1、INTERNATIONAL TELECOMMUNICATION UNION)45G134 TELECOMMUNICATIONSTANDARDIZATION SECTOROF ITU).4%2.!4)/.!,G0G0!.!,/5%G0G0#!22)%2G0G03934%-3%.%2!,G0G0#(!2!#4%2)34)#3G0G0#/-/.G0G04/G0G0!,!.!,/5%G0G0#!22)%2G1342!.3-)33)/.G0G03934%-3-%!352%-%.4G0G0/ further amended)The CCITT,considering that(a) it is desir
2、able to measure the performance of cable systems for frequency-division multiplex telephonyunder conditions closely approaching those of actual operation;(b) a signal with a continuous uniform spectrum (white noise) has statistical properties similar to those of amultiplex signal when the number of
3、channels is not too small;(c) the use of a signal with a continuous uniform spectrum to measure the performance of such cable systemsis already widespread;(d) it is necessary to standardize the frequencies and bandwidths of the measuring channels to be used forsuch measurements;(e) for reasons of in
4、ternational compatibility it is necessary to standardize the minimum attenuation and thebandwidth of the stop filters which may have to be used in the white-noise generator;(f) the CCITT has indicated, for the planning of telephone circuits, a mean value of signal power in thebaseband of a multiplex
5、 telephone system to be taken into consideration during the busy hour(Recommendation G.223),recommends that1 The performance of frequency-division multiplex cable systems should be measured by means of a signalwith a continuous uniform spectrum in the frequency band used for the telephone channels.2
6、 The nominal power level of the uniform spectrum test signal should be in accordance with the conventionalload, specified in Recommendation G.223. If applied at the point of interconnection of the system corresponding to Tof Recommendation G.213, the absolute power levels of interest are shown in co
7、lumn 4 of Table 1/G.228.2.1 The sending equipment should be capable of providing, at the output of an inserted bandstop filter, a loadinglevel at least up to + 10 dB relative to the nominal power level defined above.2.2 Within the bandwidth corresponding to the baseband of the system under test, the
8、 r.m.s. voltage of the whitenoise spectrum measured in a band of about 2 kHz should not vary by more than 0.5 dB. This degree of spectrumuniformity should be met in the level range up to +6 dB relative to the nominal power level, indicated inTable 1/G.228, column 4.2.3 The white noise test signal sh
9、ould be available at the output of the sending equipment with a peak factor ofabout 12 dB with respect to the r.m.s. value.3 The nominal effective cut-off frequencies (the cut-off frequencies of hypothetical filters having ideal squarecut-off characteristics and transmitting the same power as the re
10、al filters) and tolerances for the bandpass filtersproposed for the various bandwidths of systems to be tested, should be as specified in Table 2/G.228. To reduce thenumber of filters required, compromises have been made between the nominal effective cut-off frequency and thesystem bandwidth-limitin
11、g frequency in some cases. The tolerances ensure that consequent calibration errors do notexceed 0.1 dB and errors in measurement of intermodulation noise do not exceed 0.2 dB assuming system pre-emphasis of about 10 dB.2 Fascicle III.2 - Rec. G.228TABLE 1/G.228123 4Numberof telephonechannelsRelativ
12、e powerlevel at point T(dBr)Level of theconventional load(dBm0)Nominal power levelof the test signal at point T(dBm)60 -36 6.1 -29.9120 -36 7.3 - 28.7300 -36 9.8 -26.2600 -36-3312.8 -23.2-20.2960 -36-3314.8 -21.2-18.21 260 -33 16.0 -17.01 800 -33 17.5 -15.52700 -33 19.3 -13.73 600 -33 20.5 -12.510 8
13、00 -33 25.3 -7.7TABLE 2/G.228Fascicle III.2 - Rec. G.228 33.1 The discrimination of a lowpass filter should be at least 20 dB at a frequency more than 10% above nominalcut-off and at least 25 dB at frequencies more than 20% above nominal cut-off. The discrimination of a highpass filtershould be at l
14、east 25 dB at frequencies more than 20% below nominal cut-off.3.2 To limit discrimination against measuring channels, the spread of losses introduced by any pair of highpassand lowpass filters should not exceed 0.2 dB over a range of frequencies which includes the upper and lowermeasuring channels.4
15、 Values of the characteristics for the discrimination in each stop-band at the output of a sending equipment aregiven in Table 3/G.228. These characteristics are intended to apply over a temperature range from 10 C to 40 C.5 When the receiving equipment is connected directly to a sending equipment p
16、rovided with bandstop filterswhich only just meet the requirements of 4 above, the ratio of the noise power indicated by the receiving equipmentwhen the bandstop filter is bypassed, to that indicated when the filter is in circuit, should be a minimum of 67 dB; thisrequirement applies when a conventi
17、onal load is applied. The minimum effective bandwidth of the receiver should be1.7 kHz; the maximum reading of absolute noise power arising from leakage given by a receiver of 1.74 kHz effectivebandwidth and which just meets the foregoing leakage requirement is -85.6 dBm0p.6 Additional measuring cha
18、nnels may be provided by agreement between the Administrations concerned.Note - In Annexes A and B some general information is given on the measuring procedures, the choice offilter characteristics, correction methods and accuracy objectives.TABLE 3/G.228Characteristics of bandstop filtersCentrefreq
19、uency f0(kHz)Bandwidth (kHz), in relation to f0,over which the discrimination should be at least:Bandwidth (kHz)in relation to f0, outsideof which the discriminationshould not exceed:70 dB 55 dB 30 dB 3 dB 3 dB 0.5 dB702705347701 2482 438 1.5 1.5 1.5 1.5 1.5 1.5 2.2 2.3 3.5 3.8 4.0 4.5 15.0 3.5 2.9
20、7.0 8.0 11.0 19.0 30.0 12 8 15 21 35 60 110 18 24 48 70 110 220 3503 8865 3407 60011 70016 40035 74855 548 1.5 1.5 1.5 1.5 1.5 1.8 2.5 1.8 2.2 2.4 3.0 4.0 2.2 3.5 3.5 4.0 4.6 7.0 9.0 3.5 6.0 8.0 8.5 9.5 11.0 14 5.0 9.5 12 14 16 20 30 20 30 100 150 200 300 500 150 200Note 1 - The characteristics reco
21、mmended for the filters 70 kHz to 2438 kHz inclusive are based on coil-capacitor type filters.Those characteristics recommended for the filters at 5340 kHz and above are based on crystal-type filters. Optionalcharacteristics are recommended for the 3886-kHz filter to permit a choice of design betwee
22、n a coil-capacitor type or crystal-type filter. The crystal-type filters at 35 748 kHz and 55 548 kHz are assumed to be operating in a higher harmonic mode of thecrystal resonators. That is why the relative bandwidths of these filters show a discontinuity compared with the crystal-typefilters up to
23、11 700 kHz.4 Fascicle III.2 - Rec. G.228Note 2 - The discrimination values quoted are relative referred to the minimum attenuation of the bandstop filters within thebaseband frequency range defined by highpass and lowpass filters in Table 2/G.228. This implies that a bandstop filter suitablefor meas
24、urements on one system is not necessarily suitable for measurements on a system of larger bandwidth.Note 3 - The design of the receiver selectivity of 3886 kHz should be related to the characteristic of the crystal-type bandstopfilter.Note 4 - Due to spurious resonances, narrow spikes of the discrim
25、ination may occur in the upper passband of crystal-type bandstopfilters. When resonators are operated in a higher harmonic mode narrow spikes can also appear in the lower passband. Thosespikes of about 10-dB peak attenuation within 1 to 5 kHz bandwidth are admissible because they do not affect the m
26、easuringaccuracy.ANNEX A(to Recommendation G.228)Outline of the white noise measuring methodA.1 General principleThe principal components of the measuring setup are shown in Figure A-1/G.228.FIGURE A-1/G.228Principle of the measuring set-upA.2 Measuring proceduresTwo methods for assessing the noise
27、performance of a transmission system are in widespread use:Fascicle III.2 - Rec. G.228 5A.2.1 Measurement of noise power ratio (NPR)The noise power ratio(A-1)is measured at various levels of Ps. The r.m.s. level meter serves as an indicator only. The value WAis the noise powerin the measuring channe
28、l without taking account of the effect of frequency gaps between groups of channels in actualoperation.In an N-channel system the following definitions are introduced:Ps= N PCHPCH= variable signal power per channelpCH= -15 dBm0 + p = load level per channel-15 dBm0 is the conventional load per channe
29、l according to Recommendation G.223 for systems withN 240 p (dB) is the excess load relative to -15 dBm0pn= weighted noise power level (dBm0p) measured at point T in a 3.1 kHz telephone channel.The measured NPR values are usually plotted, as shown in Figure A-2/G.228, as a function of the excesschan
30、nel loading p.FIGURE A-2/G.228NPR curve versus channel loading6 Fascicle III.2 - Rec. G.228The relation between NPR values measured on a channel and the weighted noise power level referred to a zerorelative level point is:pn= (- NPR - 18.6 - 10 log k + p) dBm0p (A-2)k = B/4N(B in kHz) is a correctio
31、n factor which takes account of the effect of the frequency gaps betweengroups of channels in the transmission system.Table A-1/G.228 gives examples of the correction for some N-channel systems:TABLE A-1/G.228N 300 960 2700 10 80010 log k(dB)0.14 0.22 0.46 1.08A.2.2 Direct measurement of weighted no
32、ise power levelWith the particular choice of the effective receiver bandwidthf = 1.74 kHz (= 3.1 kHz .10-0.25),the weighted noise power Pnin a telephone channel is:Pn= WB(see Figure A-1/G.228)and the weighted noise level pn, referred to a point of zero relative level becomes:In this case the receive
33、r (component 7 of Figure A-1/G.228) must be a calibrated power level meter.A.3 Examples of investigations using the white noise measuring methodTwo kinds of investigations can be made on a system (with length L) between flat relative level points T andT. The one case a) investigates the effect on th
34、e noise performance of load deviations at the input of the system,whereas the other case b) indicates the influence of level misalignments along the transmission line:a) The test signal noise power PSis varied and the weighted noise level pnis determined in dBm0p. Theresult is plotted as indicated i
35、n Figure A-3/G.228.Alternatively to the indication of the noise level for system length L in dBm0p, the noise power couldhave been indicated in pW0p/km.Fascicle III.2 - Rec. G.228 7FIGURE A-3/G.228Weighted noise power level versus system load (relative levels of the system fixed according to plan)b)
36、 The relative levels on the transmission line are varied by insertion of attenuators -n and +n at the inputand output of the system as is illustrated in Figure A-4/G.228 which is an excerpt of Figure A-l/G.228.FIGURE A-4/G.228Variation of the internal relative levels of the systemThe test signal noi
37、se power Psis set to the conventional value (-15 dBm0/4 kHz) at point T and is keptconstant. The noise power level in the measuring channel is determined at point T as a function of the relative level atthe repeater output, for example. The result is plotted as shown in Figure A-5/G.228.8 Fascicle I
38、II.2 - Rec. G.228FIGURE A-5/G.228Weighted noise level in measuring channel versus relative level at repeater outputANNEX B(to Recommendation G.228)Measuring accuracy considerations affectingthe design of the measuring equipmentB.1 IntroductionThe Recommendations relating to the measurement of circui
39、t noise in systems artificially loaded with uniformspectrum random noise simulating FDM telephone signals were agreed after carefully coordinated studies by three CCIStudy Groups concerned. The different Recommendations provided for the application of the white noise measuringmethod to cable systems
40、 (CCITT Recommendation G.228), radio-relay systems (CCIR Recommendation 399 1),satellite systems (CCIR Recommendation 482 2) and translating equipments (CCITT Recommendation G.230). Theobjective of the coordination was that the separately recommended measuring equipments should conform withcommon me
41、asuring accuracy objectives and, as far as possible, be compatible and interchangeable.The overall accuracy objective of the measuring equipment when used for routine maintenance measurementsis 2 dB. A higher accuracy of about 1 dB is desirable when measurements are made for the purpose of assessing
42、 thenoise performance of a system in relation to required performance. This can be achieved by following certainprocedures and applying corrections as described in B.4 and B.5 below.This Annex states how certain characteristics of measuring equipments were related to measuring accuracyobjectives; an
43、y future extensions of the Recommendations to provide for measurements on new transmission systems,as yet unstandardized, should take account of those relationships.B.2 Bandstop fillersB.2.1 Choice of centre frequenciesIn all cases the choice of nominal centre frequencies of band-elimination filters
44、 (i.e. of measuring channels)should take account of the need to minimize the combined discrimination of the pair of bandpass filters used when theFascicle III.2 - Rec. G.228 9bandstop filter provides a lower or upper measuring channel. Therefore, as a rule the centre frequency of a lowermeasuring ch
45、annel should be at least 15% above the effective cut-off frequency of the highpass filter and the centrefrequency of an upper measuring channel should be more than approximately 5% below the cut-off frequency of thelowpass filter involved. Under 3.2 of the text of this Recommendation it is prescribe
46、d that “the spread of lossesintroduced by any pair of highpass and lowpass filters should not exceed 0.2 dB over a range of frequencies whichincludes the outer measuring channels“.B.2.2 LeakageThe discrimination of a bandstop filter in the neighbourhood of the centre frequency determines, jointly wi
47、ththe receiver selectivity the smallest noise-to-signal ratio that can be measured accurately, i.e. the “leakage“ effect. Thebandstop filter discrimination of 70 dB (Table 3/G.228) results in a ratio of the order of -67 dB-being measured whenthe noise is actually negligible. Leakage effect in the re
48、ceiver is adequately limited by requiring (see 5 in the text ofthe Recommendation) that the NPR value should be a minimum of 67 dB when connected directly to a send equipmentwith bandstop filters which only just meet the discrimination requirements of Table 3/G.228 and when a conventionalload of -15
49、 dBm0/4 kHz is applied.Note - According to Formula (A-2) of Annex A this value of NPR = 67 dB corresponds to a residual noiselevel of -85.6 dBm0p (i.e. 2.8 pW0p) at the most.B.2.3 Effective bandwidthThe basic requirement for the stopband is the condition that the discrimination should be at least 70 dB in abandwidth of at least 3 kHz. The effective bandwidths (
