is it necessary to use 50om-port when designing a filter?

[hakunamatata]

hello!everyone !i am desiging a harpin microstrip filter with SONNET now!
generally the port of the filter must be 50om but i have designed a 8poles harpin filter with very good performance using the ports which is about 20om,that means the port is wider than 50om-port!all the work i have done is completed by SONNET,i want to know whether it is right ?
is it necessary to use 50om-port when designing a microstrip filter?
can the filter i designed in the SONNET really behave a good performance in the practice?
thanks in advace!!!!!!

 

[Gemini1706]

if the 20ohm filter will be connected to 50ohm system (which is must probably will be), then:
1-mismatch loss will happen, and the loss of the system will appear very high, which could be unacceptable (check your specs, and overall system gain/performance requirements).
2-The filter will not behave as you see it in your simulation, becasue it is not presented with its optimum impedance (20ohms), but with a different impedance (50ohm, from the system)..so it will change its response...

In short: check what system will be used (usually, 99% is 50ohms), then: NO, you will not get what you see in simulations.

If you design the filter with 50ohms, then simulation vs real perofrmance is very comparable. You may need to tweek some stuff after measurements if the filter in real-world is off a little bit from simulation...It also depends on your frequency..

Cheers...

 

[BigBoss]

In your simulation, if you use 20 Ohm instead of 50 Ohm , you may not compare simulation result and measurement result due to characteristic impedance of the measurement sytems.
To compare measurement results and simulations , it should be terminated neccesserily 50 Ohm. If you see the real behaviour of your filter , putting measured or simulated S-parameters is sufficient in a linear simulator's "black-box" with proper termination ( e.g. 20 Ohm).

Because I have tried in a project that has complex input and output impedance . I have simulated in EM with 50 Ohm and compared with measurment results and found good match between them. And then I have put these results in a black box in a linear simulator and obtained real result. It has meant that my circuit works well..

Rgrds

 

[mwpro]

if you are designing something that is going to be implement in a system, conforms to he system requirement of input and output impedance. otherwise, the behaviour may not be the same as simulated.

if this filter is just an academic excercise, it is fine. even though the equipment (network analyzer) are 50ohm, you can always renormalize to 20 ohm.