Phase Comparison Relaying

MULTILINGER-2681B

PhaseComparison

Relaying

GE Power Management

1FUNDAMENTAL PRINCIPLE OF PHASECOMPARISON

The basic operation of a phase comparisonscheme requires that the phase angle of two ormore currents be compared with each other. Inthe case of transmission line protection, thesecurrents may originate many miles from eachother so, as noted above, some form ofcommunication channel is required as part ofthe scheme. If a two-terminal line is considered,the relays located at terminal A can measure thecurrent at that terminal directly. The phase angleof the current at the remote terminal (B) mustsomehow be communicated to terminal A.Since the current sine wave is positive for onehalf cycle and then negative for the next halfcycle etc., it may be used to key a transmitterfirst to a MARK signal for a half cycle and thento a SPACE signal for the next half cycle for aslong as the current is present. Such a signaltransmitted at B and received at A can becompared with the current at A to determinewhether the two quantities are in phase or outof phase with each other. Conversely, the currentat terminal B may be compared with the signalsreceived from terminal A.

It becomes apparent that a comparison such asthat described above must be made on a single

phase basis. That is, it would not be possible tocompare all three phase currents at terminal Aindividually with all three at terminal B over onesingle channel and one single comparing unit.Thus, in the interest of economy, all three phase-currents are mixed to produce a single phasequantity whose magnitude and phase anglehave a definite relation to the magnitude andphase angle of the three original currents. It isthis single phase quantity that is phasecompared with a similarly obtained quantity atthe remote end(s) of the line.

While there are many variations on the basicscheme (and these will be discussedsubsequently), the general method employed tocompare the phase angle, or phase position of thecurrents is always the same. The left side of Figure1 illustrates the conditions for a fault internal tothe protected zone. The top sketch shows about 1 cycle of the current flowing into terminal A. Thesecond sketch down indicates the current flowinginto terminal B. These are in phase since the faultis internal. The third sketch down represents thereceiver output at terminal A as a result of thetransmitter at B being keyed from a signalproduced by the current at that terminal.

The MARK-SPACE designations given to thereceived signal are for identification and have no

PHASE COMPARISON RELAYING

INTRODUCTION

Phase comparison relaying is a kind of differential relaying that compares the phase angles of thecurrents entering one terminal of a transmission line with the phase angles of the currents enteringall the remote terminals of the same line. For the conditions of a fault within the protected zone(internal fault), the currents entering all the terminals will be in phase. For conditions of a faultoutside the zone of protection (external or through fault), or for just plain load flow, the currentsentering any one terminal will be 180 degrees out of phase with the currents entering at least one ofthe remote terminals. The phase comparison relay scheme makes this phase angle comparison andtrips the associated breakers for internal faults. Since the terminals of a transmission line arenormally many miles apart, some sort of communication channel between the terminals is requiredto make this comparison.

2special significance. If the communicationequipment happened to be a simple radiofrequency transmitter-receiver, and if thepositive half cycle of current keyed thetransmitter to ON, then the MARK block wouldcorrespond to a received remote signal while theSPACE block would correspond to no signal.Conversely, if the negative portion of the currentwave keyed the transmitter to ON, then theSPACE block would represent the receivedsignal.

With a frequency-shift transmitter-receiver as thecommunication equipment, the MARK blockwould represent the receipt of the hi-shiftfrequency and the SPACE block the lo-shiftfrequency if the remote transmitter were keyed tohigh from a positive current signal. The conversewould be true if the transmitter were keyed to highfrom a negative current signal. In any case theMARK block received at A, whatever it represents,corresponds to positive current at B while theSPACE block corresponds to negative current at B.

If we consider an internal fault as shown on the leftside of Figure 1, the relay at A would be comparingquantities illustrated in the top and third-from-the-top sketches. If these two signals at terminal Awere to be compared as shown in Figure 2A over afrequency-shift equipment, a trip output wouldoccur if positive current and a receiver MARKsignal were both concurrently and continuouslypresent for at least 8.33 milliseconds. The tripoutput would be continued for 18 milliseconds toride over the following half cycle during which thecurrent is negative, and the half cycle after thatwhen the pick-up timing takes place again.

Assuming that the MARK and SPACE signalscannot both be present concurrently then it mightbe argued that a comparison could be madebetween the positive half cycle of current and theabsence of a receiver SPACE output, Figure 2Billustrates this logic.

If the communication equipment happened tobe a frequency shift channel so that both theMARK and the SPACE signals were definiteoutputs, Figure 2A would represent a trippingscheme since tripping is predicated on thereceipt of a remote MARK or tripping signal. Onthe other hand, Figure 2B would represent ablocking scheme in as much as it will blocktripping in the presence of a SPACE or blockingsignal. It will trip only in the absence of thissignal.

The right side of Figure 1 illustrates theconditions during an external fault. Referring toFigures 2A and 2B it will be noted that neitherapproach, the blocking or the tripping, willresult in a trip output for this condition since theAND circuits will never produce any outputs tothe 8.33/18 timers.

At this point it should be explained that theconditions illustrated in Figure 1 are ideal. Theyseldom, if ever, occur in a real power system.Actually an internal fault would not produce areceived signal MARK-SPACE relationship thatis exactly in phase with the locally contrivedsingle phase current. This is true for a variety ofreasons including the following:

(a) Current transformer saturation.

(b) Adjustment differences in the current mixingnetworks in the relays at both ends of theline.

(c) Phase angle differences between the currentsentering both ends of the line as a result ofphase angle differences in the driving systemvoltages.

(d) Transit time of the communication signal.

(e) Unsymmetrical build-up and tail-off times ofthe receiver.

5Thus, the logic shown in Figures 2A and 2Bwould rarely, if ever, produce a trip output on aninternal fault because the 8.33 milliseconds(which is the time of a half cycle on a 60 hertzbase) requires perfect matching. In actualpractice a 3-4 millisecond setting is used ratherthan the 8.33 setting illustrated. This makes itmuch easier to trip on internal faults. It alsomakes it much easier to trip undesirably onexternal faults. However, experience hasindicated that with proper settings andadjustments in the relay such a timer settingoffers an excellent compromise. This may bebetter appreciated if it is recognized that item(a) above is generally minimized and item (c) isnonexistent on external faults.

In the event that an ON-OFF communicationequipment were to be employed rather than thefrequency-shift equipment, the logic would appearas in Figures 2C and 2D. It will be noted in thesetwo figures that the reference to MARK and SPACEhave been conveniently omitted since the receiveroutput is either present or not as against the caseof the frequency-shift equipment where it could bethere in either of two states. Figure 2C illustrates atripping scheme while Figure 2D a blockingscheme. Here again, the 8.33 millisecond timer is,in practice, actually set for 3-4 milliseconds.

Figures 3A, 3B, 3C, and 3D are for three-terminallines and they correspond directly to Figures 2A,2B, 2C, and 2D. It will be noted from Figure 3 thatfor a three-terminal line, the relay at A mustreceive information from both the remoteterminals. The same applies to the relays atterminals B and C. As in the case of the two-terminal lines, the 8.33 millisecond timerillustrated in Figure 3 will actually be set for 3-4milliseconds.

While all the sketches in Figures 2 and 3compare the positive half cycle of current with areceiver output, the negative half cycle mightjust as well have been selected. However, if this

were done, in Figure 2A for example, it wouldhave been necessary to compare the presence ofnegative current with a received SPACE signalrather than a MARK signal.

It should be recognized that the abovediscussion, as well as Figures 2 and 3, arerudimentary. The complete phase comparisonscheme is considerably more sophisticated andwill be discussed in more detail subsequently.However, at this point it would be well to notethat phase comparison on a continuous basis isnot permitted mainly because it would tend toreduce the security of the scheme. For thisreason fault detectors are provided. They initiatephase comparison only when a fault occurs on,or in the general vicinity of, the protected line. Asimplified sketch of the logic of a phasecomparison blocking scheme including faultdetectors is illustrated in Figure 4. This is asomewhat more fully developed version ofFigure 2D, and the same logic is present at bothends of a two-terminal line.

It will be noted from Figure 4 that AND1 (thecomparer) at each end of the line compares thecoincidence time of the positive half cycle ofcurrent with the absence of receiver output. Thisis initiated only when a fault is present asindicated by an output from FDH. FDH is set sothat it does not pick up on load current but doespick up for all faults on the protected linesection. Thus, when a fault occurs FDH picks up,and if the receiver output is not present for 3 milliseconds during the positive half cycle ofcurrent out of the mixing network, a trip outputwill be obtained.

Of course, the output from the receiver willdepend on the keying of the remote transmitter.The transmitters at all line terminals are keyedin the same manner. They are keyed ON by anoutput from FDL and keyed OFF by the squaringamplifier via AND2 during the positive halfcycles of current. The FDL function is required at

all terminals in all phase comparison blockingschemes to initiate a blocking signal from theassociated transmitter. This is received at the remotereceiver and blocks tripping via the comparer duringexternal faults. FDL has a more sensitive setting andtherefore operates faster than the remote FDHfunction. It is obvious from Figure 4 that if anexternal fault occurred, and FDL did not operate atleast as fast as the remote FDH, false tripping couldoccur because of the lack of receiver output. Ingeneral FDL is set so as not to pick up on loadcurrent but still with a lower pick up than FDH sothat it will operate before FDH. For an internal fault,the currents entering both ends of the line are inphase with each other. Thus, during the half cyclethat the SQ AMP is providing an input to AND1, theassociated receiver is producing no output, and sotripping will take place at both ends of the line. Foran external fault, the current entering one terminal is180 degrees out of phase with the current enteringthe other terminal. Under these conditions, duringthe half cycles when the SQ AMP is producingoutputs, the associated receiver is also providing anoutput thus preventing an AND1 output. No trippingwill take place.

VARIATIONS IN PHASE COMPARISONSCHEMES

There are a number of different phase comparisonschemes in general use today and while all of theseemploy the same basic means of comparisondescribed above, significant differences do exist.These differences relate to the following:

(1) Phase comparison excitation (component orcurrent to be compared).

(2) Pure phase comparison vs. combined phaseand directional comparison.

(3) Blocking vs. tripping schemes.

(4) Single vs. dual phase comparison.

PHASE COMPARISON EXCITATION

Before discussing this subject it is well toconsider what takes place in terms of thecurrents that are available for comparison whena fault occurs on a power system. Table I belowlists the sequence components of fault currentthat are present during the various differentkinds of faults while Figure 5 illustrates therelative phase positions of the sequencecomponents of fault current for the differentkinds of faults and the different phases involved.

TABLE ISequence Components

Type of Fault Positive Negative Zero

Single-Phase-to-Ground yes yes yes

Phase-to-Phase yes yes no

Double-Phase-to-Ground yes yes yes

Three-Phase yes no no

Figure 5 shows the relative phase positions ofthe outputs of a positive sequence network, anegative sequence network, and a zerosequence network all referenced to phase A. Thetransfer functions of these three networks aregiven by the following equations.

I1 =1

(Ia + Ib/120 + Ic/-120) (1)3

I2 =1

(Ia + Ib/-120 + Ic/120) (2)3

I0 =1

(Ia + Ib + Ic) (3)3

It is interesting to note that the phase positionsof the sequence network outputs differdepending on the phase or phases that arefaulted as well as the type of fault. For example,while the positive, negative, and zero sequencecomponents are all in phase for a single-phase-A-to-ground fault, they are 120 degrees out ofphase with each other for phase-B-to-ground,and phase-C-to-ground faults.

6

It will be observed from Table I that positivesequence currents are available for all kinds offaults, negative sequence currents are availablefor all but three-phase faults, and zero sequencecurrents are available only for faults involvingground. Thus, it appears that if one singlesequence component of current were to beselected for use to make the phase comparison,the positive sequence component would suffice.Actually this is not the case in many if not mostof the applications because of the presence ofthrough load current during the fault.

For a single-phase-to-ground fault on theprotected line, the positive sequencecomponent of fault current entering one end willbe in phase with that entering the other end.This is a tripping situation for the phasecomparison scheme. However, any load flowacross the line during the fault will produce apositive sequence component of load currententering one end of the line that is 180 degreesout of phase with that entering the other end.(That is, the positive sequence component ofload current entering one end is in phase withthat leaving the other end). This is a non-tripping situation for the phase comparisonscheme. The phase position of the loadcomponent relative to the fault componentdepends on such factors as the direction of theload flow, power factor of the load flow, and thephase angles of the system impedances. Thephase position of the "net" (load plus fault)positive sequence current entering one end ofthe line relative to that entering the other endwill depend on these same factors plus therelative magnitude of the fault and loadcomponents of current.

In general, the heavier the fault current, and thelighter the load current, the more suitable is theuse of pure positive sequence for phasecomparison. Heavier line loadings and lowerfault currents will tend to make the scheme lessapt to function properly for internal faults. Thus,

pure positive sequence phase comparisonappears practical only in a minority of the casesand so is not suitable for a scheme that is to begenerally applicable.

Significant negative sequence currents arepresent only during faults, they are present in allbut balanced three phase faults, and there is nosignificant negative sequence component ofload current. All this combines to make purenegative sequence ideal for phase comparisonexcept that it will not operate for balanced threephase faults. Similar comments may be maderegarding pure zero sequence phase comparisonwith the additional limitation that it will notoperate for phase-to-phase faults. Thus, theredoes not appear to be one single sequencecomponent or one single phase current thatcould be used in a phase comparison scheme toprotect against all types of faults.

There are a number of different approaches thatare possible to provide a complete scheme.Probably the most obvious would be to makethe phase comparison on each phase separately.This is undesirable principally because the costwould be high since three separate phasecomparison relays and communication setswould be required. Another approach would beto use two separate phase comparison relaysand communication sets, one for pure positiveand the other for pure negative sequencecurrents. The latter would serve to protectagainst all unbalanced faults while the formerwould take care of three phase faults and alsoprovide a measure of back-up protection forheavy unbalanced faults. Here again cost is animportant factor.

As soon as consideration is given to the use of aseparate positive and a separate negative phasesequence comparison, the idea of switchingfrom one to the other presents itself. Suchschemes are available. They include detectorsseparate from the phase comparison function

8

9that distinguish between three phase faults andall other types. For three phase faults thenegative sequence network is unbalanced sothat it produces an output for positive sequencecurrent as well as for negative sequence current.The scheme operates normally to providenegative sequence phase comparison for allunbalanced faults. When a three phase faultoccurs the three-phase detectors at both ends ofthe line operate to automatically unbalancetheir respective negative sequence networksand make them sensitive to positive as well asnegative sequence currents. Since the fault isthree phase, there is no negative sequencecurrent produced so the phase comparison is made on a pure positive sequence basis. This is all accomplished with a commoncommunication channel for both modes.

Another similar approach would be to providetwo separate sequence networks, one purepositive sequence and the other pure negativesequence. Then use the three-phase detector toswitch the logic so that only for three phasefaults the outputs of the positive sequencenetworks at both ends of the line are comparedbut for all other faults the negative sequenceoutputs are compared. Here again all this beingaccomplished over a common channel. Thisapproach has never been used possibly becauseof the idea of using "Mixed Excitation."

Mixed Excitation is a term used to describe aphase comparison scheme that mixes theoutputs of the different sequence networks in agiven proportion and phase angle and thenmakes a phase comparison for all faults basedon this mix. Thus, all such schemes mustinclude positive sequence plus negativesequence and/or zero sequence in order tooperate for all faults. The two main questions tobe resolved are:

(1) Which sequence components should bemixed with the positive sequence.

(2) What percentages of the full magnitude ofeach sequence component of current shouldbe used.

Figure 6 illustrates a two-terminal line with aninternal phase B-to-ground fault. The phasordiagrams at the top of the page indicate thephase positions of the sequence currents atboth ends-of the line assuming the positivedirection of current flow into the line, and alsoassuming phase A reference as in equations (1),(2), and (3) previously given.

At this point it should be recognized that thepositive sequence component of current ismade up of two parts, the load component (IL1)and the fault component (IF1). By an analysisutilizing superposition the load component (IL1)may be established as the current flowing justprior to the fault. The three fault components ofcurrent (IF1, I2 and l0) are then calculated usingthe voltage that existed at the point of fault justprior to the fault. Since the load component ofcurrent is equal to the vector difference betweenBus X and Bus Y voltages divided by theimpedance of the line, and since the prefaultvoltage (at the point of fault) has a phaseposition somewhere between that of X and Yvoltages, the positive sequence component offault current will be displaced from the loadcomponent by about 90 degrees plus or minusabout 30 degrees. The phasor diagrams at thetop of Figure 6 assume that load current flow isfrom bus X to bus Y. The first row of the table inFigure 6 indicates that for the conditionsassumed, the net positive sequence currententering both ends of the line are about 120degrees displaced from each other. Heavier faultcurrent and lighter load current would reducethis angle toward zero while the converse wouldincrease the angle toward 180 degrees.

11

The second and third rows of the table of Figure 6indicates the relative phase positions of thepositive plus negative, and positive plus negativeplus zero sequence components respectively.These appear to be more unsatisfactory. Rows 4and 5 combine the components differently andboth appear to yield much better results.

It is obvious, from Figure 5 that a similar fault ona different phase would yield different results.This is illustrated in Figure 7 where a phase-A-to-ground fault at the same location is analyzed.As noted earlier, the integrator timers in phasecomparison schemes are generally set for about3 milliseconds. This will permit tripping oninternal faults with as much as 115 degreesbetween the phase angles of the currentsentering both ends of the lines. On this basis, onlyexcitation by I2 - I1/5 would prove satisfactory forthe two cases studied in Figures 6 and 7.

Actually only two simple faults wereinvestigated. It is obvious that different resultswould have been obtained for these same kindof faults if the relative magnitudes of loadcurrent, positive sequence fault current, and zerosequence fault current had been assumeddifferently. Also, for the values of currentsassumed, different results would obtain forother types of faults. In addition, if differentcombinations and weighting factors of thesequence components had been investigatedstill different answers would have resulted. Inthe proper selection of sequence componentsand weighting factors for Mixed Excitationphase comparison the following points must beconsidered:

(a) Whatever combination and weightingfactors are employed, the application rulesshould be simple enough to make theapplication practical.

(b) As a corollary to (a) above, the fewest numberof sequence components should be used.

(c) The effects of load current must beminimized. Thus, negative and/or zerosequence components should be weightedover the positive sequence components.

(d) The limits of application should be broadenough to render the scheme useful as aprotection tool.

In line with the considerations stipulated above,the best overall results using mixed excitationwould be attained by using I2 - I1/K, where K is aconstant that is adjustable within limits. While itis likely that the inclusion of zero sequenceexcitation would be helpful for one case oranother, it is not generally employed becausethe problem of evaluating the overallperformance of the scheme would be magnifiedconsiderably. This is true mainly because thecurrent distribution in the zero sequencenetwork is generally quite different from that inthe positive and negative sequence networkswhere the current distributions areapproximately the same. For any given fault ona transmission line, the ratio of IF1/I2 at anyterminal is the same as at any other terminal ofthat line. This is not true of either IF1/I0 or I2/l0. Itis this that makes the use of zero sequenceexcitation undesirable.

Mixed Excitation Phase Comparison

If the mixing network of Figure 4 were designedto produce an output that is proportional to I2 I1/K, this logic would then be a simplifiedrepresentation of a mixed excitation phasecomparison scheme. In such schemes, the pickup setting of FDH must be high enough so thatthe I1/K output from the mixing network doesnot result in continuous phase comparison onload current (I2 is normally zero during normalsystem conditions). Also, it may be desirable tohave FDL set to pick-up at some level above fullload so that channel is not keyed on and offcontinuously during normal load conditions.

12

Since FDH is set higher than FDL, thisrequirement results in a still higher setting forFDH.

Because FDH controls tripping, this arrangementlimits the applicability of the basic scheme tocircuits where the minimum three phase faultcurrent is significantly higher than the maximumload current.

The requirements for the satisfactoryperformance of a mixed excitation scheme usingovercurrent fault detectors (FDH and FDL) arenoted below.

(a) Both the FDL and FDH fault detectors mustbe set above full load current.

(b) All internal faults regardless of type or theparticular phases involved must produceenough I2 - I1/K to operate FDH at all ends ofthe line.

(c) FDL must be set with a lower pick-up thanFDH at the remote end(s) of the line forsecurity during external faults.

(d) The phase angle difference between the I2 - I1/K quantities obtained at all terminals ofthe protected line during all types of internalfaults, and for any combination of phases,must be less than 115 degrees.

Mho Supervised Mixed Excitation PhaseComparison

Figure 8 is an abbreviated logic diagramcovering a modified mixed excitation blockingscheme that provides somewhat moresensitivity than the basic scheme. In order toaccomplish this, two single-phase directionalmho distance measuring functions are used.These are both associated with the same pair ofphases which, in this case, are phases A and B.The MT function at each terminal "looks" into

the line and is set to reach beyond the remoteterminal(s). The MB function "looks" backwards,out of the line and is set to reach beyond thereach of the remote MT function(s). Thesefunctions are required to operate for three phasefaults but will incidentally also operate for someother faults involving one or both of theassociated phases.

The basic idea is to compare the relative phasepositions of the mixed excitation (I2 - I1/K) in thenormal manner but to initiate the comparisonwith more discriminatory fault detectors. Toaccomplish these ends, pure negative sequenceis used to operate FDH and FDL. Since there isno significant negative sequence currentflowing during normal system conditions, thesefault detectors may be applied with verysensitive settings. They will initiate phasecomparison for all except three-phase faults. Inthe event of three phase faults the MT and MBunits function as FDH and FDL respectively.Since these functions can discriminate betweenload currents and fault currents regardless ofmagnitudes, they can detect faults that producecurrents less than full load values. Thus, thecombination of negative sequence current leveldetectors and distance measuring functionsprovide a means for more sensitive faultdetection.

With this arrangement, the requirements forsatisfactory performance are given below.

(a) The FDH function must be set sensitivelyenough to pick up for all unbalanced faultson the protected line section.

(b) FDL must be set with a lower pick up thanFDH at the remote end(s) of the line forsecurity during external faults.

(c) The MT function must be set with a longenough reach to detect all internal threephase faults.

15

(d) The MB function must be set to outreach allremote MT units.

(e) The phase angle difference between the I2 - I1/K quantities obtained at all terminals ofthe protected line during all types of internalfaults, and for any combination of phasesmust be less than 115 degrees.

Network Unbalancing Phase Comparison

Earlier in this section reference was made toschemes that employ pure negative sequencephase comparison except for the case of a threephase fault for which the sequence network isunbalanced to produce sensitivity to positivesequence currents. Figure 9 illustrates, in anabbreviated manner, the logic of such a blockingtype of a scheme. It will be noted that unless allthree of the FD units pick up, the phasecomparison will be on a negative sequencebasis. When all three FD units pick up, as theywill for three phase faults, the network isunbalanced to produce an output for balancedpositive sequence current inputs, and phasecomparison takes place on a pure positivesequence basis. The FD units must be set to pickup above maximum full load current in order toinsure pure negative sequence comparison onall but three-phase faults.

Since phase comparison is initiated by the sameFDH for all kinds of faults, the ratio of positivesequence pick up to negative sequence pick updepends only on the amount of unbalance in thenetwork introduced by FD operation. This isusually adjustable in the relay. In this kind ofscheme FDH is set so that it picks up at a valueof positive sequence current that is somewhathigher than the three phase fault currentrequired to operate all three FD units. Theresponse of FDH to negative sequence currentsis then dependent on the network unbalancedescribed above. FDL is set with a pick up thatis below that of FDH.

With this kind of scheme, the requirements forsatisfactory performance are given below:

(a) The FD function must be set with a pick upthat is above full load current.

(b) The FDH function must be set to pick up,with the network unbalanced, at some levelof positive sequence current that is higherthan the FD pick up setting

(c) The FDL function must be set with a pick upbelow that of the remote FDH unit in order tomaintain security during external faults.

(d) FD must pick up for all internal three phasefaults.

(e) FDH must pick up for all internal faults.

Separate Positive & Negative Sequence PhaseComparison

The scheme described earlier in this section that iscompromised of two separate phase comparisonschemes may be represented in abbreviated logicby two diagrams similar to Figure 4. In one schemethe mixing network would be a pure negativesequence network. In the other scheme it wouldbe a pure positive sequence network. In order forthis approach to perform satisfactorily, thefollowing requirements must be met.

(a) The negative sequence FDH must be set sothat it operates for all unbalanced faultsinternal to the protected section.

(b) The negative sequence FDL must be setsomewhat more sensitively than the remoteFDH for security during external faults.

(c) The pick up of the positive sequence FDLmust be set above full load to preventcontinuous transmission under load.

16

(d) The pick up of the positive sequence FDHmust be set somewhat higher than that ofFDL at the remote end of the line for securityduring external faults.

(e) The positive sequence FDH must pick up forall three phase faults internal to theprotected line section.

It should be recognized that this scheme will notdetect three phase faults unless the fault currentexceeds full load currents.

Summary of Phase Comparison ExcitationConsiderations

The following is a brief summation of theforegoing discussions.

1. There are a number of different ways in whichphase comparison relaying may be arranged.However, in every case some combination orarrangement of positive and negativesequence components of current offers thebest general approach.

2. Mixed excitation schemes are generally moredifficult to apply than are schemes thatcompare pure sequence components.

3. Unless some sort of distance type faultdetectors are applied, none of these schemescan detect three phase faults that producecurrents that are below full load values.

In light of the above, the question of, "WhyPhase Comparison?" might come to mind.Some of the answers are given below.

1. Phase comparison relaying is not affected byzero sequence mutual impedances that cancause directional type ground relays tomisoperate.

2. Phase comparison relays with overcurrentfault detectors do not require a systempotential supply in order to operate.

3. Phase comparison schemes are generallysuitable for use on series compensated lineswhere distance type schemes may not besuitable.

4. In their simple form, phase comparisonschemes are relatively inexpensive.

5. Except for the mho supervised schemesphase comparison relays will not operateduring system swings and out-of-stepconditions.

COMBINING PHASE AND DIRECTIONALCOMPARISON

Since there is no single sequence componentthat could be used in phase comparisonschemes to provide protection for all types offaults it is necessary to compensate for thisdeficiency. The previous section discussesmeans for mixing sequence components,unbalancing sequence networks for certainfaults, and using two or more completeschemes each with different excitation. Anotherapproach that is possible and that has gainedacceptance is called "Combined Phase AndDirectional Comparison."

As the name implies, a combined phase anddirectional comparison scheme combines theprinciples of both phase comparison anddirectional comparison in one single schemeutilizing a common communication channel.The basic approach is to use pure negative orpure zero sequence for the phase comparisonplus a set of mho functions at each end of theline for the directional comparison portion. Onemho function (MT) operates for faults in thetripping direction. The other (MB) operates forfaults in the blocking direction. Figure 10

18

illustrates the rudiments of a combined phaseand directional comparison blocking scheme.Such a scheme is in general more sensitive thanthe schemes described earlier. This is sobecause the phase comparison, since it does notinclude positive sequence excitation, may be setto operate for low values of fault current. On theother hand the distance relays are basically ableto discriminate between fault currents and loadcurrents. Thus, the overall scheme will operatefor fault currents well below full load current.

Referring to Figure 10 and assuming that the MTand MB functions do not exist, it will be observedthat the scheme becomes the same as the simplephase comparison scheme of Figure 4. Thus, forany fault internal or external to the protectedsection for which no mho unit operates, thescheme will perform as a simple phasecomparison scheme. On the other hand if it isassumed for the moment that the faultdetectors and the squaring amplifier areinoperative, the scheme behaves as a simpledirectional comparison scheme under control ofMT and MB. For an internal fault the MT units atboth ends of the line operate to stop alltransmission via AND2. At the same time theyprovide the lower input to AND1. With noreceiver outputs, AND1 provides a signal to theassociated integrator which after 3 millisecondsproduces a trip output. For an external fault, sayto the left of circuit breaker A, MB at A wouldoperate to key the transmitter on. This results incontinuous transmission of a blocking signal tothe receiver at B. The output of the receiver at Bblocks any MT operation at B from producing anAND1 output at that terminal. This in turnprevents any trip output from the integrator. Atterminal A, since the fault is in the blockingdirection, MT does not operate, the lower inputto AND1 is not present, and no tripping can takeplace.

Since in the combined scheme it is possible, andeven likely, that the mho functions, fault

detectors, and squaring amplifier would alloperate for some faults, it is necessary to set upan order of preference between the two modesof operation. For reasons that will be discussedsubsequently, these combined schemes givepreference to the directional comparison (mhofunctions) over the phase comparison (faultdetectors), as may be observed from Figure 10.If an internal fault were to occur for which bothmodes functioned, the MT functions at bothends of the line would prevent any signaltransmission by blocking AND2. This is trueregardless of the attempt of FDL to starttransmission. Also, the lower input to AND1would be made continuously present by MTregardless of FDH and the squaring amplifier.

During an external fault to the left of terminal Afor example, MB at that terminal will operate tokey on the transmitter to send a continuousblocking signal to the remote end. At the sametime MB will block AND4 from permitting thesquaring amplifier from stopping carrier everyother half cycle. This in turn will block trippingat terminal B. MB also blocks AND3 fromallowing FDH to provide a comparer input. Thus,terminal A will not trip either.

The need for this directional comparisonpreference will become apparent if an externalfault is considered to be just to the right ofterminal B. Assume that for this fault both thephase comparison and the directionalcomparison detectors would operate. Thus, atterminal B the MB function will key itstransmitter to send a continuous blockingsignal. At terminal A the MT function will seethe fault and attempt to trip but will be blockedby a continuously received blocking signal fromthe receiver. If the phase comparison squaringamplifier at B were permitted to key thetransmitter off every other half cycle, the MTfunction at A would cause a trip during that halfcycle. This is so because AND1 would producealternate half cycle output pulses that would in

19

turn time out the integrator. For this reason, andbecause it is likely that some external faults willresult in the operation of both modes,directional comparison preference is employed.

Zero Sequence Excitation

As noted earlier, in combined phase and directionalcomparison schemes, the excitation could be purezero or pure negative sequence current.Considering pure zero sequence excitation andreferring to Figure 10, the sequence network wouldbe a zero sequence network. Actually no networkwould be required in the relay for this case becausethe wye connected CTs normally used are in fact azero sequence network in themselves. With zerosequence excitation the phase comparison portionof the overall scheme would not be capable ofoperating for phase-to-phase and three-phasefaults. For this reason the directional comparisonportion must include MT and MB functions thatcan detect and operate for faults involving any twoor more phases. This requires that the MT and MBfunctions each be three single phase mho units.

It should be noted that distance relays designed tooperate for faults involving two or more phaseswill operate for double-phase-to-ground faults andalso for certain close-in single-phase-to-groundfaults. Thus, it is reasonable to expect that boththe phase and directional comparison modes willbe activated for many faults and so the preferenceis required.

In order to obtain satisfactory performance fromsuch a scheme, the following requirements mustbe met:

(a) The FDH function must be set so that itoperates for all single-phase-to-ground faultsinternal to the protected line.

(b) The FDL function must be set somewhatmore sensitively than the remote FDH forsecurity during external faults.

(c) The MT function must be set with a reachthat is long enough to enable it to see allmulti-phase faults on the protected line.

(d) The MB function must be set to reach furtherthan the remote MT unit for security duringexternal faults.

(e) The MB function must not operate for anyinternal single-phase-to-ground fault forwhich the associated MT function does notoperate otherwise tripping will be blocked.

Negative Sequence Excitation

The logic shown in Figure 10 will apply as wellto negative sequence phase comparisonexcitation as it does to zero sequence excitation.In this case, however, the sequence networkillustrated would have to be a negativesequence rather than a zero sequence network.Also, since the negative sequence phasecomparison will protect against all unbalancedfaults, the MT and MB (directional comparison)functions are required only for three-phase faultprotection. However, if these functions aredesigned to respond to all multi phase faults,then phase-to-phase and double phase-to-ground faults will be protected by both modeswhile single-phase-to-ground faults will beprotected by only the phase comparison mode,and three phase faults by only the directionalcomparison mode.

In order to obtain satisfactory performance fromsuch a scheme, the following requirementsmust be met.

(a) The FDH function must be set so that itoperates for all unbalanced faults internal tothe protected line.

(b) The FDL function must be set somewhatmore sensitively than the remote FDH forsecurity during external faults.

20

(c) The MB function must be set to reach furtherthan the remote MT unit for security duringexternal faults.

(d) The MT function must be set with a reach thatis long enough to enable it to see all multi-phase faults on the protected line.

(e) The MB function must not operate for any internalfault for which the associated MT function doesnot operate otherwise tripping will be blocked.

Summation of Combined Phase & DirectionalComparison Considerations

1. Both the zero and the negative sequenceschemes require the use of directionaldistance measuring functions. Thus, bothschemes require potential supplies, and bothschemes are sensitive to system swings andout-of-step conditions.

2. The scheme using zero sequence excitationrequires mho functions that must operate forall multi-phase faults with the possibleexception of double-phase-to-ground faults.

3. The scheme using negative sequenceexcitation requires mho functions thatrespond to three phase faults only.

4. Both schemes require that the MB functionsdo not operate for any internal faults forwhich the associated MT functions do notoperate. Since phase distance mho relays canrespond to single-phase-to-ground faults andfaults on adjacent phases, the selection of thetype of mho functions employed for any givenapplication requires consideration. Thefactors involved in this consideration areoutside the scope of this discussion.

5. Neither scheme will misoperate as a result ofzero sequence mutual coupling between theprotected line and other parallel lines.

6. In general both schemes will have the samesensitivity. That is, the 3I0 sensitivity of thefault detectors in the zero sequence schemewill be about the same as the I2 sensitivity ofthe fault detectors in the negative sequencescheme. For some long lines with strongground sources, the distribution of faultcurrents may be such that for faults near oneterminal the I2 at the remote terminal isgreater than 3I0 at the same terminal. Forsuch applications the negative sequencescheme may be best.

7. Both schemes will operate for three phasefaults that produce currents well below fullload values.

8. These schemes are a half way point betweenpure phase comparison and pure directionalcomparison. As such they requirecoordination between the two modes ofoperation as noted in item (4) above. Thisproblem is not present in the pure schemes.

BLOCKING vs. TRIPPING SCHEMES

Earlier discussion in conjunction with figure 2provides a basis for further consideration ofblocking vs tripping pilot schemes. Figure 2Cillustrates the comparer-integrator logic for atripping scheme using an ON-OFF type of pilotchannel. In order to trip, a receiver output isrequired to be present during the half cycle thatthe local current is positive. Figure 2D isrepresentative of a blocking pilot scheme wheretripping will take place if there is no receiveroutput during the half cycle that the localcurrent is positive.

If we consider that an input to, or an outputfrom a logic box is a positive going signal, thelogic illustrated in Figures 2B and 2C assumethat a received signal at the input of a receiverwill produce a positive going voltage signal atthe output of the receiver to the relay logic. This

21

is not always true. Some types of receivers willproduce negative (or reference) voltage outputswhen a signal is present at the input, and apositive signal output when nothing is received.If this were the situation in Figure 2, Figure 2Cwould then represent a blocking scheme whileFigure 2D would be a tripping scheme. In someapplications where receiver outputs areinverted, the interface between the receiver andthe relay logic includes an invertor (INV) whichin effect inverts the receiver output signal sothat a received signal produces a positive goingsignal at the output of the invertor. The samegeneral statements regarding signal polaritiesapplies to the keying requirements fortransmitters. Some transmitters may require apositive signal while others a reference ornegative signal to key them off of theirquiescent states.

The main point to be gained from the foregoingdiscussion is that it is not always possible todetermine from a logic diagram whether ascheme is of the blocking or tripping type unlessan indication is given as to the receiver outputvoltages. This applies to frequency shift as wellas ON-OFF communication equipment.

It will become apparent from subsequentdiscussion that it is extremely difficult, if notimpossible, to provide a concise rigorousdefinition of the terms Blocking Scheme andTripping Scheme. Possibly it would be well toproceed with a discussion of the different kindsof channels, their characteristics, and theirapplication before attempting a definition.

Type of Channels

The total channel is composed of thecommunication equipment itself plus the pathor link over which the signal is sent. For relayingpurposes there are two basic types ofcommunication equipment.

1. ON-OFF

2. Frequency-shift

The ON-OFF type, as the name implies, operateswith the transmitter either being keyed on or offby the relay logic. That is, the transmitter at anygiven instant is either sending an unmodulatedsignal or it is sending nothing.

There are two kinds of frequency-shiftequipments. The most prevalent is the two-frequency kind. With this type, the transmittercan send either of two closely spacedfrequencies. When no keying signal is applied tothe transmitter it sends one of these twofrequencies. When the transmitter is keyed, itshifts to the other frequency. It is alwayssending one or the other. The frequency-shiftreceiver has two separate outputs one for eachof the two transmitted signal frequencies. Thus,if the transmitter is sending the MARKfrequency the MARK output is present in thereceiver. If the transmitter is sending the SPACEfrequency, the receiver SPACE output is present.These types of receivers are basically FMreceivers and utilize discriminators. Because ofthis the SPACE and MARK outputs from thereceiver cannot both be present simultaneously.Also, broad band noise at the input to thereceiver tends to provide a balanced signal tothe discriminator which forces its output towardzero. If the noise is severe enough to swamp outthe real signal it can cause random receiveroutput or all output to disappear.

The other kind of frequency-shift equipment is athree-frequency type. When this type oftransmitter is in its quiescent state it sends thecenter frequency. It has two separate keyinginputs so that it can be keyed to shift high orlow (MARK or SPACE) from the centerfrequency. The three-frequency receiver receivesall three frequencies but provides only twooutputs to the relay logic, the high shift and low

22

shift outputs. When the receiver receives thecenter frequency neither the high nor lowoutputs are present. Here again the MARK andSPACE outputs (high and low) cannot both bepresent simultaneously, and severe broad bandnoise at the receiver inputs can result in receiveroutput.

There are several characteristics ofcommunication equipment directly related tophase comparison relaying performance thatmight well be discussed. Phase comparisontypes of schemes compare the phase angle of acurrent derived at one end of a line with acommunication signal received from the remoteend. The communication signal arrives in aMARK-SPACE arrangement that shouldrepresent the positive and negative half cyclesof current at the transmitted end of the line.Actually this is not possible for several reasons.

1. There is a time lag from the instant atransmitter is keyed until the output reflects achange. This build up is generally a very shorttime and is usually insignificant.

2. There is the propagation time from the instantthe transmitter sends until this signal arrivesat the remote location, approximately 1 milli-second for every 180 miles of distance. Thesame applies from the instant the transmitterstops until the remote signal is gone.

3. There is the build up time in the receiver fromthe instant the signal appears at its inputuntil the output reflects the change of state.This time plus the build up time in thetransmitter is called the channel operatingtime.

4. There is the tail off time in the transmitterfrom the instant the keying is removed untilthe output signal changes or disappears. Thisis generally very short and is usuallyinsignificant.

5. There is the tail off time in the receiver fromthe instant the input changes until the outputchanges accordingly. This time plus the tailoff time of the transmitter is called thechannel release time.

6. In ON-OFF channels the operating and releasetimes are not generally the same. They canvary with frequency and attenuation.

7. In frequency-shift channels the discriminatoremployed in the receiver can be balanced sothat build up and tail off times are equal, or itcan be unbalanced (biased) to the MARK orSPACE side. For example, if it is biasedtoward MARK and the input signal issymmetrical (half cycle MARK and half cycleSPACE), the output will be more than a halfcycle MARK and less than a half cycle SPACE.

8. In general wide band channels tend tooperate and release faster than narrow bandchannels. That is, faster channels use morespectrum than slower channels.

It is obvious from the foregoing that thereceived signal at any given terminal is not anexact analog of the remote current. There aretechniques used in phase comparison schemesto compensate for this and they will bediscussed subsequently. Until then it should beassumed that the received signal provides a truerepresentation of the phase position of theremote current.

Types of Communication Links

The communication link over which thetransmitted signal is propagated to the remotereceiver can take several forms. These are notedbelow.

1. Directly over the power line (Power linecarrier)

23

2. Multiplexed over the power line (Single SideBand Carrier)

3. Multiplexed over microwave (Microwave)

4. Pair of Wires (Pilot Wire)

5. Leased Facilities (Telephone Company)(a) Metallic pilot wire(b) Microwave(c) Cable

A distinction is made between leased facilitiesand the other (power company owned) facilitiesbecause in many cases the telephone companydefines the characteristics of the channelwithout defining the type of link.

The ON-OFF type of communication equipmentis used exclusively over power line carrier links.The transmitted signal is propagated along thepower line between the transmitter and theremote receiver. These equipments usuallyoperate in the frequency range of 30-200 kHz.

The frequency-shift equipments are available inseveral frequency ranges. First there are those inthe audio range. These are generally employedover single side-band, microwave, pilot wires,and leased facilities. There are also frequencyshift channels in the power line carrier frequencyrange. These are employed directly over thepower line as are the ON-OFF types ofequipment. Finally there are the frequency shiftequipments that operate in and occasionallyoutside the power line carrier spectrum. Theseare employed over microwave and leasedfacilities.

Power Line Carrier Links

It is obvious that the performance of anychannel that utilizes the protected power lineitself as a link will be affected in some way byfaults on the power line. A fault on atransmission line can attenuate or completelyblock a signal, transmitted at one end of theline, from being received at the remote end.Faults external to the protected line have noeffect on the signal attenuation sincetransmission lines that incorporate power linecarrier channels are trapped at each end (SeeFigure 11).

In the case of ON-OFF power line carrierchannels, the operating frequencies of theequipment at all terminals of the protected lineare generally the same. Thus, a signaltransmitted from any terminal is received at allterminals. This is not a necessary requirementfor using this kind of equipment. Rather it isdesirable because the protection schemes thatuse ON-OFF channels can accommodate asingle frequency arrangement and thisconserves the carrier spectrum.

When frequency-shift equipment is used overpower line carrier, the frequencies of eachtransmitter on the line must be different from allthe others on the same line. For example, if thecommunication equipment in Figure 11 is of thefrequency-shift type, the transmitter at the leftend must operate at the same frequencies asthe receiver at the right end. Also, the right endtransmitter and left end receiver must operateat the same frequencies while the frequencies ofthe two transmitters must be different. This isnecessary because with frequency-shiftequipment the transmitters associated with agiven line protection scheme are not allgenerally sending the MARK or the SPACEfrequencies at the same time. Thus, if a receiverwere able to receive more than one transmitter,it could be simultaneously receiving a MARK

25

signal from one and a SPACE signal fromanother. This would not result in a workableprotection scheme.

When power line carrier channels are used,significant losses are present in the couplingequipment and the line itself. Depending on theselosses and the ambient noise on the line, thetransmitter power required may vary from about 1Watt to 10 Watts and even more in extreme cases.

Consider an ON-OFF tripping type of scheme asdefined by Figure 12. For a moment assume thatFDL and NOT1 do not exist in the logic. Duringan internal fault, the currents out of the mixing(or sequence) networks at both ends of the lineare in phase with each other so that the outputsof the SQ AMP are in phase at both ends of theline. The transmitters at both ends of the lineare keyed on during the same half cycles thattheir associated SQ AMPs are attempting totrip via AND1. Thus, the receivers will besupplying the bottom input to AND1, andtripping will take place when FDH operates toprovide the third input.

For external faults, the currents out of themixing networks at the two ends of the line willbe 180 degrees out of phase with each other.Therefore, during the half cycle that the SQ AMPat one end of the line is producing an output, theone at the remote end is not, so no tripping willtake place. It should be noted that a trippingtype of scheme over an ON-OFF channelrequires transmitters of different frequency at each end of the line so that no receiver can receive the locally-transmitted signals;otherwise tripping would occur during externalfaults. For this reason, such schemes are notgenerally applied.

It appears that the tripping scheme as describedabove has no need for an FDL function since noblocking coordination is required as is in ablocking scheme. However, this is not the case.

The FDL and NOT1 functions provide a meansfor tripping when one end of the line is open aswhen picking up a faulted line from one end. Forsuch a condition, the SQ AMP at the open endreceives no current and so produces no outputto key its transmitter. Without a received signalthe closed end of the line cannot trip under anyconditions even in the presence of a fault. TheFDL function acts as a current detector. It is setwith a very very low pick up so that anysignificant output from the mixing networkcauses it to produce a continuous output. Whenthe mixing network outputs goes to zero, FDLdrops out causing an output from NOT1 whichin turn keys the transmitter on continuously.This is received at the remote end to provide acontinuous signal at the bottom input to AND1.Any fault that picks up FDH will then be trippedat the closed end of the line.

If the mixing network includes a positivesequence output, load current will keep FDLpicked up continuously. If the mixing networkincludes only zero and/or negative sequenceoutputs, load current will not keep FDL pickedup. Thus, with zero or negative sequence phasecomparison the receivers at both ends of theline will be producing outputs to AND1continuously. When a fault occurs, FDL picks-upvery fast to restore the keying function to SQAMP. This operation resembles a blockingscheme, although it is often called a permissivetripping scheme.

Another scheme to facilitate tripping on singleend feed uses a circuit breaker 52/b switchrather than FDL and NOT1. When the breaker isopen, the 52/b switch closes and keys theassociated transmitter on continuously. Whenthe breaker is closed, the 52/b switch is openand keying is under control of the SQ AMP.While on the surface the use of 52/b appearssimple and direct, the following problems arisethat can require more complex logic and stationwiring:

26

1. The 52/b contacts do not generally operate insynchronism with the main poles of thebreaker so some timing functions must beincluded with the logic to compensate for this.

2. In multi-breaker schemes, such as ring buses,two breakers at each terminal are associatedwith each line so 52/b switches from eachbreaker are required in series.

3. In multi-breaker schemes one of the twobreakers may be out of service but in theclosed position. This would require a bypassof its 52/b switch which is open.

Regardless of which tripping scheme is used, itis obvious from Figure 11 that in order to tripeither circuit breaker A or B for an internal faultat P it is necessary to get a carrier signal throughthe fault. If the fault attenuates the signal sothat this does not happen, no tripping can takeplace. The amount of attenuation in signal thatis produced by the fault will depend on the typeof coupling (single phase, interphase, etc.), thetype of fault, the phase involved, and thelocation of the fault on the line. The evaluationof these factors is outside the scope of thisdiscussion.

Figure 13 illustrates the same tripping schemeas Figure 12 except that it utilizes a frequencyshift rather than an ON-OFF communication set.The same comments apply to this scheme as doto that of Figure 12.

A tripping scheme that operates over a powerline carrier channel runs the risk of a failure totrip on internal faults because of signalattenuation. During external faults the line trapsisolate the signal on the protected line from thefault. This is of no significance becauseattenuation or loss of signal on external faultscannot result in any misoperations. Conversely,a blocking scheme is unaffected by loss orattenuation of signal during internal faults

because absence of a signal is required in orderto trip. During external faults it is important thatthe blocking signal be isolated from the faultbecause loss of the signal can result in a falsetrip. The line traps provide this isolation.

Figures 4 and 14 illustrate phase comparisonblocking schemes with ON-OFF and frequency-shift channels respectively. Figure 4 wasdiscussed earlier and Figure 14 is exactly thesame except for the high frequency shift whichis not used in the protection scheme. While onlyone of the two frequencies of the frequency-shift equipment is used in the protectionscheme, the second frequency does perform auseful function. It provides a means forcontinuous monitoring of the channel. Sinceone of the two frequencies is always beingtransmitted, it is possible to monitor the signalat each receiver continuously and incapacitatethe protective scheme and/or provide indicationat that terminal if the signal is lost.

Most schemes that use an ON-OFF channel arearranged so that no transmission takes placeduring normal conditions (no fault). This doesnot lend itself to continuous monitoring.However, schemes are available that periodicallystart transmission of a signal at one end of a linewhich, when received at the remote end,initiates a return transmitted signal. Suchschemes can be started manually orautomatically on a time schedule. They arecalled carrier check-back schemes. They can bearranged so as not to affect the normaloperation of the scheme even in the event of afault during a check-back operation.

For the most part, phase comparison blockingcarrier schemes use ON-OFF rather thanfrequency-shift channels, possibly for one ormore of the following reasons:

1. The overall speed of the protective scheme isdirectly related to the speed of the channel.

28

Until recently high speed frequency shiftcarrier channels were not available. Eventoday the ON-OFF channel is somewhat fasterthan the fastest frequency-shift channel.

2. Noise at the input of an ON-OFF channelreceiver would tend to produce a blockingsignal output. Noise at the input of afrequency-shift channel tends to drive itsoutput to zero which is a tripping condition (ina blocking scheme). This tends to make thefrequency-shift blocking scheme less secureagainst false tripping during external faults. Itis possible to build channel conditiondetectors (signal to noise, loss of channel,etc.) into frequency-shift channels and blocktripping when these detectors indicatetrouble, but these features increase thecomplexity and the cost. This approach tendsto make the blocking scheme resemble thetripping scheme since the receiver must nowindicate an intact channel in order to trip.

3. Aside from the ability to accommodatecontinuous monitoring, the frequency-shiftchannel provides little advantage over theON-OFF carrier channel.

There are very few if any phase comparisontripping schemes in service over carrier channelsmainly because of the fear that it will not alwaysbe possible to get a trip signal through a fault.

There is another type of scheme that has recentlybeen gaining some favor. This is called Unblocking.It is a cross between blocking and tripping in thatit operates in the blocking mode but the blockingsignal is sent continuously even in the quiescentstate (no fault), and so it must be turned off inorder to trip. Thus this scheme, as in the trippingschemes previously described, must include somemeans to stop the blocking signal from beingtransmitted at an open terminal in order to permittripping of the closed remote terminal in the eventof a fault. Here again the FDL logic of Figure 12 and

13 or the circuit breaker auxiliary 52/b switch couldbe used.

In general, unblocking utilizes frequency-shiftchannels because this permits monitoring of thecontinuous blocking signals. As they are usuallyapplied, ON-OFF channels do not lend themselvesto monitoring because the single frequencysystem transmits the same frequency from alltransmitters and the loss of any one transmittercould not be detected. If applied in a normalduplex frequency basis (one in each direction) theON-OFF channel would provide the monitoringfeatures at the cost of carrier spectrum. However,this disadvantage can be overcome by the use of anew application of ON-OFF equipment where thetransmitters at the different terminals areoperating at frequencies offset from each other yetclose enough to be nominally a single frequencysystem. This application permits monitoring, andat the same time has the advantage of a higherchannel speed than the frequency-shift channels,while utilizing less channel spectrum in threeterminal line applications.

Single Side Band Carrier Links

Like the carrier channels discussed above, thesingle side band communication equipment iscoupled directly to the power line throughcoupling capacitors and utilizes the power lineto propagate the signal. In general, the outputfrequency of the single side band equipment isin the 30-200KC carrier spectrum, while themultiplexed channels are of the frequency-shifttype that operate in the audio range. The onlyadvantage of this type of arrangement is that itprovides a multi-channel equipment over aminimum of channel spectrum with a costeconomy over single channels.

Single side band carrier links have not beenused extensively in this country for protectiverelaying. When they are used for this purposethey usually include a relatively high speed

29

frequency-shift channel for the relaying functionplus a limited number of narrow bandequipments for control functions. On someoccasions a high speed ON-OFF channel hasbeen multiplexed over single side band. Thesehigh speed channels usually operate in thekilohertz range, and because of the speedrequirement are of the wide band variety.

The final R.F. amplifier of the single side bandtransmitter has, by design, a certain maximumpower capability. Each of the signals that ismultiplexed over this equipment uses a fractionof that total power depending on the percentmodulation as indicated by the followingequation.

Pn = PtM 2(100)

where:

Pt = total capability of the output amplifierPn = Power output of a given multiplexed

signalM = Percent modulation by the given

multiplexed signal

It should be noted that the total percentmodulation for all multiplexed signals must notexceed 100 percent. Thus, if only two signals aremultiplexed, each at 50 percent modulation,each signal will have one fourth the total power.For four signals equally muitiplexed only onesixteenth the total power will be available toeach. When relaying is used in conjunction withvoice and control functions on single side bandit is often necessary to exhalt the relayingchannel signal when it is called on to block ortrip. This results in the particular signaltemporarily usurping most, if not all, of the totalavailable power at the expense of the otherfunctions.

Phase comparison relaying has been employedover single side band links. In general, the highspeed frequency-shift type of channel is used ina tripping scheme, and the ON-OFF type in ablocking scheme. The same basic commentsapply to these schemes as were made above forsimilar carrier schemes applied without thesingle side band. The single side bandequipment adds nothing to the overallperformance of the protective relay schemeexcept that it contributes to better spectrumutilization and is economically advantageous touse when there are several voice and controlfunctions in addition to relaying requiredbetween the same terminal points.

Microwave Links

Microwave links are quite commonly used forprotective relaying including phase comparisonschemes. However, because of the high cost ofthe microwave equipment, the applications aregenerally limited to cases where a large numberof control and/or monitoring functions areneeded between the same terminals as therelaying.

Since microwave links propagate through theatmosphere, rather than over the power line,they are generally unaffected by faults and noiseon the power system. Thus, with a microwavelink there is no problem of getting a signalthrough the fault, so tripping type schemes arevery acceptable. On the other hand, since thereis a possibility of fading of the microwave signal,there is some reluctance to use it in blockingschemes for fear of false tripping in the event ofa fade during a nearby external fault. However,blocking schemes are used occasionally mainlybecause the tripping scheme requires specialcircuitry (as described earlier) in order to trip onsingle-end feed to a fault.

The communication equipment multiplexed onto a microwave system for protective relaying is

30

invariably of the frequency-shift type, andusually of the high speed variety. Figures 13 and14 are representative of the tripping andblocking schemes respectively. Since, asmentioned above, the microwave signal canfade, some of the frequency-shift receiverequipments include channel status detectorsthat operate into the relay logic to incapacitateall tripping when the channel conditions are notnormal. The ability to trip is then automaticallyreinstated when normality returns. With such anarrangement, complete loss of receiver outputwould incapacitate tripping. If the scheme werea blocking scheme similar to that of Figure 14,complete loss of channel during an externalfault would permit a false trip unless anincapacitating feature were included in thescheme.

The receiver has only two outputs (high andlow). Since the scheme trips on internal faultsduring the absence of the low-shift output, andsince the absence of both the low and high shiftoutputs incapacitates tripping (where used), theimplied requirement for tripping is the presenceof the high-shift receiver output. While such ascheme is called a blocking scheme it appears tobe, at least by implication, a tripping scheme.

In any case, there is nothing about a microwavechannel to alter the previous discussionconcerning phase comparison protection. Thesame basic schemes may be used with theunderstanding that the microwave signal canfade on occasion. For the most part, phasecomparison relaying schemes over microwavechannels have been of the tripping types.

Pilot Wire Links

There are very few, if any, privately owned pilotwires in this country that are used as a link inphase comparison schemes. However, suchapplications would require a frequency-shiftcommunication equipment used in either a

tripping or blocking mode as indicated inFigures 13 and 14 respectively. Aside from theconsiderations involved in tripping for a faultwith single-end feed, which were discussedpreviously, the selection between a blocking anda tripping scheme will generally result from acompromise between security and reliability. Inorder to make such a selection, consideration ofthe pilot pair, its protection, and its physicallocation in relation to power conductors mustbe evaluated.

In general, a high speed channel would requirepilot wires that have a frequency response thatis somewhat better than the standard telephonevoice circuits.

Possibly because of the uncertainties of channelcharacteristics, plus the availability of pilot wirerelays that are much lower in overall cost, phasecomparison over privately owned pilot wires isnot a common application.

Leased (Telephone Company) Facilities

There has been some use of phase comparisonrelaying over leased facilities including voicegrade pilot wire circuits. In general, if acustomer requires or specifies thecharacteristics of a leased channel, the localtelephone company could provide this link overmicrowave, cable, even pilot wires, or acombination of these. In such cases theselection between tripping and blockingschemes will depend on the performance of thechannel as specified. The same basic schemes ofFigures 13 and 14 would apply.

Summation of Blocking vs. Tripping Schemes

The foregoing discussion of blocking andtripping schemes was presented without thebenefit of a concise definition of these terms. Asindicated in the discussion, the difficulty ofmaking such definitions which would always

31

apply is brought about by the channel statusfeature used in some frequency-shift blockingschemes. Such arrangements tend to behybrids. Thus, the following simple definitionsexclude any considerations of channel statusfeatures:

A blocking scheme is one that requires aspecific output signal from the associatedreceiver in order to block tripping. Trippingcan only take place during the time that thissignal is absent.

A tripping scheme is one that requires aspecific output signal from the associatedreceiver in order to permit tripping. Trippingcan only take place during the time that thissignal is present.

Where channel status logic is used, thesedefinitions will have to be modified to meetthe exact logic of the scheme.

In general, the selection of a blocking or atripping scheme is one that should be made inconjunction with the chosen channel and with aknowledge of the channel characteristics in theface of system noise. Many differentcombinations are possible, but of these, only aselected few will meet any given set ofrequirements.

SINGLE vs. DUAL PHASE-COMPARISON

In all the phase-comparison schemes describedso far, a trip attempt is made only every otherhalf cycle. In the examples illustrated, this wasevery positive half cycle. Such schemes aretermed single phase-comparison as against dualphase-comparison where a trip attempt is madeevery half cycle, positive and negative.

The only advantage of dual-comparison is thatits maximum operating time to trip on internalfaults will be a half cycle faster than the

maximum time for the single phase-comparison. The minimum times for bothschemes will be the same. This difference inmaximum time results because a fault couldoccur at such an instant in time when thecurrent is just going negative. Under suchconditions, the single phase-comparison wouldhave to wait till the next positive half cyclewhile the dual phase-comparison could trip onthe upcoming negative half cycle.

While, as a general rule, high speed operationand security are on opposite sides of the coin, itis possible to design dual phase-comparisonschemes that can provide the added speed withlittle or no loss in security. However, theseschemes are somewhat more complex thanequivalent single phase-comparison schemes.Figure 15 illustrates the dual phase-comparisontripping scheme that is the counterpart of thesingle phase-comparison scheme of Figure 13.The differences are noted below:

1. The dual scheme uses two separatecomparer-integrator combinations, one forthe positive half cycle and the other for thenegative half cycle.

2. A three-frequency, frequency-shift channel isused in dual phase comparison. The high-shiftoperates in conjunction with the positive halfcycle while the low-shift works with thenegative half cycle. When the channel is notkeyed to either high or low, it operates on thecenter frequency. There is no center frequencyoutput from the receiver into the relaytripping logic.

3. AND3 is included to make it impossible to keyboth frequencies simultaneously. It also givespreference to the low-shift which is sentcontinuously when FDL is dropped out. Thus,on single-end feed tripping can take placeonly on the negative half cycle.

32

The center frequency, while not actually used inthe relay tripping logic, adds security to thescheme during transient conditions.

The dual phase-comparison scheme of Figure 15could be modified to operate over a two-frequency, frequency-shift channel byeliminating AND3, FDL, NOT1, and the centerfrequency. The transmitter could then bearranged to send the low-shift frequencycontinuously except when keyed by the positiveSQ. AMP to the high shift frequency. Thisarrangement, though simpler than the three-frequency scheme, is deemed to be less secure.

Figure 16 illustrates a dual phase-comparisonblocking scheme using a two-frequency,frequency-shift channel. Since one or the otherof the two frequencies must be on at all times,and since both are blocking frequencies, thereappears to be little need for an FDL function.Thus, it is not included. When the transmitter isnot keyed, it sends low-shift continuously andwhen it is keyed by the negative squaringamplifier, it shifts to high for the negative halfcycle. This scheme is simpler than that of Figure15 but probably is not as secure.

There does not appear to be any good purposefor a three-frequency channel in dual phase-comparison blocking schemes since the centerfrequency would not add to the security, orotherwise improve the performance.

It is interesting to note that a dual phase-comparison scheme using an ON-OFF channelwould have to be a combined blocking andtripping scheme. During one polarity of halfcycle, it would have to trip on absence of anyreceived signal (blocking), and on the otherpolarity of half cycle, it would have to trip in thepresence of the received signal.

In general, it may be concluded that dual phase-comparison may be accomplished in the

blocking and in the tripping modes. The overallperformance of the scheme will be dependenton the characteristics of the channel selected.While dual phase-comparison will reduce themaximum tripping time, it does so at theexpense of simplicity and possibly somesecurity depending on how it is accomplished.

REFINEMENTS TO BASIC SCHEMES

There are a number of standard refinementsthat are required and normally included in allphase comparison schemes. These will bediscussed in terms of the basic blocking schemeof Figure 4, but will apply generally to allschemes, sometimes in a somewhat differentform.

SYMMETRY ADJUSTMENT

As was noted in a previous section, receivers arenot always symmetrical in their response. Thatis, if a transmitter is keyed on and offsymmetrically every half cycle, the remotereceiver output would not necessarilycorrespond exactly to the keying signal. Forexample, if an ON-OFF transmitter were keyedon for a half cycle and then off for a half cycle,and so on, the remote receiver output might beon for more than a half cycle and off for lessthan a half cycle. This effect is primarily due tothe filter response in the receiver and iscommon with ON-OFF type of equipment. It isnot a constant value but rather depends onoperating frequencies as well as received signalstrength. Thus, this asymmetry may vary fromequipment to equipment and from time to time(as atmospheric conditions change) in service.

Frequency shift channels are generallysymmetrical in their response when thediscriminator in the receiver is balanced. If thediscriminator is biased to one side or the otherthe receiver output tends to favor the side towhich it is biased.

34

Because of this, all phase comparison schemesthat may operate with asymmetrical channelsare equipped with a symmetry adjustment.Figure 17 illustrates the same scheme as that ofFigure 4 except that it includes a symmetryadjustment and a phase delay adjustmentfeature. The phase delay adjustment will bediscussed subsequently.

The symmetry adjustment (0-3/0-3) is in thekeying circuit to the transmitter. It is set witheither a time delay pickup or a time delay dropout depending on whether the receiverelongates or shortens the received signal. Thetime setting is made in the field after thetransmitters, receivers, and coupling equipmenthave all been tuned and adjusted for propersensitivity. The proper setting is obtained bykeying the transmitter on and off by means of asymmetrical sinusoidal output from the mixingnetwork. Then, while this is taking place, thetime delay pickup or dropout of 0-3/0-3 isadjusted so that the remote receiver yields asymmetrical output.

The results of the symmetry adjustment aretwofold. First, the remote receiver output issymmetrical, and second the keying signal maybe phase shifted in the lagging direction fromthe actual squaring amplifier output. This latterresult is not desirable, but fortunately it may bemitigated. It is obvious that if the keying signalis shifted in phase from the local squaringamplifier output, the remote received signal willbe phase shifted from the current at the keyingterminal. In addition to this there is thepropagation delay in getting the communicationsignal from the transmitter to the remotereceiver (1 millisecond per 186 miles) plus thedelay in the receiver itself. All of these add toeach other to produce a receiver output thatmay be significantly phase delayed from thecurrent at the remote end of the line. This isundesirable because it introduces an error in thephase comparison. There is no way to eliminate

this phase delay but there is a way tocompensate for it. This compensation isaccomplished by the phase delay timer (0-8/0-8)in the comparer input circuit.

PHASE DELAY ADJUSTMENT

The phase delay timer (0-8/0-8) is a timer that isset with a pickup and a dropout delay that areequal to each other so that it introduces a phasedelay. Its output is the same shape as that of thesquaring amplifier but delayed in time by thesetting. This time delay setting is made in thefield to be just equal to the sum of the threedelays (symmetry adjustment, propagation, andreceiver) discussed above. Thus, with thisarrangement in the blocking scheme of Figure17, an external fault would produce a receiveroutput exactly in phase and symmetrical withthe output of the phase delay timer. This isnecessary for proper blocking. For internal faultsthe output from the phase delay timer would besymmetrical with, but 180 degrees out of phasewith the receiver output. This is necessary fortripping. It should be recognized that any errorsin these adjustments can reduce the trippingmargins for internal faults and/or reduce theblocking margins during external faults.

It is interesting to note that the setting of thephase delay timer is dependent on the channeloperating time, and that the total tripping time ofthe scheme is affected by this timer setting.Thus, the tripping speed of the scheme is to thatdegree dependent on the channel operating time.

TRANSIENT BLOCKING

Transient blocking is a feature that is included inall phase comparison schemes. It adds to thesecurity of the scheme during and immediatelyafter the clearing of external faults. Figure 18 isa representation of Figure 17 except with thetransient blocking logic added. This consists ofAND6 and the 20-40/40 timer.

35

The basic logic of the transient blocking scheme issuch that if a fault is detected, as indicated by theoperation of FDH, but no trip takes place, asindicated by no output from the integrator (3/18)timer, AND/6 produces an output to the transientblocking timer (20-40/40). If this condition persistsfor a time that is long enough for the transientblocking timer to produce an output, tripping isblocked via the NOT input to AND5. This blockingof a trip output persists for 40 milliseconds afterAND6 output disappears as a result of FDHresetting or 3/18 producing an output.

The pickup time delay setting of the transientblocking timer is somewhat longer than theexpected time difference between FDH pickupand a 3/18 output during an internal fault. Thisinsures no delay in tripping in the event of aninternal fault, as well as prolonged blockingduring the clearing of an external fault duringwhich transient power reversals may tend tocause false tripping. Actually this transientblocking feature provides an added margin ofsecurity since the scheme is designed to besecure without this circuitry.

SUMMATION

It should be recognized that while thediscussions in this section were stronglyoriented toward a blocking scheme over an ON-OFF channel, the same or similar comments andcircuits apply to all the other schemes. In somecases, for one reason or another, these functionsmay differ somewhat in arrangement from thelogic illustrated in Figure 18. For example, it isobvious that the symmetry adjustment may beleft out of the keying circuit if it is included inthe output of the receiver, and in someapplications this is done. In some cases offrequency-shift equipment, the symmetryadjustment is an integral part of the receiver,and so would not be shown on the logicdiagram. In any case these functions areincluded in all schemes in one form or another.

MULTI-TERMINAL LINES

Up to this point these discussions havepertained principally to two-terminal lines.Phase comparison schemes are often applied tolines having more than two terminals and thoseapplications differ somewhat depending on thechannel equipment.

ON-OFF CHANNEL

The ON-OFF channel equipment is invariablyused in blocking type carrier schemes similar tothat of Figure 18. Since this type of schemeutilizes only one common frequency for all thetransmitters and receivers, Figure 18 will applyto multi-terminal lines as well as two terminallines. A blocking signal sent from any terminalwill be received at all the other terminals toprovide the necessary blocking via the singlereceiver at that terminal.

Since all the receivers and transmitters operateat the same frequency, each receiver will receiveits own local transmitters as well as all theremote transmitters. It will be recalled that thesymmetry adjustment in the keying circuit isbased on the response of the remote receiver.Since the attenuation between a transmitterand its local receiver may be quite different fromwhat it is between the same transmitters and aremote receiver, it is reasonable to expect thatthe output of a receiver responding to a localtransmitter will not be symmetrical. Thus,during internal faults, even assuming that thecurrents at all the line terminals are exactly inphase, it is quite possible that the receivers willproduce outputs somewhat longer than a halfcycle. This is true for two-terminal lines as wellas multi-terminal lines but it can be moreexaggerated on multi-terminal lines because ofthe added transmitters and adjustments thatmay not be exactly correct.

37

While the situation described above will reducethe margin for tripping on internal faults, it isnot generally a problem if proper sensitivitysettings are made on the receivers, and carefuland proper symmetry and phase delayadjustments are made on the relays as well. Theuse of hybrids to attenuate the strength of thetransmitted signal as received by the localreceiver can also be helpful. These are availableon some of the newer carrier sets and they tendto make the received local transmitter signalsappear weaker than the remote transmittersignals.

It is of interest to recall in passing that in mixedexcitation (positive plus negative) schemesduring internal faults, the mixing networkoutput currents at all terminals are not exactly inphase with each other. This in itself cuts downthe tripping margin, and any additional loss ofmargin from the items discussed above is thatmuch more undesirable.

FREQUENCY-SHIFT CHANNEL

Frequency-shift channels are generally used intripping type schemes. Figure 19 illustrates athree-terminal line tripping scheme using afrequency-shift channel. This arrangementrequires two receivers at each terminal. Onereceiver is required for each remote transmitterbecause each transmitter is operated at adifferent frequency. In order to trip, a high-shiftoutput is required from both receiversconcurrently to AND2. A two-terminal linescheme would require only one receiver whichwould operate directly into AND1 without theneed for AND2.

Frequency-shift channels are inherentlysymmetrical, and they do not receive their ownlocally-transmitted signals. Because of this,schemes using this type of channel are notgenerally troubled by the symmetry problems ofthe ON-OFF carrier channels.

STANDARD SCHEMES

In the foregoing discussions it was pointed outthat phase comparison relaying can beaccomplished in many different ways. Thesedifferences relate to such factors as:

1. The type of excitation that is used.

2. The type of transmitter-receiver that is used.

3. The type of channel link that is used.

4. The operating mode blocking vs. tripping,single vs. dual, mho supervision vs. no mhosupervision.

While all the different combinations werediscussed and illustrated to some degree, not allof them are used in practice. For the most partthere are some so-called standard schemes thatare most popular because of a combination ofsuch factors as overall performance, cost, andsimplicity. It should also be recognized that thediscussions and logic diagrams discussed werebasic and generally rudimentary in order topresent the fundamental concepts. Actually thevarious schemes often involve more detailedconsiderations and may appear somewhatdifferent when their logic diagrams arecompared to those discussed. However, thebasic concepts are the same.

Table II presents a listing of typical schemes inuse today. It will be noted that all the schemeslisted in this table except the last two providecomplete protection against single-phase-to-ground as well

Documents

Phase Comparison Relaying