NXDN Signal and Interference Contour Requirements An Empirical Study

Transcription

1 NXDN Signal and Interference Contour Requirements An Empirical Study Icom America Engineering December 2007

2 Contents Introduction Results Analysis Appendix A. Test Equipment Appendix B. Test Methodology Appendix C. TIA/EIA Measurement Method Test Results Appendix D. Maximum Distortion Method Test Results Appendix E. Comparison with Theoretical Calculations Introduction This paper presents research that was initiated to answer basic questions of what service and interference criteria should be adopted for the new NXDN 6.25kHz system. Since the bottom line is what counts, the analysis and conclusions from the measurement results are presented first. Then in Appendices A & B the measurement method is documented and the full test results are presented in Appendices C & D. Finally, some comparisons between these measurements and previous theoretical analysis are presented in Appendix E. Comparing the range and interference performance of digital and analog radios is a bit like comparing apples and oranges. Analog FM radio co-channel and adjacent channel interference is usually measured by using pure tone modulations and adjusting power levels for a 3dB degradation of SINAD per the TIA/EIA-603 test methods. However, TIA/EIA-603 does not specify NXDN measurement techniques. Moreover, empirical results have shown that for digital radios with companders (like the AMBE compander used in NXDN), the tones used in TIA/EIA SINAD measurements do not give meaningful results. For radios like NXDN that use Forward Error Correction (FEC), Bit Error Rate (BER) measurements has been shown to be a much more reliable indicator of performance. The particular implementation of FEC used in NXDN is capable of correcting all transmission errors up to a BER of approximately 5%. Therefore this BER was used in comparing performance. The resulting interference power measurement is perhaps too generous as the impairments generated at this level are judged in listening tests to be worse than the audio quality of 12dB SINAD FM reception. However, because of the all or nothing quality of FEC digital radio, only a 2-3dB correction would have to be applied and so this effect was ignored in the TIA/EIA measurements. A second set of measurements was made by a measurement method we called maximum distortion method that casts NXDN in the worst possible light. It assumes that 12dB SINAD is acceptable for analog FM radio and makes no allowance for the fact that the D/U ratio giving this minimum performance is not satisfied over the entire service area.

3 This maximum distortion comparison is made because it is possible for trained operators to use analog radios under extremely poor SINAD conditions where digital radios will not work Results Analysis Before presenting the measurement results, a word of caution is necessary. These results can only be interpreted as pertaining to the Icom F3061, the model radio used in this test. Other models from Icom or other manufacturers may have somewhat different performance. Probably the greatest area of concern is receiver selectivity, which can vary greatly with the tradeoffs made by the design engineers. Major Conclusions: The data collected for this report supports the following conclusions: The same signal and interference field strength can be used for NXDN signals spaced 6.25kHz apart as are used for Narrow FM signals 12.5kHz apart. When NXDN is co-channel with Narrow or Wide FM, current interference field strength should be used to protect the analog radios. (NXDN can tolerate higher co-channel interference and so its protection is not an issue in this case.) TIA/EIA 603 Comparison Table 1 below summaries the TIA/EIA measurements. The 0dB reference point for this table is co-channel interference, Wide FM transmission and Wide FM reception. A received power of -100dBm for the desired signal was used. The results are not sensitive to desired signal level. This particular level is based on the calculation of receiver input power versus Electric Field Intensity given in Appendix B. Under these conditions, the level of Interfering Signal for 3dB SINAD degradation is -118dBm on this radio. (D/U =+18) Table 1: Interfering Signal in db above D/U= +18dB Interfering Radio Target Receiver Step 6.25 Step 7.5 Co- Sensitivity dbm Step 12.5 Analog 25 Analog Analog 12.5 Analog Analog 25 Analog Analog 12.5 Analog Analog 25 NXDN Analog 12.5 NXDN NXDN Analog NXDN Analog NXDN NXDN The conclusions presented above are detailed below with graphics of this table.

4 Conclusion 1) The same signal and interference field strengths can be used for NXDN signals spaced 6.25kHz apart as are used for Narrow FM signals 12.5kHz apart. Interfering Radio Target Receiver Step 6.25 Step 7.5 Co- Sensitivity dbm Step 12.5 Analog 25 Analog Analog 12.5 Analog Analog 25 Analog Analog 12.5 Analog Analog 25 NXDN Analog 12.5 NXDN NXDN Analog NXDN Analog NXDN NXDN Interfering Radio Target Receiver Step 6.25 Step 7.5 Co- Sensitivity dbm Step 12.5 Analog 25 Analog Analog 12.5 Analog Analog 25 Analog Analog 12.5 Analog Analog 25 NXDN Analog 12.5 NXDN NXDN Analog NXDN Analog NXDN NXDN (Actually, according to these measurements, the Interference Field Intensity for NXDN could be increased by db.) Conclusion 2) When NXDN is co-channel with Narrow or Wide FM, the current interference field strength should be used to protect the analog radios. (NXDN can tolerate higher co-channel interference and so its protection is not an issue in this case.) Interfering Radio Target Receiver Step 6.25 Step 7.5 Co- Sensitivity dbm Step 12.5 Analog 25 Analog Analog 12.5 Analog Analog 25 Analog Analog 12.5 Analog Analog 25 NXDN Analog 12.5 NXDN NXDN Analog NXDN Analog NXDN NXDN It has been noted that the adjacent channel interference D/U levels are quite high compared to the level specified by the FCC. However, these results are consistent with the F3061 typical adjacent channel spec of D/U=68dB and with the TIA/EIA recommendation of D/U> 60dB for portable

5 radios. This difference between the FCC and TIA/EIA numbers can be explained by the fact that the interfering signal spectrum for TIA/EIA measurements is rather narrow due to the very low modulation frequency of 400Hz. Appendix E suggests that the adjacent channel interference levels in real world voice communications should be reduced by perhaps 30dB. Maximum Distortion Comparison As mentioned in the introduction, this comparison casts NXDN in the worst possible light. Here we assume that 12dB SINAD is acceptable for analog FM radio and makes no allowance for the fact that the D/U ratio giving this minimum performance is not satisfied over the entire service area due to seasonal variation and terrain. This maximum distortion comparison is made because it is possible for trained operators to use analog radios under extremely poor SINAD conditions where digital radios will not work. The table shows that while NXDN radios exceed the requirements of the current FCC signal and interference field intensity, analog radios can tolerate more interference under these very poor reception conditions. As with the TIA/EIA measurements above, the levels for analog interfering radios in this table are probably 30 db too high due to the narrow occupied bandwidth of 400 Hz modulation. (See Appendix E.) Table 2: Maximum Distortion in db above D/U = +18dB Interfering Radio Target Receiver Step 6.25 Step 7.5 Co- Step Analog 25 Analog Analog 12.5 Analog Analog 25 Analog Analog 12.5 Analog Analog 25 NXDN Analog 12.5 NXDN NXDN Analog NXDN Analog NXDN NXDN ) This data is a revised version of the data presented at the LMCC December 2007 meeting. The reference point of the data has been changed to make it clear that NXDN exceeds current FCC D/U standards. Also, some measurements were repeated with corrected FM deviation. These measurement changes only made small differences in the entries. 2) The 0dB reference point for Table 2 is the same as Table 1: -118dBm interference, -100dBm desired received signal power.

6 Appendix A. Test Equipment Unfortunately, no signal generators are currently available commercially that comply with the NXDN standard. Therefore it was determined to use NXDN-compliant radios as signal generators for all measurements. The test setup is shown in Photograph A.1, the test equipment list in Table A.1 and is described below. Two battery-powered radios were used to generate the desired and interfering signals and a third was used as the target (receiving) radio. The radios were set to lowest stable power (1 Watt) but careful shielding was necessary to reduce signal level to normal receive levels. Otherwise, stray radiation from the transmitters could couple into the target radio. The transmitting radios were mounted in special RF shielded boxes rated for 50 db isolation at this frequency. The transmitters and receiver were widely spaced on the test bench and double-shielded cables were used throughout, even on the audio modulation lines. Attenuators inside and outside of the shielded boxes provided a total of 60 db fixed attenuation. With the radios set to 1W output power, approximately -30dbm of power was delivered to the combiner. Power levels at the output of combiner were calibrated using a R&S Spectrum Analyzer. Precision HP step attenuators were then used to control power level to target radio. Audio modulation for the two transmitting radios was provided by two HP Communication Test sets. VOX was used to control the TX to reduce battery drain when not testing. The use of VOX eliminated an extra set of lines for PTT control that could also cause stray radiation. The HP Communications Test Set modulating the desired transmitter was also connected to the audio output of the target receiver and provided analog FM SINAD measurements of the target receiver. The receiver was programmed to display Bit Error Rate (BER) measurements required for the NXDN measurements. The performance of this system was verified by measuring SINAD in FM mode at low signal levels and comparing these measurements with the SINAD results using an HP Signal Generator. It was found that RF leakage into the target radio caused a difference in the measurements. Additional copper tape shielding was used to form an RF gasket on the shielded boxes and copper tape was also added to all RF connectors (as shown in photos) to reduce the leakage to the point where identical readings were produced by the test setup and the Communications Test Set.

7 Photograph A.1: Test Setup Photograph A.2 Details of Radio in Shield Enclosure Table A.1 Test Equipment Receiver IC-F3061T s/n Rx Radio Transmitter IC-F3061T s/n Tx Interference Transmitter IC-F3061T s/n Tx Desired Communications Test Set HP 8920B s/n US SINAD Measurement Communications Test Set HP 8920A 3141A00917 Interferer Modulation Spectrum Analyzer R&S 9kHz-3GHz FSP s/n Power Calibration 10dB Step Attenuator HP 355D s/n 1204A db Step Attenuator HP 355C s/n 2524A db Attenuator VAT-30W db Attenuator VAT RF Signal Combiner ZSC Radio Firmware: Rev 1.5, Sum CC89, DRev 1.3, DSum EEDD

8 Appendix B. Test Methodology Power Levels While the test results were not expected to be overly dependent on signal level, it was first necessary to determine the approximate signal level of the desired and interfering signals in dbm at the radio input. According to Kraus (Antennas, Second Edition, 1988) the power into a receiver from a matched antenna is: P= S*A e Where A e is the effective aperture of the antenna and S is the average Poynting vector. forward derivation from equations in Kraus yields these quantities as the following: A straight S = ½ * E 2 /377 A e = λ 2 / 4π * G where E is the Electric Field Intensity, G is the antenna gain over isotropic, and λ is the wavelength. So in dbm, P = 30 + G 20log(F/300) + (E 120) 10log(4*π*2*377) P = G 20log(F/300) +E -130 Where P is in dbm, F is in MHz and E is in dbυ (db microvolts/meter). The antenna gains for the F3061 & F4061 are specified by the manufacturer and are listed below. Mobile antennas can be expected to have significantly higher gain, but this will be true for the desired and interfering signal equally. VHF E Field Ant Gain Frequency Rx Power Service Contour 37dBυ -4.5dBi 150MHz -92dBm Interference Contour 19dBυ -110dBm UHF E Field Ant Gain Frequency Rx Power Service Contour 39dBυ -3.0dBi 450MHz -98dBm Interference Contour 21dBυ -116dBm As a check on the reasonableness of these numbers, we compare with A.G. Longley and P.L. Rice, ESSA Technical Report ERL79-ITS 67 that has been the standard propagation prediction model for the FCC. The R-6602 (Carey) Service and Interference Contours are based on this work. The

9 variability of propagation is discussed in Annex 1 of this work. We note that Figure 1.6 in that report suggests that for 90% probability of location coverage 10dB needs to be added to the minimum field intensity. Also Figure 1.4 in the report suggests that as much as another 10 db must be added to allow for propagation changes with seasonal climate changes. Since the 12dB SINAD for the F3061 and F4061 radios is about -119dBm, the above numbers for service contour are reasonable. As a further check, the measurements in Appendix C show that in wide analog FM mode, at these power levels, interference 18 db below the desired signal causes 3dB degradation in SINAD. This is the TIA/EIA-603 standard for interference rejection. Measurement Method Measurements were only done at VHF. There are no differences in the modulation methods between UHF and VHF, so it was not deemed necessary to test at other than a single desired carrier frequency, which was somewhat arbitrarily picked to be MHz. However the two different adjacent channel spacing for VHF and UHF of 6.25 and 7.5 khz were used. Because FM and NXDN waveforms are symmetrical, only high side co-channel interference was measured. Any small differences between high and low adjacent channels would be due to peculiarities in receiver design (e.g. filter asymmetry) and the objective was to compare modulations, not particular receiver performance. TIA/EIA Measurements The TIA/EIA-603 test procedure for analog FM radios specifies that the desired transmitter should have 1kHz sine wave modulation, 60% of maximum deviation. It also specifies that the Interfering transmitter be modulated with 400Hz, 60% of maximum deviation. (60% deviation is 3.6 khz for wide band and 1.8 khz for narrow band.) The interference power level is adjusted until the SINAD in the target radio is degraded by 3 db. Note that the interfering signal spectrum for TIA/EIA measurements is rather narrow due to the 400Hz modulation. This means that the adjacent channel interference levels in real world voice communications would have to be reduced by perhaps 30dB. This is further discussed in Appendix E. TIA/EIA-603 does not cover NXDN measurement techniques and empirical results have shown that for digital radios with companders (like the AMBE used in NXDN), that SINAD does not give meaningful results. When Forward Error Correction is used in the radio, Bit Error Rate (BER) is much more reliable indication. Because of FEC, NXDN is capable of correcting all transmission errors up to a BER of approximately 5%. This BER was used in comparing performance. The resulting interference power is perhaps too generous as the impairments generated at this level are judged in listening tests to be worse than the audio quality of 12dB SINAD FM reception. However, because of the all or nothing quality of digital radio, only a 2-3dB correction would have to be applied and this affect was ignored.

10 Maximum Distortion Measurements To test the worst case performance of the NXDN system against analog FM, it was decided to measure the level of interference required to degrade the FM SINAD to 12dB. 12dB SINAD is usually considered the minimum usable performance. Note that this criteria does not give any margin for locally weak signals or seasonal variations in propagation.

11 Appendix C. TIA/EIA Measurement Method Test Results Co- TIA/EIA Measurements Analog FM WIDE Level dbm F3061 to F3061 HP8920 to F3061 Co- Interference, dbm Wide Narrow NXDN SINAD SINAD SINAD ) 12 db SINAD typical sensitivity is -120dBm (from F3061 product brochure) so measurement is reasonable. 2) Radio measurement with HP Communication Test set and our setup agree to within 1 db resolution. 3) At -100dBm desired signal in Wide FM, D/U = +18 co-channel interference is required to degrade SINAD by 3 db. (Same as FCC D/U ratio.) 4) NXDN and FM Narrow create slightly more co-channel interference in a Wide FM channel than a Wide FM interferer. But difference is small enough that same interference contour could be used. Analog FM NARROW Co- Interference, dbm Level dbm F3061 to F3061 HP8920 to F3061 Narrow Wide NXDN SINAD SINAD SINAD ) 12 db SINAD typical sensitivity is -120dBm (from F3061 product brochure) so measurement is reasonable. 2) Radio measurement with HP Communication Test set and our setup agree. 3) NXDN, FM wide and FM Narrow create the same co-channel interference in a Narrow FM channel.

12 NXDN On Signal Co- Interference, dbm Level, dbm BER, % NXDN Wide Narrow -118 < < < < < < ) BER used as quality measure for NXDN instead of SINAD 2) NXDN to NXDN D/U ratio in this table is 12dB better than the NFM to NFM D/U ratio in the previous table at -100dBm desired signal level. 3) NXDN requires the same level of protection from Wide, Narrow and NXDN co-channel interferers. 4) NXDN requires about 8dB LESS protection than the standard interference contour. (D/U=+10)

13 VHF Adjacent TIA/EIA Measurements Analog FM WIDE Adjacent Interference, dbm Level SINAD Wide Narrow NXDN SINAD ) Per TIA/EIA measurement procedure, Adj Ch modulation is 400 Hz. This makes the signal bandwidth narrow and the allowable interfering signal higher than could be expected in actual voice usage. This does not affect NXDN signals, as their spectrum is independent of modulation. 2) NXDN creates LESS adjacent channel interference than wide or narrow FM. Analog FM NARROW +6.25kHz Adjacent Interference, dbm Level SINAD Narrow Wide NXDN SINAD ) Per TIA/EIA measurement procedure, Adj Ch modulation is 400 Hz. This makes the signal bandwidth narrow and the allowable interfering signal higher than could be expected in actual voice usage. This does not affect NXDN signals as their spectrum is independent of modulation. 2) NXDN creates LESS adjacent channel interference than wide FM and about the same as narrow FM. NXDN +6.25kHz Adjacent Interference, dbm Level BER, % NXDN Wide Narrow -117 < < < < < ) BER used as quality measure for NXDN instead of SINAD 2) NXDN D/U ratio 36dB better than Narrow FM /w 400Hz modulation at -100dBm desired signal level.

14 VHF Narrowband Adjacent TIA/EIA Measurements Analog Narrow FM +12.5kHz Adjacent Interference, dbm Level SINAD Narrow SINAD ) Per TIA/EIA measurement procedure, Adj Ch modulation is 400 Hz. This makes the signal bandwidth narrow and the interfering signal higher than could be expected in actual voice usage. This does not affect NXDN signals as their spectrum is independent of modulation. 2) D/U ratio for Narrow FM w/ 400Hz modulation at 12.5kHz spacing is same as the NXDN D/U ratio at 6.25kHz spacing in the previous table. VHF Splinter TIA/EIA Measurements NXDN kHz Adjacent Interference, dbm Level BER, % NXDN Wide Narrow -117 < < < < < ) Splinter data included for completeness.

15 UHF Adjacent TIA/EIA Measurements Analog FM WIDE +7.5kHz Adjacent Interference, dbm Level SINAD Wide Narrow NXDN SINAD ) Per TIA/EIA measurement procedure, Adjacent modulation is 400 Hz. This makes the signal bandwidth narrow and the interfering signal higher than could be expected in actual voice usage. This does not affect NXDN signals, as their spectrum is independent of modulation. Analog FM NARROW +7.5kHz Adjacent Interference, dbm Level SINAD Narrow Wide NXDN SINAD ) Per TIA/EIA measurement procedure, Adjacent modulation is 400 Hz. This makes the signal bandwidth narrow and the interfering signal higher than could be expected in actual voice usage. This does not affect NXDN signals, as their spectrum is independent of modulation. NXDN +7.5kHz Adjacent Interference, dbm Level BER, % NXDN Wide Narrow -118 < < < < < ) BER used as quality measure for NXDN instead of SINAD 2) NXDN to NXDN D/U ratio is about 16dB better than NFM to NFM D/U ratio from previous table at -100dBm desired signal level. 3) NXDN sensitivity is about 1 db less than Narrow FM w/ 400 Hz modulation. (Same sensitivity within measurement tolerance.)

16 Appendix D. Maximum Distortion Measurement Method Test Results Co- Max Distortion Analog FM WIDE Co- Interference, dbm Level SINAD Wide Narrow NXDN ) The channel requires the same degree of protection against all three kinds of co-channel interference. Analog FM NARROW On Signal Co- Interference, dbm Level, dbm SINAD, db wide narrow digital ) The channel requires the same degree of protection against all three kinds of co-channel interference. NXDN Co- Interference, dbm Level BER, % NXDN Wide Narrow -118 < < < < < < ) The channel requires the same degree of protection against all three kinds of co-channel interference.

17 VHF Adjacent Max Distortion Analog FM WIDE Adjacent Interference, dbm Level SINAD Wide Narrow NXDN SINAD ) Because of the FM Capture Effect and the channel spacing, for WFM to WFM D/U=0 Analog FM NARROW +6.25kHz Adjacent Interference, dbm Level SINAD Narrow Wide NXDN SINAD NXDN +6.25kHz Adjacent Interference, dbm Level BER, % NXDN Wide Narrow -117 < < < < <

18 UHF Adjacent Max Distortion Analog FM WIDE +7.5 Adjacent Interference, dbm Level SINAD Wide Narrow NXDN SINAD Analog FM NARROW +7.5kHz Adjacent Interference, dbm Level SINAD Narrow Wide NXDN SINAD NXDN +7.5kHz Adjacent Interference, dbm Level BER, % NXDN Wide Narrow -118 < < < < <

19 Appendix E. Comparison with Theoretical Calculations To further understand the high levels of adjacent channel interference presented in the measurements above, the spectrum of a transmitter was measured and is presented in Photograph E.1. Correction factors are then estimated based on these actual spectra. Photograph E kHz Narrow FM Transmitter Spectra Spectra 1) The yellow trace is the unmodulated transmitter spectrum. This was used to determine the power level for the FCC mask passband. 2) The green trace is the Narrow (12.5kHz) FM transmitter modulated by 400Hz sine wave and 60% frequency deviation. (TIA/EIA standard for adjacent and co-channel interference measurements and the modulation used in this report.) 3) The blue trace is an averaged spectrum of the Narrow (12.5kHz) FM transmitter modulated by a male voice. 4) The red trace is the FCC D mask (12.5kHz) centered on the desired channel. 5) In this photo, a 6.25kHz adjacent channel is 2 divisions away and a 12.5kHz adjacent channel is 4 divisions away. At least for the radios used in this test, we can see that under all test conditions, the actual spectrum is much narrower than the FCC D mask and the adjacent power needs to be estimated by other techniques than using the FCC mask. From the NFM data in Appendix C, we see that for 6.25kHz channel spacing the interference power necessary to degrade the SINAD by 3dB has increased 30 db over the co-channel case. Studying Photograph E.1 we can see that for 6.25kHz channel spacing, a 12.5kHz adjacent channel mask would intersect the TIA/EIA green trace and the voice blue trace at about -10dB. As this

20 interference is asymmetrical about the adjacent channel, it is best represented as an equal combination of AM and FM interference, which reduces the level of the FM interference by 6dB. Finally, the only significant interference power is at the edge of the adjacent channel. We can convert this to average power in the adjacent channel by using the ratio of the spectrum analyzer RBW to the channel width. (11kHz to 300Hz) In total this gives about a 30dB correction factor for two adjacent 12.5kHz channels spaced 6.25kHz apart. The same argument applies to 7.5kHz spaced channels with about 10dB greater reduction. correction factor) (40dB By the time we move two NFM signals 12.5kHz apart, the 400Hz modulation interference is relatively flat throughout the desired channel mask and no correction factors are needed. However, when voice modulation is used, at the band edge, the interference is about 30dB higher, so a 50dB correction would have to be applied for actual voice transmissions. Table E.1: Interfering Signal in db above D/U= +18dB, Adjacent FM s Adjusted Interfering Radio Target Receiver Step 6.25 Step 7.5 Co- Step 12.5 Analog 12.5 Analog The above somewhat loose results indicate that NFM signals spaced 6.25kHz apart should be treated as co-channel interference and when spaced 12.5kHz apart, a +14dB allowance could be made (D/U=0). Note that current standards call for +12 db allowance for 12.5kHz spacing NFM. The above presentation should certainly not be taken as a rigorous argument for adjusting the data presented in the main body of this report. It is presented to correlate the data in this report with other results and experience.