good to know. I went back and measured the model so in the case of the f20, plan/drawing => hornresp => measurement all very well.
I've monkeyed around with the various t/s specs to try to recreate the effect observed and I can't.
I've monkeyed around with the various par/exp/con settings and they are confirmed not to be the problem.
The actual build was done quite well. I followed along pretty closely and have pretty high confidence that it was built according to the plan.
Another other thing that I can think of is how I measured around the bends/turns. I simply measure using straight lines through the center of the horn. Maybe that overshoots the actual result by a bit and more in the case where there are larger bends toward the mouth. I don't think that would result in a 50% difference though, maybe 10%.
Last, as has been discussed, may be that below a certain tuning point, the size of the horn (specifically the mouth) must increase at a certain rate or something comes undone. This can be seen in the acoustic impedance increasing as tuning is lowered, but then dropping rapidly almost as though it goes off some sort of cliff. I thought it looked fine on the Submaximus, but now I see that a small increase results in significant drop.
BTW, the danley dts-10, at least as I modeled it in hornresp some time ago, appeared to have a corner target around 15hz, but the actual build came in lower, something like 11hz, and it is another longer narrow low tuned horn. honestly, I had forgotten about that when drawing up maximus.
maybe it matters where the bends occur too. perhaps bends closer to the throat don't have as much of an effect as bends closer to the mouth.