RL of a port of a filter, need a 50ohm load on the other port ?

[jujun]

Hello,

I wonder if when metering the matching with a VNA of a port of a filter, is it best to add a 50 ohm load on the other port of the filter?
I find that would be logical to do so, because I will use a near 50 ohm thing on this port.

Sorry for the stupid question.

Thanks
Julien

[jujun]

So, I quickly tried by myself, and if I put a 50 ohm load, or without a 50 ohm load it's near the same.
And if a put a short on the other input of the filter I have a very different RL curve.

But when I plug nothing it's like I have an open, so why it doesn't change a lot?

[TheUnnamedNewbie]

You always have to measure a filter with all ports terminated in their characteristic impedance. In fact, this is in the definition of s-parameters: \$S_{xy}\$ is the voltage wave out of port x, relative to the voltage wave into port y, with all other ports terminated in their characteristic impedance (note that this does not have to be 50 ohm, but whatever the impedance of the port is).

Try adding a quarter wavelength of coax to the output, and then add a short and a load. If everything is working, you should see that the short suddenly behaves like an open, and the open suddenly behaves like a short. If the length of your output just happens to line up with a stub that matches to your characteristic impedance at your target frequency, it is in essence matched, just not with a 50 ohm resistor.