The Story of the Little Computer That Could!


Revised 4/9/06


Two Stanford graduate students named Bill Hewlett and Dave Packard form a partnership called the Hewlett-Packard Company with the urging and financial assistance of their legendary electrical engineering professor Frederick Terman, who would do much to jump start the creation of Silicon Valley.

Based on a coin toss, Hewlett won the right to have his name listed first in the new company’s name. The Hewlett-Packard Company started in the garage of Packard’s rented home in Palo Alto, California and first manufactured an audio oscillator based on a design Hewlett developed at Stanford. The oscillator used a tungsten lamp as a thermally self-compensating feedback element. The first customer is Disney Studios, which uses the oscillators to help create the soundtrack and test audio systems for the groundbreaking animated film Fantasia.

Parts for the audio oscillator are made in the Packard garage and kitchen, establishing an early and important company policy: make whatever you need if you can’t get it on the open market. This “can do” attitude will be extremely important time and again in the development of the HP 9825 desktop computer, its predecessors, and its successors.



Howard Vollum, who met Bill Hewlett while working with him in the Signal Corps during World War II, visited Dave Packard to discuss his ideas for making a triggered-sweep oscilloscope based on radar techniques developed during the War. Packard perceived that Vollum really was an entrepreneur who needed to start his own company. He gives Vollum an entree into the network of HP sales representatives and sends him out into the world to make his fortune. Vollum incorporated in late 1945 and his company ultimately became Tektronix in 1946, after two renamings.

Vollum built his first triggered-sweep scope from military surplus parts and soon cornered the ‘scope market. HP eventually realized that a test and measurement instrument company that purportedly offers a complete instrumentation line must also offer oscilloscopes so the company finally entered the oscilloscope market in 1956, far behind Tektronix. HP’s analog scopes never become market leaders because HP never developed trigger circuits as good as the ones on Tek scopes. In this instance, Packard’s business largess cost HP the lion’s share of the oscilloscope market for a total of about 40 years, until the advent of digital sampling oscilloscopes (DSOs). Coincidentally, analog scopes can’t interface to computers because they don’t digitize signals. DSOs can and do.



John Bardeen, Walter Brattain, and William Shockley invent the transistor at Bell Labs. Bardeen and Brattain manage to get a crude point-contact germanium transistor working late in 1947 and Bell Labs announces the development in 1948.

Dave Packard will be averse to using transistors in HP’s product designs for another 10 years or so because they will be more expensive and less reliable than vacuum tubes throughout the 1950s, which is typical of any disruptive technology. Eventually, HP will fully embrace the solid-state revolution. At first, HP makes its own semiconductor diodes because it can’t buy diodes with the performance it wants. Later, in 1961, HP established HP Associates to perform semiconductor (and eventually LED) research. Several HP desktop calculators and computers will use these solid-state displays.



Andrew Kay founds Non Linear Systems (NLS) in Del Mar (a suburb of San Diego, California) with the intent of developing the world’s first digital voltmeter (DVM). His plan is to simplify the taking of electrical measurements so that less skilled workers can take measurements more accurately. The DVM Kay creates is really the first digital instrument. Without digital instrumentation, there’s no way for computers to automate test and measurement processes. The HP 9825 will revolutionize automated test and measurement systems in another two decades with the digital descendants of Kay’s DVM.

Andrew Kay will form Kaypro as a division of NLS in 1982 to develop and sell portable computers. Kaypro’s PC sales completely overshadow DVM sales so the company is renamed Kaypro in 1983 and NLS becomes the subsidiary. NLS files for bankruptcy in 1990 but the company reorganizes and still sells specialized panel meters. Kay starts Kay Computers in 1992 and continues to make and sell PC-based computers in the 21st century.

Meanwhile, Barney Oliver leaves Bell Labs (where the transistor was born) and joins HP. Among countless contributions at HP, Oliver will play a central role in the creation of HP’s desktop calculators and computers.



NLS introduces its first electromechanical digital voltmeter based on stepping relays that closely resemble the switches used in telephone central offices. The Naval Electronics Laboratory (now the Naval Ocean Systems Center) buys the first unit in April 1953 for $2,300.

There are no liquid-crystal displays or light-emitting diodes in 1953 so NLS uses stacks of edge-lit Lucite plates to make displays. Each plate is engraved with a numeral. Each stack consists of 10 plates, each engraved with one numeral ranging from 0 to 9. When a plate is illuminated by a small incandescent bulb along the edge, its engraved numeral becomes visible at the front of the display because the engraving scatters the light in the plate. NLS will use these “in-line but not in-plane” readouts well into the 1960s. By 1962, NLS DVMs have digital outputs to drive recording equipment such as printers and plotters. The stage is set for automated test systems but computers will need another decade or so to catch up.

Meanwhile, RCA Laboratory’s David Sarnoff Research Center demonstrates the first magnetic core memory with 10,000 bits of storage. Core memory will become a key component of computer design in the 1960s including the design of HP’s first desktop calculator, the 9100, more than a decade later.



William Shockley (along with Bardeen and Brattain) wins the Nobel prize for co-inventing the transistor. Stanford’s provost Frederick Terman lures Shockley from the east coast to Palo Alto where he and Arnold Beckmann establish the Shockley Semiconductor Laboratory of Beckmann Instruments. Transistor manufacture remains more alchemy than science at this point.



By now, HP employs 1500 people and has its first public stock offering. Hewlett and Packard realize that they can no longer efficiently run the company’s R&D lab as one monolithic organization and they split it into four semi-autonomous sections: Audio-Video, Microwave, Time and Frequency (or Frequency and Time depending on who you believe), and Oscilloscopes. In an attempt to ensure managerial uniformity across company, they hold an eventful 2-day off-site meeting with senior HP managers in Sonoma. There they create a 6-point set of rules (later expanded to seven points) that form the foundation of the world famous HP Way, widely recognized as a benevolent and successful foundation for a Silicon-Valley management style.

Meanwhile, eight employees at Shockley Transistor Laboratory reluctantly led by Robert Noyce find Shockley’s managerial style so onerous that they quit en masse. Shockley names them the “traitorous eight.” The traitorous eight quickly sign a contract with the Fairchild Camera and Instrument Corporation to found Fairchild Semiconductor, which will focus on silicon-based transistor manufacturing. They instantly create a Silicon Valley tradition for walking across the street and creating a startup when the opportunity presents itself. The traitorous eight are: Julius Blank, Victor Grinich, Jean Hoerni, Gene Kleiner, Jay Last, Gordon Moore, Robert Noyce, and Sheldon Roberts. These people will profoundly influence the growth of Silicon Valley and the entire electronics industry.

Noyce and Moore will become disenfranchised with Farichild’s eastern corporate management and will leave Fairchild in 1968 to start Intel. The new company will initially focus on the development of memory chips and will quickly become closely involved in the development of HP’s second-generation desktop calculator/computers.

Shockley Transistor Laboratory never recovers from the departure of the traitorous eight. It eventually becomes the Shockley Transistor Corporation, a Beckmann subsidiary, and later becomes the Clevite Transistor Division when Clevite Corporation buys the Shockley Transistor Corporation’s assets in 1960.



During a July holiday weekend, Jack Kilby at Texas Instruments constructs the first integrated circuit (IC) by hand from of a piece of germanium (a “bar” of germanium in TI-speak) after only two months on the job. (He’s working over the holiday because he’s too new to have accrued any vacation like his fellow TI engineers.) Like the first Bell Labs’ transistor invented a decade earlier, Kilby’s IC is crudely built by hand but it demonstrates the concept of putting several components on one chip (or bar in TI’s parlance).

In January of 1959, Robert Noyce at Fairchild will independently conceive of and document a practical integrated circuit using silicon and Fairchild’s planar fabrication process, which is based on photolithographic processes that still serve as the foundation technology for fabricating every IC made today. Like the transistor, ICs will need about a decade of gestation before transforming the electronics industry forever. Fairchild and Texas Instruments will fight in the courts for a very long time over who will have the right to be called the inventor of the IC. Eventually Kilby and Noyce will be co-credited with the invention of the integrated circuit. Kilby’s device was first, but Noyce’s was practical.



HP opens its first manufacturing division outside of California in Colorado, the state where Dave Packard was born. Initially, Boulder was considered as a location for the new division because of the presence of the University of Colorado at Boulder. However, due to the concerted sales efforts of Loveland First National Bank President Paul Rice and local appliance dealer (and future Loveland mayor) Bob Hipps, HP’s first Colorado division opens in Loveland, Colorado to manufacture test and measurement equipment designed in California. At first, the Loveland Division gets the company’s voltmeters and power supplies to manufacture. Signal generators follow shortly and eventually, the Loveland Division takes over HP’s entire audio-video product line. Loveland will quickly develop its own R&D lab and in a few years, Loveland will start producing HP’s first desktop calculator, the HP 9100, which will be developed by Barney Oliver’s HP Labs.



The HP Loveland Division starts a small R&D lab in the north end of a rented Quonset hut at First and Lincoln in downtown Loveland under R&D manager Marco Negrete. During the summers, the Quonset hut is cooled by a lawn sprinkler carefully balanced at the hut’s apex. The south end of the Quonset hut houses HP Loveland’s transformer-manufacturing operation.

HP establishes another Colorado division in Colorado Springs to manufacture the company’s oscilloscopes. The Colorado Springs Division will eventually produce a line of digital sampling oscilloscopes (DSOs) that boost HP’s share of the scope market.



HP introduces the 3440A DVM, which is based on newfangled transistor technology and Nixie gas-discharge display tubes. Significantly, the 3440A has a digital output in the back so that the measurements can be logged by a printer. That same port can be interfaced to a computer to create a modern data-acquisition system. The 3440A’s transistors are noiseless, as opposed to the clacking, clattering stepper relays used in the DVMs made by Non Linear Systems. The HP DVM is also more reliable. HP will build and ship more than 10,000 HP 3440A digital voltmeters in only five years.

The HP 3440A digital voltmeter is both directly and indirectly related to HP’s desktop calculators and computers. Indirectly, the 3440A is an instrument that marks HP’s entry into the age of digital test equipment. Digital instrumentation paves the way for a boom in test automation, made possible by the desktop calculators and computers that HP will soon develop.

Directly, the electronics of the 3440A are designed by Chuck Near and Dave Cochran. Both of these engineers will later work on HP’s first programmable desktop calculator, the HP 9100A. It’s Near’s first project after graduating from engineering school. He joined the design project shortly after it begins in 1961.

After he completes the 3440A project, Near will work on every desktop calculator and computer project at HP until 1980. Cochran had previously designed the HP 204B, a transistorized audio oscillator. Cochran will develop the firmware for the HP 9100A. He takes a then-obscure algorithm used to calculate trigonometric and transcendental functions called the CORDIC algorithm and will write firmware for both the HP 9100 and its immediate successor, the HP 35 pocket scientific calculator.

Don Schulz is the section manager in charge of the HP 3440A in Palo Alto. He moves to Loveland and becomes the Loveland Division’s section manager for digital instruments. Schulz will eventually become the first division manager of the Loveland Instrument Division (LID) after the Loveland site splits into multiple operating divisions in 1970. He will also become the third division manager of the Calculator Products Division, following Tom Kelly and Bob Watson.



In the summer of ‘65, two people visit HP in Palo Alto to demonstrate two different calculators they’ve developed. The two are a physicist named Malcolm Macmillan and Tom Osborne. Macmillan’s fixed-point calculator, dubbed “Athena,” proves buggy and impractical but he brings knowledge of a new computational technique called the CORDIC algorithm that can quickly compute trigonometric and transcendental functions. Osborne’s calculator is housed in a balsa wood case and painted green (so he calls it “The Green Machine”). It’s only a 4-function calculator but its floating-point hardware design is brilliant and unlike any other computer design of the 1960s. Today, we’d say that the Green Machine incorporated a 64-bit VLIW (very long instruction word) microprogrammed processor. Back then, it just looked like the entire machine was a processor because the computational circuits were distributed throughout the machine.

Osborne’s radical approach hardware design and Macmillan’s CORDIC algorithms form the core of the HP 9100A desktop calculator. HP’s chief engineer, the legendary Barney Oliver, realizes he potential of Osborne’s design approach and sets up a calculator project in his lab. Osborne will work on the project but he doesn’t join HP as an employee. Tom Osborne never joined HP but served as a consultant to the company for many, many years. HP 3440A voltmeter engineers Chuck Near and Dave Cochran also join the HP 9100A project.

Although HP’s Cupertino Division will start making the HP 2116A minicomputer in 1966 and will be the only HP division at the time that is building computers, it wants nothing to do with the ugly duckling 9100A because it looks nothing like a “real” computer. However, Marco Negrete at HP’s Loveland Division is eager to expand beyond voltmeters and HPLoveland will adopt the new product when it moves into production. Chuck Near will move to Colorado with the HP 9100A.



HP creates a central R&D division, the renowned HP Labs, partly as a reaction to the growing geographic distribution of HP’s R&D efforts across its rapidly increasing number of divisions. Barney Oliver becomes the first head of HP Labs. The HP 9100A project accompanies Oliver to the Labs.

HP introduces the HP 2116A instrumentation minicomputer in November, 1966. It’s HP’s first product to be based on digital-IC technology. The HP 2116A’s 16-bit processor has a very simple instruction set and, more significantly, it has a bay for plug-in cards that interface to HP instruments and peripheral devices such as paper tape punches, typewriters, teletypes, and plotters. (Some will say that the HP 2116A architecture resembles a “stretch” PDP-8 (a 12-bit machine) from the Digital Equipment Corporation in Maynard, Massachusetts.)

HP is solidly an instrumentation company in 1966 so the HP 2116A is conceived of as an instrumentation computer to make it more palatable to Bill Hewlett. However, HP soon realizes that it’s selling a lot of HP 2116As as general-purpose minicomputers that will never talk to an instrument. Consequently, HP develops the HP 2115A and 2114A reduced-cost minicomputers to sell in its growing data-processing markets.

The 16-bit instruction set developed for these minicomputers will become extremely important to HP’s desktop calculator and computer business, in about five years.



HP introduces the programmable HP 9100A desktop calculator. It combines a keyboard (for user input), a 3-line CRT display (for user output), and a magnetic card (for mass storage) into one desktop-sized package. It’s ready to go almost the instant you switch it on. An accessory printer bolts onto the top of the machine so as not to increase its footprint on the desktop. This is a revolutionary product when compared to the box-and-terminal design of minicomputer systems, which must be booted with paper tape.

One of the HP 9100A’s key components (out of many unique parts required by the HP 9100 design) is the ROM (read-only memory) that drives the machine’s 64-bit processor. The ROM must have 64 outputs and must store 32,768 bits. In 1968, this sort of storage is about 1000x beyond the abilities of the primitive integrated circuits available at the time so the HP engineers build the ROM out of a large, complex, multilayer circuit board. Storage-space requirements for the program and variables are also beyond the then state-of-the-art for integrated circuits so the HP 9100A uses magnetic core memory popular in mainframe and minicomputers of the day.

The manufacturing requirements for the ROM are almost beyond the abilities of HP’s circuit-board manufacturing department, almost. What’s required is a 16-layer circuit board with special Teflon insulation between the layers. Because the ROM-board design pushes the limits of state-of-the-art circuit-board manufacturing technology, HP manufacturing engineers will crinkle their noses in disgust for years at its mere mention. Chuck Near designs the prototype ROM circuit board (later made production-ready by Tom Osborne). Near also develops the automated test procedures for the complex ROM board. He will use the new HP 2116B minicomputer as the foundation for an automated tester for his ROM board and for the other HP 9100A boards.

The HP 9100A meets with massive acceptance in the engineering and scientific communities and HP suddenly has a tiger by the tail. As a follow-on to the HP 9100A, HP adds more core memory, introduces the 9100B. It then ponders the future.

Robert Noyce and Gordon Moore leave Fairchild Semiconductor to found Intel with Andrew Grove. Intel will initially focus on developing semiconductor memories with a stated goal of replacing core memory in computers.



HP Loveland spins out its successful desktop calculator product line and creates the Calculator Products Division (CPD), which shares the Loveland facility (no, they never made HP’s handheld calculators here).

CPD’s mission is to develop successors to the HP 9100A desktop calculator. However, the HP 9100A’s 64-bit VLIW processor architecture proves problematic and will not scale up with the rudimentary semiconductor technology of the day. R&D engineers in Loveland ponder their processor alternatives and elect to adopt the HP 2114A minicomputer’s instruction set because it need not be reinvented and because the HP 2114A minicomputer and it’s software-development tools can serve as a software-development platform for the project.

Chuck Near develops a way to serialize the 2114A processor. This design modification allows the HP Loveland engineers to fit the minicomputer processor on several small circuit boards with relatively few interconnect lines between the boards, which keeps the desktop computer small and allows the calculator to actually fit on a desktop.

The serialized minicomputer processor is destined for three (eventually four) desktop computers: initially the HP 9810A, 9820A, and 9830A. (The HP 9821A will follow later.) The 9810A will use the numerically-oriented, keystroke-programming language developed for the HP 9100A. The HP 9820A will use a more advanced algebraic programming language with alphanumeric capabilities but will still map functions to individual keys. The HP 9830A will implement a relatively new programming language called Basic. All of the machines will use new solid-state LED alphanumeric displays internally developed by HP.

These three machines will greatly increase HP’s lead in the desktop computing market. In the same year, HP introduces the Hewlett-Packard Instrument Bus (HPIB), which standardizes the connection between test equipment and computers and further cements HP’s lead in test and measurement automation.

HP introduces the HP 9810A first, in 1971.



HP introduces the HP 9820A and 9830A desktop calculators.

HP also introduces the world’s first handheld scientific calculator in 1972. It’s called the HP 35 because it has 35 keys and it employs the same CORDIC algorithm used in the HP 9100A desktop calculator.



HP’s 9810A, 9820A, and 9830A desktop calculators are remarkably successful, albeit somewhat slower than the massively parallel, 64-bit HP 9100A. To speed things up, HP decides to create a 16-bit microprocessor, still based on the HP 2114A minicomputer processor’s instruction set, using an “advanced” NMOS semiconductor process developed at HP’s Loveland Division.

HP Loveland had first started to develop a home-grown integrated circuit process technology for the 9810A, 9820A, and 9830A desktop calculators. At the time, it appeared that no “real” semiconductor vendor would be able to supply HP with the 4-Kbit ROMs and 1-Kbit RAMs needed for the 9810/20/30 desktop machines’ operating systems. So, in true HP fashion, the Loveland Division decided to start making NMOS integrated circuits, from scratch. HP did end up making its own NMOS ROMs for he HP 9810, 9820, and 9830 but a startup company named Intel was eventually able to deliver the needed PMOS RAMs, with some process-technology help from HP’s engineers (who didn’t know any better than to help Intel for free). HP bought bunches of Intel’s RAMs for its second-generation desktop computers and helped to make fledgeling Intel a raging success.

The success of its first home-grown NMOS ICs spurred HP Loveland to develop an NMOS II semiconductor manufacturing process as the foundation for the 16-bit microprocessor. But even HP’s advanced NMOS II process with 7-micron lithography couldn’t fit an entire 16-bit processor on one chip, so HP Loveland engineers designed three NMOS processor chips and combined them on a ceramic hybrid substrate. The 3-chip hybrid processor (which also included four small bipolar buffer chips) served as the processor for an entirely new line of desktop computers: the HP 9825A and B; the HP 9831A; the HP 9835A and B; and the HP 9845A, B, and C. With these machines, HP completely conquered and owned the desktop computer and automated test equipment markets for five years, until IBM introduced the IBM PC and turns the entire computing world (including HP’s) upside down.

The HP 9825A first appears in 1976. It becomes the most successful desktop computer ever developed by the various incarnations of the HP Colorado desktop calculator/computer division.



The HP 9845A debuts. It’s almost the first HP desktop with a full alphanumeric CRT display instead of an LED display. The HP 9100A also had a CRT, but it only displayed the numeric contents of the calculator’s three internal registers. The HP 9845A display shows 25 lines of 80 characters each and has a monochrome graphics capability as well. The HP 9845A architecture uses separate language and I/O microprocessors, both based on the same processor hybrid that powers the HP 9825A.

HP’s Calculator Products Division moves from Loveland to nearby Fort Collins, Colorado and becomes the Desktop Computer Division.



HP introduces the single-processor HP 9835A and B, reduced-cost versions of the dual-processor 9845A. One machine looks like an HP 9825A with an HP 9845A CRT grafted on top. The other machine looks exactly like an HP 9825A except for the nameplate.



HP rolls out the HP9825B with triple the RAM capacity. The company also introduces the HP 9845C with a color-CRT upgrade to the HP 9845 desktop computer.



HP leapfrogs the competition again with the development of its NMOS III process and the 18-MHz, single-chip, 32-bit Focus processor. (Motorola’s 16.67-MHz MC68020 will appear in 1984 and Intel’s 16-MHz 80386 won’t debut until 1986.) Focus directly supports multiprocessing primitives in hardware and becomes the core of the first desktop machine to successfully run Unix—in the form of HP’s proprietary HP-UX operating system—and the first computer to support Unix running on multiple symmetric processors.

The Focus-based desktop computer was called the HP 9020 (code named Dawn and later officially renamed the HP 9000 model 520). The HP 9020 was the technical pinnacle of HP’s desktop computer dynasty and reflected everything that HP had learned about desktop computing in the ensuing 16 years since first starting the 9100 project. Nothing could touch Dawn’s performance for several years. An HP 9020 was used in the US Navy’s TAC (Tactical Advanced Computer) program and served on US nuclear submarines well into the 1990s. Back in the mid 1980s when it was first incorporated into the TAC program, the HP 9020 was far and away the most powerful computer that could fit through a submarine hatch. It was built like a brick, like all HP equipment back then.

However, the introduction of the IBM PC in 1981 (with a wimpy 4.77-MHz, 8/16-bit Intel 8088 processor) and rapid evolution from the 8088 to the 80286, 80386, 80486, and Pentium microprocessors quickly put a stop to the home-grown microprocessor technology that marked HP’s desktop computer designs.

Prices for HP’s desktop machines climbed from about $5000 in 1968 to nearly $50,000 in 1981. A nicely configured IBM PC cost about $5000 in 1981. Like a big part of the computing world, HP became a rabid adherent of RISC processor design and HP Labs will create the PA-RISC 32-bit processor architecture, which inherited the HP 9020’s HP-UX operating system. (The Labs giveth and the Labs taketh away.) HP’s desktop computer business started morphing into the workstation business. HP girded its loins, and the company set off to engage in the workstation wars. But that’s another timeline and someone else’s Web site. HP’s desktop computer story ends here.


21st Century

Except that 30 years later, well into the 21st century, NASA, DOD contractors, and calibration labs are still using HP 9825 desktop computers because the machines continue to work well and because no one wants to port the HPL software from these machines to newer computers. Crisis Computer in San Jose, CA continues to support the machine.




HP9825.COM is in no way associated with the Hewlett-Packard Company.

Hewlett-Packard, HP, the HP logos, the HP 9825, and all other product model numbers starting
with HP are all trademarks of the Hewlett-Packard Company and don’t let nobody tell you different!

This site exists strictly for educational purposes. Nothing is for sale here. Substantial material from various issues
of the incomparable Hewlett-Packard Journal appears on this site, by permission of the Hewlett-Packard Company.

All text Copyright 2004 to 2010 - Steve Leibson

[Home] [Prehistoric Times] [The 9100 Project] [Prologues] [The 9825 Projects] [Epilogues]