Shields are your friend, except when...
Engineers just love to put shields on circuits, mostly as a defensive measure against signals on the outside getting into and disturbing our circuits, but they also keep signals inside from getting out and this really makes the folks responsible for EMI compliance happy.
Even on low-frequency circuits, shielding can take care of drift due to air currents and AC power mains pickup. At higher frequencies shielding can take care of emissions and pickup. In RF or microwave circuits shielding makes possible radio receivers, spectrum analyzers and all sorts of equipment that simply would not function without sufficient shielding and the isolation that it provides between the various RF and IF sections.
So what could go wrong? Well 10 years ago, not much. We were still working at frequencies up to 2.5 GHz for almost all the standard wireless applications. And shields generally only helped the situation. Then we considered “commonplace RF” to be anything below 3 GHz, but today we consider “commonplace” to be below 6 GHz, and we may routinely need to lay out circuits that operate to 6 GHz even in consumer circuits .
This situation demands shielding our RF circuits, not only to keep our circuits working properly, but to prevent EMI emissions failures during regulatory testing.
The usual form of shielding is some sort of conductive structure placed over our circuit. These can range from TV tuner types of cans, simple plastic moldings that have been coated with conductive material, or expensive machined chunks of aluminum, purpose made just for our circuit board.
When we place a shield on a circuit board we are creating a conductive electrical cavity and this cavity likes certain frequencies or various TEM (Transverse ElectroMagnetic) modes . In other words it becomes resonant at certain frequencies.
These resonance modes are usually bad, because we wanted our shield to stop electrical fields from escaping (and that it does), but at certain frequencies, inside the shield, the electromagnetic waves get quite vigorous at certain frequencies. This is where our problems can start.
It is possible to predict the frequencies where this starts to happen by using simple math .
For a box of dimensions H, W and L (for Height, Width and Length), and where W > L > H), the various TEM modes can be calculated by the equation:
where the dimensions are in meters,
the calculated Resonance Frequency is in MHz
and the medium inside the shield or cavity is assumed to be air.
In shielding design we are typically only interested in the lowest frequency that the shield cavity can resonate at, so Equation 1 can be simplified to,
Equation 2 represents TEM101 Mode. The 'a' and 'c' of equation 1 are made equal to 1 and 'b' is 0. As you can intuitively see, this is the diagonal of the shield's largest dimensions. And the inverse of this diagonal length would be proportional to frequency.
Equation 2 then can be used to calculate the lowest frequency at which a rectangular shield will resonate.
Testing Theory and Practice
We now have a way to see if our shield will resonate at some low-enough frequency that we care about; let's do some real testing. I routinely use these little box shields that can be soldered to a PCB . A small shield that I regularly use measures 1.5 x 1 x 0.4 inches. Using these dimensions with Equation 2 we find that the lowest-frequency TEM mode is at about 7 GHz.
To test this out, I built a test circuit that consists of an input probe, an output probe and a shield soldered to a piece of scrap circuit board (Figure 1).
Figure 1: To test this theory out, I soldered a 1.5 x 1 x 0.4-inch metal shield on a piece of circuit board material. To get signals in and out I also soldered some modified SMA PCB connectors to the inside of the shield cavity. These act as small capacitive probes and allow us to take a quantitative look inside the shield when connected to a Vector Network Analyzer (VNA).
No particular attention was given to the length of the center conductor probe, other than to make them short electrically (<< ¼ wavelength at the frequencies of interest). They were mounted just at the ends of the box somewhat randomly. The ends of the box were chosen for the probe locations as this perhaps represents the best hoped for isolation inside of a shield, the ends possibly representing the input and output points of a particular circuit, etc.