Hello Heinz and everyone!
Heinz, it's a pleasure to read you here!
Here are my measurements of SR1a distortion, 2nd and 3rd harmonics at 100dB and 90dB.
You can see markers with frequency and distortion in % of 2nd and 3rd harmonic.
Measurements are done in steady-state, after the settlement of the diaphragm.
Sure, it measures quite high, but it's not a big deal audibly and I'll explain why.
This is not unlike what speakers would do at high SPL, but speaker's distortion doesn't rise so monotonically as SR1a's distortion and that is very important for distortion audibility. Speakers "clip" much harder, so to speak, but still way softer than amplifiers.
For example, imagine these numbers in an amplifier. It would be terrible to listen to, only because it would do that too abruptly to stay unnoticed.
The thing here is that I have used a progressive compliance to hang the ribbon. The more excursion the signal demands, the tighter the suspension will get, as there was no other way to control the excessive excursion and achieve high enough SPL. Basically, there is a change in resonant frequency as SPL is increased.
The key to understand what's happening here is to watch the resonant frequency change and remember that below that frequency, the output drops at 12dB/oct slope.
You can see the sharp dip in dist.measurements at about 28Hz at 90dB (the resonant frequency at 90dB) and at 100dB, you will see a less sharp dip (the Q-factor got reduced), showing that the resonant frequency has risen to 41-42Hz. That's what a progressive suspension does.
Above the resonant frequency, the distortion peaks at ~9% at 100dB and ~7.5% at 90dB and that is not bad at all for a completely open design.
Below the resonant frequency, the output drops at 12dB/oct., so it's normal to measure higher distortion % at 20Hz because the output has dropped (because the resonant frequency has risen), but audibly, it doesn't attract much attention to itself, as the total distortion output remains low.
The main objective here was to have the distortion to smoothly increase with SPL and smoothly decrease as we go higher in frequency, as in my experience with speakers, those are the things that make it inobtrusive, if not downright inaudible in bass.
In mid-high range, things are much more demanding and while the same rules about smooth distortion progression apply, the allowable numbers are much, much lower and the graphs reflect that this was the goal.
All in all, what counts here is the distortion above 40Hz, where we have a good output at high SPL, not what happens below the -3 point of the frequency response. The numbers there are very much speaker-like, so they are nothing out of the ordinary. The lack of artifacts that are unrelated to music is very important.
Also, a very important thing about distortion audibility in transducers is a very clean impulse response.
No generated harmonics or any other artifacts should last longer than the signal that caused them. There shouldn't be distortion harmonics that can "ride" on top of ridges in spectral decay, dying out much later than the signal...due to time domain issues, same distortion percentage is not comparable between different transducers, as in some, the harmonics last longer than the cause, but that sort of discussion is for another time...
Cheers,
Alex