FOR THE MOST PART, NOISE is a dirty word in electronics, appearing even in the most carefully crafted component. The temperature of the environment causes electrons to move about randomly, a process that introduces voltage shifts that can be heard as hiss and rumble. In certain instances, however, noise can be useful, even critical, for the transmission of information. Essentially, it can boost an otherwise weak signal above the threshold of detection. During the past 10 years, researchers have begun to recognize how the phenomenon, called stochastic resonance, may play a role in many diverse events, from the occurrence of ice ages to neuron signaling [see "The Benefits of Background Noise," by Frank Moss and Kurt Wiesenfeld, page 66].

We will describe two experiments.

The first shows the utility of noise in bringing out a signal; the second illustrates the remarkable ability of the brain to home in on one conversation in a crowded restaurant-that is, to extract data from background noise.

The key device we will build is a threshold detector (which provides the simplest manifestation of stochastic resonance). Such an instrument sends out a signal only if it receives an input of sufficient intensity. The project also needs two other devices: a signal generator and a noise generator. Besides these three modules, you will need a source of music (an ordinary tape player, say) and a set of headphones. The two generators and the detector can easily be built from inexpensive electronics parts; the total cost should be less than $100. (One source is Digi-Key Corporation, 701 Brooks Avenue South, Thief River Falls, MN 56701-0677; 1800-344-4539.) The components can be assembled either on a breadboard (a plastic panel with many sockets that allow connections without soldering) or on printed circuit boards (which we recommend). The crucial components are integrated circuits called operational amplifiers, or op-amps.

The threshold-detector module requires the largest number of parts [see Figure 2]. The LF411 op-amp sums the noise and the signal ( from either the tape player or the signal generator ). This noisy signal is then applied to the LM311 op-amp, which compares the voltage of the noisy signal to a threshold voltage. This voltage is adjusted with the one-kilohm potentiometer (an adjustable resistor).

When the noisy signal is less than the threshold, the output of the op-amp is -9 volts. It rapidly swings to +9 volts when the noisy signal crosses the threshold. The LM2917 op-amp measures the average rate at which these swings occur and converts it into a voltage that is amplified by the LM386 op-amp. The output of the LM386 is connected to the headphones. The potentiometer connected to the LM386 controls the volume of the sound.

The noise generator relies on the thermal agitation of the electrons in a megohm resistor. This large resistor is connected

to the plus input of the LF411 op-amp at terminal 3, where fluctuating voltage, called Johnson noise, appears. This resistor and its amplifier must be shielded from extraneous signals, such as the 60-hertz power lines in your house. Covering them with a small piece of steel cut from a can lid should work. The shield must also be grounded: solder it to the ground plane on the printed circuit board underneath the resistor. The three LF411 op-amps function as high-gain amplifiers. The noise level can be adjusted with the 10-kilohm potentiometer.

In the signal-generator module, an ICL8038 chip generates a sinusoidal voltage signal with a frequency of about 660 hertz. The LF411 serves as an amplifier, and a 10-kilohm potentiometer regulates the amplitude of the signal.

The most convenient way to construct these modules is with printed circuit boards. (Information about obtaining these boards appears at the end of this article.) Audio cables for connecting the threshold-detector module to the tape player and the signal and noise modules are available from electronics supply stores. You will need two such connecting cables that have miniature headphone male jacks (the type that plug into portable cassette players) on both ends. Each module will need two 9-volt batteries, one for the plus and one for the minus supply voltages.

The first experiment makes use of the tape player and the noise module connected to the threshold-detector module. Connect the output of the tape player to the terminal marked "signal" on the threshold-detector module. Use the other cable to connect the noise module to the "noise" terminal on the thresh old-detector module. Finally, plug in the headphones to the output.

Turn on the tape player and increase its volume to a somewhat large value- say, two thirds of maximum. Set the volume adjustment on the threshold-detector module about halfway between zero and the maximum. With the noise level at zero, adjust the threshold until you can just easily hear the music (it will sound distorted because of the nature of the circuit). Once set, the threshold adjustment does not need to be changed again. Next, turn down the volume on the tape player until you can no longer hear the music. The sound is now a weak, or subthreshold, signal.

Now you are ready to add noise to recover the music. Slowly turn up the noise-amplitude control. At first, you will hear unintelligible sounds, but as you boost the noise level, you will begin to perceive the music. Beyond that point, however, the noise will start degrading the music until once again all you hear is the noise. You can experiment with different threshold settings. Generally, the lower the threshold, the less noise will be required to produce an intelligible response.

In the second experiment the 660-hertz signal generator replaces the tape player. The idea is to measure a person's threshold of perception of a signal embedded in various amounts of noise. You will need to add gradations on all the potentiometer dials; number the range from 0 to 10 at regularly spaced intervals. You will also need a piece of ordinary graph paper to plot the noise level against the signal.

Put the headphones on and set the noise dial to 0. Turn up the signal dial to 8. Then, without looking at the dial, I set the threshold to the point at which you can barely hear the 660-hertz tone. You can use these settings as your first data point (noise = 0, signal = 8).

Now turn up the noise dial to 1. Adjust the signal dial (not the threshold dial) until you are barely able to perceive the tone; plot that point. Repeat this procedure for all noise dial settings up to 10 [see illustration above]. On the graph, you should notice that the data pass through a minimum at a noise setting in the range of 2 to about 5. This level is the optimum noise for detection of the 660-hertz tone. Note that the detailed shape of this data set may depend on the individual ability to perceive the signal amid the noise. Try to make the same measurements on several friends and plot the data from each on separate graphs. Are there any notable differences? Does the ability to detect the signal in the noise depend on age or hearing ability? For variation, try altering the 660-hertz tone by changing the 0.1-microfarad capacitor, located between pins 10 and ll of the ICL8038 op-amp. The frequency equals 6.3829 x 10^-5 divided by the value of the capacitor (more specifically, frequency = 0.3/(RC), where R is 4,700 ohms and C is the capacitance in farads.

For prefabricated circuit boards and additional construction tips, send $12 to Wayne Garver, Department of Physics and Astronomy, University of Missouri at St Louis, St Louis, MO 63121; E-mail: swpgarv@sivaxa.umsl.edu. This offer expires January 31, 1996.

Bibliography

THE SIGNAL VALUE OF NOISE: ADDING THE RIGHT KIND CAN AMPLIFY A WEAK SIGNAL. Ivers Peterson in Science News, Vol. 139, No. 8, page 127; February 23, 1991.

BRINGING MORE ORDER OUT OF NOISINESS. John Maddox in Nature, Vol. 369, page 271; May 26, 1994.

The Society for Amateur Scientists (SAS) is a nonprofit research and educational organization dedicated to helping people enrich their lives by following their passion to take part in scientific adventures of all kinds.