When I got these speaker boxes and was gifted with the woofers to put in them, I was excited to finally hear a design vetted by the forum. I was not expecting to be so critical once I heard them. If the original design sounded great, then the TC9FD is probably not like the 10f at all.
In my case the treble was way too soft, the midbass was absent and the mids were overblown and marble-mouthed. The reason is kind of convoluted. The TC9FD has less sensitivity than the 10f but has a descending response from the bass whereas the 10f has a well damped rising bass response. In the case of the TC9FD this means the drivers cancel in the midbass and we lose it. When we raise the level of the tweeter to bring the highs to the level of the 10f, the cancellation gets worse. Both these issues together mean that ultimately you end up with boosted mids around 1KHz if you try and reach a compromise. This is also where the natural baffle step hump is... So the crossover really needed redone.
I suspect that the original suffers from this issue just to a lesser degree. I'm also extremely curious whether my crossover would change if I had the baffle with the offset tweeter...
I must have done hundreds of measurements, maybe even thousands by now. And in between those bursts of measurements, I was listening. It's not as if I am just trying to create a flat response or get a perfect impulse response. That is a good target or starting point, but despite my best efforts I have to allow some departures from what is hypothetically accurate, in order to make a sound that me and my weary test subjects (family) actually want to listen to, and resembles what instruments might actually sound like.
In this case I think it may have a lot to do with the ringing of the tweeter at 3.5KHz in combination with the diffraction from 3 baffle edges at the same distance. With my current preferred crossover this region ended up suppressed, which suggests that unevenly dispersive ringing is far worse than a deviation from a straight line in the step response.
As much as I like the idea of a good step response and time alignment, I only view that as a starting point. If the result doesn't sound good, adjustments have to be made. To come up with a good step response chart and then say I was finished would be ignoring a great deal of my own experience listening to my modifications.
But ultimately I don't have a bucket full of inductors to test with, and my design may be as much a result of my flawed listening environment as it is genuine problem solving. It would be better at this point for someone to replicate my work and add their observations, because it's not obvious whether I'm improving it at this point or just changing the flavor. And actually I need a long break. I've actually spent many, many hours on this. Here is where I'm at:
All the extra tweeter components were unnecessary as far as I can tell, and don't seem to be needed for the TC9FD. The treble was rolled off a bit in the 10f crossover which was not an advantage here where the diffraction means we need to be off-axis and still have good treble response.
Changing the value of R5 has a very interesting effect on the sound...
I don't know what this crossover would do for the 10f, but I would use the woofer portion of this crossover with either tweeter. I think the step response could likely be improved. The step response of this crossover is pretty good even though it's not the best I achieved, but it sounds so much better than the version with the best step response.
The imaging is unreal at times. All of the music sounds better, it is smooth yet detailed, all the instruments can be heard when they were intended to be heard and not climbing over each other. The imaging is incredibly solid, not exaggerated at all, just very life-like, as if something is really there. Especially with nature sounds.