ok, I did some more simulations with WinISD...
When I let the bottom two subs move 'up-down'
and the top two, move 'down-up', working against each other,
then I'm convinced I can divide the middle chamber by two, because each clamshell needs the same space.
So...
Then I get two different 4th order band pass enclosures to work with in WinISD.
The bottom one gets a peak at about 78.5hz.
The top one peaks at ±45hz.
If I look at the graphical output and imagine the two plots combined, I get a reasonable flat response.
I might just try and draw over it in a graphic program if anyone cares.
The only thing I'm not so sure about is, how to combine them correct, example:
how much is 92dB + 92dB?
is it 95dB?(+3dB) or 98dB?(+6dB) or...?
and what about 97dB + 86dB? ::)
I also did a 6th order simulation and didn't take the middle chamber in to account, because that works as a spring, flattening the peaks en dips. Now the two clamshells work the same direction.
I'm sure it has more influence than only a spring but I can't use more then what I know
Even with this simulation, it doesn't look that bad either.
B.