Rat-race coupler power dividing with open stub instead of 50ohm load resistor.

Hello all! Recently I have seen few designs where rat-race power divider is used without 50-ohm termination. Instead there was a tiny open stub, around lambda/8 or less. I tried to simulate such coupler with open stub length varying from 0 to lambda/2, but no any practical usable performance was observed.

Rat-race coupler was the same as:
https://www.microwaves101.com/encycl...-race-couplers
Rat-race coupler (equal power split), port 3 is terminated with resistor (50 Ohm to ground or RF quarterwave-stub ground). With resistor, even mismatched value of 20 Ohm provided acceptable performance.

What is the purpose of tiny stub instead of resistor? Why i did not get any acceptable performance when trying to analyze similar design?

The effect of not terminating the isolation port can be well seen at the rat-race coupler scattering matrix or the "Arithmetics with rat-race coupler" diagram. Unpaired signals feed back to the output ports won't be absorbed any more but reflected back to all ports, similar to a simple non-absorbing power splitter. https://en.wikipedia.org/wiki/Rat-race_coupler

For certain unwanted signals to be absorbed, you need an absorptive port, quite simple.

http://scholarcommons.usf.edu/cgi/vi...21&context=etd
page 102 and 52-53
Also few other sources I need to search for, at least 5 papers/photos of microwave designs, where port is not terminated, but is "o.c." stub (open circuited stub).

I have used unterminated hybrid splitters to boost the gain of video sync tips in the old days of suppressed sync pay TV since the splitter loss changes from 3.5dB to 0.5dB with a switched open 75 Ohm load.

There's no theoretical advantage of a rat-race coupler in 180 degree configuration with unterminated isolation port over a simple λ/4 power splitter + λ/2 delay line. It's just a different geometry achieving a similar function. Off-band behavior is different but surely not better than the latter.

Do you see any reference that states an advantage over other non-absorptive power splitters?



You can try this geometry in comparison

I did not found any references, but it seems that authors used simulation software and obtained acceptable results. I did not obtained any results indicating usable design, without 50 Ohm termination performance is below poor.

Doesn't make sense not using a terminated port in a coupler or a splitter.
For example in a Wilkinson splitter if you don't terminate one of the divided ports in 50 ohms, the insertion loss on the other port drops from about 3.3dB to about 0.3dB.
But the isolation of the splitter also decrease from about 35dB to less than 6dB.