Synthesis of Symmetrical Branch-Guide Directional Couplers RALPH LEVY(high pass filter)

To be serious, I don't recognize the relation between high-pass and directional coupler. In any case, it would be better to have a filter specification in commonly understood form, order and filter prototype. Getting large impedance values in transmission line filter sounds familiar, however.
Secondly, a qualified reference for the quoted table would be helpful.

FvM said:

To be serious, I don't recognize the relation between high-pass and directional coupler. In any case, it would be better to have a filter specification in commonly understood form, order and filter prototype. Getting large impedance values in transmission line filter sounds familiar, however.
Secondly, a qualified reference for the quoted table would be

Click to expand...

Symmetrical Branch-Guide
Directional Couplers by RALPH LEVY its from here

If your trace tolerances can handle thin lines at 1/8th of 50 Ohms then your calculation is correct. Here I am assuming trace Zo is proportional to trace width (approx). and if using microstrip, you compute Er Effective (air and substrate mix) accurately.

i.e. if your min trace is 2mm ($) then your max. trace width is 16mm. Failing to do that means you must alter your design specs or change technologies to laser lithography.

I just did my 1st shorted stub filter design using Falstad's filter simulator (quick and dirty).

Notice there are an odd # of stubs, one on every node and the middle two conductor paths are identical with symmetry around the middle of the filter for path lengths and Zo where the highest stub impedances are on the ends.

The plot below is 20 dB /decade . 7.88 GHz HPF 0.25 dB ripple

Loss tangents on the dielectric will limit the upper range so a LPF is needed to control it.

hellotoke said:

400 ohms

Click to expand...

Unfortunately, you can't realize 400 Ohm in microstrip.

The approximation that Tony did ("thin lines at 1/8th of 50 Ohms") is not exact.
Example: FR4 with 1.6mm thickness requires a really wide 50 Ohm line width of 2.7mm. If you reduce line width to a very narrow value of 50µm, that results in ~175 Ohm line impedance. You can't realize Zline values much larger than this.

Thanks Volker for the correction. I was thinking Admittance which is linear with trace widths above thickness and quadratic below that and this is scalable by Er and thickness is more accurate.



Same plot in log scale for thickness.



Using Saturn PCB Design calculator which I plotted in a spreadsheet
and is based on "Width and Effective Dielectric Constant Equations for
Design of Microstrip Transmission Lines".
Rogers Corp.

I suggest normalizing using a lower Zo impedance driver (e.g. 22 Ohms ) and a thinner substrate like <= 0.8mm with a layer of stiffener below ground. Then correct impedances after the filter.

This may permit the same stub admittance ratios to be realized due to the thinner substrate effects on Zo taking advantage of the quadratic effects of w/h<1 and realizable impedances easily at 50% of free space.Zo.

Low loss tangent materials are preferred for microwave.
The purpose for of using Er effective, if using microstrip is due to the effects of free space leakage on top layer lowering the Er of the substrate which is also affected by moisture content in air or condensation of water on surface if exposed in harsh cold environments.