EMC NOTEBOOK
Double Shielded Cables
By William D. Kimmel, PEand Daryl D. Gerke, PE
Kimmel Gerke Associates, Ltd
Radiated emissions and immunity are a big part of EMI control. The regulatory requirements are intended to protect nearby equipment, but don't address the subject of receiver electronics in the same package as switching electronics, where distances between source and antenna may be measured in centimeters. You have the makings of a big RFI problem.
This problem is proliferating with wireless applications and other modern electronic devices, including cell phones, GPS, RFID, AM/FM portable receivers and others. While the most prevalent problem is switching circuit interference to on-board receivers, we also experience problems with local transmitters interfering with on-board electronics ‚ the problem is bi-directional.
Let's take a look at the problem and some solutions.
The Problem
Figure 1. Coupling paths from circuit board to Antenna
The usual EMI rules of thumb are inadequate to describe this problem. Yes, the laws of physics are the same, but we are dealing with microvolt sensitivities in close proximity to volt sources. At close range, coupling factors typically fall off with an inverse square law ‚ field strengths that would be inconsequential 10cm away will become major issues when you cut that distance to one cm, or even less. Figure 1 shows the common coupling paths.
To get an idea of the magnitude of the problem, suppose we have a 10MHz clock switching three volts immediately at the die. According to Fourier, harmonics will fall off at 1/f, so the 10th harmonic will be about 0.3 volt. Suppose we have 1pF of capacitance coupling the chip to a nearby antenna, which is terminated in a 50ohm input. Simple network analysis will show that about 100 microvolts of clock noise will show up at the amplifier, more than enough to interfere with that 100MHz radio station you are listening to. Yes, we have made some gross assumptions in arriving at this number, but it does illustrate the effects of close proximity ‚ even small coupling paths are significant.
We haven't had as much trouble with transmitters interfering with on-board electronics, but it does occur, especially with more powerful transmitters, say 1 - 5 watts. The principal recipients of RF are low level audio frequency analog sensors, not commonly used with receiver/transmitter circuits. It doesn't make any difference to us, however, because the fixes are the same in either case.
Far and away the most effective method of controlling local RFI is shielding. Radio people dealing with transmitters don't monkey with circuit suppression techniques ‚ they shield the critical circuits. Cell phones are all shielded ‚ take your phone (preferably an old one) apart, you will find the package is divided into a number of shielded compartments complete with gasketing. The idea is to keep the internal interference from reaching the receiving antenna (which is obviously outside the shield) and to keep the transmitted RF energy delivered to the antenna from reaching the electronics within ‚ the shield keeps the outside energy outside and the inside energy inside.
To complete the shield, you need to filter or shield all penetrations, including the antenna lead, signal out and power in.
Is shielding too expensive? The cell phone people don't think so, and they stamp out phones by the millions, all shielded and gasketed, and do it very inexpensively. If they had a cheaper way, they would be using it. So before you abandon shielding as an option, take a closer look. If you start early in the design stage, you may be able to use shielding after all.
The Fixes
Suppose your boss tells you that cell phone type shielding is not an option: you will have to do your best without a full shield. Be sure to budget plenty of engineering and test time, as you are likely to have considerable difficulty getting the noise down low enough to satisfy the same person who wouldn't let you shield in the first place. Having said that, here are some things you can do:
Spacing. As mentioned above, electric and magnetic fields fall off very quickly, generally with an inverse square law, so even a small increase in spacing is important. Pay particular attention to critical chips and traces. For receivers, clocked chips and signals along with address and data buses are most likely to cause trouble. Low frequency power switching devices will cause problems at lower frequencies, affecting AM receivers. For transmitters, just about any circuit can be upset, but low-level analog circuits are most vulnerable. Orientation is important if you have noisy traces ‚ avoid running flex or long PCB traces in parallel with antenna elements.
Use a Multilayer Circuit Board. Those who don't want to pay the cost for full shielding are also likely to balk with using a multilayer circuit board. But the ground plane is an exceedingly important factor in your design ‚ it keeps loop areas and ground impedance low and provides a measure of shielding as well. If you must use a two-layer board for your design, budget plenty of engineering and test time, and add this to the overall schedule as well. Worse, you may still fail to achieve your goal.
Further, you need to keep major portions of the plane intact: the key chips and traces all need to be over the ground plane ‚ critical traces crossing a slot in the plane are a sure fire path to failure.
Shield the Chip. Significant E-fields appear immediately above the noisy chip. These can be contained with a conductive chip cover. The cover is to be grounded as often as possible around the perimeter, at all four corners as a minimum. The cover goes immediately to circuit ground plane, not case ground. If you are trying a two-layer circuit board, you will need two chip shields (top and bottom) in order to effectively close the shield.
Minimize Loop Areas. This is aimed primarily at high speed traces and also any magnetics associated with an SMPS. You need to minimize loop area of the signal path and return path, relatively easy to do if you have the ground plane we mentioned above. This will keep the return path immediately under the signal path. Keep the signal path referenced to the same plane from source to destination. If your return path is discontinuous, as it would be by crossing a slot in the plane, you'll create a nasty dipole antenna.
Mount on Opposite Sides. Both electric and magnetic coupling will be reduced if you can mount the noisy chips and traces opposite the antenna location. The ground plane provides some measure of isolation. This isn't a cure-all, however, as the plane itself can become an antenna.
Filter the Lines. Apply filters immediately at the noise source, generally close to the chip shield boundary. This will be at power and periodic signal pins. Make sure the filter is working at your frequency range of interest.
Avoid Problem Frequencies. Select your operating frequencies to keep harmonics out of your frequency band. This is fairly easy to do if you are dealing with a narrow frequency band. For example, you can use a 22MHz clock if you are working with the FM band. The fourth harmonic will be just below and the fifth harmonic will be just above the 88-108MHz range. Switching regulators generate nearly broad band, so avoidance is usually not feasible.
Summary
Close proximity of antennas to active electronics spells problems with local interference ‚ the big problem is periodic or near periodic waves interfering with receiving antennas. Transmitting antennas can also be a problem, interfering with sensitive on-board electronics.
Shielding is far and away the most effective method of controlling interference. If you start early in the design process, you may be able to effectively incorporate overall shielding. If not possible, there are things you can do to reduce interference, including use of multilayer circuit boards, careful placement of noisy components and traces, selective chip shielding and filtering.
But be prepared to do lots of engineering and testing.