Aha, interesting. I read somewhere that the Nichia would have a 14um ridge, and this would be consistent then. However this was a forum read as well, without source given, so I can't tell how well founded this statement was.
The reason why one sometimes sees dots near threshold may not necessarily be due to several emitters but perhaps just reflects a particular mode pattern, who knows. It appears only for one of two diodes I tested so far.
Of course one can estimate the emitter dimensions by considering the divergences of the axes. The datasheet says (assuming it is the same chip as for the Nichia NDB7352) that the ratio of the axes is like 4:1, which seems a bit low for a 14um ridge. The divergence ratio from the various pictures shown here appears larger, but this is confused by astigmatism - that is, the ratio changes with the distance, so this is pretty meaningless.
I intend to do precise measurements to determine the astigmatism and then see whether the data are consistent with the NDB7352. This is anyway needed for proper collimation. I believe a good way to collimate would be to first have a standard collimator close to the diode, followed by a long-fl cylinder lens to correct the astigmatism, then followed by a prism pair to bring the axis ratio down.
