How can you be sure you're hearing the fundamental though ? I'm sure you're hearing a
noise, but I'm rather dubious about it being anything other than harmonics. Even a good set of in-ear earphones are likely to be over 100% distortion at 17Hz at the necessary volume levels. (Eg the harmonics are higher in amplitude than the fundamental) By turning the volume up you've just turned them into a more efficient distortion generator.
There have been studies showing that below 20Hz or so not only do you not perceive the sound as a discernible tone, the bulk of your perception of those low frequencies actually comes from the pressure wave physically vibrating your body, rather than movement of the eardrum.
Given that in-ear earphones will only stimulate the eardrum and <20Hz doesn't sound like a tone, if you're hearing a tone when you drive it with 17Hz then you've got to be listening to the harmonics...
Not sure if that's another thread or the one I was participating in a while ago (it wasn't in the subwoofer section) where I posted some Fs versus drive graphs for two different drivers showing that Fs does indeed (contrary to popular expectation) drop with increasing excursion, dramatically so on some drives, less so on others. (I've re-posted those graphs here)
One is a typical 12" woofer with butyl rubber surround where the small signal Fs was 29Hz, dropping to around 24Hz at higher excursions, the other was an 8" full range driver with cloth surround whose Fs is 48Hz at very low levels and drops all the way to 34.5Hz at high levels.
While the cloth surround seems more excursion sensitive the butyl rubber surround also has a memory effect due (presumably) to the plasticity of the rubber - after large excursions the Fs would remain at 24Hz even for small excursions for a period of several seconds - over about 10 seconds the Fs would "creep" back up to it's no signal value of 29Hz.
(This would mean bass with high crest factors such as a big kick drum every couple of seconds would tend to keep the woofer's surround more compliant even when average levels were low, at least with a rubber surround)
So there is definitely a volume level dependent change in a woofers characteristics, but I think it's a bit oversimplifying things to say that more drive (below Xmax) = softer suspension = more bass.
What happens to the bass response when the surround becomes dynamically softer from excursion depends on the bass alignment. In some alignments like a closed box with a high tuning ratio (large Vas, small Vb
) even a large change in the free air Fs will have negligible change on the in-box response, since the Cms wasn't much of a determining factor in the first place.
Open baffle and bass reflex are more likely to see a significant change, and in the case of a bass reflex although softer compliance generally means more lower end bass, with many alignments it will reduce the mid bass, since the Qms will drop when Cms drops.
The way I've approached the issue is to measure the parameters for the driver at both very low ("small signal") levels, and also near but below Xmax, (typically 1/2 Xmax) save the two sets of values as two different drivers and model both responses in the same size box / port tuning to see the difference in response - now you know what the response would be at low and high levels due to changes in Cms, and you know that intermediate levels will lie somewhere in between...
I would just add this to the long list of reasons why a speakers frequency response does change somewhat between low and high volumes...