Thermocouple Junctions Are Not Voltage Sources R. P. Reed, Ph.D, PEret Thermocouples, based on the Seebeck effect, remain the simplest, most widely used, electrical sensor of temperature. Thermocouples consist only of thermoelectrically dissimilar conductor legs connected at junctions. The Seebeck emf occurs only in the legs. Therefore, commonplace calibration and thermometry errors relate to degraded thermoelements, not to junctions. A yet commonplace implicit Junction-Source Model incorrectly asserts that Seebeck emf occurs only in junctions. That erroneous concept hides problems that are commonplace in consequential thermometry. Thermocouple applications range widely. Some are of casual indication of moderate temperatures of benign environments. Many now are for critical monitoring and consequential control of extreme temperatures under harsh conditions of industrial process applications. Modern circuits now range from the initial simple loop of only two thermoelectrically dissimilar electrical conductors to distributed complex networks and of several different materials. [ 1-3 ] Properly understood and applied, thermocouples now are capable of accurate and reliable thermometry. However, as now often misunderstood, they have concealed very costly and consequential error. The physical nature of thermoelectric phenomena is complex but now is very well understood. [ 4,5 ] Peculiarly, however, the history, terminology, and simple principles essential for use in thermometry are yet very widely misrepresented in current literature at all levels of sophistication. An initial brief summary of thermocouple history suggests here why confusion has persisted for so long. This article then briefly reviews aspects of a decades-old comprehensive, simple, authentic, practical model, the Functional Model of Thermoelectric Circuits. [ 6 ] Terminology Some thermocouple terms now popularly used are inconsistent or contradictory. The following definitions apply to thermoelectric circuits of all complexity. A thermoelectric junction is an ohmic interface between two electrical conductors that are dissimilar in thermoelectric characteristic. That electrical connection can be simply by temporary point contact. Each of those joined dissimilar materials is a thermoelement. Copyright 2014, R.P. Reed, All Rights Reserved. Published with Permission of the author 1
Each leg could be in the form of rod, wire, foil, liquid, gas, semi-, or super-conductor, etc. Two such legs, electrically paired at a junction, form a thermocouple. (Note: Four all dissimilar thermoelements that share a single junction allow six alternative thermocouple pairings.) Thermoelements might be electrically joined by fusion to form an electrically conductive connecting bead. Such a bead is just an incidental thermoelement of undetermined alloy composition. (Note: A bead is not a junction. A bead or a junction is not a thermocouple nor a source of emf.) How the Technology Evolved The puzzling reason why thermocouple principles, terminology and related history have remained contradictory for almost two centuries is suggested by the following sequence of events. Seebeck s Discovery. At the very infancy of electrical technology and its terminology, T. J. Seebeck and others experimented unsuccessfully to seek a dry-cell source of electricity. [ 7 ] Seebeck s experiments were about dissimilar materials, not about relative temperatures. In 1821, he began many experiments. [ 8 ] He joined both ends of various different pairs of dissimilar thermoelements to form closed series loops. Two junctions electrically connected their ends (i.e., his loop circuit was two thermocouples, back-to-back). He first noticed, only incidentally, that a magnetized needle suspended near a circuit leg deflected while one junction happened to became warmer or cooler relative to the other. Ironically, Seebeck discovered thermoelectricity only because initially he didn t control junction temperatures. Fatefully, in his futile systematic search for a dry cell electrical source, unlike the many other experimenters he accidentally was first to discover the existence of thermoelectric phenomena. Seebeck s consequential observation provided the essential steady galvanic source that allowed G. Ohm to formulate his final of several versions of the now indispensible Ohm s Law. Also, his discovery has resulted in several modern industries. Seebeck neither applied the effect to, nor offered rules for, thermometry. He eventually did accurately rank the relative strengths of his effect for different combinations of electrical conductors. Incorrectly, Seebeck associated the physical effect with the electrical connection points ( junctions ), not the legs. Understandably, he misinterpreted the unknown phenomenon to be thermo-magnetism. He didn t recognize the basic voltaic nature of his effect. Early users associated the effect with current. Current now is suppressed in modern thermocouple thermometry. [ 1-3 ] It is the Seebeck emf that is now best applied in thermometry. 2
Soon, successive recognition, separately by Peltier and then by Thomson (Lord Kelvin), of the only two other thermoelectric effects resulted directly from Seebeck s seminal discovery. [ 9,10 ] Some still incorrectly call Seebeck s independent effect the Seebeck-Peltier Effect. The Seebeck effect wrongly is now said by some to depend on either the Peltier or the Thomson effect or else jointly on both. It depends on neither. Actually, both Peltier and Thomson effects completely vanish in the absence of current. It is only the Seebeck effect (actually best applied absent current) that is significant in thermocouple thermometry. The two other, thermoelectric effects are each currentresultant heat effects, not source emfs. The three independent thermoelectric effects are only energy-interrelated. [10] Properly, the Seebeck Effect is just the occurrence of a net emf across segments of individual thermoelements while their endpoints (not as material junctions) are at different temperatures. [ 1,3,5 ] Seebeck s understandable focus on the junctions was followed two decades later by an influential experimental paper by Magnus who concluded, also incorrectly, that the thermoelectric source definitely was not the legs. [ 11 ] He had experimentally observed the strength of Seebeck s effect to be entirely independent of any possible distribution of temperature along the legs. He therefore claimed, incorrectly, to confirm Seebeck s effect as a junction phenomenon. Magnus influential, but wrong, conclusion was because his thermoelements happened to be essentially homogeneous. The now insignificant Magnus Law is not true if thermoelements are inhomogeneous, as often occurs in service. Measuring junctions affect authenticity of thermometry only thermally, not as voltage sources. [ 1-3 ] A measuring junction might be poorly coupled to the subject, or else its thermal mass might thermally affect the temperature being measured. Apart from instrument error, most thermocouple electrical error in thermometry and calibration now is from inhomogeneity that can be introduced locally during fabrication, calibration, or application. Recognition, diagnosis, and avoidance of such thermocouple problems requires that the user understand that the voltage source is from thermoelements, not junctions, and that most problems are located apart from, but most often adjacent to the measuring junction. [ 12 ] The Seebeck Effect Applied to Thermometry. Later, as the Seebeck effect was applied by others to serious thermometry, users noted that commercial wires of the same chemical composition, even from samples of the same batch and treatment, formed thermocouples of substantially different sensitivity. Users then recognized that initially the thermocouple sensitivity usually varied substantially along even a short length of a wire. [ 13 ] The sensitivity often changed significantly in service. 3
For reliable thermometry, early experimenters had to experimentally screen commercial grade wire to select segments that were acceptably uniform. That should have instructed conclusively that the Seebeck effect definitely is a characteristic of thermoelements, not of junctions. W. P. White, at the Geophysical Laboratory, Carnegie Institute, applied thermocouples to real thermometry. He had to cope with inhomogeneous thermocouples. He described thermocouple principles realistically in a series of scholarly papers. [ 13 ] Early thermocouple users had practiced thermometry based on three observed rules of thermocouple circuit behavior that had progressively been posed by Magnus, Becquerel, and others. It was not until 1940 that Roeser, of the National Bureau of Standards, intending to simplify thermometry, formally codified those three historic but conditional rules. [ 14 ] Focused on precise laboratory application of ideal thermocouples, Roeser identified the rules as a set of necessary and sufficient thermocouple laws ( those of Homogeneous Metals, Intermediate Metals, and Intermediate Temperatures). Those rules applied to commonplace simple idealized series circuits. The rules emphasize junction temperatures. Misrepresented as separate laws, those traditional rules actually are just corollaries that follow from a single, un-named, very simple, concise, and universal law. [ 12 ] Like White, Roesser and others did appropriately caution that the traditional laws conditionally presumed ideal homogeneity. Roesser s statement of those laws collected as a set have been paraphrased without attribution by many later authors. Effectively, thermocouples since have been treated as typically homogeneous with focus conveniently only on temperatures of the principal junctions. Eventually, thermocouple materials specially processed for thermometry became available. The significant practical problem of initial non-uniform sensitivity along thermoelements seemingly was avoided. The historic problem of inhomogeneity of Seebeck Characteristic no longer was emphasized. Taught thermocouple principles then re-focused simply on junctions and on the simple series circuits then used. Why an Authentic Circuit Model Does Matter The false junction-source concept is physically incorrect. Nevertheless, it is implicit in commonplace treatments of thermocouple principles that focus on intended junction temperatures. Attribution of emf to junctions does produce a value of overall net circuit emf that only might be correct, though misleading. However, in use, thermocouples, where too long exposed to adverse environment or excessive temperature, locally change in sensitivity. The sensitivity then depends on temperature distribution along the thermoelements, rather than just on junction temperatures as expected. The narrow focus 4
on junctions diverts attention from the relevant emf source locations and even from the recognition of significant hidden error. Modern circuits usually include several thermoelements, not just two. [ 1-3 ] Like-named thermoelement pairs may be very similar in material or else, properly, very different in their Seebeck Characteristic. The flawed junction-source notion does not suggest which thermoelement pairs presently are contributing emf, in what proportion, whether properly paired, or where error is likely. The typically uncontrolled temperatures of incidental junctions often mis-pair even homogeneous thermoelements to have inappropriate relative Seebeck characteristics. The false junction-source approach conceals error as it develops. It misdirects diagnosis. It does not aid easy avoidance of many commonplace problems that occur in modern thermocouple thermometry. An Authentic Basis for Thermocouple Thermometry The Functional Model of Thermoelectric Circuits A decades-old, simpler presentation, The Functional Model of Thermoelectric Circuits, is well adapted to consequential modern thermometry. [ 1,3,6,12 ] This comprehensive thermoelectric circuit model has only three essential components: 1. A Basic Circuit Element, 2. A Law of Seebeck emf, and 3. A T/X Sketch. This authentic general model applies to thermoelectric circuits of all complexity and even to inhomogeneous thermoelements. It applies to branched thermoelectric circuits and networks. Its underlying principles have long been acknowledged. [ 13-17 ] Its distinctive form was crafted to aid practical users in recognizing commonplace circuit problems and in avoiding them. [ 6 ] The Circuit Element. In the Functional Model the basic circuit element is a thermoelectrically homogeneous segment of arbitrary length within an individual thermoelement, Fig. 1. This simple element is directly usable in conventional electrical circuit network analysis programs. The element is a conventional non-ideal emf source with internal resistance. Both the emf and the resistance values are nonlinear functions of segment s temperatures. The segment resistance is consequential only if significant Seebeck current or externally supplied current is allowed. Undesirably, current results in ohmic voltage drop, therefore, thermocouple thermometry, that uses Seebeck terminal voltage, is best conducted under null current (aka, open-circuit ) conditions. 5
An entire thermoelement leg (but only while overall-homogenous) is a single segment. An entire locally inhomogeneous thermoelement is represented as a linked series of any number of local adequately-homogenous segments. As represented, the simple element applies to the usual static or low frequency thermoelectric applications. The Seebeck Characteristic The Seebeck emf of an individual thermoelement segment is expressed as its Absolute Seebeck Characteristic, E(T). Physically, the Seebeck Characteristic is a three-dimensional transport property of individual thermoelements, not of junctions. [ 4,5 ] A corresponding Relative Seebeck Characteristic applies to any pair of segments only while they happen to share the same two endpoint temperatures. The sensitivity of each segment is the Seebeck Coefficient, σ(t) = de(t)/dt. Both the Characteristic and the Coefficient functions are of significantly, non-linear form. The expressions, standardized for thermometry by letter-type, are experimentally, not theoretically, determined. For precision, NIST (The National Institute of Standards and Technology) had to express E(T) and its inverse T(E) by lengthy polynomials of up to fourteenth degree. The polynomial expressions, not the thermoelement materials nor tables, now formally define the standardized letter-designated thermocouple types. Note that now thermoelement pairs of very different materials, but like relative characteristic are designated by the same letter-type. [ 18 ] The Law of Seebeck emf Because the Seebeck Characteristic, E(T) is significantly nonlinear the net emf between the endpoints of a thermoelement segment of arbitrary length must be determined by a very simple integration, Eq. (1), where the limits are the temperatures of the two segment endpoints. [1,3,5,12] E(T) = (1) Fortunately, despite the nonlinearity of σ(t), the result is simply E(T) = E(T2) E(T1). (2) The net segment Seebeck emf depends simply on just the two values of E(T) at temperatures of the segment end point. These are two alternative expressions of one simple universal law. 6
Note closely, however, that the simple result applies only while the Seebeck Characteristic of the segment remains a function of temperature, T, alone. That limitation is one dominant significance of spatial inhomogeneity of the Seebeck Characteristic. Excessive inhomogeneity (i.e., beyond tolerance) of Seebeck Characteristic precludes accurate thermometry. This commonplace, often implicit, law was not explicitly recognized as a law by Seebeck nor formalized as a law by the earliest thermoelectric investigators. The Functional Model distinctively identifies this classic relation as the Law of Seebeck emf, the comprehensive fundamental law that governs all thermoelectric circuit emf behavior. That law, traditionally unnamed, underlies all valid thermoelectric models. It is applied, inappropriately, even by those who attribute that emf to junctions. The law applies to individual thermoelement segments and to those of any thermally-paired thermoelements while they happen to share the same pair of temperature endpoints in all circuit configurations. Fig. 1 illustrates one representative temperature distribution along a locally homogeneous circuit element. Over the arbitrary span of the segment the illustrated temperature has both upper and lower temperature extremes. This segment incidentally experiences a very wide range of spatial temperature gradients, dt/dx. It is profoundly significant that the segment net emf is independent of extremes of temperature and of temperature gradients, dt/dx, within the segment. Therefore, net thermocouple emf relates only indirectly to literal temperature gradient. Just as important, isothermal segments create no Seebeck emf regardless of material or homogeneity. Thermoelectric inhomogeneity is spatial variation of the Seebeck characteristic, E(T), along an individual thermoelement. Actually, no real thermoelement is perfectly uniform in Seebeck characteristic along its full length. However, modern manufacture now assures that most thermoelement wire as delivered initially is reliably homogenous overall within a standardized small tolerance. [ 2 ] Many authors inconspiculously have cautioned that their model presumes homogeneity. That indeliberately suggests homogeneity to be the usual persistent state. [ 14,16,17 ] In practical severe use within their standard temperature range, 7
Seebeck characteristics of thermoelements do change. [ 15,17,19 ] Thermoelements chemically react with adjacent thermoelements or insulants. They change metallurgically. Circuits may be electrically compromised by dramatic reduction of isolation resistances as at higher temperatures. Significantly, a thermocouple that has become significantly inhomogeneous has no universal global Seebeck Characteristic or calibration. Apparent sensitivity of an inhomogeneous thermoelement depends on temperature distribution along it. A very precise, even certified, re-calibration of a significantly degraded thermocouple might be invalid. A thermocouple is useful for general thermometry just while acceptably inhomogeneous. Initially homogeneous within tolerance, while used only well within their temperature range thermocouples remain practically unchanged. However, at the higher temperatures within standard temperature range, sensitivity can change beyond tolerance progressively, irregularly, and unrecognized over durations of hours, days, or weeks. [ 1 ] The change usually cannot be noticed in service. Changes of E(T) are localized, due to strain or pressure, result from chemical or radiation exposure, etc. Changes due to strain might be reversible. Most changes are not. Sensitivity changes are because of time in adverse environment, not from time alone. Thermocouples do not passively age. Peculiarly, the usually most degraded segments often contribute no thermometry error. The most damaging thermal exposure typically is over a wide thermoelement span that includes the measuring junction. In proper thermometry the measuring junction and nearby adjoining thermoelement segments lie within a nearly isothermal region. Therefore, even while substantially degrading, the most changed, isothermal, segments nearest the junction contribute no emf. Most measured thermocouple voltage is from nearby contiguous degraded segments over that span where temperature decreases from the junction temperature toward ambient temperature. Even severe localized sensitivity change remote from the measuring junction conditionally would result in little error while the temperature there happened to remain nearly uniform across the length of the inhomogeneity. So-called Drift of sensitivity is the sign of progressing inhomogeneity. Progressing inhomogeneity is recognized during thermometry only in situations where the indicated temperature varies implausibly while the measured temperature is expected to remain nearly constant. 8
More definitely, actual sensitivity change is documented only when the actual imposed measuring junction temperature is independently known, as in systematic laboratory Drift tests. [ 1 ] Commonplace Open Circuit detectors do not warn of progressing error. Thermocouples that degrade in service usually exceed tolerance well before circuits open. Neither traditional swept-temperature-spike tests nor sampled segment calibrations are valid tests of inhomogeneity. [ 1 ] Special unconventional validation circuits can independently reveal problems and definitely diagnose many commonplace defects during thermometry. [ 1,20 ] The spatial distribution of inhomogeneity can be measured but only by special uncommon test procedures apart from thermometry. [ 19 ] The T/X Sketch. The sketch is intended to illustrate circuit function. The reader may find it helpful first, briefly, to view the illustrated sample T/X sketch, Fig, 2, before reading its description. The practical T/X sketch is very much simpler in application than in description. Several references offer much more detail and varied applications. [1,3,12 ] A thermocouple circuit traditionally is described by a conventional electrical schematic. Incidentally, the similar T/X sketch serves even better as a functional circuit diagram both for authentic instruction of thermoelectric principles and requirements as well as for description of connectivity for circuit installers. The T/X sketch portrays relative temperatures of all circuit junctions to show how they determine which segments of which remote thermoelements are thermally paired, or else mis-paired, to contribute to net emf. It is principally for visualization. It is not drawn to scale. It is not used for graphic calculation of emf. Secondarily, it does support calculation of plausible error from different assumed temperature distributions. Once the principles are comprehended, the T/X sketch is just imagined. 9
Traditional false junction-source circuit concepts impractically ignore commonplace inhomogeneity and focus attention on just the principal junctions, measuring and reference. Any of several additional, incidental junctions are other material interfaces (e.g., connector and extension endpoints, junction boxes, hidden splices, etc.) that lie between the principal junctions. Reference junction temperatures must be well known and carefully controlled. [ 1-3 ] The measuring junction temperature is carefully imposed yet its value is only deduced. Consequentially, the important temperatures of incidental junctions usually are neither controlled nor known. The practical importance of the temperatures of inconspicuous incidental junctions and thermoelements rarely is acknowledged. Most incidental junctions occur as cable interfaces with junctions coincidentally very close together, They are electrically isolated but in thermal contact. Such adjacent junctions naturally tend appropriately to share the same local temperature. However, as in distributed large scale industrial field installations, such connections may be by ordinary links or jumpers on electrical terminal blocks with related junctions widely separated and subject to a wide range of heat, cold, and drafts if not deliberately insulated and thermally lagged. Such deliberate passive control is easy and inexpensive. The underlying thermoelectric principles have long been acknowledged. [ 13-17 ] However, the distinctive T/X sketch was crafted to represents circuits in a practical way that forcefully highlights where the Seebeck emfs (and so possible error) actually occur. The sketch illustrates how thermoelement segments, even if widely separated in the circuit, temporarily thermally pair across different temperature zones that are bounded by the principal or incidental junctions. Each temporary thermal pairing of segments has an effective Relative Seebeck Characteristic, across a particular temperature zone. A T/X sketch illustrates, by identifying functional temperature zones, the critical temperature distribution around a thermoelectric circuit. It shows why control of relative temperatures of particular incidental junctions is essential to avoid hidden error. Significant error occurs when temperature distribution inappropriately pairs even homogeneous segments of thermoelements. The simple T/X sketch encourages the routine recognition of momentarily functioning pairs, particularly those that unnoticed are improperly thermally paired. The sketch aids in sensitizing users to circuit problems. It encourages visualizing, diagnosing, and avoiding thermoelectric problems 10
T/X Sketching Procedure First, as in Fig. 2, the Functional Model plots every junction temperature, Ti, against the sequence, Xi, that each physical junction appears in the circuit. Of course, actual temperatures of incidental junctions usually are not known. Rather, for a T/X sketch, an estimated or arbitrary temperature is assigned to every incidental junction to illustrate or to numerically evaluate the consequence of different plausible temperature distributions. Second, junctions are joined in sequence by lines that represent the thermoelements. Every thermoelement (even of complementary components -- connectors, extensions, spliced segments, etc.) must be shown. Every thermoelement that presently lies along an isotherm between junctions is shown even though it presently is contributing no emf because it is isothermal. Branched circuits or networks parallel some thermoelements. [ 1 ] Finally, an isotherm line is drawn through every physical junction, both principal and incidental. An intersection of any such isotherm, where it crosses a thermoelement, is a non-physical virtual junction. Virtual junctions optionally may be marked by a tic across the thermoelement merely to delineate the temperature zone. Practical Significance of the T/X Sketch Each homogeneous segment of an individual thermoelement, where it spans a temperature zone, has an Absolute Seebeck Characteristic that determines its individual emf contribution to the relative characteristic of that temperature zone. Each pair of adjacent isotherms bounds a temperature zone that thermally pairs, with a particular momentary Relative Seebeck Characteristic, two or more thermoelements that presently span that zone. Significantly, the T/X sketch reveals particular junction pairs that must be held at the same temperature to avoid unintended contribution of emf. As important, the sketch identifies the specific temperature spans over which particular non-isothermal thermoelement pairs presently are contributing part of the net Seebeck emf, whether or not as intended. Properly, each entire thermoelement should span only one temperature zone. However, when temperatures of incidental junctions are not controlled, a single thermoelement, whether or not homogeneous, might unintentionally span more than one temperature zone. Then, it inappropriately pairs with segments of two or more other thermoelements. That results in an unintended emf. Each presently thermally-matched pair has a Relative Seebeck Characteristic of the kind developed for standardized pairings of particular letter-designated thermoelements. Inappropriate pairings have a definite but usually undefined Relative Seebeck Characteristic. Therefore, each such inappropriate pairing introduces an error of indefinite amount that can only be estimated. 11
Just as critical. Unintended thermal pairing of identical, or even very similar, segments result in a gap of essential emf from a temperature zone. Such omission error can only be avoided by control of temperatures of the incidental junctions of such like pairs to make the segments isothermal. The T/X sketch depicts which pairs of junctions must be controlled to have the same temperature, independent of their unknown temperature. Corollaries from the Law A separate informative set of corollaries from the Law of Seebeck emf just complements the model. For brevity, they are presented only elsewhere. [1,3,12] The corollaries are instructive. They are neither essential to the model nor are they directly applied in analysis. The functional significance of each is just to be comprehended, not memorized. They are not used operationally in analysis. They are intended to add useful insight into the practical significance and consequences of the Law in practical application. The corollaries of the Functional Model do generally and directly apply to practical application. They do entirely supplant the more indirectly phrased, unnecessary, out-dated but familiar, thermocouple laws that just specialize to particular circumstances. Examples of the T/X Sketch One realistic simple example circuit illustrates some features of the T/X application. Figure 2 compares two different T/X plots. Both are for the same commonplace compound thermoelectric circuit but at different times. The two circumstances are for the same three intended negative/positive thermoelectric pairs, N1P1, N2P2, and N3P3). In thermometry, the Relative Seebeck Characteristics of the three pairs usually would be very similar, within tolerance, though not identical. (In some common applications, the individual thermoelements properly might all be very different. [ 1-3 ] ) Figure 2(a) shows the circuit as initially homogeneous. It displays an appropriate temperature distribution. Figure 2(b), shows later service-induced inhomogeneity. It also demonstrates a commonplace inappropriate distribution of temperatures of incidental junctions. For thermometry, the primary pair, N3P3, should have its single measuring junction, 4 (or 7), properly in a broad isothermal region of the subject, as shown. That principal pair would have a standard narrow initial tolerance and might be especially calibrated. It, informally, is perceived to be the thermocouple. Reference pair, N1P1, is a conventional thermoelectric extension with both its reference junctions properly at, 0 C. Intended as a lower-priced extension, it would have a broader tolerance and usually would not be calibrated. 12
Figure 2(a) shows intermediate pair N2P2 as appropriately isothermal. Pair 2 might be just a short connector for convenience with easily replaceable internal links. Otherwise, for flexibility, it might be a supplementary lengthy extension cable, spliced hidden within a metal sheath. In Fig. 2(a) there are only two temperature zones, I and II. The net Seebeck emf is generated in those two temperature zones. Most emf should be from pair 3 across zone I. (Often unrecognized, in calibration near room temperature most or all emf might be from uncalibrated pair 1.) Figure 2(b) represents the same physical components with a commonplace different temperature distribution as after prolonged exposure to excessive damaging temperature. Now, the same circuit occupies six junction-defined functional temperature zones, I through VI. Zones I typically span a range, effectively from reference T1= 0C, to ambient temperature, T2. In Fig. 2(b) junction endpoints of N2P2 improperly are at three different temperatures, T2, T3 and T4. Note in both figures that pair 2 had improperly been inserted in the wrong polarity. The thermoelements of pair 2 also were of a Relative Seebeck Characteristic different from those of pairs 1 and 3. Significantly, while Pair 2 remained isothermal (null emf), in Fig. 2(a), neither the wrong polarity nor the material difference affected the net emf. Connector links occasionally are improperly assembled (e.g., both of copper or of the wrong thermoelement materials, or in inverted polarity. ) In service, Fig. 2(b), the uncontrolled junction temperatures of the incidental pair 2 thermoelements were allowed to diverge to three uncontrolled temperatures, T2, T3 and T4. That introduced temperature zones II and III. Across Zone II, the inverted polarity (not detectable from the monitoring end) introduced a doubled error emf. Additionally, across Zone III the dissimilar negative legs of pairs 2 and 3, of different materials, produced an error from the temperature interval between T3 and T4. A novice, or a skilled technician unfamiliar with thermocouples, might use copper instrument cable as a thermocouple extension. If pair 2, viewed incorrectly as just a passive extension, had been of electrical grade copper, an essential emf contribution from that temperature interval would have been missing. In this example the zone III temperatures paired N2 with N3, likely very similar in Seebeck characteristic, to introduce a zone inappropriately of nearly-null emf. Also, damaged segments of thermoelements N3 and P3, were exposed to excessive temperature adjacent to measuring junction 7, as is common in thermometry. Both legs, 13
near the measuring junction, have locally changed in Seebeck characteristic. The dissimilar thermoelements have degraded, each with a different indefinite graduated Seebeck Characteristic and over a different span. Several consequences are evident from this T/X plot. In thermometry and calibration, most emf is generated in a high temperature zone like V and VI. It is that very source region subject to the greatest change that is dominant in thermometry and in calibration. Thermoelements in Zone IV span the tolerable temperature interval closer to moderate ambient temperature. It is emphasis on the temperature zones that is the important feature of the T/X sketch. Their significance is profound. The actual specific junction temperatures and locations and distribution of degraded sensitivities usually are not known. Nevertheless, plausible error limits from estimated junction temperature values or different sensitivity tolerances often can be useful in diagnosis. Most important, initial inhomogeneity is rarely significant. [ 1,19 ] An eventually degraded emf source region, Fig 2(b), almost always is localized adjacent to and near the measuring junction. In thermometry, most net Seebeck emf is generated in Zones 3 through 5 where temperatures range from that of measuring junction down to ambient temperature. It is the upper temperature zones near the measuring junction where segments are degraded and most effect error. In Fig. 2(b) the degraded portion of net terminal voltage is from abnormal regions like V and VI but a significant portion is contributed by normal sensitivity, zone IV. Paradoxically, those most degraded segments while isothermal at the junction contribute no emf. The Significance of the T/X Sketch in Calibration Of practical importance to thermometry is the authenticity and presumed benefit of initial calibration or re-calibration. Calibrating tradition for thermocouple thermometry was established by specialists who in calibration dealt mostly with pristine homogeneous thermocouples. Therefore, traditionally, a primary emphasis in calibration has been on reducing longitudinal heat conduction to the junction along adjacent ambient temperature thermoelements. That presumes calibration accuracy is improved by deeply immersing the measuring junction in an isothermal zone of known temperature. Note, however, that immersion depth that makes degraded segments isothermal can make even a severely degraded thermocouple seem nearly normal during the most precise recalibration. Where isothermal, it is not the thermocouple junction, nor the contiguous damaged thermoelements, that are being calibrated. 14
Fully immersed isothermally, error from segments that were degraded from exposure is concealed. While only portions of degraded segments adjacent to the junction are held isothermal, the most carefully conducted calibration is invalid even though impressively certified. That is apparent from the Functional Model and its T/X sketch. The Significance of Isothermal Control of Incidental Components Many thermoelements and incidental junctions of modern circuits are inconspicuous. In Fig. 2(a) neither the materials nor the polarity of incidental N2P2 affected the net emf, but only because that component happened to be isothermal. Note particularly that thermoelements and incidental junctions such as Pair 2 joined with Pair 3, often are hidden within a metal sheath. Pair 2 might reside hidden within in a connector body. Also, less costly or more flexible extension thermoelements, used with precious or refractory metal thermocouples, expose to inconspicuous error. Only as a pair, the extension is intended to have a Relative Seebeck Characteristic very similar to the costly measuring thermocouple. Individual thermoelements of such specialty extension pairs usually are of Absolute Seebeck Characteristics very different from those of the individual thermoelements they join. [ 1,2 ] Also, in Fig. 2b, the same circuit temperature zones II and III illustrate two different effects, a) insertion in opposed polarity and b) inappropriate thermal pairing of thermoelement segments. Across Zone II the intended thermoelements were paired, but inserted in wrong polarity. As inverted in polarity, if not isothermal the amount of emf error would have been doubled. Alternatively, N2 and P2 might actually both to be of the same material (e.g. commonplace electrical grade Cu/Cu links or electrical cable). Otherwise, like the similar N2/N3 pair across zone III, their sensitivities might be very similar. In either instance, mispaired they would contribute an inappropriate nearly-null emf over an essential temperature span.. Neither would have contributed any of the net Seebeck emf essential to thermometry from the uncontrolled temperature zones they happened to span. Summary Only segments of thermoelements contribute Seebeck emf. Junctions do not. Initially homogeneous thermocouples become substantially inhomogeneous only where exposed too long to excessive temperature, even within their standardized range, or to damaging environment. Pairs of homogeneous thermoelements can be thermally mis-paired to cause substantial error. 15
Inconspicuous incidental junctions typically are not controlled in temperature. They define functional temperature zone boundaries. However, appropriate pairs of junctions must be held at the same, even though indefinite, temperature. Often that occurs coincidentally. Nevertheless, deliberate attention and assurance of their proper relative temperatures is essential to authenticity as well as to accuracy. Thermoelements that are not intended as emf contributors deliberately must be forced to be isothermal to avoid error. However, if thermoelements are inappropriately isothermal a lacking emf introduces consequential error. The distinctive Functional Model of Thermoelectric Circuits aids users in recognizing, understanding, and avoiding thermocouple thermometry problems. Related thermoelectric topics are treated elsewhere. [ 21 ] References 1. R. P. Reed, Thermocouple Thermometry, pp. 70.1-70.44, In: J.G. Webster and Halit Eren (Eds.), Measurement, Instrumentation, and Sensors Handbook, 2 nd Edition, 2, CRC Press, 2014. 2. Park, R. M., Ed., Manual on the Use of Thermocouples in Temperature Measurement, Fourth Edition, MNL 12, ASTM International, 1993 3. R. P. Reed, Principles of Thermoelectric Thermometry, Chap. 2 In Ref 2, pp. 4 42, 1993. 4. Pollock, D. D., Physics of Engineering Materials, Prentice Hall, 1990. 5. Pollock, D. D., Thermoelectricity Theory, Thermometry, Tool, STP 852, ASTM, 1985. 6. Reed, R. P., The Temperature / Circuit Position Diagram A Tool for Thermoelectric Circuit Analysis, Res. Rpt. SCDC 72-1065, June 1972. 7. Finn, B. S., Developments in Thermoelectricity, 1850-1920, Ph.D. Thesis, 1963. 8. Seebeck, T. J., On the Magnetism of Galvanic Circuits, Abh. Koenig. Akad. Wiss. Berlin, pp. 265-373,1821-1822. 9. J. C. A. Peltier, New Experiments on the Heat of Electric Currents, Ann. Chim. et Phys., 56, pp. 371-386, 1834 10. W. Thomson, On the Thermal Effects of Electric Currents in Unequally heated Conductors, Proc. Edinburgh Roy. Soc., 7, pp. 49-58, 1855 11. G. Magnus, Concerning Thermoelectric Currents, Abh. Koenig. Akad. Wiss. Berlin, pp. 1-32, 1851 12. Reed, R. P., Thermoelectric Thermometry A Functional Model, Temperature *, 5, Part 2, pp. 915-922, 1982 13. W. P. White, What is the Most Important Portion of a Thermocouple?, Phys. Rev., 26, pp. 535-536, 1908. 14. Roeser, W. F., Thermoelectric Circuitry, J. Appl. Phys., 11, pp. 388-407, 1940 16
15. Fenton, A. W., The Travelling Gradient Approach to Thermocouples, Temperature* 5, Part 3, pp. 1973-1990, 1940. 16. Moffat, R. J., The Gradient Approach to Thermocouple Circuitry, Temperature *, 3, Part 2, pp.33-38, 1962. 17. Bentley, R. E., The Distributed Nature of EMF in Thermocouples and Its Consequences, Aus. J. Instr. Contr., 1982. 18. Burns, G. W., Temperature-Emf Reference Functions and Tables Based on ITS 90, NIST Monograph 175, 1993. 19. R. P. Reed and W. A. Bauserman, Measurement of Thermoelectric Inhomogeneity of Thermocouples, Proc. 39th Int. Instr. Symp., ISA, 1993. 20. Reed, R. P., Validation Diagnostics for Defective Thermocouples Circuits, Temperature *, 5, Part 2, pp.931-938, 1982. 21. Bibliography. Publications of R. P. Reed on Thermoelectric Thermometry, 1964 2014, temperatures.com, 2014. ----------------- Temperature * = Temperature Its Measurement and Control in Science and Industry, Proceedings of the decennial International Temperature Symposium, ISA/AIP. Citing this work: Reed, R.P. "Thermocouples Are Not Voltage Sources!" Temperatures.com. Temperatures.com, Inc. Tues. 21 Oct.. 2014. Web. <http://www.temperatures.com/thermocouples are not voltage sources!> 17