I have a 4-port s-parameter file from a full-wave simulation of a frequency multiplexer (3 channels) that I'm using in a simulink model. When I plot the response for all the channels from the block 's dialog box, I get precisely the channel responses I expect. When I run the simulink model, however, I see correct responses for two of the three channels.... with the single delinquent channel experiencing an unexpected ~30dB insertion loss.
All the Inport and Outport blocks are set to type Power and with a 50 Ohm source impedance.
I've included a narrow-band modulation spectrum for the three channels as they go into the channelizer and then the (unexpected) insertion loss in the middle channel. I've also included the "wiring diagram" from simulink and the simulink-produce response curves. From these you can see that the channelizer should provide fairly similar magnitude responses in each channel and are well matched at 50 Ohms.
I can re-create this behavior with any combination of channels being driven and even in the case of only the middle (effected) channel being excited.
This channelizer response has been validated (i.e. without the unexpected 30dB insertion loss) in Cadence, ADS, and HFSS. The Touchstone file format is identical for the three channels.