Reference / Paper · 1964
Friden Final Engineering Report: Magnetostrictive Delay Line (June 1964)
Friden's 48-page engineering report on the EC-130 delay line: the Ni-Span C wire, the transducers, the mounting and the measured performance.
Reference / Paper · 1964
Friden's 48-page engineering report on the EC-130 delay line: the Ni-Span C wire, the transducers, the mounting and the measured performance.
FINAL ENGINEERING REPORT on the 13C ELECTRONIC CALCULATOR MAGNETOSTRICTIVE DELAY LINE by Willard E. Wells June 1964 FRIDEN, INC. 1. 3. Ae Delay Line Wire ae TABLE OF CONTENTS Material - b. Wire Diameter - - - - --- =+- e. Coil Diameter - - - - --+-- - d. Heat Treatment - ----- - e. Line Configuration - - - - - [ine Supports - - - ------ - - a. Material .- ------+-- - b. Support Design - ----- - Transducer Coils ---+---+-+-- - a. Coil Length ee ee ee ee b. Coll Diameter - -~ - - - - =- - ‘c. Number of Turns - - - - - - - d. Wire Size ----+---+-- Magnetostrictive Ribbon - - - - - - a. Material ------ - -- - be Configuration - - - - --- - ec. Ribbon Size --- - - --- d. Annealing ----+-+---- _—-— = ~ eae ~ —~ ewe ~ = _- ee) ee ee ee — oe a ole WA oN je (e) 17 13 18 20 22 TABLE OF CONTENTS (continued) Dampening Pads - - - ---- 77-7 ---- Biasing Magnet System - a. Transmitter Magnet - - b. Receiver Magnet - - - Mode Converter Junction -. - a. Weld Joint - - - - - - b. Mode Converter Support Recommendations - - - - - - Page 31 Wy FINAL ENGINEERING REPORT on the 136 ELECTRONIC CALCULATOR MAGNETOSTRICTIVE DELAY LINE Willard E. Wells 10 June 1964 The following report is a sumnmary of the delay line design existing at the time of its production release, and the reasons for that design. The design areas to be covered in this report wiil be the delay line wire, the line supports, the coil design, tne magneto- strictive ribbons, the dampening pads, the biasing magnet system, and the mode converter function. In addition, recommendations wlll be made in areas where it is felt either initial or additional investi- gation will be productive to further delay line improvement. Ll. Delay Line Wire a. Material. The decision to select Ni-Span C as the line material was based primarily on its relatively high mechani- cal factor or low hysteresis and low dainpening character- ev ‘s lo] Oo ct bes istics. Tne Q factor is directly related to how much the stress impulses are attenuated as they travel. down the line medium. The higher the Q, the greater is the bandwidth and + output of the delay line. ‘The mechanical Q of Ni-Span C is increased by heat treatment. Maximum strength and hardness, which is indicative of a high Q, is obtained by a precipitation~hardening or age-hardening heat treatment Page 2 of the material received in a solution annealed and 50% cold-worked condition. . The proper heat-treating temper- ature, which ranges from 1150°F to 1350°F, varies with each mill run and should be selected according to the re- port sent with each mill shipment. A heat-treating time of three hours is generally recommended by the manufac- turer, However, I feel 25 to 30 minutes is adequate and will reduce heat-treating costs. This decision was based on tests that indicated no increase in hardness could be obtained by holding the material in the furnace for a longer period of time. The rate at which Ni-Span C is heated and cooled from the age-hardening temperature is not eritical. Any convenient rate can thus be used which ex- pedites the heat-treating operation. It is suggested by the literature that for precision applications it is advisable to remove internal stresses resulting from cold working by suitable heat treatment prior to age hardening. Varying internal stresses can produce varying degrees of age hardening and thus lead to slight property differences after heat treatment. A stress relieve anneal of 750°F from one to four hours, is recom~ mended to minimize this condition. This stress relieving was not specified in the final release because results were satisfactory without it and would add to the cost of the wire processing. Page 3 Music wire, another material considered as a line material, also possesses a relatively high mechanical Q factor. Some investigators found music wire to be superior to Ni-Span C in this respect, although the heat- treated condition of the Ni-Span C was not clearly defined when making this comparison. At no time in our investiga- tions did we find music wire superior to a properly heat- treated Ni-Span C sample. However, I feel further studies are justified in view of the low cost of music wire and the economies that could be realized by eliminating the heat-treating process, Ni-Span C was selected over music wire for the following reasons: (1) Corrosion resistance--- if heat treated ina properly controlled atmosphere, Ni-Span C, due to its nickel and chromium content will exhibit good corrosion resistance. Music wire, on the other hand, would require @ protective coating, which may be detrimental to line performance. ’ (2) Length of line - sound travels faster in music wire and would require longer line lengths to achieve the same delay time. This would increase pulse attenuation and the scope of problems dealing with adjustment to correct dispersion. (3) Ribbon-wire impedance match - to properly match the mechanical impedance of the wire to that of the nickel Page 4 transducer ribbons, music wire would require larger | diameters which increases dispersion and degrades resolution. (4) Goil diameter - control over the coil diameter is important because the size of the coil of wire directly affects the amount of pressure on the wire at the supports. Since music wire does not require further heat treatment, it would become necessary to control coil size at the mill, a less practical solution. (5) Constant delay time with temperature change - unlike music wire Ni-Span C is a constant modulus material, which gives it the ability to maintain a fairly constant delay time over a considerable range of temperature. Although this property is not as important for our delay line application, it does yield us a certain margin of safety. Other materials were also considered, including copper, sterling silver, aluminum and magnesium. Copper and sterling silver were considered primarily because of their relatively low sonic velocity properties---sterling silver has a sonic propagation velocity nearly one-half that of Ni-Span C. A five millisecond line composed of sterling silver requires a line length of only 25 feet compared with 47 feet for Ni-Span C.. The advantages of a relatively shorter line length, to summarize again, are Page 5 smaller packaging requirements, less pulse attenuation 7 and less dispersion. However both copper and sterling silver evidenced such low mechanical Q and high dispersive characteristics that the shorter line lengths could not compensate for the resulting poor quality and level of signal output. The copper investigated was hard drawn. The sterling silver admittedly was in a softer condition. Attempts to work harden it were tried unsuccessfully because tooling marks on the material produced high level noise effects. In addition these materials are susceptible to tarnishing, which may be detrimental to line performance. Aluminum and magnesium were considered because of their known high Q characteristics and light weight. | The. sonic propagation velocities of these materials are about equal to that of music wire. The mechanical Q values were found to be satisfactory although there was little evidence to prove they were superior to Ni-Span C. In addition both would require wire diameters twice the size of Ni-Span C to gain an impedance match at the weld with the nickel ribbon. Also, both are difficult to weld and would create additional problems at the mode converter junction. Wire Diameter. The diameter of the Ni-Span C wire was selected by considerations of the mechanical impedance match between the wire and the transducer ribbons at the Page 6 mode converter weld. A formula for matching the impedances of the wire and ribbon were derived in Scarrott and Naylor's paper on "Wire-type Acoustic Delay Lines for Digital Storage" in the 1956 Proceedings of the Institution of Electrical Engineers. The formula is shown below. A = Tr2 | Sw Su : an e rE, where A = cross-sectional area of one ribbon n <= number of ribbons used r = radius of the wire @w = density of the wire material @r = density of the ribbon material Gw -= shear modulus of the wire Er = Young's Modulus of the ribbon There was correlation between this formula and our experimental results, and a wire size was specified primarily on the basis of this formula. However, it is difficult to know whether the best choice of wire diameter could or should be based on a theoretical impedance match since there are many other factors involved. True, a perfect impedance match between the line and the ribbon will transfer energy from the one medium to the other without reflections and resulting energy losses. Yet, the formula above does not take into account the method of connecting the media, which imposes an additional impedance that in all probability will not match the impedance of the other two. (i.e. the weld puddle and the possible change of Young's Modulus in the materials Cc. Page 7 during the weld cycle, or the addition of an adhesive layer in bonding techniques.) In eddition, as the wire diameter is increased, the supports will impose larger perpendicular forces on tne wire as the coil of wire is prevented by the supports from assuming its free diameter. The supports are dampening in nature and tend to increase signal attenuation and degrade bandwidth as the support forces increase. How- ever, conversely to this, delay line investigators have stated that larger wire diameters produce higher outputs, (independent of support systems, of course), although at a sacrifice of increased dispersion and loss of resolution. In conclusion, further investigation in this area may improve performance if care is taken to lsolate each parameter from the rest and study only its effects on performance. Coil Diameter. Selection of a wire coil diameter involved two main parameters, Ry» the natural or "free" radius of the wire coil, and Re, the confined or "forced" radius Gictated by the support system, First of all, let it be understood that without additional considerations and limitations, R, and Re should both be as large as practical in theory. This is based on Scarrott and Naylor's derivation Page 8 ~ KA = + where vt= phase velocity of torsional stress waves c+ = torsional phase velocity of sound ina t straight wire K =a constant, whose value is immaterial in this discussion A xwavelength of a particular frequency component R,=and Rp as defined above if Va cy, then theoretically all frequency components of a sonic pulse traveling down a line should arrive at the end in exactly the same time phase relationship as when at started and no phase distortion, or dispersion, should result. To accomplish this feat either Ry, or Rep should be infinitely large, but in any event, both should be as large as practical. As a result of this work, a patent was granted Ferranti Electric on the use of naturally straight lines (Rzec). Friden did not choose to lease patent rights from Ferranti nor expend money on the development of a straight wire process, so the Rn of our wire coil was restricted to seven and one-half inches. Packaging restrictions limited the mean Rp of the wire coil to a maximum of four and three- quarters inches. A R, of four and three-quarters inches would minimize the effects of the forces exerted on the wire by the supports. However, a maximum R, of seven and one- half inches would minimize dispersive effects. The problem - became one of compromise. A coil Rn of four and three- quarters inches was selected because it was determined that the support forces would be more detrimental to signal amplitude than dispersion effects in the Rn o~ Page 9 range of four to seven inches. Additionally, the slight increase of dispersive effects could be adjusted out by the positioning of the receiver transducer biasing magnet. Heat Treatment. The reasons for the heat treatment of the Ni-Span C line material and the time and temperature involved have been discussed earlier in this report. How- ever, no mention was made of the importance of maintain- ing a good protective atmosphere during heat treatment. Earlier experimental delay lines composed of Ni-Span C - wire were heat treated in a normal non-carburizing electric furnace atmosphere. This heat treatment yielded wire with a dark blue oxide film. These lines performed adequately, ! but it was felt that a clean bright material surface free of oxides would improve performance as well as appearance. Thus the wire was heat treated in a hydrogen and nitrogen atmosphere - an atmosphere selected because it was already being provided at Friden. The result was successful and no further investigation into some of the other atmospheres, such as vacuum, dissociated ammonia, or helium atmospheres was planned. Although a fairly oxide-free surface can be accomplished in the hydrogen and nitrogen atmosphere, the importance of careful wire handling and cleanliness of the wire before, during and after heat-treat can not be over stressed. Page 10 ’ : e. Line Configuration. A two layer, flat spiral line config- | uration was decided on for the following reasons. First a one layer, flat spiral line would allow a reduction in package height and reduce assembly time, but had many dis- advantages. A one-spiral configuration would place one transducer assembly inside the spiral near the center of the delay line can. That is unless a special bridge could be devised to bring the wire over or under the spiral to the outside. This did not appear to be too practicable. With a transducer in the center, little air passage through the delay line package was possible and longer leads would be required from the transducer to the amplifier card. With a single spiral more diviation from a mean natural coil radius would be necessary increasing line support loading and the attendant signal attenuation. Con- sideration of three or more layers of spiral line was not considered practical since it increased package height above the maximum design limit, and required more support brackets which would increase the time to assemble the line into the delay line package. 2. Line Supports a. Material. An elastomer was selected as the line support material, because rubber was one of the few materials found which did not introduce mechanical discontinuities Page ll into the line. These discontinuities cause reflections that degrade the signal-to-noise ratio and produce dis- tortion in the output pulses. Unfortunately, rubber is also a good sound absorber. Silicone rubber was chosen over the other elastomers for the support material because of its excellent electrical properties, its temperature stability, and because of its ability to allow passage of the mechanical pulses down the line with the least ab- sorption. Under test it performed better in this respect than other "solid" elastomers, such as neoprene or gum rubber. However, loads must remain light for this to be true. Under heavy pressures silicone rubber can be an efficient dampener. That is why it was selected as the ribbon termination material. This will be discussed ra f further under the nection on dampening pads. Support design. A system employing six separate supports for containing each layer of line was specified. This number of supports was selected when it became obvious that four separate supports were not adequate for the Job. The four-support system allowed individual loops of the two layers to touch, thus producing "cross-talk" which seriously degraded the signal to noise ratio. A five- support system was not investigated, and may well prove adequate and result in a cost reduction. Such a system Page 12 \ would, of course, have to be proven acceptable under the specified shock and vibration requirements. The decision to use a combined metal stamping and molded rubber support instead of the more conventional metal bracket-tubular rubber support system was based on high-volume production considerations. Labor costs to assemble the wire into the present support system could be reduced considerably and more than outweighed the increased part cost. Transducer Coils. Two identical coils are used in the transmitter trans- ducer and two different but identical coils are used in the receiver transducer. The two-coil system is an important part of the push-pull transducer concept. With this system larger output signals can be obtained than by a one-coil and ribbon concept. The push-pull effect is obtained by a biasing magnet and by wiring the same ends of the two coils together, the other ends being wired to the input leads in the case of the transmitter coils and to the output leads in the case of the receiver coils. The design of the transmitter and receiver transducer coils is of great importance in achieving the optimum response of the system for a given input pulse, and is dictated by that input pulse and by the electrical requirements of the system. Page 13 Coil length. The physical coil length of both the transmitter and receiver coils, which in this case was one-tenth of an inch, was determined as follows. To produce an output pulse with optimum amplitude and resolution, that is a pulse which is properly coalesced, the effective length of the coils must be arrived at by, use of the formula 1,= Tv where lezreffective length of the coil. y =time for the mechanical impulse to travel the length of the coil. vr velocity of acoustical propagation in the magnetostrictive mat'l. Since there is a certain amount of flux fringing at the ends where the ribbon leaves the coil, the physical length of the coil must be made smaller than the effective length to compensate for the fringing effects. Fringing can be reduced by fitting ferrite cups around the coils to more sharply define the edges of the field. However, it was not necessary to resort to ferrite cups in our applica- tion since an adequate physical length could be maintained to suit the electrical needs of the system. Theoretical formulas are available for determining what the physical coil length should be for a given coil design. Rather than rely entirely on theoretical data, various lengths of coils were made up and tested for Page 14 resolution by inserting them in a delay line system and varying the input pulse width. When the output pulse reached coalescence with the input pulse adjusted to the width requirement of the delay line design, we knew that our coil length was correct. Coil diameter. The requirements for the coil diameter are basically simple. The mean diameter of the coil should be made as small as possible by winding the turns as close as possible to the magnetostrictive material. The reason for this is to avoid as much flux fringing as possible and closely couple the coil to the ribbon. The object of the transmitter transducer is to convert the induced magnetic energy into mechanical acoustic energy. Therefore good electromechanical coupling is desirable. The same is true of the receiver transducer, which converts mechanical energy that has been propagated down the line into electrical energy. This demands that the magnetic flux link the whole winding and again a closely wound coil is desirable. | Consideration was given to winding the coils directly on the nickel ribbons, but there are many problems associated with this design. First of all the ribbon must be annealed, and winding the coil on the fragile annealed Page 15 ribbon without work-hardening it would be extremely difficult tc achieve. Second, each coil must be wrapped around two ribbons which must be in perfect alignment for welding. Third, the sharp edges of the magnetostrictive ribbon cut into the coil insulation and cause shorted turns. An insulated layer could be used but then a bobbin might as well be utilized. In addition the wire must be wound under some tension which may hamper magnetostrictive movement. In addition the coils must be in longitudinal alignment within ten per cent of the coil length, (* .005 for our design), to provide adequate push-pull performance. This will add to the problems of ribbon alignment. A bobbinless coil design would correct most of the above disadvantages. However, during ribbon insertion there is still the danger of cutting into insulation and causing shorts. Also the mounting of the coils rigidly to prevent lead wire breakage and maintain alignment would pose difficult design problems. The final design consisted of a conventional round bobbin having a center hole slightly bigger than the magneto- strictive ribbon width. The diameter on which the coil is wound is larger than the bore only by an amount necessary. to maintain rigidity and minimize production problems. | Designing the shape of the bobbin to that of the ribbon Rage 16 configuration was considered, but the bobbin is so small and the wall so thin that the part could not be molded. Number of turns. In general, increasing the number of turns of magnet wire on the transducer coils will increase the amplitude of the output signal. This is true because the magnetostrictive ribbon is not in saturation, and with proper magnet biasing, an increase in field intensity will increase the magnetostrictive effect of the nickel. It is important to understand, however, that design limitations were present which restricted the number of turns that could be used, and a compromise had to be made. To begin with, as the number of turns is increased the inductance of a coil will go up. An increase in the inductance of tne transmitter coils increased the rise time of the input current pulse. A final value of 180 turns was specified for the transmitter coils. This was felt to be the maximum number of turns permitted with the existing transducer and amplifier design, and yet achieve the rise time required for a properly shaped trapezoidal input pulse. The sharper the rise time, the less effect will noise originating on the line have on the output signal-to-noise ratio of the recovered pulse. However, the sharper the rise time, the more average power is necessary to drive a pulse of the Page 17 Same peak amplitude through the line and l1ts associated electronic circuitry. Since the two limiting cases are a rectangular input pulse and a triangular input pulse, a suitable compromise is the trapezoidal pulse. The upper limit to the number of turns on the receiver coils is such that the resonant frequency of the colls together with their stray capacitances is appreciably above the operating frequency of the delay line. In the case of our delay line design, the upper limit for the nunber of turns on the receiver coils was 350. Once a satisfactory transducer design was achieved it was found that operation could be improved by tuning the electro- mechanical circuit with a capacitor shunted across the transducer coils. A shunt damping resistor was then used across the receiver coils to provide critical damping. Wire Size. A number 48 AWG magnet wire was chosen to wind the coils. This size of wire was chosen as a com- promise between maintaining wire strength to reduce pro- duction winding problems present when working with fine wire, and minimizing coil height to insure tight magnetic coupling between coils and ribbon to reduce flux fringing. Further consideration of finer wire sizes is recommended and would show that there will be less inductance for a given number of turns of coil winding and a resultant faster Page 18 rise time. Thus, it would be possible to increase the output of our existing delay line by placing additional turns on the transmitter coils, and yet retain the re- quired rise time of the input pulse. Belden's Celenamel magnet wire, which is essentially a cellulose acetate insulated copper wire, was chosen for the delay line coil winding because it can be easily soldered without removing the insulation by stripping. This wire is reputed to be a fast soldering wire at the lowest solder temperatures. It is superior to plain enameled wire with respect to abrasion and solvent re- sistance, and is resistant to thermal degradation at 105°¢, Its heat aging characteristics are equal to those of Formvar, and it has excellent insulation resistance. Another wire that should be investigated is Rea's Solvar magnet wire, which is available at the same price. It is also self- stripping, is generally equal to the other properties | mentioned above, and is capable of operating at temperatures up to 120°C, continuously. 4, Magnetostrictive Ribbon. as Material. Pure nickel was selected as the magnetostrictive . medium because of its availability and its relatively good magnetostrictive properties. In addition it is a good corrosion resistive material. It was originally planned to investigate other magnetostrictive materials, such as Page 19 Permendur and Hiperco, but such plans were dropped when it became apparent that nickel was entirely satisfactory for this application. Most of the work was confined to "200" nickel, a highly refined, relatively pure nickel (99+%). Nickel "205", a similar 99% nickel material, which was inadvertently selected due to confusion in manufacture identification systems, was not considered as good for this application because of its lower resistivity. Heavier currents and temperatures were consequently required to produce optimum annealing conditions, and led to the forma- tion of ex:essive scale on the surfaces of the nickel. Since this condition was considered detrimental to per- formance, "200" nickel was specified for the final delay line release. Configuration. Because of skin effects discussed below under section c, the need for a thin magnetostrictive nickel material limited configuration to ribbons, thin- walled tubing or bundles of fine wires. Fine-wire bundles achieve highest magnetic efficiencies, but their use would create problems in attaching them to the delay line wire and in dampening the unwanted signals. Thin-walled tubing have the same disadvantages to a lesser extent. Ribbon was selected because of its availability, lower cost, and because of the ease with which it can be welded and dampened. Page 20 However, because thin-walled tubing has many advantages over ribbon, an investigation of this magneto- strictive shape is recommended. First it is easier to get a close fit to a round coil aperature. Second it would be mechanically stronger and more self-supporting. Third, nearly 100% more nickel could be placed inside the same diameter coil designed for our two-layer ribbon system, This would result in increased efficiencies and higher autput. The ends of the tubing could be flattened to pro- vide an area for welding and dampening without theoreti- cally producing impedance discontinuities. Yet, a few disadvantages to tubing as a magnetostrictive medium should be mentioned. Investigators in this field claim that ribbon-shaped elements have inherently better frequency response than do tubes, although the frequency response can be improved by slotting the tubes. In addition, the frequency response of a tube will exhibit attenuation bands. The width and position of these bands are a function of the radius of the tube and the Poisson ratio of the tube material. Ribbon size. Two mil thick by twenty mil wide nickel ribbon was specified as the final size of the transducer ribbon. In the final design four ribbons were specified: two attached to the top of the end of the line wire and an Page 21 identical arrangement at the bottom of the wire to produce the push-pull effect. <A ribbon thickness of .002 was specified to insure good frequency response by reducing eddy current losses. If the thickness of the ribbon inside the transmitter and receiver transducers is great in com- parison to the skin depth, eddy currents will modify and distort the resulting output waveform by a factor proportion- al to 1fs with a phase lag of 90°. Annealing the ribbon increases the permeability of the nickel and enhances this effect. This phenomena will be discussed further under the section on annealing. Studies conducted with 5 mil thick > ribbon were finally abandoned because of the difficulties encountered in eliminating this eddy current effect. Better results were achieved by controlled annealing of the two mil thick nickel ribbon. In addition, higher pulse ampli- tudes were achieved. Another distinct advantage of tne two mil thick ribbon over the five mil ribbon is the reduced ratio of cross-sectional area to periphery. Since dampening of unwanted and reflected pulses at the ribbon ends can be achieved only on the periphery of the ribbon, the lower the ratio of cross-sectional area to periphery, the more efficient the damping of the mechanical signals. The selection of the 20 mil width for the nickel ribbon was a compromise between good magnetic coupling Page 22 . between ribbon and coil for high pulse repetition frequencies with narrow tapes and high output amplitudes with wider tapes. The 20 mil width allowed a coil design which was well matched to the pulse width and pulse repetition rate required for the application. The use of multiple ribbons---two in this case through each coil---increases the magnetic efficiency of the transducer operation and increases output amplitude. The amplitude increases in proportion to the number of ribbons in the coil, and although not in a direct propor- tion, is nearly so. . Additional increases in output amplitudes could be achieved by the use of three or more layers of ribbon. The results desired, of course, must be weighed against the multiplying difficulties to be en- countered in welding techniques, ribbon alignment, and dampening of unwanted pulses. Annealing. The annealing of the nickel ribbon is without a doubt one of the most important aspects to be considered in achieving successful delay line performance. This is certainly not difficult to comprehend when it is under- -stood that most of the losses that occur in the delay line system are in the vital area of the transducers. The efficiency of the transducer action between coils and Page 23 ‘ribbon is determined by the electro-mechanical coupling factor. This figure of merit for a magnetostrictive material is a function of the permeability and the magneto- striction coefficient. Both are affected to a great extent by the annealing parameters and the control of these areas, The final selection of the annealing method, temper- ature,time and other prevailing conditions, was specified only after a great deal of time and effort had been expended. Yet it is felt that many additional gains may be realized with additional intensive investigation. First of all it must be understood that an annealing process after receipt of the raw material is mandatory. Although the nickel is ordered in a bright soft-anneaied condition, it is not at all satisfactory for our applica- tion. Even if the nickel had been properly annealed prior to our receipt, the work-hardening affects of winding the ribbon onto the shipping spools would be disastrous to good performance. Satisfactory performance of our pilot production lines only after a thorough education of the workers involved and a persistent overseeing of the methods used in handling the annealed ribbons will attest to that. In addition it is also mandatory for the ribbon strips to possess an unusual degree of straightness to prevent the possibility of work-hardening as they are O © Page 24 assembled and threaded through the small coil apertures and positioned in the dampening pads. This straightness is not provided in the conventional reel form received by the supplier, and it is doubtful that we could achieve this unless we ordered the nickel cut in straight strips at additional cost. Thus, during the annealing process we apply a small tension on the ribbon to straighten it. Two processes to anneal the ribton were studied. The first annealing process consisted of passing an A.C. current down the ribbon while applying a light tensile load to orovide the n D Q essary straightness. The end areas of the ribbon to be welded to the delay line wire were < provided with heat-sinks to draw off the heat and prevent oxides from forming. By this technique these areas remained bright and were suitable for resistance welding without extensive surface preparation. The oxides that are produced on the surface of the nickel during air annealing, although not suitable for welding, are acceptable, providing the oxide does not become flaky as it can at higher temperatures, The specified annealing conditions were arrived at by varying temperature, time, and tension in carefully con- trolled tests until the nickel achieved that combination of permeability and magnetostrictive coefficient that produces 3 is Page 25 opt Lum ribbon performance. Optimum ribbon performance is here defined as that ribbon performance in which the mechanical pulses produced on the ribbon by the transducer coils have the maximum amplitude attainable with negligible or no distortion due to the skin effect. In general, lower annealing temperatures do not remove the stresses in the work-hardened raw material. The result is a ribbon with remanant characteristics, or a me:cry, which when assembled in a delay line will remain magnetized even after the biasing magnet is removed. The end result is a transducer assembly that is difficult te adjust for Optimun performance and one which is detrimentally affected oy degaussing and stray magnetic fields. As the annealing temperature is increased the work-hardening stresses are renoved, permeability increases, and as would be expected, increased pulse amplitudes can be achieved in the nickel, As higher and higher annealing temperatures are reached, permeability increases with corresponding increases in signal amplitudes, However, the higher permeabilities of the nickel, as mentioned earlier in this report, increase the eddy current skin effects and pulse degradation results. As temperatures are raised further, this effect increases and there is actually an amplitude decrease due to the extreme distortion of the pulse. There is a ‘Page 26 temperature somewhere between these two extremes where optimum ribbon performance is realized. Tensile loading applied to the ribbon during the annealing process actually reduces the permeability that is achieved at a given temperature. A low tensile load just capable of producing a straight ribbon is considered best for this application. However, optinum ribbon per- formance, as defined above, can be achieved at higher tensile loads, but higher temperatures must be used. This 3 Ss bec joe LUS f (ay) tension not only reduces permeability, it reduces skin effects proportionately. Thus, &@ heavier load applied during a corresponding higher temperature can still reduce the higher permeability down to the optimum ribbon per- formance point. Generally this its not good practice for two reasons, First, the higher temperatures increase the danger of excessive scale on the ribbon, Second, the ribbon begins to creep at higher temperatures and heavier loads, which produces a permanent stretch that results in loss of signal amplitude. The time of anneal becomes a factor for two cpposing reasons. The longer annealing time increases the erfrect of creep, mentioned above, and slows down the annealing precess which costs money. Conversely, a longer annealing time enhances larger grain growth, which at a specified annealing temperature increases permeability and improves Page 27 magnetostrictive properties. Thus, a compromise must be made and will affect all that has been considered above. The second process considered was a bright anneal in our hydrogen-nitrogen atmospheric furnace. This process utilized a special tensioning fixture to achieve the re- quired straightness, Temperature and time of anneal were based on data from earlier tests. It was determined that the hydrogen-nitrogen atmosphere provided a surface with adequate brightness, free of oxides, to readily weld the nickel onto the Ni-Span C wire. Unfortunately, it was found that none of the samples that were annealed were suitable for our application. Different annealing temper- atures were tried and tensions varied, but little corre- lation could be found between the results of these combi- ations and those from the first annealing process discussed. Many of the nickel ribbon samples evidenced an excessive amount of remanance. Pulse distortion was exeessive in some samples and most showed somewhat disappointing pulse amplitudes. The annealing was conducted by the Production Engineering Department; thus it was difficult to evaluate these studies and assign contributary causes for the overall poor results. Proper attention may not have been given to the control of the various factors, which would make "zeroing in" virtually impossible. Perhaps the tensioning fixture, which consisted of a relatively large Page 28 mass, was not in the furnace for a long enough time to come up to the annealing temperature. Thus it would act as a heat sink for the nickel ribbon, and produce erroneous results with respect to the existing temperature data. It may be that the nickel was subjected to rela- tively slow heating and cooling rates which may have pro- duced poor results, This normally is not believed to have any effect on the properties of annealed nickel, but annealing under tension may produce changes. There is a theory, simply stated, that the passing of a heavy current down a magnetostrictive material such as nickel ribbon creates a concentric magnetic field surround- ing the ribbon which would tend to align the magnetic domains of the nickel in a direction favorable to magneto- striction. If this theory were correct the lack of an advantageous magnetic field within the furnace during annealing could explain the lower pulse amplitudes achieved in the nickel with this second process. However, I feel @ great more investigation is necessary in this area before any conclusions can be drawn and the validity of this theory can be proved. For example, this theory is more easily understood if the current in the ribbon is d.c.; more difficult to comprehend if the current is a.c. Most of the work accomplisned in the area of current annealing at the time of the delay line release utilized a.c. currents, although a few d.c. studies were run. No differences between the two types of current annealing could be detected in the few tests conducted. TI understand Bob Rageri’s group has gone further into this area; perhaps they have reached some conclusions. In summary, the furnace anneal has several attrac- tive advantages. Namely, it produces ribbons relatively free of oxide, and it can handle more ribbon at one time, making it a practical economical production process. The big disadvantage is in control capabilities. However, before this process can be discarded as unworkable, I feel more careful and intensive study should be made. A third annealing process was considered, but little effort was applied to determining its feasibility. This process consisted of the induction heating of the nickel strips, There are a number of areas directly related te the annealing process in which investigation is strongly recommended. Cursory probes in some of these areas indicate further work would yield impressive improvement in line performance, These areas are briefly discussed below. (1) In the current annealing process, ti:e, current, and tension were specified on the basis of producing a balanced non-distorted pulse in the nickel. It 1s known O Page 30 that the peak amplitude of this pulse is about half the peak amplitude of the distorted pulse at higher per- meabilities. Also it has been found that this distortion can be compensated for by a slight adj