In this blog post I test smooth transition throat adapters for the Yuichi A290 horn. I've offered 3D CAD files for the Yuichi A290 for about four years now. The horn files have been popular and so I felt the need to provide a throat adapter design that has been optimized for both 1.40" and 2.00" throat diameters. Both size adapters transition to the standard 50mm x 50mm square entrance on the A290.
I decided to test both 1.40" and 2.00" compression drivers to see how they perform. Unfortunately I don't have any other adapters to compare against, so you'll have to compare my test data against your own test data if you are using more basic adapters.
For the test I designed a new throat adapter that is optimized for a smooth throat transition. I designed both a 1.40" and 2.00" version so that I could test both size of drivers. For reference these are adapters No.2322-1.4 and 2322-2.0.
The 3D CAD files in .STL format can be purchased here.
For dimensional drawing click here.
The 3D CAD file for the Yuichi A290 horn can be found here.
Drivers Tested
I decided to use the RCF ND950 1.40" compression driver because I have good test data on the RCF with my own ES290 Biradial. The only issue is that the RCF has breakup starting at 10kHz which will interfere with my polar map results.
To gain accuracy in the polar map's treble region I decided to test the SB Audience 65CD-T compression driver. This is the ceramic version which retails for $156 USD each compared to $210 USD each for the neo version. I have not tested the the ceramic version yet but on paper it looks nearly identical to the neo version. The $108 USD savings on a pair of compression drivers is welcome savings for most DIY enthusiasts, so I figured this version was worth testing. In other words, why buy neo if you don't need the weight savings?
To lower the driver's FS I used rear cover No.1993. (see below).
To purchase the 3D CAD file for the rear cover click here.
Measurements
I began by testing the frequency response at 1m for the RCF ND950. In my space I was able to gate the measurement cleanly for 4.38ms. This gives us anechoic resolution down to 225Hz.
SB Audience 65CD-T
Next I measured the 65CD-T with the following result.
I decided to measure the response with the mic at 30cm instead of 1m ungated so that I could see a little more resolution near the horn's Fc. I also lowered the sampling rate from 192kHz to 44.1kHz which will improve LF resolution slightly.
Measuring this way brings to light a peak near Fc which we did not see before due to the gating within ARTA. I will discuss this later but just know that this is related to the driver and not the horn.
Polar Map
I then measured the off-axis for the A290 with the RCF ND950.
The RCF appears to be having severe breakup in the upper treble which is effecting the color grading of the polar map. So I decided to re-plot the map limited to 15kHz.
Below is the same data presented as a waterfall-1 plot. This is simply another way of presenting the same data within the ARTA software.
Shown below is the same presented as a Waterfall-2 graph.
And finally, the same presented as a contour polar map. You can click on the image to enlarge.
Polar Map for 65CD-T
I then re-measured the polar map using the 65CD-T driver. This driver provides cleaner extension which should eliminate any variable to do with diaphragm breakup affecting our polar map.
Reducing the same to 15kHz is shown below.
Shown as a waterfall-1.
Shown as Waterfall-2.
Time Domain
For the time domain I decided to apply some EQ to flatten the response to a typical use case. I went with a -8.5dB shelf centered at 6.5kHz with a Q of 0.50.
Below is an overlay of the affect of the shelf.
Below is the burst decay for the ND950.
Below is the burst decay for the 65CD-T.
Below is the CSD plot for the ND950.
Below is the CSD plot for the 65CD-T.
Harmonic Distortion
With the same EQ applied as above, I measured harmonic distortion for the ND950 using an 85dB SPL test signal at 1m using a 30cm mic distance.
I then measured the 65CD-T under the same conditions.
Intermodulation Distortion
Using a 12 band per octave test tone signal for the same 85dB 1m SPL test level is shown below. The same EQ as above was used for this test as well. Below is the result for the ND950. IMD is around -70dB near Fc rising to -60 at 10kHz.
Below is the result for the 65CD-T. IMD is -60dB up to 12kHz. We then see IMD rise at the diaphragm breakup.
Subjective Listening
My listening was with the EQ in place that we mentioned earlier. The listening was conducted over a course of a few days at my friend's place, setup as a stereo pair.
My listening started with the ND950 which had excellent authority near Fc (300Hz). I got the sense that the A290 loaded very well near Fc and presented a very powerful male vocal. However upper treble detail suffered in the sense that it sounded slightly diffuse. I thought this was attributable to the horn however things changed for the positive when I switched to the 65CD-T which offered precise upper treble detail. However the strong clear male vocals diminished with the SB Audience driver and I was left with a resonant type of sound character (for male vocals) in the 300Hz region. Again, if I had just listened to the 65CD-T I would falsely attribute this to the horn, however I did not get this resonant character with the ND950. I should mention as well that the EQ was absolutely required. Otherwise the horn had too much midrange presence. It made me wonder how many leave the horn like this and then decide to add a super tweeter without consideration of a contour filter. There does not seem to be anything wrong with the upper treble detail if paired with an appropriate driver such as the 65CD-T with rear chamber No.2376.
What stood out for me was the wide horizontal off-axis coverage. As I moved about the room there was definitively some noticeable changes in the frequency response however this was offset by the added benefit of a more spacious sound character. (if that is your listening preference)
When listening to the 65CD-T I decided to increase the LR4 300Hz HPF to 500Hz. This completely alleviated the resonate sound character I mentioned earlier. I would suspect that if pairing with a very good 15" woofer, the 65CDN-T would be my choice. However if I was keen on using a 300Hz crossover point then the ND950 would be my choice. This is a very viable option that one should consider. You maintain that strong and powerful male vocal character however the upper treble was noticeably softer compared to the 65CD-T. So I feel that ultimately the ND950 with a super tweeter such as the Fostex T90A would yield the most enjoyable sound quality.
Lastly I tried the B&C DCM50 16ohm using the 2.00" version of the 2322 adapter.
While this driver performed admirably on my ES290, this driver did not fair too well on the A290. The sound was too diffuse and blurred. I feel like the combination of paper diaphragm and falling response of the A290 made the upper treble too soft, even with my EQ applied. For reference, the measured frequency response is shown below.
I should mention as well that I've recently tested the DCM50 on the Martin Horn AH-340 and it sounded excellent, with stunning clarity and soundstage depth. So if you are looking for something really special you should check out this combination.
Conclusion
The A290 provides wide coverage albeit with some off-axis anomalies. The wider coverage comes at the cost of a falling response similar to what you would see with a constant directivity horn. So a contour network is required to flatten the response, or implement a super tweeter. Applying a contour filter will increase distortion by a proportionate amount, so this is just something to keep an eye on depending on the application and driver choice. The A290 was happy to play down to the Fc without any trace of first order reflections, however the driver needs to be up to the task. I found that the RCF ND950 and DCM50 was perfectly happy playing down to 300Hz. The measurements did not show any issues in the time domain (burst decay, CSD) for the horn.
Lastly, one thing I would like to point out is the very steep cutoff at Fc for use with the ND950 shown below. This measurement was done using a 30cm mic distanced ungated, allowing me to see more detail on what's happening near cutoff in my listening space. I can't measure outdoors this time of year in Canada (snow).
I'm not saying this is a negative thing, but something that would need to be factored in when developing the LPF for the LF. I suspect a LR4 or even a LR6 LPF is required to match slopes at the crossover point when running the horn down to 300Hz.
Future Testing
I've ordered a pair of 18Sound ND1480BE (3" beryllium) compression drivers which I plan on testing.
Also, I customer sent me a new-in-box TAD TD-4003 which I plan to test as well, however I need to print 1.53" (39mm) diameter throat adapters for the test.