Yeah, it's pretty common in wider directivity designs, but again its usually due to the verticals. If we isolate the horizontals we've got:
Similar story for the polk:
Still not perfect but as I always say, the one big issue with the spinorama is the lack of separation between horizontal and vertical directivity. Most of the time people talk about directivity, they're talking about it in terms of soundstage and imaging quality, and that really seems to be mostly in the horizontals. For example, the Revel M106 doesn't fare much better despite the waveguide:
Of course, you can then get more granular and look at the ER horizontal breakdown or individual angles, but just looking at the overall horizontal seems to coincide with my impressions of soundstage usually.
------------------------------------------------------
One that I want to illustrate about my interpretations of the spinorama is that the total DI curves aren't actually very important for the type of in-depth assessment we do here, except for EQ-ability. It is more important that both the on-axis/LW and ER/PIR be assessed
independently -- not by their relationship to one another. If you look at the R200s LW alone, it is fantastic. If you look at the PIR alone, it is great, certainly comparable to some of the best speakers measured here.
If you read the Olive preference paper part 2, you'll see that the DI curves were found to have little correlation with user preference.
"The two directivity indices generally yield poor correlations regardless of which metric is applied, with the exception of NBD."
Of course, the DI curves are defined by the LW and ER, but all that really tells us is that in
the very best possible speakers, the DI curves would be smooth. But short of complete optimization of the spinorama, which very few non-coaxial speakers actually have, it is better to look at the direct sound and off-axis curves separately and independently.
For example, a speaker can achieve a decent score by having a very good PIR even if it has a mediocre on axis. Likewise, a speaker can have a perfect on-axis and a mediocre PIR.
In the case of the Polk R200, even if the relationship between the on-axis and PIR isn't perfect due to some bunching in the presence region at 5kish, both the on-axis and PIR perform strongly when assessed separately, which is why it scores so highly. And separate from the preference score studies, which
did not look at horizontal data separately from the verticals, we know from other research that the horizontal data is far more important for assessing soundstage performance, and that we want smooth performance here for decent imaging. The R200 performs well here too.
Let's look at it another way: What if the R200's ER/PIR/SP dipped in the presence region instead of rising a little? (And let's assume it was entirely due to the verticals) The DI curves would be prettier, but the score probably wouldn't change much, and the speaker probably wouldn't actually sound any better.
This is why I find it invaluable that spinorama data include a horizontal component or be accompanied by a more detailed horizontal breakdown. Which luckily it usually is in the case of Me, Erin, Amir, and Audioholics.
In the Spinorama's existing form, there is no way for us to know if anomalies in the DI curves are due to the horizontal data or the vertical data, other than making some educated guesses based on the speaker design and the difference between the SPDI and ERDI, as they weigh verticals differently (if the SPDI is worse than the ERDI, it usually means the verticals are to blame). The spin is super useful, but again, it could use the addition of a horizontal ERDI or DI =] if it really wants to be an all-in-one image. Which luckily, Harman has started to include recently, so I hope it is a change considered for CTA-2034B
P.S. This isn't to say vertical data isn't important at all. There's just a lot less conclusive research about it. I suspect that lackluster vertical data contributes to per-room variability a decent amount.