During the development and testing of signal processing systems, it is useful to generate test signals in the laboratory in order to quantify system parameters such as dynamic range, processing gain, or frequency resolution. Furthermore, it is convenient to have two test signals that are calibrated and easily variable over a range of signal-to-noise ratios (S/Ns), with the noise in each test signal uncorrelated from the other. To satisfy these and other requirements, the Dynamic Range Tester (DRT) is designed as an inexpensive test instrument and constructed using off-the-shelf components. The DRT is used for a wide range of testing and demonstration purposes.

The DRT is a patented prototype system designed at the Army Research Laboratory. This tool emulates field conditions in the laboratory in order to characterize signal processing hardware with repeatability and reliability. The DRT is a multipurpose test instrument that simulates two antennas receiving the same or different signals. As noted above, it generates two signals that are calibrated and whose signal-to-noise ratios are easily variable. User-supplied signals are injected into two highly isolated channels containing uncorrelated additive white Gaussian noise (AWGN). Precision attenuators allow the user to independently vary the S/N of each generated or "real-world" signal. These signals can be used in the test and measurement of dynamic range, processing gain, and interference in correlators, acousto-optics, telecommunications, telemetry, and other signal processing applications. Although the prototype system frequency range is 30-400 MHz, the design is extendible to other frequency ranges to produce useful instrumentation for other applications.

Prototype of the Dynamic Range Tester.

The DRT is an especially useful tool for testing dual-input systems, such as correlators, and also systems that process signals close to and below the noise level. The wideband noise in each channel, which spans the entire frequency range of the DRT, can be limited with a frequency-selective filter. The power level of the noise is variable, so that both positive and negative S/Ns may be generated. Using a single user-supplied signal source, the DRT can be used to evaluate processing gain of correlators — how far below the noise level the correlator can detect a signal in both of its inputs. Processing gain is the correlator's dynamic range, the difference between the maximum and minimum signals the system can process simultaneously, with the maximum signal being the noise level. The DRT can also operate using two separate signal sources, one for each generated noisy or "real-world" signal, in order to perform interference and bandwidth tests.

Normally, when measuring S/N, a true measurement of "signal only" cannot be made directly because some thermal noise is always present. Therefore, S/N is calculated as follows:

S/N dB = 10 × log [((S + N) – N) / N] (W).

This superposition of signal and noise in the output is a valid assumption, since thermal noise has Gaussian-distributed randomness and is therefore independent of the signal. This theoretical calculation would be used if the signal and true noise floor level were on the same order, since there would be no way to turn the noise off.

The DRT design, however, is based on having an artificial noise floor in each of its outputs that may, in fact, be turned off and on using front-panel switches. This artificial noise floor may be set to various power levels, many well above the system noise floor. Initially this excess noise power and the signal power in the DRT outputs should be set to high levels, so that direct measurement of "noise only" and "signal only" can be made by using front-panel switches to cut off signal or noise power respectively. The maximum excess noise levels and signal levels, +30 dBm, are much larger than the actual DRT noise floor, -50 dBm in a 20-MHz bandwidth. Therefore, when signal and noise power levels are well above the noise floor, direct measurement of signal only and excess noise only can be made, ignoring the addition of the DRT noise floor, which becomes negligible. All noise power measurements are based on average power and may be verified with a high-quality power meter using a thermocouple sensor. Reducing the S/N of the output signals then relies on the calibrated attenuators of the DRT. Signal power may be measured as peak power or average power, depending on the signal type supplied and the power sensor used.

The total power, composed of equal parts signal and noise in each output of the DRT, is 2 W. As noted above, the noise floor of the DRT in a 20-MHz bandwidth is -50 dBm, so that a user can vary the S/N of the generated wideband signals in the range from +80 to -80 dB. By using an external filter with a bandwidth smaller than 20 MHz, the system noise floor is further reduced so that proportionally lower S/Ns can be generated. Because correlation is a multiplicative operation, the following identity applies:

log (a × b) = log a+ log b

Therefore, using the 80-dB range for a 20-MHz bandwidth in each channel, correlator processing gains of up to 160 dB may be evaluated. The channel-to-channel isolation of this "calibrated noisy signal generator" is approximately 90 dB.

The dynamic range tester is useful for testing one- and two-input systems such as correlators or dual-channel receivers. It is used for testing systems that process signals close to and below the noise level. All signal inputs, outputs, and controls are accessed through the front panel, making the DRT easily configurable. Its design can be extended to other frequency ranges to create useful instrumentation for other applications, including telecommunications, telemetry, biomedical instrumentation, and others. Its design may also be miniaturized and enhanced with computer bus control. Control over the phase of the signals through the DRT may be added.

This work was done by Anne Marie Petrich Marinelli and Michael S. Patterson for the Army Research Laboratory . Inquiries concerning rights for the commercial use of this invention should be addressed to the U.S. Army Research Laboratory, Attn: Norma Vaught, AMSRL-CS-TT, 2800 Powder Mill Rd., Adelphi, MD 20783-1197.