Shields are your friend, except when... (Part 3)
Real Measured Data
To verify that this simulation is in the ballpark of acceptability, I used a 20-GHz network analyzer with the ability to do Time Domain Reflectometry (TDR) analysis to measure a 2-inch test sample of a properly designed CPWG trace on a 62-mil substrate. Figure 7 shows the results of the measurement with no shield top cover and with a top cover in place .
Figure 7: A 20-GHz Equivalent Bandwidth TDR Test was performed on a 2-inch-length CPWG transmission line built on a 62-mil substrate. The Blue trace is before and the Orange trace is the same sample with a 0.1-inch-tall top cover shield in place over the transmission line. As predicted the transmission line's impedance is lowered because of the added capacitance of the top cover. The "bump" from time 60 to 80 is the effect of the Edge Launch connector; first it "looks" inductive, then capacitive (bump, then dip). The flatter line portion (from time 80 on) is the CPWG transmission line by itself.
The difference between having a top cover and no top cover is as predicted. With a top cover shield over the CPWG trace the TDR plot gets lower verifying that the top cover added capacitance to the structure, hence lowering its impedance.
The top cover shield was 0.1-inch tall and the delta TDR reading as read from Figure 7 at the 110 time point is about 3 mV. These TDR plots are shown in millivolts on the old analyzers and a conversion factor must be used to convert the mV to Ohms. After carefully calibrating my network analyzer I have found that the conversion factor is 125 Ohms/Volt. So the change in Figure 6 corresponds to 0.003 * 125 = 0.375 Ohms.
Comparing this measurement to the data for Figure 4 finds that the measured results compare favorably to the simulation, i.e., not much change at all. The simulated data was 49.5998 Ohms and the actual measured was 49.625 Ohms.
I also TDR tested a similar section of Microstrip, built on the same thickness substrate, etc. Using the same shield top cover I found that the transmission line delta was 6.75 Ohms. Comparing this to the data from figure 4 yields the following:
Data for Figure 4 with 100 mil top cover = 41.82 Ohms
Actual Microstrip Data Measured = 43.25 Ohms
The delta from simulated to measured was just over 3% for the microstrip confirming that simulation is in the ballpark of actual measurements.
The Take Away
This is something to keep in mind for that new "Thin" design on the drawing board. While we can't change the underlying physics of wave propagation, with improved understanding and perhaps using a new transmission line design method we can be successful.
The next issue that you will then be facing is how that shield interacts with the components, especially unshielded inductors. But, that is a topic for another time.
Remember, as designers, we have to be careful when we scale our designs, because that's where we get in trouble – by scaling too much without understanding the limits of the underlying design interactions.
Shields are your friend, except when...
Shields are your friend, except when... (Part 2)
No Article Found