as one can get, so they were used in BBP as the reference. The speaker that matched the W15s was the PA380-8 by Dayton, available at Parts Express (P-E) for a modest $80.  Their stock number is 295-034.

The Squawker

This unit was a little easier to find as the main concern was the high frequency response.  As can be seen on the SFB3 web site, many ten inch units have a rise in output above 2 khz, some as high as several dB and over an octave in width.  The SFB10 is no exception although not so bad.  Initially, guitar ten inch units were considered as low frequency response was of minimal concern but damn near all had that nasty peak in the upper register.  I realise that musicians want that as the speaker is part of their instrument.  They are the artists so they can choose their preferred sound.  In designing a sound reproduction system, a speaker as close to neutral as possible is preferred.  OK.  Personal preference is allowed.  After all, sound reproduction is very subjective.

Eventually, the search was narrowed down to 5 drivers.  Two Daytons, two Celestions and a Faital.  Despite my attraction to Daytons, their high frequency output between 4khz and 5khz peaking at about 4.2khz by several dB kind of disturbs me.  That coupled with the rather slow roll-off from 5khz to 8khz compounds that concern.  So, they are being kept on a back burner.  One may survive the criteria if for no other reason than to try it, finances permitting.  The two Daytons are P-E stock numbers 295-030 and 295-532, the latter being preferred.

Of the two Celestions, one seems a very likely candidate.  Their P-E stock numbers are 294-2066 and 294-2070.  Here is a dichotomy.  The 2027 response falls quickly after 5khz but has a nasty peak around 12khz.  The 2066 response falls quickly after 7khz  and has a lesser nasty peak around 12khz.  One would think that a low pass filter at 5khz or maybe even 4khz would suppress that peak, but experience has proven that not to be necessarily true.  Of these two, the 2066 is preferred. See photos 13 & 14.  https://www.ln271828.net/bostwickfountekvifa.html

That system required a weird stunt to eliminate the effect of that peak.  The mid-range cuts off at just under 3khz and the tweeter comes in around 8khz, using second order filters.  The resultant response can be seen in photo 16 here.   https://www.ln271828.net/bostwickfountekvifa.html   The two Celestions also rest on a not so far back burner.  They are competing with the Faital, P-E stock number 294-1201.

This fella also has a 12khz peak about 7dB below that at 3khz, 3dB higher than the Celestion 2066 whose 12khz peak is 10 dB below that at 3khz.  So, it's a toss-up between the Celestion 204-2066 and the Faital 294-1201.  The peak at 12khz seems common to just about every 10 inch unit I've observed within that price range and power handling capacity, the latter being one parameter that can increase a price tag.

The Tweeter

There are several tweeters here of various manufacturers and types.  Some are soft dome, ceramic dome, horn and ribbon. All will be examined before making a purchase.  I've used the ribbons which are flat within 5dB to 40khz, the Beston RT002A (P-E# 277-112)  Its sensitivity is also a modest 92dB, 1w1m.  They also have a di-polar tweeter for the same price with the same response. (P-E# 277-116)  I wonder how much difference there is in their sound.  May be time for a double blind test.

Comparing the Wharfedale W15 to the Dayton DC380

This speaker, the Dayton DC380 hasn't yet been auditioned for a very good reason, I don't have one, YET.  However, IMHO, it seems worthy enough to make a purchase of two along with a pair of tens, most likely the Faital 10FE200

Of these three designs, 2 and 3 are those of a W15CS-OOC and a W15CS-TJ.  Both have original cones and annuli, the OOC having an older style cone with 6 ribs evenly spaced while the TJ has 10 ribs in decreasing spacing from the apex to the annulus.  Both W15s are 15 ohms and the Dayton DC380-8 is 8 ohms.

The Normalized Amplitude Response (NAR) shows the identical W15s, green and blue and the Dayton, yellow. This compares the relative responses of the units. The next one, Custom Amplitude Response (CAR) compares their actual SPL at 1m with 50 watts applied.  The larger cone area of the Dayton is the primary reason for the increased low frequency output.  The quicker high frequency roll-off is due to the higher inductance of the its voice coil.

The Maximum Acoustic Power, (MAP) shows the relative outputs at 1m with the diaphragm excursion within Xmax.  Here again, the Dayton performs better due to the larger cone area being 30 sq.in. larger, about 33%.

The Maximum Electrical Input Power (MEIP) shows the maximum electrical power applied before the diaphragm reaches Xmax.

(continued under design tables)

Figure 1

Cone Displacement (CD) is how far the cone will move in one direction with 50 watts applied.

System Impedance (Z) should be obvious and interesting.  Many assume the nominal impedance when calculating power.  Even with a scope, despite that the bass may be driving the trace larger in amplitude than the rest of the music, the actual power delivered can be very small, comparatively speaking.  Let's assume 20 volts applied at 500hz.  Here, all drivers are close to 10 ohms, therefore 40W is delivered to the Dayton and the W15.  Assuming an 8 ohm load for the Dayton would erroneously say 50 watts and assuming 15 ohms for the W15 would erroneously say 26.6W.

Now, at 40hz, that same 20 volts will produce 26.6W to the Dayton (it's 15 ohms at 40hz) but only 8.9W to the W15s (it's 45 ohms at 40hz)

It is true that the volume of the open back is essentially a partially enclosed space created by the floor behind the baffle and the two sides, but its effect on the loading of the speaker is negligible, although odd.

Thos oddity was noticed in two impedance curves, one with the speaker in free air, facing forward and another in the SFB enclosure, also facing forward. Now, one would expect the impedance to rise due to the loading of the air that is seemingly trapped in the chamber behind the speaker.  This is true of any sensibly sized closed box but is this SFB really closed?  It's more like a vented box with a very large area vent and a vent will split the point of resonance of the speaker in free air into two smaller resonant points, one below and one above the free air resonant frequency.

But, this large vent splits that free air resonant point into two very close points that interfere, resulting in a single resonant point.  That's my theory and until a better one comes along, I'm stuck with it.

Another way possibly detect the two resonant points may be by using a sensitive ammeter.  With a constant voltage applied to the voice coil, as the impedance rises, the current will drop.

The measured impedances are:

free air  21.643 ohms @ 124.1hz

in SFB  19.457 ohms @ 113.3hz

With 0.5v applied, the currents would be in free air, 0.0231A and in SFB, 0.0256A.  The difference being 0.0025A can be detected with my Fluke bench meter.  A scope would be used to measure the phase shift.  Remember, for an AC circuit, Ohm's Law for current becomes E/Z = I cos q

Figure 3 

The extended high frequency response of the W12 (red) is due to lower voice coil inductance and the lighter cone and coil mass as compared to the others.  These curves are all determined from the Thiele-Small parameters and therefore, are theoretical.  The W15 can perform well up to 5khz.  (See curves 18 & 20 in Fig.5)

The opposite seems to be expected of the W15.  The lower level is noticeable compared to that of the MCM but the W15 seems to give a better response in the very low register.  What is even more perplexing is that the W15 holds up very well with a full bass boost on the preamp.  At a level of 92dB, which takes about 16 watts, the diaphragm is only moving about 2mm in the bass.  Here's where the room acoustics come into play.  In this room which is actually a living room adjoining a dining room with the long dimension of 37 feet, it can support a frequency of that length, 30hz.

As for the ten inch squawker (midrange) see the section above titled The Squawker.  There's a few suggestions there at modest prices.  The 55-1555 works very well.

Other ten inch units I found to work well can be seen in the section above tithed The Squawker.  Those units can be found at Parts Express and their stock numbers are provided. 

Above 300hz, the responses are 1w1m, gated.  Below 300hz, the responses are near field.

The curve pairs are as follows. Near field, left of 300hz;   1w1m, right of 300hz 

15 blue - 14 red;      17 lilac (violet) - 16 fuchsia (hot pink);      19 gold - 18 khaki (olive green);      21 light pink - 20 black 

The 10 inch MCM (55-1555A hot pink 16) does well to about 5khz.  What is surprising is how well the W15s performed 5khz, which prompted the test shown in fig. (next)

The purple trace is the sum of the other two,  It's 3dB higher because XdB + XdB = X+3dB.  I left it there for clarification purposes.  It doesn't necessarily resemble a trace with two SFBs but when I get the second one built, that will be clarified and added here.