Oopsie hardware design
While I was technical editor of CADalyst (1985-1991), I reviewed an awful lot of hardware -- graphics boards, monitors, display-list software, pen and electrostatic plotters, dot-matrix plotters, laser printers, hardware and software plot cachers, hardware and software scanners -- because at the time CADalyst was at first the only game in town.
Those were remarkable years, in that I could have hands-on experience with hundreds of thousands of dollars of equipment related to CAD. (The most expensive was a $70,000 E-size super-fast Vidar scanner, that in today's dollars would be closer to $200,000.)
I developed testing methodologies to ensure every product in a class was tested identically, and established a rating system that ran from Not Recommended (almost never awarded) to Highly Recommended (rarely awarded); most products were Recommended, because they worked. I had been influenced by the technical editor at Stereo Review magazine, who was hard-core about consistent and repeatable testing that could be validated. This made some vendors unhappy at him, and at me.
The DraftMaster-line of pen plotters from Hewlett-Packard, for instance, usually received the High Recommended rating from me, because they were very, very good.
One of the hardware products to which I gave the dreaded Not Recommended label was a pen plotter from Houston Instruments. This company also produced contact lenses. Their plotters were considered mid-range, doing a reasonable job at a reasonable price. HI plotters I had tested in the past did a decent job.
Then a new model arrived with a bizarre design flaw. First, though, let me explain how rollerbed pen plotters worked. You manually afixed a sheet of paper (D- or E-size) on the plotter's full-width roller with roller clamps that typically were covered with soft rubber or grit for better gripping power. Remember the word "grit." This model boasted a full-width grid wheel.
When plotting the drawing, the pen head moved side to side, while the large roller pulled the paper back and forth. Diagonal lines were made by moving both at the same time, at varying speeds. The figure shows a typical HI rollerbed plotter of the late 1980s:
Sometimes, a roller would accidentally spit out the paper, due to an error in the movement codes sent by the CAD software to the plotter, or for some other reason.
As I tested the new arrival, it spat out the paper and continued plotting. Here I discovered the horrific design flaw: the pens were located over the grit wheel, so as the plotter continued its work, it methodically wore down the pen tips on the coarse grit wheel.
Other plotters, including those from HI, located the grit wheel elsewhere in the design. So I had to give this model a Not Recommended rating.