|On my PC110 (Small DSTN LCD screen) I find I *have* to change contrast |quite a LOT, temperature changes the contrast a lot. | I like the PTC Thermistor idea, these have an FN-key driven contrast |voltage, though (Might be able to dig in there & fix it.) | When I do CAD work on mine, it really changes where the contrast has |to be set a LOT, and as the screen warms up from suspend I have to |change it more to see the background grid. An annoyance, BUT, it IS |nice to be able to do CAD work on a palmtop while on a bus or waiting |for a ride - or to look up & modify a schematic while buying parts to |prototype a project with! All in all, I'm pretty addicted On multiplexed LED display, the controller can energize the rows in sequence and, when each row is energized, energize the columns whose lights should appear in that row. Simple, easy, and no need to worry about "bleeding" among the pixels on a column; starting at any row driver, there is only one possible current path to any particular column driver. On an LCD, things aren't so simple. Between any row and column, there are many signal paths: in addition to the direct one there are paths of the form [your row to some other column] [that column to some row] [that row to your column]. If care is not taken, the pixels along all the alternate paths will be darkened, and the dis- play will appear "smeared". To minimize these effects, it's necessary to use a technique called "display biasing". Essentially what happens is that inactive rows and columns are set to specific voltages to ensure that the excita- tion of unintended pixels is uniform. Unfortunately, even though the display background is uniform it still isn't transparent. The pixels there are partially switched on, and if the display drive voltage is very high they may become quite dark. Note that for most of the alphanumeric displays, the ratio of RMS drive voltage for the light and dark segments is the square root of: [note: 1/16 multiplex display with 1/5 bias] energized non-energized 15/16 * (1/5)^2 + 1/16 * (5/5)^2 :: 15/16 * (1/5)^2 + 1/16 * (3/5)^2 [multiply both sides by 400...] 15 * 1 + 1 * 25 :: 15 * 1 + 1 * 9 40 :: 24 i.e. sqrt(5/3)::1 or 1.3:1. Note that compared with that, most lap- top displays are almost ridiculously bad (though the LCD material is a bit better, so things aren't as bad as they sound...) [note: 1/240 multiplex display with 1/16 bias] energized non-energized 239/240*(1/16)^2 + 1/240*(16/16)^2 :: 239/240*(1/16)^2 + 1/240*(14/16)^2 [multiply both sides by 240 and 256...] 239 * 1 + 1 * 256 :: 239 * 1 + 1 * 196 495 :: 435 i.e. sqrt(99/87) or 1.06:1. Assuming polarization twist proportional to RMS voltage and transmission proportional to cos^2(twist), the contrast ratio will equal (cos(1.06*biastwist)/cos(biastwist))^2.