These do look interesting. The dual top plate setup is essentially a way to use the underhung approach with a high-excursion driver. Woofers generally use the overhung approach, with a long coil that extends well beyond the top plate in both directions. Tweeters are generally underhung, and with midrange drivers, some are underhung and some are overhung.
Conventional wisdom has long been that it needs to be one or the other, in order that the overlap between the coil and the gap will remain essentially constant over a large range of excursion. In one of the Klippel papers that discusses the various causes of distortion, there is a small graph and some accompanying text that briefly explores the possibility that it might actually be preferable for the length of the coil and the height of the gap to be the same. This is counterintuitive because it obviously means that the coil-gap overlap will begin to diminish as soon as the coil/diaphragm assembly has moved even a slight amount. The rationale, as briefly discussed in the Klippel paper, is that when it is done this way, only the low-order harmonics of the fundamental are stimulated at low excursion, while the high-order harmonics will actually be diminished and will remain weaker (vs. one of the conventional approaches) until excursion has nearly reached the mechanical limit. (To complete the picture of why this would be desired, we need also to consider that our hearing is apparently more sensitive to high-order harmonics than to low-order harmonics - because the masking effect of the fundamental isn't as strong for frequencies further away from the fundamental.)
Anyway, what we have here appears to be a solid effort to use the underhung approach with high-excursion drivers. The voice coil is presumably very short, however the length of the coil isn't stated, and one question that is raised is how the height of the gap should be measured. It isn't apparent whether the coil will remain fully within the space between the two top plates even at the mechanical limits of excursion, or whether, at high excursion, the coil will overlap one of the two top plates in one direction and the other top plate in the other direction. If this is what happens, then it is probably more like the approach suggested by the Klippel paper, i.e., the coil length and the gap height being approximately equal. In this case, the measurement of distortion for the driver will be tricky, and there is huge potential for misinterpretation of the distortion measurements. If the Klippel paper is correct, then we would possibly see that the low-order distortion components don't look very good at all, but that the high-order distortion components are exceptionally good even at unusually large excursion. And in this case it will be particularly important to interpret the distortion measurements in accordance with how we hear the individual harmonic components.
I've noticed in some of Erin's excellent and incredibly detailed analyses of drivers that he measures linearity separately from distortion. He measure both linearity and distortion in accordance with established standards for the respective measurements. I'm much too lazy to have bothered to read those standards, but I thought it was interesting that driver linearity and distortion would be treated as two separate things - I would have thought that they are so closely linked that each is merely a different way of looking at the other, and that the measurements would show this to be true. If Erin decides at some point to apply his driver analysis skills to one of these drivers, it will be an opportunity to delve a little deeper into the question of how the measurements of linearity and distortion each reveal something not revealed by the other. And it will be an opportunity to explore the question raised by that Klippel paper, if the drivers exhibit unusually high low-order distortion at low excursion but unusually low high-order distortion for most all of the excursion range.
One thing that I find interesting about these drivers is that the excursion capability seems to be beyond what would ordinarily be deemed useful in relation to the drivers' frequency range. I use a very formula to estimate what the F3 will be for a sealed speaker using the driver. For the smaller driver, my estimate is about 70 Hz, and for the larger driver, my estimate is about 55 Hz. These F3 values are not particularly low, and as such, it isn't apparent whether the unusually great excursion would matter much. Not to suggest that lower high-order distortion for a moderately large range of excursion isn't something to appreciate, but only that the extreme excursion, that these drivers are evidently capable of, might not be an important advantage in a typical application. One scenario where this should be a real and strong advantage is in a self-amplified speaker that uses EQ to extend the bass response to lower frequency. Nowadays it is fairly common for this to be done with small active monitors. It allows a small speaker to emulate a much bigger speaker, however one price that is paid when this is done is that driver excursion increases such that distortion at low frequency can go through the roof. We've seen this effect with some of the small speakers that Amir and Erin have examined. This suggests that one particular application where these drivers may prove to be highly desirable is in small active monitors that use EQ to extend the low bass. It may turn out that when speakers of this variety are built using one of these drivers, that these speakers will epitomize the very desirable property of the small speaker that sounds like a much bigger speaker. The smaller driver may be particularly desirable for use in small self-powered speakers to sit on a desktop, i.e., high-end computer speakers and the like. And both drivers may be particularly desirable for use in automotive applications, for anyone who has the skill to build the custom adapters required for automotive installation of ordinary drivers not intended for automotive application.