jerry@pioneer.arpa (Jerry Scharf) (05/21/88)
Never letting a little knowledge stop me, I'll pass on what I know about laser printer resolutions. If some of the actual interface designers would like to comment, that would be even better (Chen, are you out there.) As with everything else, different laser printers work differently. The two types I am familiar with are the Canon CX type, and the Xerox XP10 type. On the CX, the resolution in the paper motion (or slow) direction is set by the engine in the form of a signal for start of scan. The mirror is spinning at a fixed rate, and the projection is compensated by a cylindrical lens to make all the pixels the same width. The exact resolution in the scan (or fast) direction is controlled by the rate at which the bits are clocked out to the printer by the controller. This rate is some form of divided down crystal clock. It turns out that Imagen couldn't buy (I don't think they make) a crystal that would divide down to give an exact 300 dpi in the scan rate. My memory of the laserwriter was that they used the same crystal for both cpu clock and video output clocking, so they also have small resolution problems. The bottom line on CXs that that they are approximately 300x300 dpi. Once you factor in the gear wear and paper slippage in the paper direction, again they can only be approximate. The XP10 type systems provide clocks for both pixel and scan line timing, so I don't know the exact accuracy details, but I would assume them to be the same because of the constraints being similar. As for tuning up the accuracy of the printers, the crystal set speed makes it difficult to do anything about it for the CX scan direction. My guess is that it is as hard or harder to make it accurate in the paper direction. Remember that in many cases, you are dealing with a marking engine that probably very few dollars to construct (pre yen inflation), so the final exactness of the resolution is not that of your Linotype or Hell typesetter. Jerry Scharf NASA-Ames
keithl@vice.TEK.COM (Keith Lofstrom) (05/23/88)
> ... This rate is some form of divided down > crystal clock. It turns out that Imagen couldn't buy (I don't think > they make) a crystal that would divide down to give an exact 300 dpi > in the scan rate. This may be more relevant to sci.electronics, but over the frequency range that oscillator crystals can be cut, you can purchase small lots (or large) of crystals at any frequency you like. I once had a couple made for a test fixture; the pair cost less than $50, setup, special handling, and all. They just dial a different number into the crystal manufacturing machinery. I was amazed. That was 10 years ago, though; You may have to learn a Asian language to order special crystals these days. Of course, if you want crystals REALLY REALLY cheap, you buy frequencies like the color TV subcarrier frequency, or some other high-usage frequency; you probably wouldn't save the cost of a non-binary clock divider circuit. (Uh - what did that have to do with Postscript? Oh well...) -- Keith Lofstrom ...!tektronix!vice!keithl keithl@vice.TEK.COM MS 59-316, Tektronix, PO 500, Beaverton OR 97077 (503)-627-4052