A new fast & accurate narrow-band measurement method to measure the GSM mobile's out of channel spectrum spectral energy.

By Michael Leung, Agilent Technologies

"The Output RF Spectrum (ORFS) due to modulation measurement is important because it defines how much a transmitter will interfere with other users. For this reason this measurement is commonly used in both BTS and MS R & D and manufacturing. Usually, due to time constraints, only a subset of the prescribed list of offsets is used. For example, in manufacturing, choosing an appropriate frequency offset list depends greatly on the transmitter design [1]." A new fast digital method about the Output RF Spectrum (ORFS) due to modulation measurement of the GSM Mobile Station transmitter is presented in this paper. Figure 1 shows the ORFS due to modulation measurement, which measures out of channel spectrum energy due to GMSK modulation with a 30 kHz, 5-pole, and synchronously tuned filter (as per ETSI 05.05 and 11.10 standards). The traditional standard measurement takes sample data after the mid-amble during a modulation measurement. In this paper, it discusses new measurement method. It takes the sampling data before and after the mid-amble. The measurement speed is much faster. The experiment results obtained from the application of the actual GSM signal.

Figure 1. ORFS measurement due to modulation

Measurement Overview
The Output RF Spectrum (ORFS) measurement is described in ETSI 05.05 and 11.10 as a zero span spectrum analysis measurement of the energy in different parts of the spectrum of a GSM transmitter's output. This measurement is taken in Time and Frequency domain, which is made by measuring the power averaged over a segment of the time response of a 30 kHz bandwidth filter placed at the center of a GSM burst and comparing it to a time segment of the response of the same filter placed at some frequency offset. The result is a relative power measurement using the 30 kHz bandwidth power at zero offset as a reference in dBc.

The measurement is complicated in that time domain analysis is performed on the signal passing through the filter; so, the measurement results do not represent energy per unit bandwidth. Instead they produce numbers that are only meaningful in relation to the limits described in the ETSI standards. Since the ORFS filter bandwidth is much less than the bandwidth of the GSM signal its output will vary with time even though the GSM signal itself is nearly constant amplitude.

Figure 2 shows a typical filter output with the filter tuned to the center of the GSM signal (zero offset). The rise and fall of the burst ends can be seen at each end of the trace, though the slope is much more gradual due to the limited bandwidth of the ORFS filter. The ripples in the response during the burst can be thought of as the effect of the signal sweeping back and forth through the filter as the signals frequency changes due to modulation. At any given point the trace's value is the average of the effect of several bits worth of modulation. The actual pattern is a function of the data being transmitted and will vary greatly from burst to burst when an actual GSM phone is being tested.

Figure 2. ORFS filter output at zero offset

Figure 3 shows the type of trace that will be obtained when the filter is tuned away from the center of the GSM signal. All of the values are much closer to the noise level. In many cases the switching transient caused by the rapid rise and fall of the burst edges will cause a momentary response greater than that seen during the burst. At very large offsets the signal may be difficult to distinguish from the noise floor. This is why a trigger other than the rise of the output of the ORFS filter is needed.

Figure 3. ORFS filter output at larger offset

Measurement Period
For the ORFS due to Modulation measurements ETSI 11.10 requires measuring at least 40 bits within the Back ORFS measurement period shown in Figure 4. To increase measurement throughput, it measures 40 bits worth of response in both the Front and Back. It will measure ORFS due to Modulation over bit 15 to 54 in the front portion of the burst and bits 93 to 132 in the back. These bits are the ones farthest from the Midamble within the 45 bit window after the midamble specified in ETSI 11.10 and within the front window used by new measurement to increase throughput.

Figure 4. Measured Portions of GSM burst

Filter Impulse Response
The ORFS measurement filter has a 30 kHz bandwidth. It is composed of 5 synchronously tuned sections. With a bandwidth much less than the GSM bit rate the filter produces an output that is the average of the effect of several successive bits of GMSK modulation. The time response of a filter is known as its impulse response; this is the filter output that would be obtained with an infinitely short impulse as the input. The impulse response of the ORFS filter is shown in Figure 5.

Figure 5. ORFS filter Impulse ( μ sec)

From this figure, it can see that the filter introduces a significant delay and that it spreads the effect on any one bit over several adjacent bit periods. The effect of this for the ORFS Measurement is that small variations in triggering will have little effect on the measurement results. It also shows why ORFS due to modulation measurements cannot be made too close in time to the ends of the burst where the large transient due the burst ramping on or off will spread to many adjacent burst periods.

Measurement Calculation
An ORFS due to Modulation Measurement is a calculation done over one of the ORFS Due to Modulation Measurement Periods shown in Figure 4. Up to two modulation measurements can be made per each GSM Burst. If the number of measurements is even then 2 measurements will be made on each burst; if the number is odd, then 2 measurements per burst will be made until only one measurement is left and that remaining measurement will be made on the back part of the burst. For each measurement made over one of the ORFS due to Modulation Measurement Periods shown in Figure 4 the relative power is calculated as:

The average of the Average Log Detector output for 0 offset must be found that in order to calculate the relative power at all other offsets.

At each offset except 0 offset the Average Log Detector output for each measurement is converted to relative power as follows:

For each offset other than 0 the mean of the average relative power is calculated:

Measurement Results
The objective of the experiment is to compare the ORFS due to modulation measurement results based on the ETSI method and the new measurement method. 10 sets of 50 averages at ± 400 kHz offsets were taken. The input power level is at power level 0 (33 dBm) with timeslot 4 at channel 698. Relative levels were converted to absolute levels. Absolute levels were averaged and compared in summary table.

Total testing time: The initial burst of an ORFS measurement due to modulation at two offsets ( ± 400 kHz, – 400 kHz) is timed at 130 ms, with subsequent bursts timed at 13 ms/burst.

Result Analysis
In order to improve ORFS measurement speed, it makes measurements at the front and back of the burst (that is, the measurement is performed over 40 or more bits in each of the regions from bit 15 to 60 and bit 87 to 132). If there was some unwanted perturbation in the signal at the front part of the burst, this could influence the measurement results. Based on the Summary table, there were small changes in the measured values of 30 kHz BW power (0.34 dB), 400 kHz offset absolute power (0.14 dB) and – 400 kHz offset absolute power (0.18 dB). For this test, the differences are not significant.

The Output RF Spectrum (ORFS) measurement measures the out-of-channel emissions of a GSM mobile phone. These emissions can be caused by modulation. It must be kept to low levels to prevent interference to adjacent channels. Because of the desire of GSM system providers to provide optimum performance in their networks, there is increasing pressure on GSM manufacturers for ORFS measurement data. In this paper, it discusses the new ORFS measurement process, which takes the sampling data before and after the mid-amble. The new ORFS measurement method is significant breakthrough process to increase the testing speed for mobile manufacturing and hence to increase the mobile production throughput volume.

Michael Hf Leung is a Senior Application Consultant at Asia Marketing Center of Agilent Technologies. He received his BSEE in 1995 from City University of Hong Kong and later his MSEE from the Hong Kong Polytechnic University.

[1] Understanding GSM Transmitter Measurements for Base Transceiver Stations and Mobile Stations, Application Note 1312, Literature number 5968-2320E. Agilent Technologies [2] Testing and Troubleshooting Digital RF Transmitter Designs, Application Note 1313, Literature number 5968-3578E. Agilent Technologies [3] GSM Measurement Basics Symposium paper, 1994 Agilent Technologies [4] A measurement method based on time-frequency representations for the qualification of GSM equipment, 187-92 vol.1, 24-26 May 1999, IMTC/99. Proceedings of the 16th IEEE Instrumentation and Measurement Technology Conference [5] GSM Standard: GSM-05.05 (ETS 300-577), 14th Edition, December 1998, ETSI [6] GSM Standard GSM-11.10 (ETS300-607-1), 10th Edition, December 1998, ETSI [7] 8960 tester reference manual, Version D, E1960-90001, April 1999 Agilent Technologies [8] Increase Your Throughput with Reduced Instruction Parallel Processing, 5968-1757E, February 1999 Agilent Technologies.