The gory details of the low frequency ability variability of the RS28A.
First, these tests measure the extreme bottom end, and, in a sense, are not entirely fair. Most tweeters can't handle 1k at all. Also, this tweeter has other redeeming features.
How to look at these graphs. First, the thing to look at is the higher order distortion products. At -60 and above, they are pretty audible to the naked ear. Really, you'd like a little margin, 10-20dB. . The first production pair (graph below, red and aqua) had one reasonably good one (aqua) and one not very good one (red)
The prototype (graph above, black) and the second production pair (graph below) do very well
Even the "bad" unit at 1k in the first graph above, does reasonably well at 1.5k, as shown below.
However, Zaph's set is more problematic. One is flat out unuseable(the red graph below). The other (black graph below) is probably is a bit better than the "bad unit" above.
Really, one of the units is unuseable.
So, of the 7 I've tested, 3 are A+ and useable very low. 2 are useable above ~1.6-2k. 1 is defective. I haven't had a chance to fully test the other unit from Zaph, so I'll leave that as a ?
I still think these are one of the best tweeters around. The distortion at higher frequencies is extraordinary for all the units. (Though I didn't test Zaph's higher up, I have no reason to think that at 3k, 5k and above they do well. (except maybe the single abject failure...)
As I mentioned though, Darren is aware of this and they are checking each individually, so I think that we won't be seeing any rejects, and, at the states xover frequency of 1.6k on the PE webpage, these will be extraordinary. If you are trying for a crossover point below 1.5k, you may want to individually test these.