HP (Hewlett-Packard) 1200c User Manual

       February 1994 Hewlett-Packard Journal

stepper at slow speeds. For the first time we could measure
pinion quality after the gears had been mounted on the motor
shaft. With this measurement ability, we were able to work
with the vendor and resolve tricky problems such as damag-
ing the pinions and bending the motor shafts while mounting
the gears on the shafts. The resultant print quality of the HP
DeskJet 1200C printer is good enough that few swath ad-
vance accuracy problems have occurred in production.

The media advance accuracy is an important metric of the
performance of the media drive system. As media moves
through the printing region it is stopped, printed on, and
then advanced to the next printing location. The distance it
moves each time depends upon the print mode and what is
being printed. Print quality for both text and graphics de-
pends on advance accuracy. However, graphics print quality
is more dependent on advance accuracy because there is no
blank space in which to hide the advance error. Every ad-
vance made during graphics printing has the potential to
show a print quality error, whereas during text printing it is
possible to avoid splitting text between advances and
thereby avoid showing advance error.

In the early stages of development the design team needed
an estimate of the drive accuracy of the printer. A mathe-
matical model was constructed that simulates an advance
made by the worst-case components in the worst possible
orientation. This model provided a good guide to what sort
of tolerances we needed in our manufacturing processes, but
was too conservative to simulate what we expected most
advances to look like. The next model built was a Monte
Carlo model that chooses components from simulated dis-
tributions and orients the parts randomly, much like a manu-
facturing process. When this model is run for a large number
of cases, a good approximation of the expected mean and
standard deviation of an advance is produced. Once the
mean and standard deviation are known, process control of
advances is possible. The goal of the modeling effort was to
simulate advances and establish tolerance limits on advance
accuracy. This goal was achieved and the process limits
were confirmed by measuring drive components, building
machines, and measuring their swath advances.

The DeskJet 1200C project was able to use tools previously
developed for media advance measurement for other print-
ers. The best tool we found is an optical vision system which,
when given a specific plot, measures a series of advances
and reports the data. For our purposes, predominantly
graphics print quality, measuring 32 nozzle advances pro-
vides all the information we need about the drive system.
The 32-nozzle advance data can easily be extrapolated to the
other advance distances we are interested in.

Fig. 7 shows media advance measurements down the length
of a page. This data was processed to find the mean and

Fig. 7. The measured media advance down the length of one page
of a single machine is shown along with the predicted accuracy cal-
culated with measurements of the drive components as inputs to the
model, which is shown as a distribution. The actual output of the
model is the mean and standard deviation for a given machine.

s

s

standard deviation, which were compared with the model.
First, components were measured and used to build a set of
machines. Then the swath advance of the machines built
with these components was measured. Finally, the statistics
of the measured swath advance were compared with the
results from the model with the measured components as
inputs. Fig. 7 also shows the results of the simulation: the
distribution of swath advances predicted by the model. As
can be seen, the measured advance fits within the predicted
distribution and the mean and standard deviation of the pre-
dicted distribution agree well with the measured mean and
standard deviation. Thus, we confirmed the accuracy of our
fabrication processes, the simulation model, and the printers
we manufacture.

As with any large development effort, there are innumerable
people without whose contributions this product would not
exist or function as smoothly as it does. We would like to
thank Gerold Firl, Mike Green, Joe Milkovits, Tim Zantow,
and Tom Halpenny for their contributions to the media path.
We would like to thank Ron Kaplan for his valuable assis-
tance in the early characterization and tooling development,
Frank Nasworthy for his expertise in designing the pinch
wheel, Steve Witte for his assistance with the hardware con-
nectivity and programming skills, Damon Broder for his
overall support of the paper advance system, Jeff Wilson
and Dennis Culver for their design and fabricating assis-
tance in tooling design, and our staff of lab technicians—Tau
Ngo, Nhuong Nguyen, Robert Schmidt, Damon Johnson, and
Danny Zepeda—for their many long hours of testing. Thanks
also to Steve Gaddi for his support of paper advance testing.

HP-UX is based on and is compatible with UNIX System Laboratories’ UNIX* operating system.
It also complies with X/Open’s* XPG3, POSIX 1003.1 and SVID2 interface specifications.

UNIX is a registered trademark of UNIX System Laboratories Inc. in the U.S.A. and other countries.

X/Open is a trademark of X/Open Company Limited in the UK and other countries.