Click here to go to our main page on printed circuit boards
Click here to go to our main page on hybrid microcircuits
What do you do when you are setting out to design a printed circuit board or a hybrid microcircuit, and your design includes S-parameter blocks which have no shape associated with them? Eventually you will have to insert all details of the packages that house the components, but on Day One of design you don't need that level of detail, you just want a 2D sketch of your circuit. Design is a journey, don't get ahead of yourself!
Below is a schematic of a piece of a notional feed network, which includes a digital attenuator, an amplifier, a simple high-pass filter, and a two-way Wilkinson power divider. Sporty!
In Microwave Office, you can right-click a schematic and generate its layout. When you do this to a circuit board that has S-parameter blocks, MWO has no information on how to lay out those parts, and the result is a disjointed layout of only the physical pieces.
Below, we introduce the "Layout Thing". It is a two-port schematic of a physical microstrip thin-film resistor, where the resistivity (1 billion ohm-centimeters) is so high it looks like an open circuit at both ports. Notice that the length and width are parameters that pass upward in a schematic (this is done by using "<<" instead of "=" in their equations. Do we even care about the parameters of the microstrip substrate? Probably not, just leave the defaults which are conducive for a GaAs MMIC.
You can strap the Layout Thing across the S-parameter blocks, and specify the lengths and widths of their packages. The attenuator is a 4x4mm land-grid array, and the amplifier is a 3x3mm QFN.
While we are at it, that high-pass filter could use some via catch pads on the shorted stubs. We will throw them into that sub-circuit model as well.
Now let's admire the improved layout. Is it 100% accurate? Ha! But is it useful for a design study? Yes!
One last "thing"... does the Layout Thing actually have no effect on the model? As they say in Missouri, "show me." Let's compare predicted results with and without the Layout Thing. Here is the result without:
And here is the result with the Layout Thing:
We dropped a couple of markers on the charts, which reveal that the Layout Thing actually affected the reflection coefficient to the tune of a few hundreds of a dB. Oh, know, whatever will we do? We could increase the resitivity to a trilllion ohm-centimeters, reduce the microstrip dielectric from 12.9 to 1 and make the substrate a lot taller, and run the simulation again, or just make a mental note to tweak it next time and move on. Let's move on.