@thewas
Don had positive results with dipole CBT implementations before we connected and published the results in one of his papers (below). In terms of how he feels about the H1 in particular, good enough that we're working on a scaled up, floorstanding one at his request.
I'm happy to pass on him any specific questions you have and post his response.
In terms of your certainty of beaming, I've already shared vertical and horizontal measurements. If it helps, here is the off-axis response of the drivers used. Beaming isn't an issue.
View attachment 110741
This is Don's dipole paper:
https://faculty.tru.ca/rtaylor/publications/cbt_dipole.pdf
Conclusions reproduced below with brief comments on the advantages of dipole CBT highlighted in bold:
We have shown that a constant-directivity source can be formed by a circular-arc array of dipole source elements with frequency-independent amplitude shading. The theory developed here is a natural extension of that presented in [1] for circular arrays of monopole elements, which in turn is an adaptation of the corresponding theory for spherical-cap arrays [7, 11]. An appropriate choice of shading function leads to constant-directivity behavior. Several suitable shading functions appear in the literature, giving the designer access to a variety of beam shapes and widths. The shading function directly determines the radiation pattern in the plane of the array and, together with the arc radius, also determines the cutoff frequency above which a frequency-independent radiation pattern is achieved.
In terms of managing directivity, a dipole CBT array has several advantages over previous CBT designs based on monopole elements [1, 2, 3, 8, 9]. A conventional CBT array becomes omnidirectional below its cutoff frequency (when the array arc is smaller than the acoustic wavelength). This necessitates very large arrays if constant directivity is to be achieved over the whole audio band. By contrast, a CBT array of dipole elements radiates with a dipole pattern (hence with 4.7 dB greater directivity) at low frequency. This makes it possible to achieve broadband constant directivity with small arrays. At high frequency, a conventional CBT array presents a strong amplitude peak (tens of dB relative to on-axis) along the axis of the circular arc. Although this peak radiates into a small solid angle, and so has little effect on overall directivity, it may nevertheless be undesirable in some applications. A dipole CBT avoids this issue, by placing the dipole null of individual source elements where these peaks would otherwise occur. Dipole sources are very inefficient radiators, with a response that falls off at 6 dB/oct at low frequency. In a practical implementation this must be compensated by equalization, together with a large radiating area (e.g. in the case of electrostatic panels) and/or large linear displacement (e.g. in the case of conventional piston drivers in an open baffle). This leads to considerable engineering challenges, since large displacement typically incurs high distortion, while to maintain a frequency-independent radiation pattern one requires that the source be acoustically small. CBT dipole arrays address both these issues: being acoustically large by design, a dipole CBT provides a scalable way to increase radiating area without compromising the radiation pattern. Indeed, making a CBT array larger actually increases the bandwidth over which constant directivity is achieved. The low-frequency roll-off of a dipole CBT array must be compensated by equalization if the goal is a flat magnitude response. A naked dipole requires 6 dB/oct equalization at low frequency, which quickly runs into practical limits on driver excursion and signal headroom. However, the raw responses shown in Fig. 3 give an indication of the milder equalization required by a dipole CBT array: above cutoff the slope is only 3 dB/oct. Only below cutoff does the slope increase to 6 dB/oct; with larger arrays the bandwidth of this more demanding regime is reduced. The equalization required for a dipole CBT is quite different from that for an array of monopole elements, which requires only a +3 dB/oct boost above cutoff. A practical device implementing our theory could be formed by a discrete array of conventional drivers, much like that in [3] but with an open baffle. Such a device would necessarily be an approximation of the continuous line source considered here. Several engineering issues arise that are beyond the scope of the Page 8 of 9 Taylor, Manke and Keele Circular-Arc Dipole Line Arrays present work. These include effects of discrete sampling of the continuous shading function, spatial aliasing due to finite source spacing, the finite size of both source and baffle, mutual coupling, and the departure of radiating elements from ideal dipole behavior. Much of the relevant theory is presented in [4], and we plan to address these practical issues in subsequent work.