These speakers were first envisioned back in 2006 and were built in 2008-2009. The goal was to have a highly configurable test bed for evaluating drivers and different crossover topologies, with the additional ability to switch quickly from 3-way to 4-way to 5-way using a remote control.
Part 1 of this article described the updated electronics and provided an overview of the new libraries that were being developed for the ADAU1466/1467 DSP. Part 1 was written about 3 years ago, when I first decided to resurrect these speakers. There are a lot of things that were still in progress at that time, and most of those upgrades have been completed. However, I didn’t do any measurements or critical listening until this year (early 2026), so that article was never featured on this site.
I still had to develop some additional code to allow switching between different DSP configurations and to allow echoing commands from the right speaker to the left so that both received configuration updates at the same time. This additional code has made evaluating these speakers much easier, as I can use the cell phone app to switch between 4 different configurations. As a result, I can now provide some initial impressions, along with some measurements.
Impressions
First, these speakers have a “big” sound. There are 5 tiers, and each of the upper tiers is a horizontal “line “ring array” with 8 drivers. The multiple drivers provide excellent coupling to the room, which results in these speakers filling up the room in a way that feels natural and effortless. I know that is a strange way to characterize a speaker, but these speakers really sound much different than simple multi-way speaker boxes, and the effect is nice.
Second, with these drivers, the difference between using 2-pole Linkwitz-Riley (LR) crossovers and 8-pole LR crossovers is not as pronounced as I expected. The 2-pole crossovers sound a bit “muddy” compared to the 8-pole or 4-pole crossovers, but the difference is subtle. I’ll continue to evaluate different combinations of driver delays, EQ and crossover types in search of the most refined sound. However, I doubt that I will attempt to experiment with 3-way vs 4-way vs 5-way, as I’m not sure how I could use that information for future designs. The choice between 3-way vs 5-way is more dependent on driver characteristics, so testing with just these specific drivers might be interesting but not particularly useful.
Third, they sound good. They reproduce complex choral music with impressive clarity, and they seem low-distortion and accurate.
Fourth, the difference between monopole mode and dipole mode is massive, and dipole mode clearly sounds much better. But before going further we need to clarify how these terms are being used here, as the terms “monopole” and “dipole” are not being used in the conventional way.
“Monopole mode” for this speaker means that all of the drivers are in phase, and since they are arranged in a ring, they radiate from a roughly cylindrical shape (it is actually an octagonal prism, oriented horizontally). In what I call “dipole mode”, the back 4 drivers are 180 degrees out of phase from the front 4, which is similar to the radiation pattern from a dipole woofer, but the quasi-cylindrical shape makes it different. I’m calling this radiation pattern “dipole mode” for lack of a better name. I believe that there is some discussion of ring-array configurations with variable phase in studies by Mellow, but I haven’t pursued this research.
Whatever we want to call this “dipole mode”, it is truly impressive, at least in the room where I have these speakers. The sound is much more open and “natural” sounding, and it is much more pleasant to listen to. By comparison, the monopole mode sounds “boxy”, where it is easy to localize the sound as coming from a specific place. The dipole mode includes frequency compensation that is derived from the discussion on the Linkwitzlab web site, and the corrected frequency response measurements of monopole and dipole are similar. The dipole mode sounds far superior, and the difference between monopole and dipole mode is not subtle: it is jaw-dropping.
The room where these speakers are being used has a 12′ cathedral ceiling with 45-degree walls converging on the ceiling, which may accentuate the effect of reflected sound from the dipole. However, I noticed similarly improved “spaciousness” when testing these speakers in a much smaller room with parallel walls. I have to conclude that there is something about the reflected sound from dipoles that is quite satisfying.
Some measurements
Measuring speakers that radiate from a large area and that are intended to exploit room reflections is problematic. There are no simple answers as to where to place the measurement microphone, and it is difficult to interpret the reflections and diffraction effects from the complex enclosure. The starting point for measuring these speakers is to measure each set of drivers separately, without any crossover or other filtering. There is a user interface screen that allows adjusting the volume of each set of drivers, and one of the volume settings is “OFF”. The measurements below isolated each ring array separately, and all measurements were taken at about 7 feet away using a sine sweep signal with 1/6 octave smoothing. The speaker was approximately 2 feet from the wall. Each measurement was performed with all drivers in phase (“monopole mode”) and repeated with the rear 4 drivers out of phase (“dipole mode”).
Super-tweeter
The super-tweeter response doesn’t appear to be affected much by the dipole mode, at least in the frequency region in which it is used (8000Hz and above). With a suitable crossover, there will be no frequency compensation required for dipole mode.
Tweeter
The tweeter array illustrates the difficulty of getting meaningful measurements from this type of speaker. The distance between the drivers in the front and rear is on the order of 4.5″ for this ring, so we would expect some backwave reinforcement in the monopole mode for frequencies around 3000Hz. That peak does show up in the monopole measurement, and it appears as a peak at half this wavelength (6000Hz) for the dipole. Those peaks make sense, as does the gentle roll-off at lower frequencies for the dipole. The dipole compensation DSP blocks could be configured to provide a notch for the dipole mode, but there is no easy way to add a notch for just the monopole in the current SigmaStudio design, so I need to think about this.
The tweeter is intended to operate between 3000Hz and 8000Hz in the current crossover designs, and the measurements are not as “flat” as desirable in this range. Some of the uneven response is most likely due to diffraction effects from the ring enclosures, as the tweeter array is 2″ tall, with the midrange enclosure about 2.5″ from the center of the tweeters. The diffraction could no doubt be minimized with absorbent materials, but then the calming effect of the green LED’s would be hidden 😘.
Midrange
There is a broad peak at 400-500Hz for the monopole mode that I can’t explain yet, as this would be a reinforcement wavelength of around 30″. It’s not a big issue since the midrange crossover was expected to be around 750Hz. That peak might be due to the reflection from the wall, but I’ll need to move things around and remeasure. There is also a large peak at around 2200Hz for the dipole mode, which doesn’t make sense, as the out-of-phase back wave from the 6″ midrange ring should result in a dip at around 4000Hz rather than a peak at 2200Hz. These measurements may be suggesting a lot of diffraction from the large woofer panel below the midrange tier, but I will need to get some heavy felt to isolate the source of these peaks and dips.
Woofer
There are two things that stand out in the woofer measurements. First, there is bass roll-off starting at 200Hz that is probably due to the enclosure being too small for those 8 5″ woofers. Second, there is a clear roll-off from cancellation of the back wave for the dipole, starting at 300Hz. The dipole compensation uses a low pass shelving filter that will correct this roll-off when set to 300Hz. Using the calculations provided by Linkwitz, the shelving filter will need to have a gain of approximately 15.
Unfortunately, the peaks and low points of the frequency response above 300Hz don’t line up well with predictions based on using a tape measure to estimate the path from the rear drivers to the front. There is a broad peak around 300Hz for the monopole that corresponds to 45 inches, which might be due to wall reflection, but additional measurement will be needed to determine the cause.
Subwoofer
The peaks and dips at the lower frequencies are going to be caused by the room. However, the general shape of the dipole response is clearly showing the cancellation from the back wave at frequencies below 100Hz. This roll-off required setting the low pass shelving filter in the dipole compensation to a Gain of almost 20. With that compensation, the bass response was similar to the monopole mode at a measurement distance of 7ft.
Overall Response
The overall response with LR2 crossovers at 120, 750, 2000, and 5000Hz shows that there is still a lot of work needed to adjust the EQ and custom filters that are available from the cell phone user interface. This is just one speaker in a corner of the room measured at 7 feet. What I think is most interesting about this measurement is that the monopole and dipole modes have a similar frequency response, yet the difference in sound quality is huge.
What’s next?
These speakers are set up in a room where I can enjoy them for a while. I might make a few more measurements and tweaks, but I need to return to some line array projects, so it might be a while before any more work on this is completed. Also, we are planning on returning to the west coast in the next two years, and there is a lot of work I need to do on this house before we can move. So, for now, these are mostly “done”.
Ideally, these speakers need to go on the road where others can play with them. They were completed to this point as a learning experience, and there is still a lot that can be learned by tweaking the DSP and measuring the response and listening to the results. The speakers are heavy and it takes about an hour to dismantle the top from the bottom and reassemble them, but it is not impossible to relocate them. Who knows…maybe they will be coming to an audio show near you.