Anyone who has tried to troubleshoot an interference problem knows it can be a lot like chasing ghosts. Just about the time you think you have it, all trace of it disappears or, worse yet, the solution doesn't provide a noticeable change in the problem. After many years of field service and troubleshooting, I've learned to improve my chances for success by taking an organized approach. Here's how you can do the same.
The first and foremost step is to look at the installation. A significant number of interference problems are the result of incorrect installations. Oftentimes, the solution is no more complex than revisiting the design drawings to find the installation error.
If you're dealing with a high-frequency interference problem, inspect the grounding system before you get too carried away with special monitoring. Just like with installation errors, you can solve many interference problems by addressing the grounding system. The IEEE's Standard 1100-1999, the Recommended Practice for Powering and Grounding Sensitive Electronic Equipment, has extensive information on correcting and improving these systems.
A common grounding solution for high-frequency interference problems, particularly when multiple pieces of equipment are involved, is to install a ground grid under the affected equipment. The grid provides an equipotential ground reference that has low impedance at high frequency. This provides a path for the interference to flow away from the signal path, and it greatly reduces any voltage gradient between the grounds on the various pieces of equipment.
As you may have guessed by now, it's not always necessary to find the exact cause of an interference to fix the problem. You only have to keep the interference from reaching the piece of affected equipment.
There are four ways in which interference signals can get into circuits: conducted, capacitively coupled, magnetically coupled, and radiated signals. These are listed in the order they are most likely to occur [1].
“Conductive” means that the interference signals travel on the power or signal conductor to the device or circuit element that is being adversely affected. In many cases where the interference is introduced into a circuit by one of the other three ways, it's also conducted a short distance to the device. In addition, as the frequency of the interference increases, the distance it will be conducted decreases. Wire is inherently inductive and, therefore, its impedance increases in direct proportion to the frequency.
Normally, capacitively coupled (see Fig. 1, on page 24) and radiated signals are relatively high in frequency, and magnetically coupled signals are usually at or near the power frequency. Let's take a closer look at these three interference types now.
Shielding to prevent capacitive coupling is much easier to achieve than shielding to prevent magnetic coupling. One method for preventing capacitive coupling is to put the control circuits in a separate conduit from the power circuits. The conduit acts as the shield, and the interference is coupled to the conduit/shield instead of the wire inside. When control-circuit-to-control-circuit capacitive coupling is a problem, you also can purchase control cables with individual shields for each pair of twisted wires.
One of the most common magnetic interference problems is when the screen on the monitor of a personal computer shakes. This happens when a magnetic field comes in close proximity to the monitor.
The first thing you should do is look for a power panel or a bus duct behind, below, or above the wall. The physical dimensions of panels and bus ducts are too large for the magnetic fields to cancel completely.
If you find a panel or bus duct behind the wall, try turning the monitor (because the fields are directional). If that doesn't help, try moving it across the room. If moving out of the field isn't an option, there are special boxes available to shield computer monitors from magnetic fields. In general, though, successful magnetic shielding is not easy to achieve and may require the help of a specialist.
If no such panel or bus duct exists, the next most likely culprit is a miswired neutral. Lighting circuits with as little as 10A have been known to cause this problem when the light is connected with a phase from one panel but a neutral from a different panel. The current circulates in a loop instead of flowing back through the same conduit and canceling most of the magnetic field.
For magnetic interference problems in control circuits, the best solution is to use control cable with twisted pairs. By twisting the wire, the polarity of the magnetically coupled interference is reversed and cancels itself out.
This type of interference can be in the form of actual radio and/or microwave transmissions, such as those found with hand-held radios or in close proximity to a commercial broadcast. A common example is the AM interference when you drive under a transmission line.
The cause of radiated interference problems may be emissions from high-voltage power lines or arc discharge from welders or similar equipment. Engineers can solve such problems by shielding the affected device from the emission. This may be as simple as keeping the doors on electronic equipment closed when using hand-held radios, or as complex as placing a full Faraday cage around the equipment.
Another important tool you'll need for troubleshooting interference problems is a power quality monitor. However, such a device works best if you know its limitations and understand the data it provides.
All power quality monitors are not equal. When you're dealing with interference, you need a monitor with good high-frequency capture that accurately shows the wave shapes. The cause of the transient can be determined by looking at the wave shape.
Fig. 2 and Fig. 3, on page 26, are wave shapes taken by a high-quality monitor. The impulses, which came from the commutating of the fan motor in a heat-shrink gun (one that looks like a hair dryer), helped us solve a big problem for one client. Although we had connected another monitor (a different brand) to the same location, its wave shapes looked nothing like those shown in Fig. 2. Without the better monitor, we may not have found the problem at all.
Finally, when working with monitoring equipment to measure high-frequency signals, consider the test leads and the reference you are using for the signal. The electrical noise shown in Fig. 1, on page 24, is a capacitively coupled signal into the leads of the oscilloscope. A separate reference lead was being used to measure the signal to a ground that was several feet away. The oscilloscope's instruction book states that the separate reference lead can only be used in low-frequency measurements. When the proper shielded-reference lead was connected (a short tail that connects to the probe), the noise disappeared.
We also found that, with this particular oscilloscope, it was necessary to unplug the battery charger because the power cord could also provide a path for capacitively coupled interference.
When you're in the process of troubleshooting an interference problem, first inspect the grounding system to make sure it's adequate for the type of power system. Of course, it's not always necessary to find the source of an interference to mitigate the problem. Knowing the possible methods in which interference can be introduced into the equipment will provide a framework for an investigation. If your investigation includes using a power quality monitor, then make sure you know the limitations of that piece of equipment and its connections. Following these steps may not eliminate the interference ghosts, but it will certainly help keep them at bay.
Robert J. Schuerger is a Senior Associate at EYP Mission Critical Facilities in Los Angeles, Calif. You can reach him at bschuerger@eypmcf.com.
Van Doren, Dr. Thomas. “Grounding and Shielding Electronic Systems.” Short course. University of Missouri, Rolla, Mo.