The success of the multipoint ground (MPG) system requires having an extremely-very-low RF impedance grounding connection to an extremely-very-low RF impedance equipotential reference plane. It also helps if the conducting materials are homogeneous and the bad things are kept separate from the good ones. Does that sound easy? You’re right – it’s not!
A couple of posts ago, I noted that short-wide-flat conductors have lower impedance than wires (even with the same resistance) because they have lower inductance, not because they have no inductance. Plus, I have stressed that inductance (L) is directly related to length (L). Still, by extension, a large highly conductive ground plane represents the lowest impedance conductor arrangement. Thickness does affect resistance and inductance somewhat, but it is not a major factor at high RF frequencies because of skin effect.
That being said, when the systems speeds/frequencies increase to the point where a multipoint ground system is required, it is implemented by attaching components and subsystems directly to an underlying ground plane using the shortest-widest possible connection. Component placement is important because of inductance and resonance.
The impedance between two points on a ground plane as a function of frequency is given by:
Z = RRF (1 + | tan 2 π d / λ |)
Where:
RRF = 0.26×10-6 (μf / σ)0.5
f = Frequency (Hz)
μr = Relative permeability
σr = Relative conductivity
d = Distance between points (m)
λ = Wavelength at frequency of interest (m)
Obviously, at high frequencies it is not an equipotential plane!
In the past, we used to tell PCB designers to configure their boards the way microwave aerospace systems are configured. Now, it’s easier to tell large scale microwave systems designers to configure their systems like PCB designers do with high speed multilayer boards. Both approaches are similar, even though one is steady sate and the other is transient; one uses bond straps and the other vias; and one is big and the other is small.
Most aerospace widgets are built in rectangular hogged-out aluminum boxes. They may not be as cute as the plastic ones with a lot of curves and they may be more expensive, but they are more robust and they also provide better shielding. We’ll cover shielding sometime in the future.
Ideally, RF grounding should be done by placing the widget side with the largest area against the equipotential ground reference plane and pulling the surfaces tightly together to assure close intimate contact. If surface-to-surface contact is not possible, weld the widget completely around its perimeter to the surface. If that’s not possible because grounding wasn’t considered until the last minute, then a bonding strap can be used. This is a poor quality extension of the ground system from the plane to the widget, but it is way better than a wire.
There are a number of EMC design guides that call out conductor length to width ratios. The required 4:1 or 5:1 L/W ratios are an attempt to maintain somewhat of a plane configuration between the connection points instead of just a wire. The truth is a grounding/bonding conductor should have zero (0) length and be infinitely (∞) wide—or at least as wide as the widget being grounded.
The contact surfaces should be bare but coated with something conductive to minimize corrosion, such as iridite or alodine—but not anodize, because it is nonconductive. MIL-C-5541F, Chemical Films, covers these types of materials. Contact surfaces should always be measured to assure conductivity. This prevents Murphy from painting the surfaces with non-removable, nonconductive epoxy paint.
– Ron Brewer
