Horn Loading the Faital Pro 3FE22-16 - Addendum

 

The previous study of this horn and driver was mostly academic. There was various listening tests done in the testing room but those auditions were done with one system, mono.  A few were curious as to how they'd sound in stereo and in a more representative room.  Hence, this addendum.

The previous study can be found HERE

There are substantial differences in these two rooms.  The testing room has no heavy drapes but does have 2" acoustic foam over about 6% of the ceiling.  It measures 20ft by 17.5ft  the speaker was measured under quasi-anechoic conditions, simulated by CLIO and LMS.  This will give a better approximation of the speaker's performance but says little about how it may sound since the latter is highly subjective.  The auditions were done by myself so they are based mostly on personal preference.  For the most part, many seem to prefer a system that tests as flat within about 6dB.  I am one of them but again, subjectivity is forever present.  Also, at my age, my hearing sensitivity to higher frequencies is far from one who is half my age.  I can still hear up to about 11khz from a tweeter with a sensitivity of about 93dB, 1w1m driven by 1 watt.

The living room opens into a so called dining room and combined, they measure about 17ft by 36ft.  It also opens into a hallway There are heavy drapes on the wall to the right of the right speaker.  The carpet is plusher.

Also, the bass section of the test model was a Wharfedale W12RS.  The current model uses an Eminence Basslite S2512 which has a stronger neodymium magnet.  This unit outperforms the Wharfedale in the higher register above 2khz.  This was by design as the Basslite is a guitar speaker.  However, in the bass region, according to BassBoxPro, v6, both are close to identical.  However, BBP does not measure transient response but my aged ears do and the Basslite does have a tighter bass.  This is noticeable in transient bass such as that from a kick drum.  Also, the Basslite will handle much more power than the poor old Wharfedale not to mention that it is still available whereas the Wharfedale has all but faded into oblivion outside the set of nostalgic old buzzards, like myself.

It should be noted that responses 1, 2 and 3 were run at 1w1m, on axis.  Responses 4, 5 and 6 were run with the microphone 64" above the floor, ear level when standing and 30 degrees off the horizontal axes and 20 degrees off the vertical axes.

 

 

 

PHOTO 1

Looking from the hallway opening which is about 13ft; the wall to the right is only about 5ft, if that much.

the microphone is several feet from the speakers, my usual listening position which is usually for short durations.  Mostly, I'm elsewhere when playing music.  The mic, therefore, is at ear height, about 64".

 

 

PHOTO 2

This view is from inside the dining room

 

 

 

 

PHOTO 3

This and the following two photos show a closer detail of the chamber for the horn driver

They should have been included in the previous study but that was an oversight

The purpose of the crossbar will be evident in the following 2 photos

 

 

 

PHOTO 4

Not as good a view of the gap created by the speaker's gasket being slightly above the front plate.

PHOTO 5

This is a better view.  The gap is about 0.010"  When the horn is attached, the gasket will compress and form a seal.  The resultant Vfc, the front cavity volume will be the volume of the cone frustrum.  That's as small as can be made without using a phase plug.  This worked out much better than expected as the high high frequency response of the system with and without the horn is close to identical.  Too large a front cavity will adversely affect the ultra high frequency response.

 

 

 

 

The Crossover

The crossover using the Wharfedale W12RS had to be redesigned as that woofer doesn't go as high as the Eminence.  In fact, the Eminence has so much high frequency output that conventional crossover design had to be abandoned. The low mid frequency output of the horn was also high enough to cause response problems at the crossover point which, initially, was chosen at 500hz.  This created a wide band at and above the crossover point several dB higher than the rest of the response.  This band was about 2 octaves wide, starting around 400hz and extending to close to 2khz.  A wide band notch filter was considered but that isn't the easiest thing to design and get working.  The values of the inductor and capacitor in such a filter deviate from standard values requiring series and/or parallel coils and capacitors.  So, another approach was used, one that was used years ago in the Bostwicks (photo21)  What was done there was to spread the low pass woofer section from the high pass tweeter section.  Admittedly, with better speaker choices, this may not have been necessary but in a two-way system, the high frequency unit is expected to match the woofer as well as replace the tweeter.

Filter theory works well on paper, where such things as reactive loads are considered to be uniform which is anything but true for a loudspeaker.  The same logic is applied to loudspeaker radiation.  They are assumed to be radiating into 2p steradian, a hemisphere, a situation that doesn't occur in a reverberant field, such as a room.  It can be approached in an anechoic chamber where the speaker is far from any reflecting boundary; even the floor is a grate.  In a room, the wavefront from a loudspeaker strikes the floor, then the walls, ceiling and rear wall, each of which creates a reflection which interferes with the reflections from the other boundaries and this activity continues.  In a typical room, it can persist for about half a second or slightly longer.  Inn a room with a lot of sound absorbing materials, it will fade fast and is usually called a dead room.  Without such materials, the reflections can persist for longer than a second, often referred ro as a live room.  It's called, T60 which is then length of time for a pulse to decay by 60dB.  In a typical room, a decent T60 can be about 0.625 sec. to 0.75 sec.  Anyway, 'nuff digression.

After a day or two of experimentation, a satisfactory crossover was designed. It consists of a second order 200hz low pass filter for the woofer and a first order 4200hz high pass filter for the high frequency section.  the result is shown in fig 1 below along with the crossover circuit, fig 2  Note the tweeter polarity being reversed.  Admittedly, the difference is probably subtle.  It amounts to a dB or two decrease at 850hz and a rise of same around 1500hz.  The reversing of polarity was decided upon by a flatter response curve as no A-B comparison was performed.

In a designed done years ago for a friend as old as I am, a switch was placed on the rear to flip the polarity as the difference was noticeable.  He has admitted to flipping the switch depending on the music.  I have the option of digital graphic and parametric equalizers with memory storage.  For example, I couldn't resist playing with an EQ just for schitts 'n giggles to obtain the response shown in fig 2

 

 

FIGURE 1

Keep in mind that this response curve was performed at 1w1m and midway between floor and ceiling and it is quasi anechoic.  It in no way reflects that which will be seen/heard when two systems are playing at a distance of several feet and at a height of 64 inches, ear level, standing

The trick now is to duplicate this in the living room with two speakers playing. See figs 4 & 5

FIGURE 2

 

 

 

 

FIGURE 3

GREY=NO EQ   RED=EQ

The difference here is slightly audible depending on the music.  That audition was performed with a single speaker system, mono.

This is a classic example of aural perception being influenced by visual perception.  Tests have been done over the decades and proved that the colour of the grille cloth can affect what is heard.  Also, the cost of the speaker will influence what we hear and like.

Floyd E. Toole did tests whereby the listeners had no idea of the cost of the speakers being played.  It turned out tye least expensive speakers were preferred.  Then, the test was repeated and the listeners were told the cost of the speaker before the audition.  The results reversed.

If you have time and are interested, that 1 hour and 14 minute presentation can be seen here  Floyd Toole

 

 

 

FIGURE 4

An abandoned attempt at equalizing the bass (red)

Using a parametric equalizer (Behringer DEQ2496), the curve of the bass in the living room was smoothed to that which 'looked good'.  However, the sound was disappointing.  It had somewhat of a 'honk' to it, not much but enough to be unfavourable.  That was attributed to the boost in the octave between 100hz and 200hz.  By removing that boost, the obnoxious sound vanished.  The boost below 50hz was mostly unnoticeable with most music when cancelled with a bypass switch.

That substantiates the claim that the best equalization is done by attenuation of the peaks rather than boosting the nulls.  There are equalizers available that have a unity gain with 12dB attenuation.

The nulls in the higher register can be caused by absorption, in which case they can be boosted. Peaks in the upper register can be caused by room reflections and loudspeaker resonances, the latter of which can be attenuated In the bass are caused by interference as the wavelengths are longer than internal room dimensions.  Consider a 50hz tone which is 22 feet long.  In a room with smaller dimensions, that pressure wave will reflect back on itself.  Depending on room dimensions relative to wavelength, this reflection can cause a peak or a null or anything in between.  Keep in mind that for any source, there will be 6 reflections in a room with 4 walls.  Now, add two sources to further complicate this.

As for subwoofers, using two can actually improve the sound if the subs are strategically located.  This does not necessarily render a near flat bass response but it will relocate the peaks and nulls as well as change their amplitudes

 

Boosting such a cancellation is close to impossible because as the null is boosted, so are the reflections.  All that is achieved is heating the voice coil, the amplifier and possibly the speaker wire.  It's tantamount to driving a car with one foot on the brake and as the accelerator is further depressed, so is the brake.  The end result is heating the brakes (voice coil) not to mention the increased load imposed on the engine and transmission (amplifier and speaker wire)  Remember, as boost is applied, power transfer increases along with the current in the speaker wire. You can see where this is going.

The peaks can be attenuated easily simply b y attenuating the frequency/frequencies at or around the peak.  A 31 band graphic equalizer will work but not as well as a parametric equalizer.  A ten band equalizer is essentially a waste of time and those 4 band bass equalizers are nothing less than a bad joke.

In the end, bass equalization below 200hz was abandoned

 

 

FIGURE 5 (cross reference fig 1)

This is an equalized response (BLACK) of both speakers playing in the living room at a distance of about 9 feet and at a point 64 inches above the floor which is ear level while standing.  The RED is without EQ

Two equalizers were used in the Behringer DEQ2496, the graphic below 315hz and the parametric above 299hz

the reason for using both equalizers is that either can be independently turned on/off.  That option is usually used for the graphic as certain music sounds better with the heightened bass.

No attempt to boost above 15khz as my hearing stops at about 11khz.  Also, the mic is about 30 degrees off the horizontal axes and about 20 degrees off the vertical axes.

This took the better part of an afternoon

See table 1 below for the settings

 

 

TABLE 1

equalizer center frequency bandwidth (octave) dB
graphic 80 1/3 -8.5
graphic 160 1/3 +4
graphic 200 1/3 +5.5
graphic 250 1/3 -7.5
graphic 315 1/3 -5
parametric 299 1/8 +3
parametric 843 1/6 -7
parametric 1552 3/4 +8
parametric 2432 1/4 -7
parametric 2890 1/10 -4.5
parametric 3767 1/3 +4

 

 

FIGURE 6

Just for schitts 'n giggles, here's another equalized response run on LMS.  Compare this to the black trace of fig 5.

The slight differences are attributed to the possibility of CLIO being internally compensated to produce quasi anechoic results as measured at a distance of 1m and about 48" above the floor.  The LMS will run gated responses but this isn't one so in this  ungated mode, it picks up that which the microphone hears, including room reflections.

 

 

Back to the loudspeaker main page

Table of Contents