Additive white Gaussian noise (AWGN) has proven itself over the years as a simple yet extremely valuable tool in measurements ranging from BER and SINAD testing to wireless link emulation and the overall characterization of communications receiver performance. The unique characteristics of AWGN are becoming even more useful now that higher order digital modulation techniques are almost universally employed in wireless communications systems. Digital communications systems require noise-generating instruments to be more accurate than ever before, and to maintain their accuracy and repeatability over their entire operating frequency. Some of the noise generators traditionally used have not delivered this capability. The WGN Series AWGN generators has been designed to address the need for a noise-generating instrument with performance in keeping with the demands of digitally-modulated systems for terrestrial and satellite communications.


Standard models in the WGN Series cover 1 MHz to 4.2 GHz in bands tailored to the needs of specific applications, such as satellite communications, first- through third-generation cellular/PCS, CATV and WLAN. However, specialized models with coverage to 18 GHz are available as well. Key performance specifications include overall absolute power accuracy of 0.5 dBm (typically less than 0.2 dBm) and dynamic range of 95 dB.

All of the instruments are equally suited for stand-alone bench-top operation or as part of an ATE system, in which the WGN instrument is interconnected with other instruments via GPIB. Unique to this instrument is the ability to set the output power in terms of power spectral density or total noise power (noise power is based on the user-defined bandwidth of interest). Other noise generators can only control an attenuator within the instrument. Actual noise power is not indicated and must be measured by external equipment. In addition, the WGN's output level can be stepped in programmable increments.

Performing measurements is extremely easy, and dedicated keys for each function are employed rather than multiple-level menus, which simplifies operation. The user is guided through setup and measurement via help screens and a bright LCD display. The display indicates center frequency, noise power, noise density, bandwidth, signal attenuation, signal step size and noise step size. The user can store up to 10 instrument states in nonvolatile memory for use at a later time.

The instrument essentially generates precise levels of AWGN in conjunction with noise power in a specific bandwidth. A signal combiner allows the user to inject a signal and add it to the internally generated noise to set different carrier-to-noise (C/N) ratios. A signal path attenuator can be specified that allows signal power and noise power to be set independently. Once the C/N ratio is set, the instrument's noise attenuator is used to vary the noise power to set new C/N ratios without the need for recalibration. A summary of the WGN Series noise generator's key specifications is outlined in Table 1 .

Table 1
WGN Series Noise Generator Specifications

Output

Calibrated white Gaussian noise

Crest factor

At least 15 dB

Frequency range (MHz)

1 to 6000

Maximum noise power (dBm)

-5 typical

Attenuation range (dB)
Noise
Signal


0 to 95, 0.016 dB resolution
0 to 63, 1 dB resolution (optional)

Relative attenuation accuracy (dB)

0.2

Power spectral density uncertainty (dBm/Hz)

<0.5 (<0.2 typ)

Interface

IEEE-488.2

Stored measurement states

Up to 10

Dimensions (in)

2U rack mount (17 W x 3.5 H x 21 D)

Powers

90 to 264 VAC autoranging, 100 VA max.

Operating temperature range (°C)

0 to 35

Tight Tolerance Control

Design of a basic noise-generating instrument is quite simple. However, creating a noise generator that delivers a consistently high level of performance over all combinations of frequency and noise density poses a formidable challenge. Compensation must be provided for all possible sources of error, from variations in noise density and signal path flatness to step attenuation resolution. To address this challenge, dBm employs a different design methodology than previous noise generators, as well as key components chosen for their superior performance.

In some cases, this has required the company to design its own components instead of using "off-the-shelf" types. For example, one of the most critical components in the instrument is the noise attenuator, which to a large extent determines the accuracy of noise density and noise power. To obtain the desired accuracy and resolution, the company designed its own attenuator, which has a resolution of 0.016 dB. Accuracy is compensated with a dedicated microprocessor in 1 MHz steps at every attenuation setting, and each attenuator is temperature compensated, calibrated and verified over its range of operating frequency and attenuation. The result is an attenuation uncertainty of less than 0.008 dB and a typical error of less than 0.02 dB.

The power supply in the WGN Series was also designed in-house to achieve the required level of shielding. In contrast to most commercially available supplies, the unit powering the instruments is completely enclosed, with a medical-grade power module that delivers a high level of EMI/RFI suppression. The instrument case itself is steel and fitted with EMI gasketing to reduce emissions. Even the GPIB interface is home grown, which allows the company complete control over instrument drivers in order to maintain reliable performance and compatibility. The interface embodies the error-reporting structure and mnemonics of the IEEE-488.2 standard, with which it is fully compliant. Finally, the company chose a microcontroller designed for industrial applications and proprietary software over a general-purpose embedded PC and operating system.

Accuracy of a noise generator can only be ensured if it accounts for both variations in noise density over its frequency range and the bandwidth of the instrument. Variations in noise density are inevitable over a broad frequency range, so calibration must be performed at frequent points throughout this range. If noise density is specified at a single frequency and calibration is not performed, it is very likely that the actual signal density is not correct at any frequency other than the single specified frequency. In addition, slope and ripple of the noise within the signal bandwidth can cause errors in total noise power. Without calibration, these errors can significantly increase measurement uncertainty.

The WGN Series addresses these problems by calibrating noise density in 1 MHz increments throughout the instrument's range of operating frequency. The noise density displayed on the front panel is the true generated value at that frequency. The calibration data is also integrated over the actual noise power within the signal bandwidth, which maintains a high level of accuracy.

The architecture of the instruments departs from the conventional approach for noise generation, in which all components are bolted to an aluminum plate and connected with semi-rigid coaxial cable. This type of construction makes it difficult to replace components in the field and requires tedious adjustments that can usually only be performed at the factory. In contrast, the WGN Series instruments are completely modular, and all subassemblies are contained on removable trays and have their own calibration look-up tables. As a result, the user can replace a subassembly, and software calibration techniques eliminate the need for adjustments. Critical circuitry is ovenized, which virtually eliminates long-term drift with changes in ambient temperature.

Annual calibrations can be performed at the factory or by users or third parties using a PC driven calibration test station designed specifically for accurate measurement of noise density. Worst-case uncertainty is less than 0.1 dB. Calibration is performed with software correction factors stored in the instrument and is completely automated. Noise density is measured in 1 MHz increments with a precision reference frequency converter so that all power measurements are made at 50 MHz, the frequency at which power sensors are generally calibrated, which eliminates power meter uncertainty versus frequency. The instrument collects and stores 96 calibration factors for each 1 MHz of operating bandwidth. Accuracy of each instrument is verified at the factory over its operating frequency range.

dBm Corp., Wayne, NJ (973) 709-0020, www.dbmcorp.com. Circle No. 305