Wideband noise measurements on TETRA transmitters

The ETS EN 300-394-1 standard defines the radio test methods to be used in TETRA type testing. The specification includes a definition for the minimum performance, the test method to be used and the acceptable uncertainty level of the measurement equipment.

For wideband noise, the standard requires that the measurement be made through a TETRA filter. This is a Nyquist filter with an 18kHz bandwidth. The wideband noise-performance limit varies as a function of both transmitter power class and offset from carrier.

The measurement method specified in the standard states that the test system will calculate the average noise power over at least 20 TETRA bursts. The measurement samples must be taken synchronously with the transmitted burst. Furthermore, sample data must cover at least 200 of the 255 symbols within each burst.

The measure of average transmitter power level will be used as the 0dB reference for the wideband noise measurements.

Within the bands specified in Table 1 at the left, measurements must be made at 112.5kHz, 262.5kHz, 512.5kHz and frb 1 12.5kHz.

The conformance specification for acceptable uncertainty introduced by the measurement equipment used to test wideband noise is:

RF power relative to 0dB reference (0dBc to 245dBc) 5 61.0dB RF power relative to 0dB reference (<245dBc to 2105dBc) 5 61.5dB

Limitations of test methods

General-purpose spectrum analyzers suffer from a number of deficiencies that make them unsuitable for this measurement. Typically, spectrum analyzers do not incorporate a TETRA filter. This filter is not possible to implement accurately as an analog filter. Instead, a spectrum analyzer uses swept measurement with a narrow IF resolution bandwidth. The responses are weighted and summed to approximate an otherwise true Nyquist response.

For a measurement made using the TETRA filter, the dynamic range requirement is 100dB. Reducing the resolution bandwidth for a swept measurement approach increases the dynamic range requirement (e.g., a 180Hz resolution bandwidth implies a dynamic range of 120dB). This presents a significant challenge for the log amp and detectors of any spectrum analyzer.

TETRA requires that the wideband noise measurement be performed while synchronized to the bursted carrier. A normal spectrum analyzer cannot maintain synchronization with the carrier while tuned to the offset frequency. Why can’t you just sweep the spectrum analyzer slowly across the measurement channels? First, swept spectrum analyzers only see one portion of the spectrum at a time and therefore might miss the occurrence of the burst. Second, a slow sweep will give a small video filter bandwidth. Video filters are applied after RF detection and log conversion. Applying a filter to a noisy power series, after log conversion, will return a measurement error.

Furthermore, any rise in noise level during the burst period will be rejected by the action of the video filter. The default video filter setting can usually be replaced by a user-specified value. If a video bandwidth much greater than resolution bandwidth is selected, then this introduces another variable — namely, many video samples for each pixel will be on the spectrum analyzer screen. Most spectrum analyzers use a selection algorithm, which attempts to pick out the sample peaks. Again, this leads to ambiguity when trying to measure noise power.

Spectrum analyzers may provide external gating capability so that the analyzer only sweeps when the TETRA transmitter bursts. This overcomes some of the drawbacks of using a spectrum analyzer, but a gated sweep requires the addition of a second receiver to be tuned to the carrier or the supply of external triggering from the device under test.

All of these problems make general-purpose spectrum analyzers unsuitable for TETRA wideband noise measurement.

Creating specialized equipment

To address this specific testing need, IFR Systems developed a TETRA-specific signal analyzer. The Model 2310 (Figure 1) uses a single-stage downconverter to create an IF of 10.71MHz. The IF signal is amplified to match the operating point of the instrument’s analog-to-digital converter. The analyzer uses a single-bit bandpass delta sigma converter. This patented converter architecture was developed to obtain a dynamic range suitable to the rigorous demands of TETRA transmitter testing. Once digitized, the IF is passed to dedicated signal-processing hardware that generates a filtered I/Q data stream. The I/Q data are passed to an Analog Devices SHARC DSP that runs the TETRA measurement algorithms.

TETRA wideband noise

The analyzer performs synchronization by demodulating the signal to find the training sequence. This is used to set an accurate internal-timing reference that enables the measurement software to predict when the transmitter is bursting and to avoid the need for a second receiver or external gating input.

When configured for the majority of its TETRA measurements, the analyzer has a noise floor density of 2128dBc/Hz, equal to 285dBc in a TETRA filter.

To extend the dynamic range to more than 100dBc, as required for wideband noise measurement, the analyzer incorporates a switched-IF bandpass filter. The filter is used to pass the noise channel and to reject the carrier. This allows the IF sensitivity to be increased without fear of saturation by the carrier.

For carrier offsets greater than 250kHz, the dominant factor becomes the noise floor of the local oscillator. Reciprocal mixing between the local oscillator and the incoming TETRA carrier ensures that the IF noise floor does not fall below the local oscillator noise floor. However, the local oscillator noise floor is more than 2105dBc when measured through a TETRA filter and therefore exceeds the transmitter specification.

Table 2 above shows the variation in instrument sensitivity against offset from carrier. The analyzer’s residual noise is below that of the TETRA specification for all offsets.

Enhanced wideband noise

The analyzer’s option 01 is designed to increase the maximum measurement range from -105dBc to -110dBc and thereby to provide a 10dB measurement standoff against the ETSI requirement.

The analyzer provides no image rejection. The IF frequency is 10.71MHz, giving an image response 21.42MHz away from the desired input signal. If the transmitter under test has a flat noise floor extending beyond 26.42MHz (5MHz plus image frequency) from the carrier, then the noise floor will be raised by 3dB. Option 01 removes image noise from contributing to the measurement.

Wideband noise measurement

The function of the external components is to produce an IF in the range of 80MHz to 200MHz that is then fed to the analyzer as the RF input. In this way, sufficient selectivity is obtained to prevent device-under-test image noise from compromising the measurement (i.e., with an externally generated IF of 80MHz, the image product seen by the analyzer is 160MHz away). The noise profile of a TETRA transmitter will diminish sufficiently at that offset, and consequently will not raise the noise floor seen at the analyzer’s 10.7MHz IF. External components are not supplied as part of the option.

It is important to minimize the SSB phase noise of the external LO to avoid generating noise contribution through reciprocal mixing between the LO and the TETRA TX input signal. Tests have shown that to achieve reliable results, the LO SSB phase noise should be better than -155dBc/Hz at 5MHz offset and -133dBc at 100kHz offset.

With an appropriate LO frequency setting, the analyzer can be instructed to synchronize to the TETRA carrier with a single key press. In addition to setting a timing reference, the analyzer’s synchronization process measures the amplitude of the carrier to act as a reference point for the measurement.

The next step in the measurement process is to make a noise measurement by retuning the external LO frequency to mix the noise sideband of interest to the center frequency of the external IF. For each frequency offset specified, the noise power is averaged across 20 data blocks that have been captured synchronous to the transmitter burst.

Results

The noise floor of the measurement system depends on both the analyzer and the specification of the external IF, as shown in Table 3 above. These results show that the enhanced setup always gives at least 10dB headroom between the ETSI specification and the measurement noise floor.

Ridgewell is principal design engineer and Carey is product marketing engineer for IFR Systems in Stevenage, England.