Filter PCB Track Impedance

[HardwareChap]

I realise that for maximum power transfer, the input and output impedance of filters should match the impedance of whatever they're connected to (in this case 50 Ω in and out). My question is, should the PCB tracks between the Ls and Cs, that make up a complete filter, use track widths that equate to 50 Ω? My thought is that if you follow the filter round a Smith Chart, it will deviate from 50 Ω, but work its way back there in the end. Is the deviation from 50 Ω negligible, so the rule of thumb is to control track impedance and set it at 50 Ω?

 

[FvM]

the input and output impedance of filters should match the impedance of whatever they're connected to (in this case 50 Ω in and out)

It's not the actual filter in- and output impedance but its designed or nominal impedance.

The design equations of a lumped filter don't assume transmission line segments inside the filter. But in a real filter, there's nonzero distance between LC elements. The best fit is to estimate the unavoidable PCB trace segments inside the filter as series inductance and parallel capacitance and put it into the filter calculation. Choosing the track impedance around the nominal filter impedance isn't a bad choice, making the TL segments as short as possible is better.