I just thought I'd use some sims to explain a bit better what I mean here.
All the following is based on a 5.5" SB Acoustics SBNRXC30-4 woofer in a 6-litre ported box with a tuning frequency of 55Hz. This is an affordable, well-performing small midwoofer designed to be used in compact standmount speakers.
Shown is the frequency response of the woofer (black) and port (grey):
And now the summed response (black) and port (grey):
As you can see, the summed response actually rolls off earlier than the port does. This is because, below the port tuning frequency, the output of the port is out of phase with the output from the woofer. There is absolutely no escaping this effect in a ported speaker.
You can also see that the -3dB point is at about 58Hz, which is just a shade above the port tuning frequency.
Next, here is the diaphragm displacement vs frequency at this SPL (96dB). This woofer has an Xmax of 5mm, so at 96dB, we are just a shade below Xmax at all frequencies from the port tuning frequency up.
However, although the woofer does not displace at all at the port tuning frequency, is displaces
hugely below it. This is inevitably the case with any passive ported speaker, and is the main argument against them IMHO. Note, however, that displacement at the port tuning frequency and above is actually significantly less than it would be if the box were sealed - but that's another story.
Now this huge displacement below port tuning is definitely a problem. Fortunately, we can tame it quite easily with DSP, firstly by implementing a high pass filter just below the port tuning frequency. In red we see an example effect on our amplitude response of a 4th order high pass filter at 37Hz. Our -3dB has changed very little, but there is now far less stress on the woofer (an excellent thing of course).
And here is this filter's effect on diaphragm displacement (as you can see, the woofer will now not extend past Xmax at any frequency at our orginal SPL of 96dB).
So we've now used a high-pass filter to hugely reduce diaphragm displacement and resultant distortion, with almost no cost. That's the first and greatest advantage of using DSP to manage the woofer in a ported speaker.
But what if we want to actually lower the -3dB below what it was when the speaker was passive?
We could move our DSP high pass filter lower in frequency and then use some EQ to boost the frequencies between the high pass filter and the port tuning frequency to improve the bass extension. Perhaps we want to aim for a -3dB that is 18Hz lower in frequency, at about 40Hz (this is less of an extension than Devialet claims on the similarly-sized LS50, FWIW):
Here's our resultant SPL:
That's great at first glance. But now look at the diaphragm displacement (which FWIW is basically a direct indicator of nonlinear distortion):
That type of filter/EQ - even after high-pass filtering - results in a diaphragm displacement of 25mm in the region between the high pass filter and the port tuning frequency (i.e. where we've had to apply the EQ boost). 25mm is 5x this woofer's Xmax. Obviously this is an impossible amount of stress on the woofer.
Our only option if we want to keep the -3dB at 40Hz and not massively increase distortion and almost certainly permanently destroy the woofer, then, is to reduce SPL until the driver stays within Xmax.
Reducing the voltage to the driver by a factor of about 5 or 6 now gives us this diaphragm displacement:
That translates into an SPL around 15dB lower than we had with a -3dB point of 58Hz:
In other words, we've gained 18Hz of bass extension at the expense of 15dB of SPL. Or, alternatively,
our speaker will now produce the same amount of bass distortion at 81dB as it did previously at 96dB.
We're not finished though. Sophisticated DSP can raise and lower the bass extension (i.e. change the corner frequency of the high pass filter) depending on the voltage of the input signal. This is (I suspect) exactly what the Devialet does. But it can't do much, if anything, to reduce the massively increased distortion
at any given SPL that results from pushing the -3dB point below the port tuning frequency.
To illustrate, here is the driver displacement at 81dB
before we lowered the -3dB point with a lower filter and bass boost:
That's 5 or 6 times less displacement, which will translate into 5 or 6 times lower distortion. Such is the cost of trying to make a ported speaker produce sound below the port tuning frequency.
In other words, if this is going to be attempted, it will have to be done extremely conservatively and carefully to avoid increasing the distortion the speaker produces by 5-fold or more.
So, when I see Devialet claim that their SAM system can push the bass extension of an LS50 (which is also a small ported speaker with IIRC a 5.5" woofer) down from 44Hz to 28Hz, I do believe them, but at the same time, I can be certain that the price paid for this is a
massive increase in distortion at any given SPL, plus either (a) a huge reduction in max SPL or (b) a -3dB point that rises rapidly as SPLs reach even moderate levels.
If we choose (b), at low levels the speaker will play lower (but with higher distortion), while at high levels the speaker's -3dB will return to what it would have been with our original high-pass filter just below the port tuning frequency.
Personally, if Devialet's statements about e.g. the LS50 are true, I think they have either gone too far for the sake of specsmanship, or (more likely I hope) they are only true at
very low SPLs.
(IIRC, Devialet also implements limiting to ensure the driver never exceeds Xmax. This and a voltage-dependant variable high pass filter can basically be used complementarily to achieve the same purpose of protecting the woofer at moderate-high SPLs while letting it play lower at lower SPLs. The downside of limiting is of course that it is a form of compression and its implementation will therefore reduce dynamic range.)
In all cases, of course, we have a speaker that performs better than it would have if it were completely passive. But the ported design places a pretty hard limit on just how much extra performance can be squeezed out of it.
My issue is with the examples you've given and what you attribute to them (e.g. that the same drivers in the same box with the same port - but with an active crossover - can achieve a full octave of extra bass extension, which is frankly impossible).
