Managing the development of the HP DeskJet printer - Hewlett-Packard - contains related article on market research as a design tool, and one on human factors and industrial design of the DeskJet printer - technical

Managing the Development of the HP DeskJet Printer

THE CREATION OF A HIGH-TECHNOLOGY PRODUCT is an enterprise that requires the contributions and skills of many people. The articles on the HP DeskJet printer in this issue mainly explore the technical engineering problem solving that is essential to new-product development. But there is a bigger picture. There is much more to engineering than solving equations and setting up experiments.

In the case of the DeskJet printer, our organization and planning encompassed all of the functional departments of HP's Vancouver Division. Our development and management teams included members from R&D, manufacturing, marketing, and quality assurance, with special assistance from personnel and finance.

The core development teams, which worked on the product from its inception, consisted of about 25 engineers, split into three project teams--firmware, electronics, and mechanical--with a project manager leading each team. The three core project managers reported to a laboratory section manager who served to coordinate the entire project.

Within the lab, midway through the development process, additional teams of two to five engineers were formed. These teams developed character fonts, emulation software, and application drivers, and performed extensive verification and performance testing.

During the first year of development, the core management team's role was directed towards technical guidance, resource organization, planning, and progress tracking. In the second year of development, the management team's emphasis shifted to coordination and prioritization as the circle of people involved in the program grew larger, and as the date for product introduction grew closer.

Our project started with a loose collection of specifications--a list describing what features we felt customers needed--tempered by what we felt we could technically achieve given our time, people, and financial resource levels. In the broadest sense, the goal of the project teams was to transform that list into an engineering specification.

DeskJet Printer Features

The HP DeskJet printer, Fig. 1, is a personal-convenience printer that produces laser-quality output at a price comparable to other low-cost personal printers. Among its features are 300-dot-per-inch resolution, merged text and graphics, multiple fonts, two slots for font or personality cartridges, 120-character-per-second letter-quality speed, built-in cut-sheet feeder for common office paper, desktop design, and quiet operation.

The DeskJet printer comes with Centronics parallel and RS-232-D interfaces. It is supported on HP Vectra, Portable, and Touchscreen personal computers, HP terminals, the Apple II series, IBM PC/XT, PC/AT, and PS/2 computers, and compatibles. It is also supported on HP 3000, HP 1000 A Series, HP 9000, and HP 260 systems.

Many applications software packages, such as spreadsheet and word processing programs, support the DeskJet printer. For other packages, the HP LaserJet Series II printer driver will work well because the DeskJet printer uses the HP PCL (Printer Command Language) Level III command set. An Epson FX-80 driver will also work if used with the optional HP Epson DeskJet personality cartridge, which fits into one of the DeskJet printer's option slots.

Technical Challenges

The development path is never smooth or level. There are steep hills and traverses across unexplored territory. The technical challenges on the DeskJet printer project were many:

* Extend the 96-dpi inkjet printing technology to 300 dpi, keeping the printing speed above 120 cps

* Ensure that the operation of the ink delivery system is totally transparent to the user, eliminating the messy image that inkjet printing had inherited from its early days

* Make this printing technology available on all standard office papers, eliminating the dependency of inkjet printing on special papers

* Provide these product features in a small-footprint package that does not dominate the desk on which it sits

* Make this product of traditional HP quality, with reliability unrivaled by any other printer

* Accomplish all of this within a tight development schedule of 22 months with a design that can be built, distributed, and profitably sold for a low target price.

Design for Reliability

The hallmark of a successful project is careful risk management, for tough or recalcitrant problems require intensive resources to solve. If the development team is considered as a problem-solving engine, then that engine has a specific capacity, and for a given complement of engineers, there is a limit to the number and difficulty of technical problems the engine can solve in a given time period.

With follow-on products, the design task is that of interpolating from a well-understood basis, peaking performance, adjusting features, or reducing costs.

In breakthrough products, the design task must include solutions that use unfamiliar technologies. It is these forays into unexplored regions that must be carefully limited to essential development, since there is little experience to guide progress or to gauge the potential difficulties. In other words, the design teams must carefully choose which problems they are going to solve.

An example will help to illustrate this point. Early in the DeskJet development project, the mechanical design team elected to use filled thermoset plastic for the major structural part (the chassis). This decision was based on experience with similar structural plastic parts in several HP Divisions. In fact, the material set chosen for the structure and gears was identical to that successfully used in the PaintJet printer. The DeskJet team sought to reduce its design load by using an existing and well-understood technology. Or so we thought! The first prototype printers assembled from the molded plastic parts showed rapid deterioration of bearing materials with resultant squeaking, galling, and seizing--often within a few tens of pages. It turned out that we had exceeded a critical PV (pressure-velocity) point in the bearing loads. Until solutions were found (it took several intensive weeks), all design teams were hampered in their development by a lack of working prototype printers.

In our laboratory, team techniques are an important contributor to rapid progress and reliable design. Development tasks always have a principal designer and a subsidiary designer. The principal designer has part responsibility, while the subsidiary designer is a valued consultant. This designer pairing is based on interacting part/subsystem functions. Thus, a web of pairings exists, connecting the designer teams.

This pairing has several advantages over solitary design. The synergism of two (or more) designers working on the same problem is remarkable, and the quality and quantity of potential solutions is superior. In addition, solutions always have two committed designers to argue their merits with the rest of the design team. Furthermore, the principal designer has a backup in the event of illness, a trip, or reassignment to another design task.

We also encourage informal and frequent design reviews. Typically, these occur when the designers have a concept worked out that is supported by preliminary analysis. The designers meet with those who have expertise in the area (in mechanical designs usually including a procurement engineer and a manufacturing engineer), and walk through the design with their peers. It is important that these reviews occur early in the design so ideas and suggestions can be incorporated easily.

These design reviews occur close in time to the development of the first crude prototypes. Initially, the early mockups examine a subsystem or specific function, such as picking paper with a platen roller. Later refinements are incorporated in a product breadboard--a working printer that demonstrates all of the critical subsystem functions.

Testing the Design

Coincident with the emergence of the prototypes is testing. First the basic concept is examined to see if it addresses all of the design constraints. Next, normal operations are explored to find where the design is deficient. (It always is!) Finally, the design limits are probed through accelerated tests or abuse testing.

Throughout, the engineer repeatedly cycles through the test-analyze-fix process. Initial testing yields many easily discovered defects, which can be quickly resolved with engineering analysis. Subsequent testing is aimed at improving ruggedness and reliability; these test scenarios usually require many unit-hours of experience, and the conclusions must be reached by careful statistical inference.

Compounding the statistical problem is that of securing representative parts. Much of the initial testing is performed with parts from prototype tools or processes. The design limits are not well-understood at this point, and the parts are varying because the process is still unstable and undeveloped. Tolerance analysis of the design is useful, but not sufficient, since the called-for tolerances must be satisfied by a production process. Much effort goes on at this time to allocate the tolerances and allowances between parts and process.

The final phases of testing involve tests under controlled conditions by impartial quality assurance engineers. Here, testing to rigorous HP standards is completed, including temperature and humidity excursions, shock and vibration tests, and transportation and use/abuse tests that seek out the weak links in the design.

Life testing proceeds under accelerated and nonaccelerated conditions, probing the design for deterioration, wearout, and contamination.

In summary, reliable design requires more than theoretical design skills and analyses. Although good first-round designs are an essential foundation, the bulk of the engineer's efforts go into executing well-thought-out testing programs whose intent is to stress the design and uncover its limitations so that improvements in the subsystems and the integrated product can be made.

Acknowledgments

Thanks to Tom Braun, the DeskJet section manager, whose drive and insights guided the entire project, Bob McClung, under whose capable management the electronics team delivered their reliable, high-performance design without ever being on the critical path, and Mark DiVittorio, whose firmware team simultaneously satisfied the requirements of two printer projects. Special thanks to Susan Hoff, the DeskJet project coordinator, for her great skills in maintaining documentation order in the midst of development chaos, and to our model makers, who contributed valuable suggestions and worked many hours of overtime producing the mechanical prototypes of the printer. Special credit goes to all the members of the design team for their enthusiastic dedication to the development of the DeskJet printer and its predecessor projects. Finally, I want to acknowledge the outstanding contributions of Bill Buskirk and Niels Nielsen of our Corvallis Inkjet Components Operation, without whose professional and personal commitment we could never have effectively coupled the development of the printer and printhead.

COPYRIGHT 1988 Hewlett Packard Company
COPYRIGHT 2004 Gale Group