10th WSEAS Int. Conf. on MATHEMATICAL METHODS AND COMPUTATIONAL TECHNIQUES IN ELECTRICAL ENGINEERING (MMACTEE'08), Sofia, Bulgaria, May 2-4, 2008 A Novel Correlation Pulse Echo Methodology for Transmission Line Fault Identification and Location using Pseudorandom Binary Sequences Richard A. Guinee , MIEEE Cork Institute of Technology, Cork, IRELAND Abstract- A novel pulse echo test methodology, using pseudorandom binary sequence (PRBS) excitation, is presented in this paper as an alternative to Time Domain Reflectometry (TDR) for transmission line fault location and identification. The essential feature of this scheme is the cross correlation (CCR) of the fault response echo with the PRBS test input stimulus input which results in a unique signature for identification of the fault type, if any, or load termination present as well as its distance from the point of test stimulus injection. This fault identification method can used in a number of key industrial applications incorporating printed circuit boards, overhead transmission lines and underground cables in inaccessible locations which rely on a pathway for power transfer or signal propagation. As an improved method PRBS fault identification can be performed over several cycles at low amplitude levels online to reject normal signal traffic and extraneous noise pickup for the purpose of multiple fault coverage, resolution and identification. In this paper a high frequency co-axial transmission line model is presented for transmission line behavioural simulation with PRBS stimulus injection under known load terminations to mimic fault conditions encountered in practice for proof of concept. Simulation results, for known resistive fault terminations, with measured CCR response demonstrate the effectiveness of the PRBS test method in fault type identification and location. Key experimental test results are also presented for a co-axial cable, under laboratory controlled conditions, which substantiates the accuracy of PRBS diagnostic CCR method of fault recognition and location using a range of resistive fault terminations. The accuracy of the method is further validated through theoretical calculation via known co-axial cable parameters, fault resistance terminations and link distances in transmission line experimental testing. I. correlation of the input PRBS stimulus with the fault echo response, over several PRBS cycles in averaging out noise pickup and normal signal traffic in the online mode, to identify the nature or characteristic signature of the fault. Based on the distinct spike-like attribute of the PRBS autocorrelation (ACR) function, fault location measurement relies on the time displacement of the transmission link conditioned PRBS cross-correlated echo response from the ACR peak for accurate (CCR) fault/load positioning and identification. This measured time translation of the correlation peaks can be subsequently used to determine the propagation delay of the reflected response from the fault/load-termination of the unit under test (UUT). This procedure not only results in fault/load parameter identification but also of the reflection transit time from the fault interface and thus the distance of the fault from the point of stimulus insertion. In this paper the novel application with experimental validation of the PRBS perturbation method of transmission line testing for fault identification with location is presented. A simplified distributed model is employed to simulate transmission line behaviour [10,11] for a range of known load terminations mimicking fault conditions encountered in practice for comparison purposes with experimental results. Essential test results, which were not available for [10], are presented here for the first time for experimental verification of the accuracy of the PRBS diagnostic method of fault identification and location in co- INTRODUCTION Transmission line testing and fault finding is a mature and well established test and measurement trouble-shooting strategy deployed in the telecommunications and electrical power utilities industries to guarantee quality of service to telephone subscribers and a stable electrical energy supply to consumers. Time domain reflectometry (TDR) as a test strategy [1,2] has been at the forefront of this discipline for many years in the development of high frequency test and measurement network analysers for a range of HF industrial applications encompassing IC [3] and PCB design layout [4] for minimization of track parasitics. However if has the disadvantage [1,2] of relying on a single pulse echo for transmission link fault finding that is susceptible to measurement inaccuracy resulting from link attenuation with fault distance and phase change distortion with frequency as well as resolution error in the presence of noise pickup. The alternative improved strategy proposed in this paper is the application of a bipolar PRBS stimulus, which consists of a random time arrangement of pulses, resulting in a sustained pulse echo sequence with an accumulated CCR response build-up, which overcomes the difficulty of uncorrelated link noise, at low pulse energy-to-noise ratio conditions, in fault location measurement. This stochastic method, which is a well known system identification tool [5,6,7] in optimal control, can also be used via cross ISBN: 978-960-6766-60-2 155 ISSN: 1790-5117 10th WSEAS Int. Conf. on MATHEMATICAL METHODS AND COMPUTATIONAL TECHNIQUES IN ELECTRICAL ENGINEERING (MMACTEE'08), Sofia, Bulgaria, May 2-4, 2008 A transmission line can be modelled by a lumped parameter ‘T-section’ unit equivalent circuit in Fig. 1 at low frequencies (LF) [12,13] with a uniformly distributed series inductance L, resistance R and a shunt capacitance C in parallel with a conductance G over elemental distance dx. The T-section can be suitably scaled to the required dimensions axial cable testing under laboratory controlled conditions using a range of known resistive fault terminations. Further validation and substantiation of accuracy of the method is provided through theoretical calculation via known co-axial cable parameters, fault resistance terminations and link distances in transmission line experimental testing. II. OVERVIEW OF TDR AND PRBS TESTING R dx 2 TDR is based on single pulse propagation down a cable UUT such that when it reaches the cable fault some of the pulse energy is reflected back to the TDR test instrument. Since the propagation velocity is assumed constant for UUT the pulse transit time is a measure of the fault distance. TDR displays the fault information as a waveform and distance reading. This test technique is not a perfect fault location method because the transmitted pulse is progressively broadened and made less sharp as it propagated down the line as a result of pulse distortion due to phase change. Pulse resolution is essential in TDR where narrow pulses give rise to very sharp trace features that are ideal for measurement. Narrow pulses are easily attenuated with signal path frequency response rolloff, due to reciprocity of pulsewidth with bandwidth [2], which causes reflected pulse amplitude definition loss and as such narrow pulse TDR is useful over short distances only. Alternative TDR wide pulse usage produces wider and more rounded echo trace features with leading edge transitions that are difficult to gauge and lead to inaccurate fault distance resolution. These wide pulse stimuli are, however, not so quickly attenuated which make them suitable for long distance measurements. The TDR technique is susceptible to link noise interference, which can mask out weak long distance fault reflections [1], with resulting pulse definition loss for accurate fault location measurement. Pulse stimuli can be sent repeatedly but they are uncorrelated and the fault information is contained in unconnected pulse echoes which renders repeated pulse injection unproductive in noisy link fault measurements. The alternative PRBS line fault location method employs a random series of bipolar pulses that are reflected by the impedance mismatch as a correlated response build-up over the entire test sequence. CCR evaluation of the fault response with the incident PRBS and comparison of its time displaced peak with that for the ACR peak can identify the fault location. The echo CCR amplitude profile yields a characteristic signature, which identifies the type of fault present. Link noise can be averaged out through CCR evaluation over multiple PRBS cycles [6,7] which accentuates the fault signature. This ‘magnified’ fault characteristic can be built up over a number of PRBS lengths, dramatically reducing the impact of noise. L dx 2 R dx 2 L dx 2 G dx 2 C dx 2 dx Fig.1: Full Model Lumped Parameter ‘T’ - Section for overhead line and underground cable modelling and PCB operation at high frequency (HF). The T-section can be simplified for transmission line operation at high frequency ω with the elimination of the distributed resistance and conductance in accordance with the condition [12] ωL>>R and ωC>>G (1) This reduced ‘T’ model is simulated below and compared with experimental results for HF co-axial cable operation for various types of line faults, known apriori, with PRBS injection in order to gauge the accuracy of the CCR process in identifying the type of fault and its location. A. Full Model Example for Low Frequency Line Operation The T-section model in Fig.1 is first exercised with lumped parameter values per-loop-km of R = 10.15Ω, L = 3.93mH, G = 0.29uS, C = 0.00797uF at a line frequency of 5000rad/sec. Longer line lengths can be obtained by chaining several sections in tandem. The model characteristic impedance Zo and complex propagation coefficient γ are given by [12] Z0 = ( R + j ωL ) ( G + j ωC ) (2) γ = ( R + jωL)(G + jωC ) = α + jβ (3) The propagation coefficient for this line is with attenuation coefficient α in Nepers/km and phase-change coefficient β in rads/km. The propagation velocity is determined from the line frequency 5000 rads/sec and phase-change coefficient as vp = f ⋅(2π/β) = (ω/β) =173611 km/s γ = α + jβ = 0.00712 + j 0.0288 III. TRANSMISSION LINE ‘T-SECTION’ MODELLING FOR FAULT SIMULATION MEASUREMENT ISBN: 978-960-6766-60-2 156 ISSN: 1790-5117 10th WSEAS Int. Conf. on MATHEMATICAL METHODS AND COMPUTATIONAL TECHNIQUES IN ELECTRICAL ENGINEERING (MMACTEE'08), Sofia, Bulgaria, May 2-4, 2008 The pseudonoise (pN) stimulus has a unique delta function like ACR defined for a time shift τ = kΔt with 0 ≤ k ≤ (L-1) over one period T, as per Fig.3, by 1.0 0.8 0.6 PRBS_ACR =1 B. Reduced Model Example for High Frequency Line Operation PRBS_ACR Stimulus Reference Peak ACR =0.5 0.4 0.2 Repeat PRBS ACR Peak Transit Delay ττl 0.0 0.2 V. FAULT FINDING WITH PSEUDONOISE SEQUENCES PRBS Period T = 6.25μs When a transmission line is mis-terminated in a load impedance ZL ≠Z0 an echo reflection Y(t) will be present on the line at any point along with the incident wave X(t). The worst cases of line fault mismatch occur for (i) a short circuit fault with ZL = 0 and (ii) an open circuit fault with ZL=∞. Besides these well known line faults other types of partial line discontinuities result in reflections caused by [1] joints, splits and waterlogged zones, which are all characterised as minor mismatches. The degree of reflection at the load termination ZL ≠Z0 can be determined from the incident wave by the reflection coefficient ρ [12] as Time μs 0.26 0.24 0.28 Fig.2: Open Circuit Line – CCR Fault Signature For HF ‘lossless line’ operation in (1) with α=0 the real characteristic impedance Z0 and propagation coefficientγ, as the phase change coefficient β, are given respectively by Z0 = L C (4) γ = jβ = jω LC (5) If for example, a 100m HF line operating at 20MHz based on the pSpice distributed ‘lossless’ model [11] with transmission wavelength λ=5.1862m and Z0 = 30Ω is simulated for open circuit conditions with a PRBS stimulus the CCR echo response along with the PRBS ACR are shown in Fig.2. The propagation velocity is calculated from vp = λ×f as 103724 km/s. 1.2 Z − Zo ρ= L Z L + Zo (7) The resulting voltage standing wave ratio (VSWR) s is determined from ρ, with −1 ≤ ρ ≤ 1 , as s= V2 0.8 0.6 0.4 - V2/L T = LΔt Delay τ = kΔt 0.2 0 -4 -2 0 -0.2 2 4 6 8 10 Δt Fig. 3: PN Autocorrelation Function (ACR) IV. TRANSMISSION LINE TESTING USING PRBS PRBS stimuli x(t), employed in transmission line testing, change logic state pseudo randomly between prescribed voltage levels +V and –V at discrete time intervals Δt. The bipolar test signal is generated from a specially configured n stage linear feedback shift register [8,14] and has a maximum sequence length L =2n-1 with period T=LΔt. ISBN: 978-960-6766-60-2 1+ ρ 1− ρ (8) Four general cases of line termination can arise to influence the values [12,13] of ρ and s which will provide an indication of the type of line fault present: (i) Matched load conditions with Zo = ZL: ⇒ ρ = 0 with no reflection and s =1. (ii) Open-circuit line ZL = ∞: complete incident wave reflection occurs without phase reversal ⇒ ρ = 1 as per Fig.2 (iii) Short-circuit line ZL = 0: complete incident wave reflection occurs with phase reversal ⇒ ρ = - 1 as per Fig.4 (iv) Mismatch termination ZL ≠ Z0: incident wave reflection occurs with or without phase reversal depending on the relative sizes of ZL and Z0. (1) If ZL < Z0 ⇒ ρ < 0 and s = Z0/ZL. (2) If ZL > Z0 ⇒ ρ > 0 and s = ZL /Z0. The phase relationship between the reflection echo response and the incident PRBS, determined through ρ in the PRBS-CCR process along with the VSWR, indicates the type of load termination present. For PRBS X(t) = {x(1), x(2), …, x(L)} injection into a faulty line a conditioned echo response will result as Y(t) = PRBS Autocorrelation Function (ACR) RXX(τ) 1 ⎧⎪+ V 2 1 L for k = 0 ∑ x( j ) x( j + k ) = ⎨ 2 L j =1 ⎪⎩− V L for k ≠ 0 (6) The PRBS ACR along with the CCR fault response is used to determine the transit time delay τl and thus the fault distance l. Repeat Echo CCR Peak Reflected PRBS Echo Correlation Peak O/C Echo Response CCR 0.22 R xx (k ) = 157 ISSN: 1790-5117 10th WSEAS Int. Conf. on MATHEMATICAL METHODS AND COMPUTATIONAL TECHNIQUES IN ELECTRICAL ENGINEERING (MMACTEE'08), Sofia, Bulgaria, May 2-4, 2008 {y(1), y(2), …, y(L)} which can be cross correlated with the incident disturbance X(t) as R xy (k ) = 1 Alternatively if the 100m HF line has a s/c fault the pN input test signal will undergo phase reversal upon reflection as evidenced by the negative CCR echo response in Fig.4. Again simple calculation reveals that the relative time displacement τl of the ACR and CCR peaks provide an accurate estimate lˆ of the ‘known’ fault location l, which is identical to the o/c position in Fig.2, and its identity. Lossy line simulation demonstrates the effects of PRBS stimulus attenuation as it is propagated down the line under test. However if multi PRBS injection is used the effect of pN pulse attenuation, unlike TDR, is surmounted with multiple cycle correlation [7]. If the lossy line in § III-B is simulated for an o/c fault, at a distance l =100km, pN fault diagnosis similar to that for Fig.2 results in an ACR to CCR peak displacement or round trip propagation delay τl = 1.1507ms. Thus the fault location estimate is easily estimated, via (10), with vp= 173611km/s, as lˆ = (173611 km/s)⋅(1.151ms/2) ≈ 100km which is identical to the known value l used in simulation and as such validates the PRBS test methodology for fault detection on lossy lines. Further pN testing [10] for a range of lossy line fault impedance terminations yield estimates which are practically identical to those used in line simulation which further establishes concept validation. Lossy line simulation demonstrates the effects of PRBS stimulus attenuation as it is propagated down the line under test. However if multi PRBS injection is used the effect of pN pulse attenuation, unlike TDR, is surmounted with multiple cycle correlation [7]. If the LF lossy line in § III(b) is simulated for an o/c fault, at a distance l =100km, pN fault diagnosis similar to that for Fig.2 results in an ACR to CCR peak displacement or round trip propagation delay τl = 1.1507ms. Thus the fault location estimate lˆ is easily estimated, via (10), with vp= 173611km/s, as lˆ = (173611 km/s)⋅(1.151ms/2) ≈ 100km which is identical to the known value l used in simulation and as such validates the PRBS test methodology for fault detection on lossy lines. Further pN testing [10] for a range of lossy line fault impedance terminations yield estimates which are practically identical to those used in line simulation which further establishes concept validation L ∑ x(i) y (i + k ) L i =1 (9) to yield a characteristic CCR signature of the line fault present. The CCR process yields a correlation peak at some shift time τl, as per Fig.2 for an open circuit fault, which is indicative of the line fault distance l from the test stimulus input X(t). The time displacement τl of CCR peak is measured from incident PRBS reference ACR peak as per Fig.2 which when divided by two and multiplied by the line propagation velocity vp will give the fault distance l to the source of reflection as l = vp⋅τl /2 (10) VI. SIMULATED HF AND LF LINE FAULT DIAGNOSIS The HF model simulation in §III–B for open circuit (o/c) and short circuit (s/c) termination faults employed a 127 bit PRBS input with frequency f PRBS =1/Δt=20Mhz, as the reciprocal of the chip time Δt, and period T = L/fPRBS = LΔt = 6.35μs. The ACR reference peaks with period T, as shown in Figs.2, 3 & 4, are generated via the autocorrelation of the incident PRBS at the stimulus source. For an o/c line fault the incident PRBS stimulus is reflected back to the source without phase reversal. If the combined incident and reflected components are cross-correlated with the incident PRBS the resultant non inverted CCR peak indicates the presence of a open circuit fault ‘echo’ via (9) as the line fault signature along with the reference ACR in Fig.2. The relative displacement τl ≈ 1.9μs of the CCR and ACR peaks results in the propagation delay for the PRBS stimulus to traverse the line from the input to the fault and back with a total distance 2l. For the 100m open circuit HF line, known apriori, the measured ACR-CCR displacement τl in Fig.2 along with the known phase velocity vp in § III – (b) provides an accurate gauge of the fault location l and its identity as l = (τl/2)*vp = 0.96μs*103724km/s = 100m. Hence knowledge of the CCR peak displacement τl and link propagation velocity vp is all that is needed for accurate estimation of the fault location l. VII. EXPERIMENTAL HF LINE FAULT DIAGNOSIS A 50m roll and 500m drum of URM-43 HF co-axial cable with Z0 = 50Ω and distributed capacitance C = 100pF/m was tested using PRBS stimuli for various ‘known’ discontinuities apriori in order to experimentally validate the fault location and identification capability of the PRBS CCR technique. Assuming lossless HF line behaviour, at the pN test frequency fPRBS =100MHz, the other line parameters can be derived for fault location estimation during test. Using (4) for HF line operation the distributed inductance L can be determined as L=CZ02 = 0.25 μH/m along with the line propagation velocity as v p = 1 LC = 1 Z 0C =2×108m/s. This value of vp can be used with the echo fault response transit time τl /2, back to PRBS Autocorrelation PRBS – ACR =1 1.0 0.5 ACR Reference Peak due to Incident PRBS ACR = 0.5 τl =1.9282us 0.0 Cross Correlation Peak due to Fault Reflected PRBS S/C Response CCR -0.5 0.25 0.26 0.27 Time (μs) 0.28 Fig.4: Short Circuit Line – CCR Fault Signature ISBN: 978-960-6766-60-2 158 ISSN: 1790-5117 10th WSEAS Int. Conf. on MATHEMATICAL METHODS AND COMPUTATIONAL TECHNIQUES IN ELECTRICAL ENGINEERING (MMACTEE'08), Sofia, Bulgaria, May 2-4, 2008 line input test point, to determine the fault location distance as per (10). A. PRBS Stimulus Characteristics and Fault Distance Resolution 1.00E+00 Experimental PRBS - CCR Fault Diagnosis 8.00E-01 Incident PRBS ACR Reference Peak 50 m HF Co-Axial Cable Z0 = 50Ω 6.00E-01 4.00E-01 63 Bit Bipolar PRBS Stimulus Injection X(t) S/C CCR Response 2.00E-01 τl =506.5 ns PRBS Amplitude (Volts) 0.00E+00 1 76 151 226 301 376 451 526 601 676 751 826 901 976 1051 1126 1201 1276 1351 -2.00E-01 PRBS S/C Fault CCR Peak Time (× 0.5 ns) -4.00E-01 Fig.7: Experimentally derived S/C Fault Signature HP Test Pattern Generator O/P: 100 MBPS ⇒ Chip Time Δt = 10ns A 50m reel of co-axial cable was initially terminated under o/c and s/c conditions to gauge the accuracy of the PRBS CCR method of fault identification and location as depicted in Figs. 6 & 7 respectively. The location l of o/c and s/c faults can ascertained from the relative displacement τl =506.5ns of the CCR fault response from the ACR reference peak via (10) as l = vp⋅τl /2 = (2×108 m/s)⋅(506.5×10-9s/2) =50.65m which is the same as the measured line length used. Additional testing with resistive fault terminations in 10Ω steps beginning at 10Ω up to 100 Ω and thereafter in 100Ω steps, with little observed difference from o/c conditions, further validated the accuracy of pN–CCR method as per Figs.8 & 9 which enhances confidence in this trouble-shooting technique of fault identification. The importance of the CCR fault observations in Figs.6 & 7 is that the existence of correlated echo peaks and their polarity provides an indication of the type of line fault present. The first observation that should be made is whether or not a CCR fault peak is present other than the reference ACR spikes. Time (Secs) Fig.5: Observed Bipolar PRBS Stimulus Injection A 63 bit PRBS test pattern, shown in Fig.5, was employed at various polarities (unipolar and bipolar) and voltage levels for test stimulus injection with a 10ns bit duration Δt which coincides with the co-axial cable specified operational frequency of 100 MHz and also within HP IC Test Generator (TPG) limits. Using the bit duration Δt = 10ns with the line propagation velocity vp= 2×108 m/s the fault distance resolution accuracy Δd can be determined as Δd = vpΔt = 2m. An 8-channel Agilent mixed storage oscilloscope with sampling frequency 2GHz was used at the line i/p end for simultaneous data capture of the i/p test stimulus along with the delayed fault echo response for later post test data analysis and cross correlation signal processing. The higher sampling frequency results in an improved fault distance resolution accuracy of Δd = vpΔt/20 = 0.1m. 1.00E+00 Experimental pN - CCR Fault Analysis: ZL≠Z0 8.00E-01 B. Co-Axial Cable Test Fault Results 1.00E+00 Experimental PRBS - CCR Fault Diagnosis 8.00E-01 Incident PRBS ACR Reference Peak 6.00E-01 4.00E-01 50 m HF Co-Axial Cable Z0 = 50Ω CCR Responses: ZL> Z0 0.00E+00 1 -2.00E-01 4.00E-01 -4.00E-01 2.00E-01 τl =507.5 ns 79 157 235 313 391 469 547 625 703 781 859 937 1015 1093 1171 1249 1327 CCR Responses: ZL< Z0 Time (× 0.5 ns) Fig.8: Experimentally Determined ACR and CCR Fault Responses for Termination Conditions ZL≠Z0 0.00E+00 1 τl =506.5 ns 2.00E-01 PRBS O/C Fault CCR Peak -2.00E-01 Repeat pN – ACR Reference Peak Incident pN - ACR Reference Peak 6.00E-01 76 151 226 301 376 451 526 601 676 751 826 901 976 1051 1126 1201 1276 1351 O/C CCR Response Time (× 0.5 ns) Fig.6: Experimentally derived O/C Fault Signature ISBN: 978-960-6766-60-2 159 ISSN: 1790-5117 10th WSEAS Int. Conf. on MATHEMATICAL METHODS AND COMPUTATIONAL TECHNIQUES IN ELECTRICAL ENGINEERING (MMACTEE'08), Sofia, Bulgaria, May 2-4, 2008 4.00E-01 3.00E-01 0.6 O/C 2.00E-01 Z L > Z0 1.00E-01 0.00E+00 300Ω 200Ω 99Ω 82Ω 68Ω 30Ω Z L < Z0 -2.00E-01 -3.00E-01 Fault Termination Response CCR ZL≠Z0 -4.00E-01 150Ω 91Ω 62Ω 0 -0.2 > max R xy 50 100 150 200 250 300 350 -0.6 40Ω 20Ω Theoretical Reflection Coefficient ρ Estimated Reflection Coefficient ρ̂ -0.8 -1 10Ω -1.2 S/C Fig 10: Reflection Coefficient Experimental Estimates Relative Time (× 0.5 ns) PRBS fault diagnosis can also be used to estimate the reflection coefficient ρ for a given RL as ρˆ = max R xy CCR max R xx ACR (12) from the ratio comparison of ‘reflected’ CCR to the incident ACR peaks in (12) for each of the resistive fault RL cases in Fig.9. This information can then be used to estimate the actual resistive fault manifestation RL [13] from the expression (1+ ρˆ ) Rˆ L = Z0 (1− ρˆ ) (13) and the VSWR s, from (8), as 1 + ρˆ sˆ = 1 − ρˆ (14) Comparison of the reflection coefficient estimates ρ̂ , scaled shifted by 2.5, with those values ρ from theoretical considerations using (12) show a good fit when plotted in Fig.10 for various line fault terminations which validates the PRBS test strategy. Similar pN test fault results, which are ongoing, have been successfully obtained for a drum of HF co-axial cable of nominal length 500m. These results have returned a consistent fault location estimate lˆ =529m for various resistive terminations in 10Ω steps beginning at 10Ω up to 100 Ω using PRBS lengths L = 1023 bits at test frequencies of 100 MHz. VIII. Z L1 CONCLUSIONS In this paper the PRBS strategy of fault finding and identification on HF co-axial transmission lines has been experimentally validated for accuracy and examined as a competitive alternative to the industrial TDR standard. This novel trouble-shooting mechanism relies on the unique randomness attributes of maximal length pN sequences and their distinctive delta/spike-like autocorrelation function for faultfinding. Hence it can be deployed in the impulse response estimation of a faulty transmission line, in order to identify the fault type and its location, through cross correlation of the reflected response with input pN test (11) for RL2>RL1. Similar conclusions prevail for the converse case in Fig.9, which depicts a negative polarity change with RL<Z0 and increased absolute CCR value with reduced fault termination resistance. ISBN: 978-960-6766-60-2 0 -0.4 Fig 9: Enlarged View of pN Echo CCR Responses in Fig.8 Z L2 Fault Termination Resistance RL 0.2 If no CCR peak is present then matched conditions prevail with no line reflections for ZL=Z0. If, however, a positive peak exists then ZL>Z0 and a possible open circuit or high impedance fault is present as in Fig.8. Conversely, if a negative peak is present then ZL<Z0 and a possible short circuit or low impedance fault exists on the line as shown in Fig.8. The presence of high or low fault impedances besides o/c or s/c types, can also be deduced from the CCR of the pN echo response experimentally as in Figs.6 to 9 for a range of load terminations ZL≠Z0. Co-axial cable termination with fault impedance values from 10Ω to 100Ω, in steps of 10 Ω, and cross correlation of line fault response reveal the existence of CCR echos with peak magnitudes and polarities commensurate with the termination values as per Figs.8 & 9. All CCR peaks occur as expected at the same time shift in Figs.8 & 9, because the impedance faults are located the same distance l = 50m away from the pN source stimulus. The CCR peak amplitude varies with ZL and the degree of mismatch with Z0 in terms of the reflection coefficient ρ in (7), which is proportionally passed to the correlation peaks. If fault resistance terminations (RL1,RL2)>Z0 are employed for example with RL2>RL1 then the polarities of the CCR peaks are positive in both instances. Also a comparison of the CCR peak amplitudes illustrate increasing CCR values with fault termination RL, as per Figs.8 & 9, resulting in increased reflected echo coefficient ρ, that is, max R xy Ref. Coef. ρ 0.4 1 18 35 52 69 86 103120 137 154 171188205 222 239 256 273 290 307324 341 358 375 -1.00E-01 Reflection Coefficient ρ Vs Termination Resistance 0.8 Fault CCR for 50m Co-Axial Cable with Z0 = 50Ω 160 ISSN: 1790-5117 10th WSEAS Int. Conf. on MATHEMATICAL METHODS AND COMPUTATIONAL TECHNIQUES IN ELECTRICAL ENGINEERING (MMACTEE'08), Sofia, Bulgaria, May 2-4, 2008 stimulus. The line fault echo signature is the magnified collective response to a time arranged pN sequence of random pulse stimuli, propagated down the line towards the fault termination, and as such is the main advantage of using PRBS testing in preference to the single pulse method in TDR. This novel test strategy, which is incorporated as a BIST feature for CPU and digital IC ‘health’ conditioning monitoring and functionality in complex integrated systems, has been validated experimentally for HF co-axial transmission lines for a range of fault impedance terminations ranging from open to short circuit types. Further confidence enhancement of the method has been provided by the success in the identification of a range mismatched fault impedance terminations. ACKNOWLEDGMENT The author wishes to acknowledge research funding from the Science Foundation Ireland (SFI) - National Access Program for test equipment usage and technical support at the Tyndall Research Institute, Cork for experimental validation of the PRBS test strategy. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] B. Clegg, Underground Cable Fault Location, McGraw Hill, 1993. Time Domain Reflectometry Theory, Application Note 1304-2, Agilient Technologies, Aug. 2002, www.agilient.com. A. Grochowski, D. Bhattacharya, T.R. Viswanathan and K. Laker, “Integrated Circuit Testing for Quality Assurance in Manufacture: History, Current Status and Future Trends”, IEEE Transactions on Circuits and Systems – II, Vol.44, No.8, AUG. 1997. Printed Circuit Test (PCB) Methodology User Guide, Rev. 1.6, Jan 2000, Intel Corp., www.intel.com/design/chipsets/ J. Schwartzenbach and K.F. Gill, System Modelling and Control, Arnold Publishers, London, 1992 K. Godfrey, Perturbation Signals for System Identification, PHI, New York, 1993. R.A. Guinee, “Variable Speed Motor Drive Testing and Parameter Identification using Pseudorandom Binary Sequences”, IEEEMelecon 2000, May 29 – May 31, 2000, Cyprus. B.R. Wilkins, Testing Digital Circuits, Van Nostrand Reinhold, 1986. R.A. Guinee, “Digital Circuit Testing with Signature Analysis Employing Pseudorandom Binary Sequence (PRBS) Generators”, Proceedings of the IASTED International Conference Applied Modelling and Simulation, Sep1-3, 1999, Cairns, Australia D.M. Horan and R. Guinee, A Novel Pulse Echo Correlation Tool for Transmission Path Testing and Fault Finding using Pseudorandom Binary Sequences., IEEE International Symposium on Defect and Fault Tolerance in VLSI Systems, pp. 229-237, 03-05 Oct. 2005. MicoSim pSpice, Circuit Analysis Software, Ver.8.0,June 1997, MicroSim Corp. W. Fraser, Telecommunications, Macdonald and Jane’s Publishers, London, 1978. F. R. Conner, Waves, Arnold publishers, Great Britain, 1972. M.G.Hartley, Digital Simulation Method, Peter Peregrinus Ltd., England, 1975 ISBN: 978-960-6766-60-2 161 ISSN: 1790-5117