The Big BenD Bass Horn
Those who have followed the evolution of my system, will have noticed that I haven't had proper bass horns since 2010, before I moved to Trondheim. In Trondheim I had Altec 816A bins, but they don't count as bass horns. They have a cutoff of about 160Hz, and create a bump at about 200Hz, which made them a bit challenging to cross over to my midrange horns.
Proper bass horns is something different. Having experienced horn loaded bass down to some 35Hz before, I miss the physical impact, detail and texture you get with proper horns. So after the Horn Book was completed, and I had recovered from the experience and enjoyed some spare time (although most of it in lockdown), it was time to bring a bass horn into the Kolbrek household again.
Having coauthored the book on horn speakers, I knew I couldn't take shortcuts. Like building someone else's design! There are many clever and experienced horn designers out there, but in addition to using the knowledge gained while researching the book, I also wanted to make a horn that would match my room and system.
I started thinking about this several years ago. Some quick calculations showed that a 2.5m long horn with a 7000cm2 mouth, using an Altec 515-8G driver, would produce decent bass down to about 40-45Hz. It could also be built in the form of a J, with the mouth at the floor and the throat near the ceiling. But at this stage I was still workig on the book, and had neither the time nor the tools to start such a project.
A second idea was to build a midbass horn covering down to about 100Hz, and making a subwoofer for the frequencies below 100Hz. The concept I was toying with was using two 12" drivers in a straight midbass horn, with about the same mouth size as the bass horn outlined above, and some sort of Slot Loaded Open Baffle (SLOB) built into a platform the horn could stand on. This would make the horn design fairly straightforward, but had the complication that I needed another amp and two more DSP channels. While I plan to upgrade my DSP, this approach required too many changes to the system to be convenient. Another problem was also showing up.
When the first lockdown started in 2020, with furloughs and working from home, I was asked to worked reduced hours. The extra time was spent extending my horn simulation software to approximate the reflections from a rear wall. Hornresp assume a corner horn to be placed in the actual corner, essentially the horn is built into one of the walls. In my PhD I extended this modelling to allow the horn to be moved away from the corner, but still being mounted in one of the walls. This does some rather unpleasant things to the radiation impedance, creating dips that create larger impedance ripple and reduced output.
Using the image source method and a simple low frequency approximation for diffraction, I was able to model the horn in front of a rear wall in a way that matched reasonably well with BEM. All good so far. Except this placement had the same shortcomings as when a horn is moved away from the reflecting surfaces. The added time delay and phase shift creates dips in the radiation impedance and response. For the placement I had planned for the midbass horn, the dip came right at the lower end of the band, effectively pushing the lower 3dB frequency up by a significant amount.
So after some simulation and thinking, I abandoned that idea too, and began looking at the J-curved horn again. And I realised that to make a horn like that, I had to upgrade my curving tool. That took a few months, but then I was ready to design my new bass horn.
Before starting a project, it's good to make a list of requirements. Then work from the requirements towards a design that will fulfill them. In my case, I wanted the following:
- Bass down to at least 40-45Hz
- Not (too) undersized mouth
- Not critically dependent on driver parameters
- Easy to cross over to the midrange horn at about 300Hz
Low Frequency Extension
The low frequency extension depends on the horn lenght and mouth size. Essentially we want the throat impedance of the horn to follow that of the infinite horn as closely as possible, but this is impractical for bass horns. We should still try to avoid impedance peaks. The lenght also needs to be sufficient, so that the first impedance peak isn't placed too high. This becomes more critical the smaller the mouth becomes.
Not Undersized Mouth
This is important for to reduce the impedance ripple. A good match between driver and horn (set by driver parameters and throat area) will minimise response variations in the face of impedance variations, but the horn can often end up fairly narrowbanded, especially of it is folded.
But mouth size is also set by practical limitations. In my case a mouth size of about 100 by 70 cm was the largest I could allow. This is a fairly substantial mouth size, nearly the size of the baffles I currently use, but still relatively small for a 40Hz horn. I would have to accept a compromise here.
If throat impedance ripple is reduced, so is sensitivity to driver parameters. This means that several drivers are possible, and that I can experiment without having to do major modifications.
Crossovers in horn systems can be tricky. Direct radiators may have some bandwidth outside the crossover frequency, but are still often pushed quite far. Textbook crossovers require the response of the drivers to be flat way past the crossover frequency, otherwise the response of the drivers have to be factored in when designing the network. Or, if using an active crossover, EQ applied to flatten the response so that the acoustic slopes are as desired. For horns, things often seem to be pushed even further. Horns need to be a certain size to be usable down to a given frequency, and although they have high group delay near cutoff, people often want to use as much of that hard-earned bandwidth as possible. Bass horns are often extra tricky to cross over, as they may be folded and have response irregularities, sharp dips and peaks that makes the phase go haywire.
I usually cross my midrange horns at around 300Hz, so the horn should behave well up to at least 400Hz, preferably higher. When testing horn folding methods for the Horn Book, I noticed that a fold typically creates a sharp loss of resistive loading above about 200Hz in typical bass horns. This would not be good for a wide band bass horn.
Curving, on the other hand, while having its own challenges, doesn't create the reflections and cancellations of a folded horn. Therefore the horn would need to be gently curved, not too sharply.
In order to get the horn into the house, it would have to be divided into sections. This also makes it possible to experiment with inserting other sections, removing sections and so on. It creates more possibilities for problems, like leakage, but the flanges also act as bracing.
The "Blameless" Concept
"...the concept of what I have called a "Blameless Amplifier", the name being chosen to emphasise that the remarkably low THD comes from the avoidance of errors rather than from fundamental advances in circuitry."
I have a great deal of respect for Mr. Self and his approach to audio design, and especially the blameless concept is something well worth adopting for more than amplifiers. Oftentimes we want to make "the best" speaker or amplifier or turntable by applying some new and inventive concept, while very good results can be had by identifying problems in old and "boring" technology and fixing them. Self identified a series of distortion mechanisms in a classic, standard amplifier topology, and through simulations and practical tests reduced or eliminated these mechanisms.
In the same way, I will try to eliminate or reduce as many problems as possible in this bass horn. Since it is my first design using this approach, and since the thought came up after I had started building the actual horn, it will not be completely blameless. But I will, as far as possible, do scientifically and rationally guided choices, backed up by simulations and measurements.
Since this design will take some time to complete, I will post updates in my blog, using this article as an index page to the blog entries.