We've already realized that information about the "correct time" is useless unless it's instantly available. But what, exactly, do we mean by "instantly," and how is it made available to everyone who wants to know what time it is? What can happen to this information on its way to the user, and what can be done to avoid some of the bad things that can and do happen?

CHOOSING A RADIO FREQUENCY

The frequency of a radio signal primarily determines its path. The signal may bounce back and forth between the ionosphere and the surface of the earth, or creep along the curved surface of the earth, or travel in a straight line-depending on the frequency of the broadcast. We shall discuss the characteristics of different frequencies, beginning with the very low frequencies and working our way up to the higher frequencies.

Very Low Frequencies (VLF)-3-30 kHz

The big advantage of VLF signals is that one relatively lowpowered transmitter could provide worldwide coverage. A number of years ago, a VLF signal broadcast in the mountains near Boulder, Colorado, was detected in Australia, even though the broadcast signal strength was less than 100 watts. The VLF signal travels great distances because it bounces back and forth between the earth's surface and the lowest layer of the ionosphere, with very little of its energy being absorbed at each reflection.

Another good thing about VLF signals is that they are not strongly affected by irregularities in the ionosphere, which is not

true in the case of shortwave transmissions. This is so because the size of the irregularities in the ionosphere "mirror" is generally small compared to the length of the VLF radio waves. For example, at 20 kHz the radio frequency wavelength is 15 kilometers. The effect is somewhat the same as the unperturbed motion of a large ocean liner through slightly choppy seas.

But VLF also has serious limitations. One of the big problems is that VLF signals cannot carry very much information because the signal frequency is so low. We cannot, for example, broadcast a 100 kHz tone over a broadcast signal operating at 20 kHz. It would be like trying to get mail delivery ten times a day when the mailman comes only once a day. More practically, it means that time information must be broadcast at a very slow rate, and any schemes involving audio frequencies, such as voice announcement of the time, are not practical.

We mentioned that VLF signals are not particularly affected by irregularities in the ionosphere, so the path delay is relatively stable. Another important fact is that the ionospheric reflection height is about the same from one day to the next at the same time of day. Unfortunately, though, calculating the path delay at VLF is a complicated and tedious procedure.

One last curious thing about VLF is that the receiver is better off if he listens to signals from a distant station than from nearby stations. The reason is that near the station he gets two signalsone that is reflected from the ionosphere (sky wave) and another that is propagated along the ground. And what he receives is the sum of these two signals. This sum varies in a complicated way as a function of time and distance from the transmitter. So the listener wants to be so far from the transmitter that for all practical purposes the ground wave has died out, and he doesn't have to deal with this complicated interference pattern.

Low Frequencies (LF)-30-300 kHz

In many respects LF signals have properties similar to VLF. Of course, the fact that the carrier frequencies are higher means that the information-carrying rate of the signal is potentially higher also.

These higher carrier frequencies have allowed the development of an interesting trick to improve the path stability of signals. The scheme was developed for the Loran-C navigation system (See pages 161-163) at 100 kHz, which is also used extensively to obtain time information. The trick is to send a pulsed signal instead of a continuous signal. A particular burst of signal will reach the observer by two different paths. He will first see the ground wave signal that travels along the surface of the earth. And a little later he will see the same burst of signal arriving via reflection from the ionosphere.

At 100 kHz, the ground wave arrives about 30 microseconds ahead of the ionospheric wave, and this is usually enough time to measure the ground wave, uncontaminated by the sky wave. The

ground wave is quite stable in path delay, and the path-delay prediction is considerably less complicated than when one is working with the sky wave.

Beyond about 1000 kilometers, however, the ground wave becomes so weak that the sky wave predominates; and at that point we are pretty much back to the kind of problems we had with VLF signals.

Medium Frequencies (MF)-300 kHz-3 MHz

We are most familiar with the medium-frequency band because it contains the AM broadcast stations of this country. During the day the ionospheric or sky wave is heavily absorbed, as it is not reflected back to earth; so for the most part, during the daytime we receive only the ground wave. At night, however, there is no appreciable absorption of the signal, and these signals can be heard at great distances.

One of the standard time and frequency signals is at 2.5 MHz. During the daytime, when the ground wave is available, the Japanese report obtaining 30-microsecond timing accuracy. At night, when one is receiving the sky wave, a few milliseconds is about the limit. It is fair to say, however, that this band has not received a great deal of attention for time dissemination, and there may be future promise here.

High Frequencies (HF)-3.0-30 MHz

The HF band is the one we usually think of when we speak of the shortwave band. Signals in this region are generally not heavily absorbed in reflection from the ionosphere. Absorption becomes even less severe as we move toward the upper end of this band. Thus the signal may be heard at great distances from a transmitter, but it may arrive after many reflections, so accurate delay prediction is difficult.

Another difficulty is that in contrast to VLF waves, HF wave lengths can be of the same order, or smaller, than irregularities in the ionosphere. And since these irregularities are constantly changing their shape and moving around, the signal strength at a particular point will fade in and out in amplitude. Because of the fading and the continuous change in path delay, accuracy in timing is again restricted to about 1 millisecond, unless one is near enough to the transmitter to receive the ground wave.

Most of the world's well-known standard time and frequency broadcasts are in this band.

Very High Frequencies (VHF)-30-300 MHz

From a propagation point of view, one of the most important things that happen in the VHF band is that the signals are often not reflected back to the surface of the earth, but penetrate through the ionosphere and propagate to outer space. This means that we can receive only those stations that are in line of sight, and explains why we do not normally receive distant TV stations, TV signals being in this band. It also means that many different

signals can be put on the same channel, and there is little chance of interference as long as the stations are separated by 300 kilometers or more.

From a timing point of view, however, this is bad; for if we want to provide worldwide or even fairly broad-coverage, many stations are required, and they must all be synchronized. On the other hand, there are advantages to having no ionospheric signals because this means that we can receive a signal uncontaminated by sky wave. We can also expect that once we know the delay for a particular path, it will remain relatively stable from day to day.

A third advantage is that because carrier frequencies are so high, we can send very sharp rise-time pulses, and thus can measure the arrival time of the signals very precisely. Because of the sharp rise time of signals and the path stability, timing accuracies in this region are very good. Microsecond timing is relatively easy; and if some care is taken, even 0.1 microsecond can be achieved.

At the time of this writing the master clock at NBS in Boulder, Colorado, is used as a reference clock for the standard time and frequency broadcasts from WWV near Fort Collins, some 80 kilometers away. A TV signal is used to maintain the link between Boulder and Fort Collins with an accuracy of a small fraction of a microsecond. The system is explained more fully at the end of the next chapter.

Frequencies Above 300 MHz

The main characteristic of these frequencies is that like VHF frequencies the signals penetrate into outer space, so systems are limited to line-of-sight. There may be problems caused by small irregularities in the path-or "diffraction effects," as they are commonly called-similar to such effects at optical wavelengths. Nevertheless, if a straight shot to the transmitter is available, we can expect good results.

Above 1000 MHz weather may produce problems; this is especially significant in broadcasting time from a satellite, where we wish to minimize both ionospheric and lower atmosphere effects.

INFRARED LIGHT

106

108

NOISE-ADDITIVE AND MULTIPLICATIVE

We have been discussing different effects that we should expect in the various frequency bands. We should differentiate explicitly here between two types of effects on the signals. They are called "additive noise" and "multiplicative noise." "Noise" is the general term used to describe any kind of interference that mingles with or distorts the signal transmitted and so contaminates it.

Additive noise is practically self-explanatory. It refers to noise added to the signal that reduces its usefulness. For example, if we were listening to a time signal that was perturbed by radio noise caused by lightning or automobile ignition noise, we would have an additive noise problem.

Multiplicative noise is noise in the sense that something happens to the signal to distort it. A simple illustration is the distortion of one's image by a mirror in a fun house. None of the light or signal is lost-it is just rearranged so that the original image is distorted. The same thing can happen to a signal as it is reflected by the ionosphere. What was transmitted as a nice, clean pulse may, by the time it reaches the user, be smeared out or distorted in some way. The amount of energy in the pulse is the same as if it had arrived cleanly, but it has been rearranged.

NOISE FROM
LIGHTNING

TIME SIGNAL

RECEIVER

ADDITIVE NOISE

IONOSPHERE

RECEIVED

TRANSMITTED

RECEIVER

MULTIPLICATIVE NOISE

How can we overcome noise? With additive noise, the most obvious thing to do is to increase the transmitter power so that the received signal-to-noise ratio is improved. Another way is to divide the available energy and transmit it on several different frequencies at once. It may be that one of these frequencies is extremely free from additive noise. Another possibility-quite often used-is to "average" the signal. We can take a number of observations, average them, and improve our result. This works because the information on the signals is nearly the same all the time, so the signal keeps building up; but the noise is, in general, different from one instant to the next; therefore it tends to cancel itself out.

With multiplicative noise it doesn't help to increase the transmitter power. To return to our previous illustration, the image from a fun-house mirror will be just as distorted whether the