In this blog post I test the Audience A3S in the Sabourin Horn No.1300. Detailed measurements were conducted on the A3 back in 2009 by Audio Xpress Test Bench which can be found here. So I am not going to conduct non-horn loaded measurements.
Audience sent me these drivers to test in the Sabourin Horn. I don't believe they want them back so full disclosure on any biases I may have.
The drivers are a drop in fit with no modification to the existing Sabourin Horn No.1300. I used the thick rubber gasket supplied with the drivers to front face mount into the horn which provided enough physicals clearance for the diaphragm movement outwards.
The Audience A3S has dual voice coils with each voice coil having a nominal impedance of 8ohm. So I wired the two drivers in series-parallel to sum to a nominal load at the amplifier. The amplifier used is the Hypex FA501 along with the Scarlet Solo as a DAC, running to my laptop via USB using ARTA measurement software. For distortion measurements I used the Shure SM58, and for frequency response measurements I used the Dayton Audio UMM6 Reference Microphone along with the Shure SM58 for distortion.
I began my testing by conducting a frequency response test at 1m. I conducted my testing with rear chamber's back cover removed so that the drivers were exposed open air at the back of the horn. I set the voltage to 2.83V and the achieve 93dB Sensitivity at 3kHz.
Below is the burst decay result with no gating applied. The first room reflections were about 2m away.
Gating out the first room reflection is shown below.
Step response is shown below.
I then decided to implement a 350Hz fourth order high pass filter and conduct distortion measurements. I used the active filtering capability within the Hypex FA501 plate amp to achieve this. Below is the frequency response with the high pass filter in place.
I then measured harmonic distortion setting the output voltage to 2.83V which produced 93dB at 3kHz.
Harmonic distortion is extremely low in the 3kHz region:
H2 0.04%
H3 0.02%
H4 0.009%
Translating these values to dB scale we have:
H2: -68dB
H3: -74dB
H4: -80dB
If we look at the harmonic distortion near the horn's cutoff frequency, we have distortion values at 400Hz:
H2 0.09%
H3 0.04%
H4 0.004%
Translating these values to dB scale we have:
H2: -60dB
H3: -69dB
H4: -88dB
I decided to increase the test SPL to 100dB and achieved the result below.
H2 0.1%
H3 0.02%
H4 0.003%
Translating these values to dB scale we have:
H2: -60dB
H3: -73dB
H4: -89dB
Looking at the 400Hz region (100dB) we have the following results
H2 0.2%
H3 0.05%
H4 0.007%
Translating these values to dB scale we have:
H2: -54dB
H3: -67dB
H4: -82dB
These results are exceptionally low even at very high SPL. The Audience A3 is pretty much immune to a high output level of 100dB. I did not want to test at any higher output levels due to others in the house.
I then looked at intermodulation distortion and produced the following results. These tests were done without the high pass filter. I used a 12 band per octave test tone from 300Hz to 20kHz.
Distortion is -65dB at 3kHz with a test signal SPL of 80dB.
Increasing the test SPL to 85 produced the following result.
We lose -5dB at 3kHz resulting in distortion at -60dB.
Increasing the test SPL to 100dB we see distortion rise to -55dB at 3kHz. The rest of the spectrum varies with the 2kHz region being +10dB better distortion.
Application
These drivers seem to do very well in the Sabourin Horn No.1300. I would suggest that these drivers be used as midrange only covering from 300Hz to 5kHz where you can pass it off to your favorite tweeter. In this configuration the drivers reproduce a very flat frequency response, very low distortion, as well as good sensitivity of 93dB with an 8ohm load.
Passive Crossover
I decided to rig up a passive crossover to see what the results would look like. We need to implement a contour circuit to flatten the natural transfer function provided by the horn. This is a 6dB per octave falling response starting at the horn's Fc of 300Hz. The response drops from 102dB at 300Hz to 92dB at 5kHz.
I began by importing the ZMA and FRD files into Xsim3D which is shown below as a baseline.
I then messed around to see what would work and came up with a single 22uF capacitor which is shown below.
Actual test data is shown below with a 22uF capacitor in place. I've overlaid the raw response for comparison.
Showing just the resulting response is shown below.
Now, if we wanted to use these drivers as full-range extending into the treble would would have to implement a high frequency contour network.
I modelled what this would look like in Xsim3D which is shown below. I also decided to add an inductor L1 to kill off the bump at 400Hz.
Actual test results are shown below from with the schematic wired up from what is shown above. The overlay is shown with and without the HF contour comprising R1 and C2. Although we do lose sensitivity since we are having the match the MF with the HF. We are now at around 88dB sensitivity however this is now a point source solution covering 300Hz to 20kHz.
Showing the final response without overlay is shown below.
Testing distortion again at 100dB shows the following.
If we compare these numbers to the GRS PT6825-8 planar driver we can see that distortion is about +10dB improved with the dual Audience A3S. 
Subjective Listening
I setup a side by side comparison between the GRS PT6825-8 planar and the dual Audience A3S. I listened to a variety of music starting with Dave Matthews 'Lie in our graves' live at Radio Station. which has a very dynamic acoustical guitar starting at 3:05. I used this track to see if there was congestion in the strumming or compression of the plucking of the strings.
The Audience driver sounded more dynamic through the midrange frequencies however it was somewhat rolled off in the highs with detracted from a sense of dynamics in the upper treble. I found however that the Audience had better noise floor retrieval and overall sounded less harsh than the PT6825-8.
Moving onto Mazzy Star 'Fade into you' I found her vocals to really project with zero distortion on the Audience A3S. There was a tendency to just want to increase the volume higher and higher, where the PT6825-8 had a definitive limit on how loud I wanted to take it.
I then played Etta James 'At Last' and found again that the vocals on the Audience A3S exceeded the planar driver reproducing her vocals with startling presence. This song is also a good test for harshness, being almost intolerable on some speakers. Here I found that the Audience was less harsh sounding than the planar which may have been due to the rolled off highs. Even if I adjusted the EQ +2dB in the treble I was not able to achieve the type of treble presence that I was looking for. I couldn't help but wonder that perhaps a super tweeter was all that is needed.
I then tried Billie Jean and this song somewhat sealed it for me. This song has the keyboard synth which sounded crystal clear even through the strong bass line and whack of the snare. The electric guitar came through sounded very smooth and realistic.
Final Remarks
Overall this has been an interesting test. I shows that the Sabourin Horn No.1300 can be achieve a very flat frequency response through the midrange. One idea is to rotate the Sabourin Horn sideways so that the horn is wider than it is tall. This allows closer driver spacing to a tweeter. The Sabourin horn still produces excellent off-axis pattern control when laid on its side. In this orientation you could go with a dual woofer cabinet where the woofers are flanking side by side, perhaps two 10" or 12" drivers.
It also has inspired me to look at other drivers that may be more affordable and with a lower QTS which is more suitable for horn loading. The Audience A3S has a QTS of 0.32 which is pretty good. Finding something lower than that may be difficult.
Overall I can highly recommend the Audience A3S in the Sabourin No.1300 horn but I do suggest a super tweeter.