Co-Entrant Horn No.1416

In this blog post I feature the test results from Co-entrant Horn No.1416. I was commissioned to design and build this horn for a customer who is using it for his personal system. The horn is part of a complete system shown below. The customer requires a vintage appearance and so this is what I came up with. Despite this, the speaker uses the latest technology from Hypex, ScanSpeak, and Viawave.

Horn No.1416 Design Features:

  • x1 Viawave SRT7 pure ribbon tweeter 
  • x2 Scanspeak 10F/8424G00 midrange drivers 
  • ES circular horn using a 300Hz cutoff frequency (Fc) 
  • Horn physically measures 60cm in diameter 
  • Special co-entrant throat 

I chose the Viawave because it differs from other pure ribbon tweeters in that it actually has a surround. This means that the diaphragm does not short circuit the air flow to the rear of the diaphragm increasing power handing and lowering distortion. Sensitivity after horn loading the tweeter ends up being 100dB which is just slightly lower than a small format compression driver. 

The ScanSpeak 10F/8424G00 woofers are wired in parallel for a nominal 4ohm load. The nominal sensitivity of each driver is 87dB. Since these are wired in parallel this raises the sensitivity +6dB to 93dB. Horn loading the drivers increases this again another +9dB which results in a overall horn sensitivity of 102dB. I also chose these drivers based on success with previous projects. They respond extremely well to horn loading due to the powerful motor which uses neodymium ring magnets. The motor also utilizes copper shorting rings for reduced hysterisis distortion. 

Below are pictures of the cherry hardwood horn in sanded pre-stained condition. I hope to publish pictures of the finished speaker soon. 


The special throat design shown below. The design uses special elliptical hole shapes allowing the midrange frequencies into the horn throat. The holes are sized and shaped to provide just enough opening to not restrict output from the midrange drivers but not too big to interfere with the wave propagation from the tweeter. It took building four prototypes before finalizing the design. Previous designs resulted in frequency response dips caused by cancellations. The end design is highly optimized in the sense that nothing can really change or else the overall performance would quickly deteriorate.
My messy setup as I try various crossover topologies and values. Crossover development was rather straightforward since the raw response from each driver was well behaved. 

Below is the computer model of the design. The midrange drivers are open back. This subjectively provided the most “open” sound although closed back would improve distortion at extremely high SPL. 

Below is the section view showing the internals. The midrange drivers need to be as close as possible to the horn throat and ribbon tweeter. Extreme effort was made to achieve this including pocketing out for the midrange drivers’ surrounds. 

Shown is the throat adapter piece which is precision CNC machined from solid hardwood. 

Measurements 

I began by measuring the frequency response of the midrange drivers. The horn achieves a flat response from 350Hz to 1.8kHz which is 2.5 octaves of bandwidth. Most horn designs can only cover this amount of the spectrum which is a well known technical limit in horn theory. 

I then rigged up a passive crossover and optimized the response. The crossover schematic is shown below. The tweeter is a simple first order with a contour network that adjusts the falling response from the horn. The mid frequency gets a large value capacitor to really kill off the peak at 6.5kHz. The inductor L1 is a very small value of only .33mH. So we are not using anything too aggressive, just some gentle coaxing. 

Below is the final frequency response of the passive crossover. The end result is within +/-1.5dB! 

I then measured the step response which is shown below. As you can see, the tweeter and midrange show good physical time alignment. The step response settles quickly without any issues. 

The impedance sweep is shown below. This includes the passive crossover. The midrange’s FS is at 80Hz. This ensures a flat phase response at the horn’s low frequency cutoff of 300Hz. The impedance sweep is smooth indicating the absence of any resonances or issues in the system. Impedance dips to 3.8ohm at 3kHz. 

I then conducted burst decay which is shown below using a 25dB vertical scale and no gating. The burst decay shows that the horn is free from any resonances.

Lowering the noise floor to -35dB  and gating out room reflections produces the following burst decay (see below). There is a small broad Q resonance at 10kHz however it is lower than -25dB in the floor.

I then looked at the cumulative spectral decay waterfall plot which is shown below. There are some small resonances that correlate to frequency response bumps which is most often the case. One of these bumps is at 3.8kHz which is a 90mm acoustical wave length. Dividing this by four is 22.5mm which is the width of the ribbon diaphragm. So perhaps this can be addressed with some foam or other object that prevents the resonance from forming. 

I then measured the horizontal off-axis response which is shown below as a coloured polar map. We can see where the midrange drivers become somewhat directional at 1.8kHz where the tweeter then takes over at 2kHz. The directivity pattern is then widened again after 2kHz due to the narrow throat size of the ribbon tweeter. The midrange drivers are directional due to the center to center spacing between the two openings into the throat. 

I then measured the vertical off-axis response which is shown below. The midrange driver's opening into the horn throat are now arranged vertically which contributes to the good horizontal coverage up to the 2kHz crossover point. The tweeter's vertical directivity becomes quite narrow starting at 6kHz which is expected because of it's tallness. Reviewing these graphs in detail it becomes obvious that the midrange drivers should have been configured vertically instead of flanking either side of the tweeter. This is an example on how test data like this can help guide future design changes. Regardless however, the current design is still excellent, especially when comparing against other competition’s test data which I can show later for comparison. 

Distortion 

I measured harmonic distortion at 85dB which is shown as a percent below.  

Harmonic distortion is extremely low at only 0.072% for the second harmonic. Third harmonic is even lower at 0.033%, certainly inaudible. 

 

Intermodulation distortion is shown below. Using an 80dB test signal distortion is extremely low at -70dB through the midrange and treble region. 

Raising the test SPL to 85dB increases distortion to -55dB. (1.5kHz region) which is a little higher than I’d like to see, but a good result. By comparison the Danley SH50 Synergy Horn has THD as high as -42dB for the same test SPL. 

Raising the test SPL to 90dB produces the distortion profile shown below. Distortion is the worst at the 2kHz region showing -50dB down. But again, this is at a SPL level that would be typically at 100dB+ for a complete system producing the entire frequency spectrum. 

Conclusion

Subjectively I am happy with the overall sound. This design certainly captures the coherence only available from point source, co-entrant or concentric configurations. The 1416 horn provides extreme levels of clarity however we do see some compromise in the off-axis directivity characteristics compared to a dedicated horn for each part of the frequency spectrum. Is the trade-off worth it? I believe there is a strong case for the co-entrant concept in terms of simple musical enjoyment. There is an ease when listening to point sources I find. The step response is basically textbook perfect ensuring seamless integration of the midrange and treble. In terms of distortion, the 1416 has lower distortion than any compression driver I've measured when comparing the same SPL of 90dB. 

Further Comparison for Context 

I thought it would be useful to compare my test results against a highly regarded co-entrant design, the Danley SH50 which was recently tested by Erin’s Audio Corner using the Klippel near field scanner system. Now granted, the SH50 is designed for the professional sound market, but it should offer some context on the challenges faced in designing a co-entrant horn. 

Manufacturer’s Published Data 

Test results from Erin’s Audio Corner 

I’ve stretched the image aspect ratio of the results below to match the 4:3 aspect ratio used in my measurements.

For quick reference my results are shown below.  

 Another example of a concentric point source option is the coaxial drivers available from nearly every professional sound driver manufacturer. An example of which is shown below, the DIYSG Volt-8 speaker. 

Here is the step response of the Genelec 8331A studio monitor. 

Now, granted the Genelec has pretty good on and off-axis response as shown below. 

 I show all of this to simply say that designing a speaker of this nature is challenging. These graphs give context to where the industry is at from a test data perspective. There remains strong interest on the subject especially in the DIY scene. Many are 3D printing their own creations following extensive AKABAK simulations. For me I will continue to develop Co-entrant designs in parallel with other configurations such as the Nighthawk which will be published in the fall of 2022. For commercial or personal project inquiries please email me at Joseph_crowe@josephcrowe.com

Back to blog