aKI Co p- ENGINEERING EXPERIMENT STATION STRUCTURAL RESEARCH SERIES S-15 ENGINEERING USRARY UNIVERSITY OF ILLINOIS NOTES ON THE APPLICATION OF SR-4 STRAIN GAGES By H. C. ROBERTS UNIVERSITY OF ILLINOIS URBANA, ILLINOIS TECHNICAL REPORT Project N6onr»71j Task Order V Project Designation No, NR 031-lfi2 NOTES ON Tlffi APPLICATION OF SR-^ STRAIN GAGES by Howard C, Roberts Research Associate Professor of Civil Engineering An informal discussion of some of the experimental difficulties which may be encountered in the application of wire-type resistance strain gages with the commercial- ly available instruments for their use,, A technical repov t ±n coC'pe^i.-.-.^-.. ^-. ..... THE UNIVERSITY OF ILLINOIS, DEPARTMENT OF CIVIL ENGINEERING and THE OFFICE OF NAVAL RESEARCH Ur : ^iDois 20 July 1949 ' Revised s 10 February 1951 I I i\ ^ :2a I (I? Br,5^f- Co p. Z2 NOTES ON THE APPLICATION OF SR-4 STRAIN GAGES. CONTENTS Page The General Problem o.^ooooooooooooooooo 1 The Available Instruments „ o » o » » „ » . o » « <> » „ » o 3 S^L 0(w-dL1|JLX Ooooo i;;oeoooo oooooooo o«ooco XXii LIST OF FIGURES Fig, No, Page lo D-Co Bridge Circuit oo<.ooo^oooo..o<.oool8 2o A-C„ Bridge Circuit » , o o o » o o <. » , o » » .. o » , 18 3. Photographs of Typical Instrxments o » » » o o » » » >, 19 4.0 Diagram of Carrier System o » <. o <> o » « » » . o » « « 20 5o Photographs of Switch Units ooo<,ooooo.<,oo<,20 i J ij NOTES ON THE APPLICATION OF SR=4 GAGES and the associated instrumen'ts . This is a revision of a report of the same title, first issued in Julys, 194.9o No material has been deletedo The new material includes some not available at that time; and some already^^fvailable material requested by users of the first edition of this report. The General Problems The vdde application of the wire-type electric resistance strain gage is due largely to its low cost and its relatively good electrical stability when properly applied,, Unless used with suitable precautions, however, these gages way give quite misleading and inaccurate results tinder certain circumstances. During operation, the change in ohmic resistance of such a gage is quite small, so that a very sensitive instrument is required as a reading auxiliaryo Instruments of such high sensitivity often have other characteristics interfering with their convenient use. Many workers have encountered experimental difficulties, but comprehensive discussions of these dif f icultles ^ and of the probable troubles to be encountered with commercially-available instruments, have been absent frcm the literatureo This di.scus3ion cannot be rcmplet- ^ '- -^ible to anticipate here all the factors whj. ■:':.. m.-g. ^_ ... .. u_- -z test. At the risk of seeming to "talk dovn* . coo ...tsr of the instruments, the discussion has been made as non-techni c»le. It is felt that perhaps many of those in need of this ... are not specifically trained in alternating-current bridge For most field work the resistance .. i'r gage ii . _.— e^ted in a bridge circxiit. (For some applications , v ^^ynamic effects are ©f imterestj, a simpler circuit is used, i here,) With such a bridge circuit, direc . .... c... -.■ . .. — _ _ sensitive galvanometer can be used| but because of the many ej.per imental difficulties encountered in using sensitive i,, it has become almost standard practice to ec^'' ._ .., :ridge with an amplifier and a more rugged indl. The frequency of the alternatirf ■— -t;. . v.; — ; -^--Ite these strain-gage bridges is almost invariably than the oO-cycle power ciirrent. There are two principal reasobo for thiss fixstj the earcit- ing current is usually generated in the equl rr--'^ -■••'■■ >v ^ ••-■.tji-- ^-i,.-..,,r -:• /i-^ oscillator, which may occupy little spat e * i high| andp second, in order to avoid con^ currents (which are everywhere) it is dee >u- mote from the 60"Cyw^le current or its nr.- ^ tamination of the bridge-circuit out p. mmon practice to introduce filtering or dit .■ .>. - ^c.^. ^.- ihls trouble. For dynamic records, with ec . . arrler^type Digitized by the Internet Archive in 2011 with funding from University of Illinois Urbana-Champaign » http://www.archive.org/details/notesonapplicati15robe ! 2o amplifiers, it is necessary to use carrier frequencies trom four to ten times as high as the maximum strain frequency to be measuredo Instrxunents intended for static tests only, such as those described herep employ similar amplifiers for convenience. The installation and the balancing of a direct»current bridge is a straightforward and relatively simple matterp but this is not true of alternating-current bridges » In a d-=c bridge only steady currents flow, and their flow is controlled only by tne resistances of the various parts of the circuit o In an a-c . bridge the current flow is affected also by in- ductance and by capacitanceo Consequently;, a bridge circuit which is per= fectly balanced for direct current may be hopelessly off balance for 1,000 cycle alternating currento Gage bridges excited by alternating current re<= quire balancing for resistance^, for capacitance, and for inductance^ and the several balances must be established simultanecusly, Thfere is often some inconsistency, even in the published litera- ture, in the statements of exact balance conditi©ns for a^c •, bridge circuits. In a direct=current bridge, like the circuit of Figo 1, the bridge is bal- anced when the resistances of the arms of the bridge are such that no potential difference appears between the points A and B^ across the indicate- ing galvanometero In an alternating-current bridge^, the bridge is balanced when no potential difference appears across the indicating instrument at any point in the alternating curren t cjeleo The average current through the in- dicating instrument may be zero without a true balance having been establish- edo A d=c bridge is usually balanced by observing the deflection of its galvanometer 5 these deflections „ by their direction, indicate the polarity of the unbalance ^ and thus the direction in which adjustment should be madeo In an a-c orldge there is an analogous change in •polarity* of the output of the bridge j on one side of the balance condition the output potential is in phase with the exciting potential, while if the bridge is un- balanced in the other direction the output potential is 180 dego 2^ of phase with the excitation. If there is a reactive unbalance (that is, one involv- ing capacitance or Inductance) there will be some "phase-shift* in addition to the 180 dego "phase-reversal* found at the balance point. With bcth d -c and a-c bridges, this change in "polarity" at the time of passing through the balance point is often employed as the criterion for balance,, because of its convenience o It imst be recognized that in ©very a=^ bridge there are capaci- tances and inductances, not shown on the diagram,' These are strays—due to imperfections in the components composing the bridge, unwanted coupling be- tween wires, etc. Perfect balance for the bridge circuit can be reached only when all of the circuit parameters— resistance,, capacitance, and in- ductance— have magnitudes in the proper ratios. The drawing and formula of Figo 2 Illustrates this, A minimum (b-ut uct lero) bridge output may be secured by adjusting any one of these three parameters; this minimum should not be confused with the true balance. In Fig, 2, all of the resistances ir,. each arm of the bridge are considered as "lumped" and occurring in the positicns shown;; the same is true of inductances and capacitances, Ot '.j this is net true, nor is » I j i it true that any circuit parameter can be altered with no change in any other parameter.. Because of thiSp the most stringent requirements as to balance conditions must be imposedo The user of resistance=type strain g&ges wishes his data to indicate only the strains occurring in the specimen. Any resistance change from other causes will give erroneous datao Likewisep any changes in bridge-balance caused by capacitance or iadttetance can produce errors unless means are provided to eliminate themo For entirely satisfactory results, the gages should experience change in resistaace only ftrcm strains, and the bridge circuits should give indications proportional to changes in resistance only— not to either inductance or capacitance changes o Tests continued over a long period of time will inevitably be subjected to distrubing factors which might not appear at all in a shorts- time test. More stringent requirements must be imposed on both equipment and techniques if long=time stability is required. The A vailable Instruments s Portable instruments used with wire^type resistance strain gages are usually alternating-cwrrent bridges p equippe, as in- dicated by the meter, there is some relatively large source of error within the strain indicator. Since the ordinary laboratory oscilloscope requires 110 volt power for operation, this method for adjusting reactive balance is often inconvenient or impracticable for field use. In such circuits stances, it may be possible to substitute a sensitive a-c voltmeter for the oscilloscope (especially if the instrument is connected to the plate of the output stage of the amplifier instead of the intermediate stage) 5 such sensitive meters are not easily obtainable as catalog items, but may be easily constructed with i small dry-disc rectifier • I f^ lie of the Instrument type and a 0-50 micrc-ampere d-c instnmento Alter- natively, a battery-operated vacuxun-tube voltmeter may be used, Ob- viously, with either of these instruments the balance adjustment is con- tinued until the instrument indication is a smll minimam.g obviously^ too, such instruments will indicate contaminating signals as well as desired signalSj, and will not discriminate between them. Ideally, the most accurate system would be one in which the total output of the strain-gage bridge is observed, and all adjustments continued until the bridge output is truly aero. This can be accom- plished only if provision is made for resistive, capacitive, and induc- tive balance^ and with a sensitive and stable indicating instnimento For most purposes it is sufficiently accuraite (and more convenient) to make the initial reactive balance with the cathode-ray oscilloscope, adjusting both capacitance and resistance simultaneouslyj then to finish the resistive balance with the indicating meter. It is not necessary to take down any data on the -values of capacity^ or its ad- justment; it is often convenient, however, to make notes of the values so that any sudden changes may be observed. It is desirable to note occasionally the number of divisions on the slide-wire dial between the resistance-and-capacitance balance point, and the instrument balance point indicated by the meter. If this difference changes appreciably from one time to another, it is an indication of some change within the instrument, and any such sudden change must be viewed as a source of error. One of the instruments pictured in Fig, 3 was not available at the time this report was first prepared; it is the Model BA-1 amp- lifier of Ellis Associates o This instrument operates on a principle en- tirely different from that employed by those previously described. In- stead of a carrier-type system, the Ellis instrument contains a re- sistance-capacity coupled amplifier, a direct-current bridge circuit, and a mechanical modulator. It is intended as a general-purpose instru- ments for both static and dynamic measurements. With the Ellis amplifier, strains are measured by using the instrument as a simple amplifier, with its output applied to a cathode- ray oscilloscope; calibration data are secured by occasionally introduc- ing known bridge unbalances through the mechanical modulator (a self- contained electrically-dri-ven vibrator or "chopper"). Static or ex- tremely 3 lowly- varying strains are measured by the d-c bridge circuit, with its output fed through the mechanical modulator and amplified for display on the oscilloscope screen. Switching devices and calibrated resistors are contained in the case of the instrument, for calibration purposes, and an indicating instrument is also provided, for battery- checking and similar applications. This system is subject to certain fundamental difficulties, partlcularyl (l) the maintenance of vibrator-type amplifiers is often difficult, (2) the system does not permit simultaneous observation of low and high-frequency phenomena, (3) the d-c bridge circuit used is subject to errors from thermal e.m.f's., electrolytic contamination. 'W'VER.SITY Of ,u,;VO;S I I f 12, and from stray pick-up which may be partially rectified in imperfect connectionsp and (4.) the use of an oscilloscope as an indicator is un- desirable both because of the relatively low reading accuracy and be- cause it normally requires line power for operationo The final verdict as to the desirability of an instrument must, however, come from its users, and the Ellis amplifier has been used only a very little in this laboratoryo During the actual reviision of this report^ still another in- strument of this kind has appeared on the marketo It is the Model 129 Dyria^Mike made by Industrial Electronics c, Inco While this instrument was intended for use with mutual-inductance gagesj, it is understood that it can also be used with resistance^type wire strain gages. Also, while it operates from 110 volt line currents it seems probable that a modi- fied version may be produced to operate from self-contained power supplies. To date only catalog information has been available j no actual performance data can be given o In addition to the carrier-type instruments (and one self- contained d-c„ bridge) for portable usBc there are a number of direct- current-excited bridge units available for laboratory use. No attempt is made to present comprehensive discussions of these instruments j as prev- iously stated, the emphasl. however, any change 13 < is noted, some further check must be made to determine whether the change has occurred in the gage or in the instrumento It is therefore advisable to maintain three pairs of gages n and so long as two of these pairs show little or no change, any change appearing in the third pair may reasonably be considered a change in the gages rather than in the instrument. In applying this test, however, it is almost essential that the gage circuits contain reactance comparable to that occurring in the circuits at the test locationo It is also essential that reactive balance be provided. If it is used carefullyp this check can be considered sufficient for practieally all short-time tests „ It does not, of ccurse, indicate the cause of any drift or other change which may be discoveredo More reliable checks can be made if a dummy gage circuit is provided^ with which small known resistance changes may be applied to the bridge circuit. Again, provision for reactive balance must be made, and it is not sufficient merely to make the reactive balance complete— it is necessary to insert capacitances, in parallel with the bridge arms, and of magnitudes comparable to those found in the test circuits » There is no other convenient way to make the loading on the exciting oscillator similar to that when under actual test conditions o In such a checking devicBp great care must be taken in construction and in operation, since the unit becomes a secondary standard, and if improperly used can actually introduce errors into the measurement » However, good quality switches and resistors are available, and a high degree of stability may be attained without great cost. For ordinary indoor short-time tests, some of these precautions are not required. Some instrument should be available for determining the point of reactive balance, when such a measurement is takeno For most applications —that is, when relatively short leads are used, under good operating con- ditions— it is not necessary to make the reactive balance for every read- ing of the instrument. For such a test it is sufficient to check the con- dition of the instrument when it is taken from the laboratory, and for thi6 check the cathode-ray oscilloscope is usually the most convenient. A vacuum-tube voltmeter may be used, but it cannot distinguish between the desired signal (1,000 cycles per second) and any noise or stray SO-cycle pick-up. Usable results may be had with a sensitive micro- ammeter and an instrument rectifier unit, but it has the same lack of discrimination as the vacuum-tube voltmeter. For field use, when high accuracy is desired, or if the operating conditions are especially difficult, very good results may be had with a vacuum-tube voltmeter fitted with a filtering circuit so that the stray voltages and the noise can be excluded. Such an instru- ment may be portable and battery-powered. If a test program is to involve long lead-wires, buried or immersed gages, long periods of time, or if the greatest accuracy is required, the plans for the test must be made carefully. No specific rules can be written^ but these points should be observed o » «.■' u. lo Matched gages should be used whenever possible. If really good temperature compensation is to be provided„ the gage circuits must be as similar as possible. This requires using gages from the same lot (and if possible from the same package) as working and as reference gages » At the best, there will be a gage=factor error of perhaps 1 per cent, and a gage-^-resistance error of perhaps i/2 per cent. 2o Matched lead-wires should be used for best results. Since the resistance of the lead-wires is effectively added to the gage re^ sistancej the resistance of the lead=wires should be as low, and as con- stant, as possible. If really long leads are used, it is advisable to take all the lead-^vdres for each gag© cireuit from the same roll of wirep and even to match their resistances with a Kelvin bridge. Any lead=- wire which shows a resistance varying appreciably from that of others of the same length should be discardedo It is definitely dangerous to use wires of different types or different makes in one installationj there may be differences in resistivity, in temperature coefficients, or in aging characteristics . 3. Lead-wires should be carefully laid outj and csLrefully tied down. As far as is possibles, all the wires leading to a pair of gages should follow the same route, so that they will be subjected to the same temperature changesj, and the same climatic effects. Wires should be tied down in such a manner that no mechanical strain can affect them. Of the three wires going to a pair of gages, the common lead is the least critical. 4., In damp locations, bundles of lead-wires should be filled with waxj, if possible? or other arrangements made so that water-pockets can not form. Such water=pockets can cause large and unpredictable changes in the reactance of a circuit. Tying wires in bundles is good practice, in general^ it equalizes temperature effects. 5. Lead-wires should be as large as possible (considering spape limitations, economies, and required accuracy). In any installation in- volving lead-=wires longer than perhaps 4.O feet, solid wire of No. lA or larger is desirable. Stranded wire is usually less stable,, apparently because of the stretching of some of the individual strands. Its in- feriority usually shows up only when the lead-wires must be tied in vert- ical runs. The insulation must be of good quality, and if it is thick there will be smaller capacity effects from moisture on the outside. Good results have been gotten with General Cable Company's "Gencaseal" wire, Noo 14 solid, with 3/64 in. insulation, used under ?00 feet of water. There are undoubtedly other makes equally good. However, at least one lot of "Habirdur* wire, made by Habirshaw and carrying specifications similar to Gencaseal, gave decidedly poor results, for no obvious reason. I 15o 60 Careful handling of both gages and vdre is required,, Sharp bends in plastiC"=insulated wire should be avoided | minor abrasions will be produced which although invisible will have an injurious effect. Rough handling of wire can cause work-=hard6ning which will cause a local change in resistance I this may later be relieved by aging„ with the result that an apparent strain will be registered. Some types of wire are especially susceptible to this effect « Wire should be purchased in standard ceils, and any coil showing signs of rough handling should be set aside for use in locations where less stringent requirements are imposed. These effects are minimised by the use of large wire;, since if the lead^wire resistance is made small enoughp small changes in it will be negligible, 7o Gages should be handled carefully,, and mounted with even greater care. Sharp bends can change the resistance of a gage, and also its gage factor. Specimen surfaces must be clean and ft:°ee from moisture; heat is the best agent for drying, Cellulose=acetate cement as supplied by Baldwin-Southwark is goed for most applications. The cement should be dried until the leakage resistance is at least 5.0OOO megohms y if possiblee A waterproofing coat of Petrosen© or of Petrolastic should then be applied, the lead~wires attached„ and the waterproofing made complete. Too thick a coating of wax is likely to be brittle, Petrolastic is rubbery, and can be used safely in thicker coats.* 8, Gage circuits when in place and wired should have leakage resistances of 50 megonms or higher^ if possible. It is entirely possible to operate gages with much lower leakage resistances than thesej, but there is greater safety in the high leeikage resistances p since even a relatively large change in an extremely high resistance will cause little change in effective gage resistance » 9, Lead-wires should be protected, so far as possible, from anything which might cams® sudden temperature changes, A bundle cf wires with the sun shining on one side may take hours to reach an equilibrium temperature (if it ever does) and during the time the temperature is changing there will be drift of the gag© circuits. Shade from the sun is desirable^ if it is necessary to immerse the wires partially in water it is better to immerse them completely. To indicate the need for uniform temperatures, it may be noted that the change in resistance of a 10- foot length of No, 18 copper wire, if its temperatiire changes 5 deg, Fahrenheit, will be equivalent to that caused by a I'j micro -inch strain in steel, 10, Switching boxes should also be kept clean and dry, and protected from direct sun and the rain. Portable Strain Indicators must be cared for as Indicated in the preceding section of this report. » Notes Petrolastic is a black., tarry, waterproofing material^ pro- duced by Standard Oil of California,, Petrosen© is a micro-crystelline wax produced by the Socony-Vacuum Oil Company, Another popular water-proofing material Is Tygon, a rubbery compound made by United States Stoneware Company, i •ri f l6o 11, Since the relatively inexpensive instruments used for most of this work are always subject to some drift „ it is advisable to set up the test schedule so that check"=readings may be made frequently. Short intervals between loadings are advisable whenever possible o 12o It is often advisable to employ mechanical strain-magni- fiers if the strains in the specimen are not large. The "reliable limit* of the ordinary installation of SR-4. gsges is 10 micro-inches or a little more J that is<, even with care in operation an xmcertainty of 10 micro- inches or more may be expected. Often, of coursec, the uncertainty is greater^ and always it will be greater if no reactive balance is used. If the total strain is not large, it is helpful to magnify it mechanically before attempting to measure it with wire=type strain gages. Many test programs require the use of more gage circuits than can be conveniently handled without some device for rapid switching of gages. There are available multi^point switching units for this purpose, and some laboratories have constructed their own. Figure 5 shows a phot- graph of a commercially-procurable 20-circuit switch, and also a unit made up in this laboratory^ capable of handling up to 100 gage circuits to be read against any of 10 compensating gages, or of handling 50 gages „ each one against its own individual compensating gage. The principal requirements imposed on these switches are (1) low, and constant, resistance, (2) uniformity of resistance. (3) freedom from sources of therassl, electrolytic, or contact potentials,, and (4.) positive action. Of these, the third point is the most important in d-c„ bridge circuits. Excepting for their bulk and appearance, good=grade copper knife switches of 30 to 50 ampere capacity are hard to beatj if kept clean and otherwise maintained. They are, howeverp bulky and subject to damage be- cause of their open construction, and consequently are seldom used for more than a few circuits. Many workers have used and recommended the use of multi-point "radio-type* switches, connecting two or more contacts in parallel for lower resistance. The writer's experience convinces him that this practice often leads to a false sense of security, especially in d-Cj^ units I it is better by far to use the best available switches and not parallel contacts. There are a number of satisfactory switches available s the writer has used those made by Communication Pro'ducts Coo„ General Radio Co.. Leeds and Northrupf Lewis Engineering Co,, and Shallcross Manufacturing Co., with entire success. There are also, undoubtedly, others not known to the writer, equally satisfactory. It should be recognized that any switching unit will contribute to the total probable error in a system, no matter how good the switch. The switch and its extra wiring will increase both the total circuit re- sistance and its reactance, as well as providing more possible paths for leakage and for contamination. Careful design, installation and maintenance of these units can almost, but not quite , eliminate errors from this source. ;%= I Battery B FIG. R], R_3>_Ci__C_3 ^ _Li=Jz3 R2 R4 C2 C4 L2 L4 A-C IN FIG. 2 In most wire strain goge circuits, inductonces (L) are ignored, as ore C_ and C , which appear inside the instrument housing. * } 19, 4 /^ r ■ 1 r 1 *. Type K. B-S. Indicators Early. Fig. 3 D-C. Bridge Ellis Young UNIVERSfTY OF ILLINOIS-URBANA 3 0112 079432438