Please note that this is my first fully RESISTORLESS crossover. I am very proud of it.

Some notes:
L3= as thick as possible and AIR type only. I use 1,8 mm motor wire.

C2 is in my case Hovland, but any good polypropylene cap for 100 V DC will do here, like North Creek Crescendo for example.
L2 is a motor wire 1 mm air coil and any good quality coil will do here. Medium importance of quality.
L1 - quality quite unimportant. I use 14 AWG Goertz foil air core inductor.

C1 is most critical. Nothing less than Hovland will do. Actually North Creek is quite near. Audio Note Copper PIO would be just perfect !!!
 

How it was done:

What are the crossover points?  Are you trying to understand the circuit ?

Well, it is not so easy to explain. First, my method does not apply to every construction.
You must first choose a FIRST CLASS midrange that deserves to be used. I know maybe 10 such units. There must be more out there.

1. I started with playing the midrange wide open straight from the amp.
The sound was really fabulous. Of course I used vocal material without any bass contents.
I must be first of all VERY HAPPY about the character of female vocals.
If it is bad, any other work is a waste of time and XO will not help it.

2. I protected the bottom end of the mid by applying various caps until it
could play clean a 100 W material. It was a 12 uF cap in series. The sound was OK, but midrange impact
transients were a little distorted (the mid has a 80 Ohm resonance at 300 Hz).

3. So in went the resonance trap - the RLC tuned for 300 Hz. Then I removed the C from that RLC
as it was unnecessary. Then off went the R. Just the coil is enough. So we are back to simple 2nd order hipass.
The remaining L is what you see.
No lopass at all!
The cap in second order XO was increased to 30 uF.

4. Then I left the top end of the ATC wide open and matched the tweeter to it:
1 uF cap was good. Sounds great. Everything AFTER listening was carefully controlled by at least three
methods of measurements and testing.

5. I filtered the bass to gently fill in the bottom. That is easy: Just play
with the L first order until it matches. Crossover point (break even) is about 500 Hz. The bass driver
plays absolutely clean until 1,5 k (!)

6. Listening to 50 albums

7. Again, VERY loud material reveals that the Raven is sweating a little at say
4 K. So I increased filtering by adding a very small coil in parallel. Increase cap to compensate
to 2,2 uF.
XO is now around 6k.

8. EVERYTHING is perfect. Clean clean clean, very open and free sound, filling
the entire room. No hardness up to 100 W of power.
What is most important, each driver is happy doing what is doing. It is not
forced to do what it cant do well, even if ON PAPER it should.

9. Keep it like that and forget.

Computers were useless this time. Book formulas were useless. Just understand R, L and C
impedance and Ohms law, and tune tune tune.

Ah! And you can admire the final XO in action: