power divider reverse engineering strategy

[yefj]

Hello , I need to recreate the following dividing network .
I only have photo.
the branches are angled and the transformers are triangles ,why is that?.
could you share some impedance logic that could fit the structure below?
Is there some theory regarding this structure?
Thanks.

 

[FvM]

Input/output impedance? Bandwidth?

 

[yefj]

input 50ohm output 140, badwidth non matter(narrow)

 

[vfone]

Tapering the branch lines of a divider improves the VSWR at the ports when they have different impedance, but isolation between divided ports remain poor (about 6dB). Compared to a similar Wilkinson divider, which have about 30dB isolation between the divided ports.
But, compared to a resistive divider which have also 6dB isolation between ports, the tapered divider have about 3dB insertion loss, when the resistive divider have 6dB insertion loss. If you are not interested in a better isolation between ports, tapered divider may be a good option.

 

[yefj]

why are there angles shape like a "sling shot"

Ill do 140->70 inparralel to 70 is 35 ,then i put transormer back to 70 and have this trick on every stage going from 70 to 35?
is its ok?
I need to keep my matching.
Thanks.