First, we are probably going to start over from scratch. Here’s a few things to consider when playing around in WinISD. First I would remove everything and take a look a the individual graphs by themselves.
Currently it looks like the tweet could crossover somewhere around 1500hz. But we haven’t taken into account the FS of the tweeter. The general rule of thumb is crossover with a second order at 2x the FS. The FS of the ND20FB-4 is 2072hz. SO a good starting crossover is around 4200hz *minimum). Now we want to look at the off axis of the DSA90-8 to make sure it can maintain directivity that high.
It is starting to lose some directivity in that region, but overall isn’t that bad. However, if you wanted to crossover even lower you could, but might want to use a third order on the tweeter.
One other thing is to guesstimate the z-offset of these driver. With something like this, I would assume the offset to be pretty small, maybe 0.5in. Although, this is entirely a guess. We will add this to the woofer as a positive number). Right click on the woofer and select tune.
Once this is done, I would guesstimate the width of your cabinet. So for a 3″ I would guess no larger than 5″. Go to this site to figure out the baffle step transition frequency based on the width of that baffle. You will need to gather some of the data from the DSA90-8.
Once you have this figure out, you can see the data. If you want to understand more of what this is doing, add the circuit to your diagram. Keep in mind, this is just to visualize what your graph should approximately look like up to the crossover point.
Now go ahead and open a new xsim file and try to mimic the slope as much as possible until you get to the crossover point. This should give you a model that better represents the baffle step loss, giving you a more even speaker in the end.