Other types of transients can be transmitted through the air and/or coupled onto nearby conductors.

In Part 1, we examined the effects of induced electrical noise within the wiring system, noting the behavior of surges (high-speed transient phenomena) as they traveled completely within the boundaries of the conductors. In the surges discussed last month (those carried on the conductors along with 60 Hz power), we concentrated on those having a very fast rise time. The easiest to understand is the spike, which travels on the power phase conductors. This high-speed impulse (or rise in voltage) can result from some form of switching of an electrical component such as a capacitor.

We also focused on common mode electrical noise (referred to as mystery noise in Part 1) and the disturbances resulting from its circulation. We noted that this type of electrical noise has no correlation to the line-to-line power-related noise known as normal mode or transverse mode noise. Instead, common mode noise is usually detected only between a phase conductor and ground, or between the power system neutral and ground. In fact, the best place to monitor common mode noise is between a panelboard neutral and the signal reference (case ground) of a sensitive load device.

Ground loops

While we might expect a neutral-to-ground or neutral-to signal reference measurement of zero volts, we actually may see as much as several hundred volts. This would indicate the presence of a ground loop, which involves the grounding of a device at two or more locations. One ground point may have a different potential with respect to the other(s). And, with some impedance between the two points, we would have a potential driving a current around the loop. Fig. 1 shows how such a loop might appear. This phenomena is one of the leading causes of sensitive signal driver/receiver disturbance and destruction.

The case of the wobbling CRT screens

Remember from Part 1, we began describing a case history involving wobbling CRT screens. We managed to determine that no transients were on the power or signal wiring because the wobbling did not cease when we made an air choke by coiling the CRT's power cord.

We suspected something being transmitted through the air, so we moved each CRT away from the outside wall of the room. Alas, the wobbling disappeared when the CRT was 7 to 8 ft away from the outside wall. What was in or outside of that wall?

Continuing our investigation, we learned that the local electric utility had made a major change of its service conductors for this building, and the change timing corresponded to when the wobbling was first seen.

Originally, the service entrance conductors approached the "suspect" wall in a "straight-in" manner, or perpendicular to the plane of the wall. The rewiring altered this layout; the service conductors now were suspended along the outer face of this wall, making them parallel to the plane of the wall. This resulted in the service conductor bundle's electric field encircling the CRT positions. The farther we moved the CRTs away from this electric field, the less the negative effect of the field.

We recommended that the utility re-configure the service conductors as per the original installation; that is, perpendicular to the outside wall. This would be much less expensive than trying to "fix" the field at each CRT.

Induced versus conducted noise

The above case history opens up a realm of disturbances, namely, those not traveling on the system conductors. Before discussing this, let's look at a corollary effect that induced noise has on the power and signal wiring of sensitive systems.

We know that a high-speed spike traveling down a wire and heading into a coil generates an inductive kick, a reaction affected in the coil that impedes the travel of the original spike. In fact, we find that this "kick back" effect actually opposes the amplitude of the spike, making it lower or of lesser effect. We're actually looking at the opposition of two forces: One coming into the coil and one generated by its travel. These two forces are in series; thus we get a canceling or reduced effect.

What would happen if we have these same two forces in parallel instead of in series with each other. Obviously, they're no longer opposing each other. Thus, each "kick" stands out by itself. This is exactly what happens when we have a noise "burst" brought side-by-side but not touching a wire.

Let's look at Fig. 2. The noise source shows only the positive spikes, coupling these impulses into the nearby wire. This wire behaves like the coil described above; however, these coupled impulses and the reaction of the wire (coil) now do not oppose each other. Instead, they show themselves as separate spikes on the wire. In other words, we now have twice as many surges because there is no canceling.

This is no longer a linear relationship, as indicated in Ohm's law. In a linear relationship, for example, 1 ohm of resistance and 10A of current results in 10V. Now, what we calculate for voltage in a resistive circuit (E = IR) is changed and the equation for determining voltage through this coil is considerably different:

E = L x di/dt

Don't let this equation scare you; the notation "di/dt" simply means the rate of change of current (i) with respect to time (t).

This "di/dt" is the critical part of the equation. Notice that time (t) is in the denominator; thus, as the speed of data signals increases, di/dt gets smaller and E gets larger. In other words, as time goes from milliseconds to microseconds to picoseconds to nanoseconds, voltage can increase to several thousand volts. Thus, we can conclude that induced noise is even more powerful (and potentially damaging) than wired noise.

Typical installation and potential problems

Let's look at a classic, acceptable panel-board/cable trough installation as shown in Fig. 3. Here we see two circuits wired into shielded isolation transformers. The transformer outputs are then run to digital telephone switches. The isolation transformers are used to keep electrical noise from the switches.

The question is: Will this arrangement do well in noise rejection? Unfortunately, the answer is NO. What is acceptable for electrical safety and convenience in wiring just will not help much in noise rejection.

First, look at the raceways holding the transformer input conductors; they also hold the output conductors. Thus, any electrical noise on the incoming conductors will be coupled onto the outgoing conductors and directly into the digital load devices. It's just not wise to allow unshielded circuits to "crosstalk" to each other.

In Part 3 next month, we'll expand our discussion and address methods of protection and techniques in noise rejection.