Wednesday, February 28, 2024
HomePakistanRemembering Knowledge Compression Pioneer Okay. R. Rao

Remembering Knowledge Compression Pioneer Okay. R. Rao

[ad_1]

This text was first printed as “The battle for the SX-70.” It appeared within the Could 1989 subject of IEEE Spectrum. A PDF model is accessible on IEEE Xplore. The diagrams and pictures appeared within the unique print model.

But this difficult system had to slot in a bundle the dimensions of Land’s jacket pocket, he decreed—a constraint that meant using ICs. However as Polaroid couldn’t fabricate ICs, the success of its SX-70 challenge lay within the arms of outsiders.

The flash management contract was given to Normal Electrical Co. Then in 1971, when GE dropped out of the IC enterprise, it was issued to Sprague Electrical Corp., in addition to to Fairchild Semiconductor Corp. of Palo Alto, Calif., and Texas Devices Inc. of Dallas, Texas. Solely Fairchild and Sprague ended up producing flash controllers.

Impartial contracts to develop the motor and publicity management modules went to Fairchild and TI. The motor management module contained a linear management IC, an NPN motor drive transistor, and a discrete PNP dynamic braking transistor, and gave the designers little bother. The publicity management module was a distinct story.

The grand problem

Included within the publicity management had been three ICs (early Fairchild variations had 4). The publicity timer used the present output of a silicon photodiode to manage how lengthy the shutter blades remained open. The delay-timing circuit generated 4 intervals: a delay of 40 milliseconds earlier than the shutter opened; the time the shutter remained open earlier than the flash was fired; the length of the flash; and the utmost publicity time given sure ambient lighting. The ability management IC drove the solenoids and motor management unit. And this all needed to match on a board that match right into a 27-by-95-by-2-millimeter area, minus a central gap for the digital camera lens.

Electrical noise was a serious stumbling block. The photocell, as an example, working with as little as 15 picoamperes, needed to keep its state in an surroundings wherein the motor, the solenoids, and the firing of the flash lamps drew amperes of present. Designers had been to take steps like inserting a delay between the discharge of the solenoids and the beginning of the photocell-timed publicity; redesigning circuitry on the facility provide line to reject noise from the motor; rising the voltage distinction between logic highs and lows, so noise spikes would now not masquerade as bits; and together with a low-pass filter.

Because it was 1969, there have been no semicustom ICs, gate array know-how was in its infancy, and solely primitive packaging was obtainable—customary twin in-line packages (DIPs) had been at the least 0.125 inch thick—whereas logic and energy transistors couldn’t but share the identical piece of silicon. And Polaroid needed to purchase this publicity controller for US $5.75.

What buddies are for

Polaroid chairman Land and TI chairman Patrick Haggerty had been previous buddies. On a weekend journey many years earlier than the SX-70 challenge, that they had mentioned how electronics may at some point make a really one-step digital camera potential. The thought was to work on this dream collectively as quickly because the know-how arrived. So it got here as no shock when TI was charged with creating the digital camera’s publicity management board. Land was relying on TI for a fail-safe design, based mostly on analog circuitry and confirmed know-how and subsequently dependable, fairly priced, and able to being produced on schedule.

Polaroid additionally requested Fairchild, which it seen because the nation’s chief in IC know-how, to sort out a design that may push the cutting-edge. Fairchild’s model was to be digital and extremely built-in, even to combining energy transistors with logic on one chip. To Polaroid the method regarded dangerous, however its engineers had been excited by its potentialities. Nonetheless, some inside Polaroid thought the Land-Haggerty relationship made nonsense of utilizing anybody however TI.

In the dead of night

The R&D contracts had been awarded in 1969, and the opponents went to work, each with the identical handicap: incomplete info. Fearing that Kodak Corp. may enter the moment digital camera enterprise, Polaroid needed no leaks—a lot in order that it talked about neither the brand new movie nor the truth that at one level the digital camera was redesigned as an SLR—and saved the design groups from seeing a prototype of the digital camera. (Though TI’s then govt vp, the now-retired Fred Bucy, noticed an illustration of the early, non-SLR SX-70 in 1969, he stated nothing about it to the corporate’s engineers.) Mentioned Peter Carcia, an engineer on the SX-70 challenge and nonetheless with Polaroid: “They’d little or no to work with”—solely stacks of specs.

white wiring diagram on blue-grey background with gates, outputs, and other features indicated, plus words "systems diagram M3 phase II G717566

When it contracted with Fairchild and TI to develop the electronics for the SX-70 digital camera, Polaroid Corp. offered this timing diagram together with 30 pages of different design specs, reliability necessities, and check info. It signifies sequences of occasions for the 4 completely different modes of operation a completely computerized digital camera required. Desk 1 signifies features for taking pictures in ambient mild, Desk 2 covers flash operation, Desk 3 calls out the sequence of occasions that’s triggered when a brand new movie pack is put in and its protecting cowl should be ejected, and Desk 4 describes the operations that happen when a pack of movie is used up.

Polaroid engineers recall that masses on the electronics had been described merely as inductive, and that particulars of the battery provide had been imprecise as a result of a brand new battery was being concurrently designed.

“We didn’t inform them whether or not a load on the electronics was from a solenoid or a relay, simply that it was an inductive load,” remembers Seymour Ellin, now a senior technical supervisor at Polaroid.

“Since we had been making our personal battery [designed concurrently], we couldn’t inform them what the battery provide could be,” stated Carcia. “I might inform them ‘I would like you to design a circuit, however I gained’t inform you what the facility supply might be,’ and they’d take a look at me unusually.”

Polaroid needed no leaks—a lot in order that it talked about neither the brand new movie nor the truth that at one level the digital camera was redesigned as an SLR.

Even worse was the “Y” delay—which Polaroid engineers instructed IEEE Spectrum got here from the “why” response given Fairchild and TI engineers every time they questioned one specification: the quick delay earlier than beginning the publicity, after the person pressed the button. This pause was to permit the mirror (which in an SLR digital camera displays the picture seen by way of the lens to the viewfinder) to cease vibrating after it snapped out of the best way of the movie to be uncovered. However that was greater than Polaroid needed to expose. The sources of the noise drawback had been left obscure, and its extent understated, stated Clark Williams, then a TI design engineer. “That motor pulled 3 amps of present and put out a wealthy spectrum of noise that performed havoc with our circuits,” he stated. (He’s now a design supervisor at Dallas Semiconductor Corp. in Dallas, Texas.)

The TI crew, unable to base a breadboard on Polaroid’s diagrams alone, despatched two engineers and several other technicians to Cambridge to work in a little bit non-public room there. Each time they wanted to check their breadboard, they might hand it over to Polaroid engineers, who would carry it to a different room and ultimately report again that, say, a sure sign wanted adjustment or a sure part didn’t operate. The TI engineers would make just a few changes, then the breadboard was carried off for one more check. This to-and-fro-ing went on for six months, whereas, stated Michael Callahan, a senior TI design engineer who’s presently govt vp of engineering at Crystal Semiconductors Corp. in Austin, Texas: “We may have finished the work in two weeks if that they had allow us to signal nondisclosure agreements.”

Designing in Dallas

A preliminary spherical had disenchanted each IC groups. In 1969, earlier than Polaroid had firmed up many SX-70 particulars, it began each TI and Fairchild creating easy publicity management chips. This early effort, stated Polaroid engineers, was additionally used to develop and check their working relationship with Fairchild. However the SX-70 challenge modified a lot, significantly with its redefinition as an SLR digital camera, that Polaroid determined to begin over. Callahan and Ken Buss, now a senior member of the technical employees at TI, recall a gathering in Dallas at which the TI engineers proudly demonstrated the working circuits—solely to have Polaroid ignore them and announce its new necessities.

“That made our chips immediately out of date,” Buss stated. At Fairchild, too, enthusiasm flagged. Coincidentally, each firms quickly after underwent a company restructuring, however whereas the modifications at Fairchild benefited its SX-70 crew, these at TI practically price it every little thing.

The TI designers, as an alternative of working straight with Polaroid, had been instructed to report back to the Assembled Capabilities Group. Missing both chip growth or manufacturing services of its personal, the Group contracted with the IC designers’ division to develop three chips—a photocell amplifier to find out the proper publicity, a chip to manage the motor and deal with dynamic braking, and a chip to deal with timing, depend the movie used, and serve different features—and with one other division to fabricate the chips. The association additional filtered the already restricted info from Polaroid.

hand holding one circuit board between thumb and forefinger, two other circuit boards of similar size on table

Three completely different designs for the SX-70 publicity management electronics had been produced. Fairchild Semiconductor Corp.’s model (prime) went into cameras in 1972 and 1973—discover the polyimide movie used to connect the ICs to the board. Texas Devices Inc. produced its ceramic board (heart) throughout 1972, then redesigned, and gained the manufacturing contract away from Fairchild with a circuit board that used miniDIP IC packaging (backside).

That left the Group itself with the job of designing the circuitry that may tie the ICs collectively. Its engineers used 13 discrete transistors, 17 laser-trimmed thick-film resistors, and a photodiode, aspiring to mount them on a printed-circuit board. Administration as an alternative mandated a ceramic substrate primarily as a result of, stated one TI design engineer, the Group reported to the identical supervisor as TI’s Hybrid Thick-Movie Group, which had extra capability.

“We knew we couldn’t meet the associated fee objectives with a ceramic substrate,” he stated. The ceramic, the dear metallic conductors, and the labor all price an excessive amount of for the substrate to function something greater than a prototype “to allow us to get all of the circuitry in a small space.” And when the design grew from 3/4 sq. inch to 4 or 5 sq. inches (from 5 to 25 or 32 sq. centimeters), the engineer recalled, he and the opposite designers predicted main manufacturing issues and urged doing a extra digital redesign with a printed-circuit board. However administration “wouldn’t pay attention,” he stated.

TI’s ceramic-based design did, nevertheless, carry out to Polaroid’s specs, and it went into manufacturing in late 1972. But it surely was certainly a nightmare. First, at $100 a unit, it was nowhere close to the $5.75 price objective. And manufacturing issues had been great, particularly with the big and subsequently fragile ceramic substrate. For example, stated TI design engineer Norm Culp: “We needed to take a chip, alloy it to a Kapton movie service [a high temperature plastic foil], then wire bond the chip to the Kapton service, then encapsulate the chip. The Kapton movie carriers had been then examined individually, then reflowed onto the ceramic substrate.”

Yield was about 1 p.c, and that one in 100 typically cracked on its solution to Polaroid.

Furthermore, stated Culp, reflow-soldering chip carriers to the substrate brought about microcracks within the ceramic, and for some time TI inspected each half for the failings. Then one engineer realized that heating all the substrate as an alternative of simply the half to be reflow-soldered would scale back the microcracks, which, nevertheless, confirmed up in different components of the method. Yield was about 1 p.c, and that one in 100 typically cracked on its solution to Polaroid.

Polaroid did order a number of hundred of those ceramic modules to get the SX-70 to market. But it surely wasn’t in any respect proud of them. Mentioned Ellin, “TI, primarily, failed to satisfy the associated fee goal.”

Competing in California

In the meantime, engineers at Fairchild had been additionally operating into difficulties, however technical ones solely. Early within the design course of, Fairchild’s company restructuring moved the R&D engineers out of their remoted laboratory into working divisions, making for higher communication with manufacturing, which “resolved numerous issues,” stated Howard Murphy, a senior member of the Fairchild analysis employees and the challenge director for the SX-70 electronics.

“We designed a die that had round 20 flip-flops on it, most likely a brand new excessive in IC complexity at the moment.”—Howard Murphy, Fairchild

One design drawback was excessive temperature. Murphy recalled that the warmth of the heavy currents drawn by the motors and the solenoids affected the management logic circuitry, which then needed to be redesigned to work at increased temperatures—the specs indicated 40 °C. One other hurdle was the photograph circuit. It needed to outing after 20 seconds, in order that photos may very well be taken in dim mild of about 0.06 candela per sq. foot (0.65 candela per sq. meter), though the circuit design crew wasn’t totally conscious of the rationale for this on the time. The circuit additionally needed to be very small and devour just some milliamperes. “So we designed a die that had round 20 flip-flops on it, most likely a brand new excessive in IC complexity at the moment,” Murphy remembered.

Frank Perrino, a Fairchild product supervisor, first turned concerned within the SX-70 challenge in Could 1971, when he oversaw its transfer into manufacturing. He recalled that the designers had been then engaged on 4 chips—a driver for the motor and solenoids, a timing chip, and the photodiode and photodiode amplification chips that later turned one bipolar CMOS IC. The cube had been to be mounted straight on an irregularly formed 1-by-4-inch ceramic substrate beforehand metalized on each side with state-of-the-art traces and areas.

The prices concerned, nevertheless, dominated the method out for manufacturing, Perrino instructed Spectrum. “The ceramic and chips all needed to be excellent,” he stated, and there was zero “likelihood of this taking place.”

He concluded a printed-circuit board was a should, however the way to mount the chips to it? Fairchild’s plastic DIPs had been too giant and expensive for the job. He had, although, learn a paper by Normal Electrical engineers on beam tape packaging (BTP), a forerunner of what’s now referred to as tape automated bonding (TAB). After investigating BTP, he instructed Fairchild and Polaroid administration, “If we don’t do it this fashion, it’s not price doing.” Each agreed.

BTP employed reels of movie with copper traces laminated on it round preexisting holes. Chips with bumps of solder on their pads had been centered beneath the holes and bonded to the overhanging copper lead frames. Particular person die/movie modules had been then encapsulated, examined, clipped off the reel, and soldered to the circuit board.

Perrino laid out the double-sided printed-circuit board at residence on paper unfold throughout his pool desk. He then visited a number of firms that made polyimide interconnect movie, contracted with 3M as a provider, and persuaded West-Bond Inc. of Anaheim, Calif., to construct tools for attaching the cube to the reel of laminated movie. The ultimate circuit board held three IC cube and two flip-chip, thick-film, laser-trimmed resistors.

Nonetheless, yields weren’t following the anticipated studying curve on two of the three ICs, the facility transistors due to excessive doping ranges and the timing chip as a result of, stated Perrino, of design errors. For instance, Jim Feit, one other engineer on the challenge, remembers a parasitic gadget affecting the flip-flops, which was mounted with the addition of a delay.

Nonetheless, although the components weren’t low-cost, costing Fairchild roughly $20 or $30 every, they had been manufacturable.

D-day

The SX-70 was launched in April 1972, along side the corporate’s annual stockholders’ assembly. A yr earlier, Land had teased the stockholders by pulling a prototype SX-70 out of his pocket and waving it within the air. That was a working mannequin, containing one in every of TI’s first profitable ceramic circuit boards. However for this assembly, Polaroid wanted 20 cameras, and John Burgarella, now retired from the corporate, needed to make a number of journeys to Texas to hand-carry sufficient working boards again to Cambridge. A few month earlier, Land had introduced Fairchild engineers Perrino, Murphy, and Will Steffe to his Cambridge workplace and demonstrated the digital camera to them. “It was clearly a technological breakthrough,” recalled Perrino, which motivated them “to return and make the factor work.”

u200bMan holds camera at chest level and waves photo above head with other hand

Edwin Land confirmed the primary working SX-70 digital camera at a stockholders’ assembly in 1971. It was solely a prototype, and contained one of many first working ceramic circuit boards produced by Texas Devices. A TI engineer had put in it the evening earlier than the assembly, working with a digital camera that was shrouded to forestall him from studying something extra about it than he already knew.

The introduction went off with no hitch. A few dozen scenes, from a poker recreation to a toddler’s party, had been enacted in a big warehouse, and well-known photographers had been capturing them with the brand new cameras whereas Polaroid stockholders circulated and examined the photographs. Polaroid engineers had been additionally circulating, with additional cameras of their pockets in case something went unsuitable.

Resting on their laurels

So Fairchild gained a contract to fabricate the publicity management modules together with the motor circuits and the flash management circuits. The commerce press touted their victory. In accordance with a January 1973 Digital Information report, as an example, this contract, “believed to be the most important ever issued by a digital camera producer to an electronics provider,” was price $19 million, and was “thought of by some semiconductor executives as an omen of appreciable future enterprise.”

Fairchild disbanded most of its design crew, happy with their success. However the manufacturing engineers pressed on, since the price of the product needed to be lowered by three-quarters or extra to satisfy Polaroid’s value goal, and contract negotiations had been to be reopened for 1974. Nonetheless, stated Perrino, two of the chips within the publicity management module had been nonetheless in bother.

C. Lester Hogan, who had not too long ago left Motorola Inc. to take over the Fairchild presidency, blames Fairchild’s then-outdated manufacturing services. He began a modernization, however he stated, “there wasn’t numerous additional money,” and it was not full till someday in 1974.

Perrino blames the IC designs as nicely. “The design guidelines utilized in these chips had been touch-and-go with the know-how,” he instructed Spectrum. Polaroid’s Carcia agreed: ‘‘We had been pushing the elemental know-how.” Redesigning the chips was talked about, however administration didn’t mandate it.

A matter of pleasure

The TI design crew was additionally disbanded in 1972. Some left the corporate, some moved on to different tasks. The failure, one design engineer instructed Spectrum, was a black mark that harm careers.

On the highest stage of TI, nevertheless, the guide was not being closed. TI chairman Haggerty reportedly referred to as his previous pal Land and stated, “We at TI don’t fail.” He assigned the challenge about $540,000 from his personal price range, and instructed his managers to do no matter it might take to succeed. The code title Challenge Alpha emphasised the significance of the contemporary begin, and Haggerty put govt vp Bucy in command of it.

The failure, one design engineer [said], was a black mark that harm careers.

As the unique TI crew had been disbanded, Bucy deliberate to assemble one other one from the semiconductor division, and to make sure that this one would talk straight with Polaroid and now have manufacturing obligations.

Dean Toombs, engineering director of the semiconductor group, held a collection of conferences and developed a proposal for the redesign that was one other break with TI’s first method: it relied not on confirmed however on state-of-the-art IC know-how and packaging. A circuit board just one/64 inch thick was to carry as much as 4 digital (not analog) ICs and eight discrete elements at most. The chips could be floor mounted to the board in a miniDIP bundle, a technique of quantity meeting then new and dangerous however low-cost. (It’s now referred to as SOT, which stands for Small Define Transistors.)

The plan was authorized by Bucy, and Henri Jarrat (then Eljarrat) chosen to move the hassle. At first Jarrat objected to the project, however gave manner when instructed it was TI’s prime precedence. Given carte blanche to assemble a crew from wherever within the group, he saved the group manageably small—solely 18 folks. They rapidly partitioned the circuitry into three ICs and introduced a six-month schedule for the redesign to Fred Bucy and Polaroid president William McCune.

Expert specmanship

Then Jarrat had his first assembly with Polaroid engineers. He instructed them he may solely combine the publicity management operate into three elements in the event that they waived a few of their specs. He started happening his listing and to every request the Polaroid engineers stated no. So Jarrat stood up, threw his papers down, and stated, he recalled, “Now I do know why this challenge goes nowhere. This may by no means work, and I don’t wish to have my title hooked up to a failure.” He charged out of the room. Toombs backed Jarrat’s risk. “We needed to get the shopper beneath management,” he instructed Spectrum.

The flexibility to barter was partially additionally as a result of availability of working cameras to check and the development of a prototype on which to check breadboards of the chips—luxuries denied the primary TI crew.

After a short adjournment, the assembly was reconvened and from then on Polaroid negotiated specs. For instance, the 20-second outing, for taking an image in a dimly lit room, had made the sign from the photodiode impossibly low for the primary design groups and this time round was lower to 10 seconds. “The large purpose for our success was Jarrat’s success at convincing them to ease the specs,” stated Clark Williams, a member of the second crew.

The flexibility to barter was partially additionally as a result of availability of working cameras to check and the development of a prototype on which to check breadboards of the chips—luxuries denied the primary TI crew. And when the primary group did increase questions out of concern for manufacturability, recalled Buss, the one TI engineer to work on each the design and the redesign efforts, they had been instructed, “Properly, your competitors can do that.” And, actually, Fairchild engineers don’t recall that the specs had been problematical.

TI started producing the Challenge Alpha boards in amount in mid-1973.

table listing part types and quantities

With the redesign, TI quoted Polaroid a value of about $4.10 a unit—nicely under the $5.75 goal. Mentioned former Fairchild president Hogan: “On the time, it price us $10. We actually believed we may get it to $6, however when TI bombed the value down to 2 thirds of the goal value, we simply needed to drop out.” As for a redesign, stated Hogan, “we didn’t have the cash to take a position that manner—we needed to spend money on the generic fixing of the manufacturing unit.”

TI created a particular digital camera division with Polaroid as its solely buyer. The corporate made about 850 000 models in 1974 and continued to supply the design till the SX-70 and the SX-70 Mannequin 2 had been discontinued in 1977. It additionally spun off just a few improvements, together with packaging for TI’s watch shows. And the engineers on the Challenge Alpha crew had been rewarded with then substantial raises of $100 to $500 a month.

West-Bond and 3M, firms Fairchild had recruited to fabricate packaging tools and movie tape, continued to profitably produce them for different firms.

Fairchild used the BTP packaging know-how it developed for the SX-70 on its high-volume plastic DIP merchandise at a number of manufacturing services. It additionally took its digital camera management know-how abroad on a tour of Japanese digital camera producers, however after a number of unsuccessful months gave up and closed down the manufacturing line for the publicity management module. It continued to fabricate flash management modules for Polaroid for one more yr, nevertheless. Inside six months to a yr of dropping the publicity management contract, at the least half the individuals who had labored on the challenge moved to different firms, Feit recalled.

Pleased ending

May the design have gone extra easily? Actually higher communications between Polaroid and the 2 semiconductor firms and amongst completely different divisions inside TI and Fairchild would have eradicated a few of the tough spots.

From Polaroid’s standpoint, the data it handed out was as full because it may very well be. In any case, a number of components of the digital camera system had been being developed concurrently, in order that the system specs couldn’t in the meantime be finalized. Additionally, stated one Polaroid engineer, unfamiliarity with images impaired the IC designers’ comprehension of the information they got.

Within the eyes of the TI and Fairchild engineers, helpful info was withheld, and Polaroid engineers do admit a preoccupation with secrecy.

Nonetheless, within the eyes of the TI and Fairchild engineers, helpful info was withheld, and Polaroid engineers do admit a preoccupation with secrecy on account of concern over competitors from Kodak. Maybe being instructed that sure design points had but to be resolved or an in depth clarification of how an SLR features would have elicited extra inventive engineering from the IC designers.

Be that as it could, the SX-70 was an excellent success. Polaroid bought some three million models of the leather-covered Mannequin 1 with its chrome-plated trim and the plastic-bodied Mannequin 2. (Mannequin 3, launched in 1975, was not an SLR.) So whereas the design issues each TI and Fairchild endured triggered tense moments in any respect three firms, their answer opened up an enormous new client market in electronics.

To probe additional

For particulars on the SX-70 circuitry, see “Behind the lens of the SX-70,” by Gerald Lapidus, IEEE Spectrum (December 1973, pp. 76-83).

Each Time and Life magazines featured the SX-70 digital camera on their covers in 1972, and mentioned it in “Polaroid’s Large Gamble on Small Cameras” (Time, June 26, 1972, pp. 80-82) and “If you’ll be able to state an issue, it may be solved” (Life, October 27, 1972, p. 48). To know how the event of the SX-70 match into Polaroid’s Jong historical past, learn The Instantaneous Picture: Edwin Land and the Polaroid Expertise by Mark Olshaker (Stein & Day, New York, 1978).

Frank Perrino’s model of tape automated bonding is described in U.S. Patent #3,868,724, “Multi-layer connecting buildings for packaging semiconductor units mounted on a versatile service,” dated Feb. 25, 1975.

From Your Web site Articles

Associated Articles Across the Net

[ad_2]

RELATED ARTICLES

Most Popular

Recent Comments