Our test suite enables us to evaluate a printer's performance (i.e., speed), quality of printed output (with separate tests for monochrome and color), usability, and features. We test and score each of these elements separately, and then we compute the overall scores by assigning a weight to each element. We assign weights as follows: 45 percent for performance, 30 percent for quality, 15 percent for features, and 10 percent for usability.
We used a Compaq Deskpro 66M loaded with MS-DOS 6.2 and Windows 3.1 for the PC platform. We connected printers that arrived with Macintosh ports to a Quadra 640AV running System 7.1. We used the printer drivers that vendors either supplied or recommended, and we disabled print servers, spoolers, and buffers during performance testing.
PERFORMANCE
NSTL's performance tests measure the
speed at which a printer can produce six different elements: dense text, sparse text, bit-mapped images, monochrome graphics, color graphics, and a minimum of two fonts. We test in three different ways for each platform: in the lowest and highest PCL resolutions (except for the Mac platform) and in the highest-resolution PostScript that each printer supports. The three platforms we tested this month were DOS/Windows, Macintosh, and a Novell NetWare 3.12 Ethernet environment.
We do a minimum of two test runs; if the results from two runs are not consistent, we perform a third run. If questions arise, we work with the manufacturer-designated technical-support person until we're confident the printer is set up correctly and is operating normally.
Text tests.
The dense-text test requires printing a 2-KB file of ASCII text with little white space. Performance in this test correlates to raw speed, since there are no fonts or graphics for the printer to interpret. A second test documen
t, designed to simulate the printing of invoices and forms, consists of ASCII text blocks broken up by interspersed areas of white space.
Graphics tests.
The graphics tests use bit-mapped images to simulate documents with custom fonts and screen shots. These tests help us determine how efficiently each printer communicates with a computer. One test measures a printer's ability to draw complex lines and filled areas. A second test concentrates on producing curves and gray scales.
Font test.
We use the font test to measure the speed of a printer's processor. The test requires printers to create Times (serif) and Helvetica (sans serif) fonts in regular and boldface in 30 different point sizes.
QUALITY, USABILITY, AND FEATURES
We judge the printers for monochrome-output quality on a series of pages that include a photographic image, text in a wide range of sizes, and a series of horizontal and vertical lines. For color printers, we add t
wo additional pages: We judge one using objective criteria, such as the output's range of hues and the absence of smearing on fine lines. We judge the second page, which contains text and complex graphics, subjectively.
This month we've introduced tests that let us objectively measure the consistency and range of printed colors (see "Measuring Color"). We used these new color tests only on the four color laser printers.
We compute usability scores based on such aspects as the ease of installing the ink media, the intuitiveness of the control panel, the ease of driver installation, and the clarity and usefulness of the provided manuals.
To arrive at the features score, we gather extensive information from vendors using a features questionnaire, verify these responses, and then compute a score using our weighting system. We assign weights for individual features based on their importance and their utility to the targeted market.
Contributors
Michele Guy, Project Manager/NSTL,
led the testing for this project. She has conducted hardware and software testing for NSTL on a variety of products over the past four years.
John McDonough, Technical Writer/NSTL,
has covered the computer industry for eight years.
Maggi Bender is a Technical Analyst for NSTL.
Tom McAndrew is a Technical Analyst for NSTL.
How We Tested
