Here is another graph,
Some explanation is in order.
Over the past few weeks I have been evaluating the 4344mk11 network to assess if its horn and slot passband filters would work for the 4345, saves the split chokes.
All the above graphs used the original 3145 mid passband filters, but the 4344mk11 horn and slot passband filter topologies.
There are some differences but its works well enough. (namely in the UHF passband.) So if you are stuck for the split chokes you can use the values shown in the 4344mk11 network for the horn and slot including the fixed resistor Pads.
However, tonight I returned to the original 3145 network (4345 original network) and experiemented with chokes of various Dcr. The design is reasonably tolerant of choke Dcr, but try and keep it as close to spec as possible, (JBL works to 5% tolerance).
The graph below represents what I consider the state of art crossover for this system (3145 with Giskard/GT alternative split choke values).
The response is on axis at 2 metres indoors and is unsmoothed, not even 1/3 octave smoothing. The divisions are 6 db, the scale is linear and not log so there is more detail at high frequencies and less at low. (The JBL graphs have a log scale).
In comparison to GT's test data, under the microscope the similarity is remarkable and the transitions are perfect.
Adjustment of the pads can fine tune the balance overall and that is up to the user.
Ian