Digital Voltmeter History ------------------------- From EE Times Gauging the impact of DVMs It all started in the San Diego area, where the DVM seed was planted in the early 1950s by a company called Non-Linear Systems. For several years, all digital voltmeters came from southern California. Then the business moved north to Hewlett-Packard Co.-in Palo Alto, Calif., at first, then to an HP plant in Loveland, Colo. Then a major part of the business gravitated even farther north, to John Fluke Mfg. Co., in Mountlake Terrace, Wash. That company's chairman and founder, Joh n Fluke, had been roommates many years earlier with HP's chairman and co-founder, David Packard, when the two were young engineers at General Electric in Schenectady, N.Y. Fluke and Packard became strong competitors and strong friends. Fluke died in 1984 and Packard in 1996. The instrument was named by Andrew F. Kay, who founded Non-Linear Systems in 1952. The first engineer he hired was Jonathan Edwards, formerly a fellow engineer at Bill Jack Scientific Instrument Co., a manufacturer of aerial-reconnaissance equipment, where Kay had been vice president of engineering and Bill Jack's first employee. Business was great at Bill Jack, which had enough contracts to last till the end of the decade, so the emphasis was on production rather than engineering. Kay liked engineering, so he left to start NLS, which would make what Kay first called a digital-readout voltmeter, then a digital voltmeter. Still later, with the added ability to measure ac volts and resistance, the instrument was renamed a digital mult imeter (DMM), though many people stuck with the old name even for the extra-capability instrument. Kay got the idea for the instrument while working some years earlier for a predecessor company, Jack & Heintz. He saw unskilled production-line people using large analog voltmeters. Because of wartime pressures, they didn't get much training, so their readings weren't dependable. Neither was their meter handling: On occasion, someone would blow up a meter with an overvoltage. And the meters were expensive and hard to come by, because of the Manhattan Project's high demand for them in building the atomic bomb. Making reading easy What was needed, Kay figured, was a voltmeter with unambiguous numbers that anybody could read-an instrument that would tolerate a reasonable degree of overvoltage, one with automatic polarity switching and automatic ranging. It shouldn't require any new design concepts, Kay believed. The only thing he had to do was to automate the Kelvin Bridge that Leeds & Northrup had been manufacturing since 1875. It should be a piece of cake. It wasn't, of course, that easy. Kay's bridge was balanced at first with cold-cathode tubes, which were notoriously non-repeatable. So he switched the design to a standard vacuum-tube triode of the day, the 201A. That determined the input impedance, 10 megohms. The first DVMs used a mechanical chopper to switch between a resistive divider, to which the input signal was applied, and a stack of batteries making a 90-V reference. Miniature, high-speed mechanical relays switched divider positions and changed ranges and polarity. These relays weren't cheap. They had to have low contact resistance, withstand 1,000 V or more and switch fast. And they had to last through millions of switching cycles. The first commercial instrument featured a four-digit display of dc voltage only and provided resolution of 0.01 percent and accuracy of 0.1 percent of reading. Later units could (and at times actually did) offer accuracy of 0.01 percent of full scale on each of four ranges to a full scale of 999.9 V, with autoranging and autopolarity. The first unit was sold to the Naval Electronics Laboratory (now the Naval Ocean Systems Center) in April 1953 for $2,300. The display used a stack of 10 Lucite plates, almost an inch deep, each plate with an engraved numeral and a tiny edge-mounted panel lamp. When the bridge reached balance, lamps were selected to illuminate the correct numerals. The display was in line, though not in plane. The edge-lit display represented a marked advance over the display using vertical rows of incandescent or neon lamps known as the columnar, thermometer or pinball display. To read a number like 4,928, one's line of sight moved up, down, then up again, as well as horizontally. Non-Linear continued to use the edge-lit display exclusively till the late 1960s, when it offered meters with the Burroughs Nixie gas-discharge tubes as well as meters wi th the edge-lit models. The Nixie readout was less expensive, but some users didn't like the neon-red glow. DVMs to PCs to out In the spring of 1981, NLS moved into the personal-computer business with an all-in-one PC package that included CPU, printer, monitor and keyboard. The company changed its name to Kaypro in 1983, while retaining the Non-Linear Systems name for the dwindling DVM business. In the year ending August 1983, when it went public, Kaypro had revenues of $75 million, with $3.5 million in DMMs and the remainder in computers. A year later, revenues climbed to $120 million. By 1990, revenues had dropped sharply to $50 million and Kaypro went into bankruptcy. In 1992, Andy Kay and his brother Stephen started again with a new PC company, Kay Computers. The DVM eventually killed the differential voltmeter, an instrument that required balancing by manually rotating a series of knobs to null an analog panel meter. The differential-voltmeter business was dominated by John Fluke Mfg., wh ich introduced its first such product in 1955. Its approaching demise drove that company into the DVM business, where it became one of the dominant factors that killed Non-Linear Systems. The first competitor In the beginning, NLS owned and was alone in the DVM business. But its monopoly didn't last long. In 1954, engineer Jonathan Edwards left the company and, with Walter East, a Non-Linear sales rep, started the first competitor, Electro Instruments (EI). MIT graduates Kay, Edwards and East all had worked at Bill Jack Scientific Instrument. East had left that firm to sell for an independent sales rep, G.S. Marshall Co. (now Marshall Industries, a distributor), which represented Non-Linear Systems. Electro Instruments was the first to use stepping switches instead of high-speed relays. So NLS added stepping-switch DVMs to its line. Then, to boost life from 100 million cycles to 400 million or more-and to reduce electrical noise and what some users described as an infernal audible racket-the com pany offered oil-immersed stepping switches. Stepping switches were less expensive than relays, but they were slower and didn't last as long. Eventually, NLS offered DVMs that used mercury-wetted reed relays-long-lived, capable of a billion cycles and fast, but expensive. Each instrument might have more than 20 relays. Electro Instruments grew rapidly. In its first year, according to East, sales reached about $300,000. In its second year, revenues jumped to $1 million, then to $3.5 million in the following year. EI acquired several other companies and started on the path to becoming a multinational conglomerate (a popular management passion at the time), not just an instrument maker. Honeywell bought the company in 1966 and shut it down less than a year later. East now spends his time as an investor. Edwards died some years ago at the age of 49. Measuring power, then voltage The third company in the DVM business, Cubic Corp., was founded in 1951, the year before NLS. It produced a calorimeter f or measuring the output power of magnetron tubes. Its founder and chairman (to this day), Walter Zable, lays claim to having developed the first transistorized power supply, as well as the calorimeter. In 1957, Cubic entered the DVM business in the belief that, as an instrumentation company, it had to be in that business. But it made no special contributions and abandoned the business in 1960 because, said Zable, it was not very profitable. Today the company is a leading manufacturer of computerized fare-collection systems. Then came Cohu, which introduced its first DVM in 1958. Like Non-Linear's first instruments, the Cohu unit used telephone-type relays, but it boosted relay switching speed to 60 pulses per second instead of the 20 in the NLS meters. Unfortunately, the relays weren't designed to take that speed; they died quickly, so Cohu retreated to the 20-pulse/s level. Cohu used a novel display, the projection readout manufactured by Industrial Electronic Engineers, which projected light from one of 10 tiny lamps through a film, one numeral lined up with each lamp. The selected numeral was projected onto a front screen, often clearly, sharply and in focus. A line of projection readouts provided not only an in-line display, but an in-plane display as well. The unit was first to do so. In time, Cohu dropped out of DVMs, opting to stick to its main arena, equipment for closed-circuit TV. The DVM business, said Jim Barnes, former chief executive officer, was getting too crowded. Among the toughest competitors in that crowded field was Dana Laboratories, one of the last southern California DVM startups. The main founders were Jack Bishop, chairman and president, and Mark Howlett, vice president of marketing. Both had been at Beckman Instruments. After earning a BS in engineering from the University of California at Berkeley, then doing a stint in the Navy, Bishop earned an MBA from Harvard Business School in 1948. When he returned to California looking for a job as an engineer, most companies, learning he had gone to a business school, wanted to know if he took shorthand. Beckman Instruments, then in South Pasadena, offered him a marketing position in 1948. He moved up quickly and was involved in acquiring Berkeley Scientific (developer of the first electronic counter) and Spinco, both in 1951. Bishop also moved the company's headquarters to Fullerton, Calif., where it remains today. In late 1959, Bishop, who was general manager, left Beckman Instruments along with marketing vice president Howlett. The two set up shop and sold a quarter of their consulting time to Boston Capital for $50,000 a year and another quarter to DuPont for $50,000 annually. That meant they could spend half their time looking for a good business to get into. In the fall of 1961, with initial funding from DuPont, they acted on an idea posed by two bright, former Beckman engineers, Norman Walker and Noel Braymer, who wanted to start a DVM compan y. The four men finalized the deal in a small restaurant in a coastal town midway between San Diego and Los Angeles-Dana Point-from which Dana Laboratories took its name. The chopper leaves The company started with a significant advantage. Walker, who became vice president of engineering (and was formerly director of engineering at Electro Instruments), had invented a technique using transistors to provide feedback from the zener reference. The scheme eliminated the need for mechanical-chopper stabilization. In fact, Dana Labs started out making chopperless low-drift dc instrumentation amplifiers because they were easier to sell, and only later added DVMs to its product mix. The company used bipolar transistors rather than relays or stepping switches for switching ladder-network resistors. Further, it isolated the DVMinput from ground. The instrument was promoted as a solid-state DVM, though it did use relays for range switching. Braymer added experience acquired as an engineer with Electro I nstruments, which he joined in December 1959. Prior to that, at Consolidated Engineering Corp. (founded by Herbert Hoover Jr., son of the former president), Braymer developed in 1948 a rack of equipment that some might consider a forerunner of the DVM. He joined Beckman in 1954 and stayed five years before launching Dana with Walker. The Dana voltmeter had another unusual feature-the display. It used the gas-discharge tubes called Nixies. Invented by two brothers, the Haydus, who later sold their company to Burroughs, these tubes had a stack of formed numerals, each on an almost invisible supporting structure. A selected numeral would shine with a neon glow. Burroughs had the Nixie at its research center in Paoli, Pa., by 1954. Though the Nixie required a high-voltage supply of at least 170 Vdc, a numeral could be switched on with as little as 20 V. The tube consumed no more than about a watt and it soon took over almost all digital-display applications. Virtually every digital instrument adopted the Nixi e until the light-emitting diode came along. It cost about $10, but the price per digit soon plummeted. There was one more factor that led to Dana's success, and his name was Jim Helfrich. In 1964, Dana Labs invited him to run its European sales. He joined the company that October, went to Europe and set up Dana corporations in Germany, France and England, making Dana a significant DVM vendor in Europe. The first handhelds Dana made another mark in DVM history by offering, in the early 1970s, the first successful handheld DVM, the Danameter. A British company, Sinclair Radionics, had introduced a handheld DVM using light-emitting diodes. But to conserve battery power, the LEDs were run with so little brightness they were hard to read. The instrument did not survive in the marketplace. The Danameter, in contrast, used a power-thrifty 3-1/2-digit liquid-crystal display. DVM manufacturers had begun to use the term "half digit" for a display with "1" as the first digit. A three-digit display readi ng up to 999 moved up to become a 3-1/2-digit display, reading up to 1,199 for 20 percent over-range. Though Dana was sailing along briskly, a new and very tough competitor, Hewlett-Packard, began making waves in the increasingly crowded DVM market. By 1979, co-founder Bishop felt that Dana no longer had anything sufficiently proprietary to compete against HP long term. He sold the company to Racal, a British concern, and Racal-Dana exited the DVM business in 1981. Bishop is now chairman of a successful microwave-counter business he acquired, EIP Microwave, which competes with HP. Hewlett-Packard began its assault on the DVM market in 1958 with the rack-mount model 405AR, designed by Ted Anderson and Noel Pace. Selling for $825, the autoranging, autopolarity instrument displayed dc-voltage measurements from 1 mV to 999 V on three one-inch-high Nixie tubes. It provided coded output for data loggers. Operation was based on a simple ramp circuit. The attenuated signal was compared to a linear ramp, and wh en balance was reached a gate closed and the number of clock pulses gated out during the rise of the ramp was counted. The count was proportional to the input voltage. In the mid-1960s, Fairchild Instrumentation introduced a small benchtop DVM using dual-slope integration, a technique developed by Fairchild's Steve Amman. Many later digital multimeters and digital panel meters adopted the dual-slope technique, but the Fairchild DVM did not survive in the marketplace. The HP DVM was developed in HP's hometown, Palo Alto. In 1960, HP opened a facility in Loveland, its first manufacturing plant outside California and the new home for the company's voltmeters. It became Noel Pace's new home, too. A DVM in a probe HP grew rapidly in the DVM business, so much so that southern California was no longer the source of all DVMs and, in time, was no longer the source of any. Pace kept working on more advanced voltmeters and was involved with the dramatic 970 Probe Voltmeter introduced in 1974. Everybody who had an interest in DVMs talked about it. The instrument got more press coverage than any other voltmeter in history, thanks, in part, to the ardent activity of Walter Skowron, Loveland's press-relations maven, but mainly on account of its unique design. It was featured on 92 trade-press front covers around the world. The handheld DVM included a probe that plugged into a banana jack. A pigtail could be tied to ground. The one feature that triggered more comment than any other was a switch that let a user invert the reading on the 3-1/2-digit LED display. If you were probing down into some equipment, you could press a switch to provide a right-side-up display. There were rumors in the industry that the folks who designed the 970 were observing with envy the astounding success of another HP product: the HP 35, the world's first handheld scientific calculator, developed by a group that included Ted Anderson, who had worked on HP's first DVM. The 35 calculator flew off dealers' shelves at $395 when it hit the market in 1972. Why, then, shouldn't the handheld DVM, at $310? Alas, it had two major flaws. First, the bright LEDs chewed up battery power. Some engineers used the 970 as a bench instrument and kept running down the battery, an HP nickel-cadmium model that could be recharged in a special charger that HP sold. But people don't always think of recharging when they're making measurements. Second, its very popularity was the cause of its demise. Industrial companies didn't want to buy 970s because they kept walking away and mysteriously disappearing. HP did offer a lockable clamp, but that defeated the instrument's appeal. HP wasn't alone in a fruitless stab at the DVM-in-a-probe. Two years earlier, in 1972, Keithley Instruments, a leading manufacturer of electrometers and other extremely sensitive meters, introduced the model 167. Like the HP 970, it was a 3-1/2-digit instrument with a LED display. Unlike the HP model, it did not depend on batteries. Instead, the probe was powered by hefty D cells i n a small box, to which it was tethered by a 14-conductor cable. For use as a bench instrument, the probe be slipped into a front-panel recess, in which case input signals were applied to a pair of banana jacks. Thus the probe DVM could become a bench instrument with a small LED display. A very small LED display. The characters, about 1/8-inch high, were adequate in a probe, which is normally used no farther away than arm's length. But many people found them too small in a bench instrument. And the tether, about two feet long, was sometimes too short. John Yeager, Keithley's first employee back in 1950, recalls an incident with a telephone-company executive who showed great interest in the 167. But he required a longer tether so that an operator could rest the box on the floor while using the probe to make tests anyplace in a relay rack. Carrier sling gets ax Why not use the carrier sling supplied with the instrument, designed to slip conveniently over an operator's shoulder? No, said the e xecutive, that wouldn't do. If the operator carries equipment, he moves into a higher pay grade. Sales of the 167 were disappointing. Joe Keithley Sr. said, "We probably sold 10 or 12 of them." Yeager, probably more realistic, added: "No, it was more than that." As digital voltmeters, then digital multimeters, continued to penetrate the market for basic measurements, they eroded the market for differential voltmeters-one overwhelmingly dominated by Fluke. The differential voltmeter was capable of far more precise measurements than many DVMs, but it required a more skillful operator. It also took longer to make a measurement, as it required rotating a series of knobs till a meter in a bridge circuit showed null. You then read from the knobs. The DVM was killing Fluke's growth and beginning to drive Fluke out of business. Fluke responded by introducing a digital multimeter of its own, the 8300, a rack-width 5-1/2-digit meter with Nixie display. That was in 1969, when everybody felt that what the world di d not need was another DMM. The 8300 challenged the main competitor at the time, Dana, whose 5-1/2-digit instrument sold in the $4,000 to $5,000 range. The Fluke came in at $2,000 and was probably the principal trigger for Dana's departure from the DVM business. The companion Fluke was the 8100, a half-rack 4-1/2-digit $695 DMM with a battery-power option. This unit evolved into Fluke's extremely successful handheld DMMs. These used LCDs, eliminating the battery-drain problem that plagued earlier handhelds. In time, Fluke brought out a $300 DMM and ran into the problem that killed earlier manufacturers of low-cost DVMs. For example, in Dayton, Ohio, United Systems had introduced the Digitec DVM in the mid-1960s for around $300. That instrument used a three-digit electromechanical display that looked like the odometer in many automobiles. A servomotor drove a potentiometer and the display till the input signal was in balance with the reference. The instrument was painfully slow. Some users commented that t hey could go out for coffee while the reading was coming up. Its low price tag couldn't save it. Another low-cost contender was ElectroLogic, a company that made oil-exploration instrumentation. The ElectroLogic DVM had a rotating drum with 000 to 999 imprinted on its edge. The wheel rotated at high speed till input-signal/reference balance was achieved, at which point a strobe light would flash at the three-digit numeral that would at that moment be passing a small viewing window. The number, representing the input signal, would appear stationary-well, almost stationary-in the viewing window. The president of ElectroLogic, Vincent van Praag, faced a major hurdle. How do you find sales representatives who will take on a $300 DVM for the usual 20 percent commission when most DVMs sell for $2,000? The answer was that you can't. The low-cost winner Fluke recognized the problem. It tried, at first, to do credit-card sales by mail order. That experiment didn't work and the idea was dropped. Then Fl uke came up with the radical idea of using distributors. That worked. Today, Fluke dominates the market for handheld DVMs while HP dominates the bench and rack market. At one time, there were 24 to 40 or more manufacturers of digital multimeters (memories and estimates vary). Today, there are no more than a half dozen significant vendors.