35

Ever since I first had my GameBoy Pocket, I've noticed that the line drawn across the screen at power off (which I'm guessing is because the CPU has stopped sending clock signals to the display driver, and it's still pumping a little power into the last row it was refreshing while a capacitor somewhere drains) is blue:

Power off line on the left, normal graphics on the right

The line on the left being a desaturated blue (#4f535c), while the normal graphics on the right are more yellow-tinted (#7c785b).

I'm curious to know how the display can produce this blue during shutdown when normally it can only produce yellow-tinted shades.

37

The gameboy has a "normally white" LCD. If the pixels are uncharged, the display is bright. The more you charge the pixels, the darker they get. Interestingly, this effect works with either polarity of charging voltage (just like you can attract a piece of iron with either the south or the north pole of a magnet). To avoid possible destruction of the LCD by electrolysis, the voltage on the pixels will be reversed each frame (60 times a second on a gameboy). So you normally see the appearance of pixels being black (charged) most of the time, but jittering around 60 times a second due to the reversal process.

When you turn off the gameboy, the final black line is charged to a DC voltage and never reversed. So the different appearance might be either to the fact that the display is now getting a DC voltage instead of an AC voltage, or it might just be related to a higher voltage level.

You could try if adjusting the contrast in a way that black gets blacker. This might also create the blue tint if it is just level related. If you can't reproduce that tint with the contrast pot, it is more likely caused by the charge being DC.

5
  • 1
    30 Hz seems like a very low drive frequency for an LCD. It would work, sure, but I would expect the driver to work at a higher frequency, uncoupled from the framerate. But maybe not! I can't say I've inspected the details of the game boy pocket in particular.
    – Hearth
    Aug 11 at 15:35
  • @Hearth 30 Hz is in the low end, but probably fine for the quite slow LCD in these devices. See e.g. PCF8551 which is configurable 32 to 128 Hz (that is for segmented displays, but the technology is similar).
    – jpa
    Aug 12 at 6:58
  • 2
    Segmented displays have a lot fewer elements than dot matrix displays, so higher drive rates are more easily implemented there. The programming model of the gameboy is (likely intentional) very much like 8-bit home computers driving a CRT: You don't have the fully rendered picture (with background/window/sprites merged) anywehere in memory, and you can "race the beam" to apply effects at certain spots. I can't imagine how you can drive the LCD at any other rate than the frame rate given these circumstances. Aug 12 at 9:33
  • @MichaelKarcher: LCD displays can be driven at any desired rate within a fairly broad range. Another convenient feature is that if one uses hardware timer to drive the "line advance" signal at a consistent rate, one can feed the data within a line at any desired rate, subject only to the requirement that one has to shift all of the data before the time expires and the display starts outputting the line that was just shifted.
    – supercat
    Aug 12 at 14:48
  • 4
    @supercat I'm not talking about constraints caused by the LCD, but about constraints rooted in the gameboy programming model. To drive a line, you need to know the pixel values for that line. The gameboy hardware only knows what to draw on a specific line 60 times a second (because software can and does move the sprites, change palette registers and changes scrolling registers on a line-by-line basis), so the scanout hardware can't drive that specific line any other time. The gameboy doesn't have a pixel-by-pixel framebuffer that can be scanned out at any rate. Aug 12 at 15:38
14

The first supertwist LCD displays had a definite blue color whose saturation varied with how hard the display was driven; some later ones use color compensation techniques to appear more grayish, but because the exact shade of blue is affected by how the displays are driven, such techniques aren't perfect. My father had a Macbook with a monochrome LCD which, probably because of a bad display driver connection, had a 64-pixel wide vertical stripe on the top half of the screen near the right edge of the screen which was bright yellow, even though the display was otherwise gray, since those parts of the display weren't being driven at all (for various technical reasons, it's impossible to prevent even blank pixels on a display from receiving a significant fraction of the drive voltage fed to energized pixels). In the scenario here, powering off the unit is likely causing the display multiplex circuitry (which would normally drive each line of the display about 1% of the time) to stop, leaving one line energized for much longer than normal. This would result in the pixels on that line being driven harder than usual. Since the color compensation is designed to white-balance the color of pixels that are driven "normal" amounts, the fact that the line is driven extra hard makes it more bluish than normal.

4
  • More precise than "around 1%": As the gameboy display timing consists of 154 scanlines, 144 of them being active scanlines, the lines are driven about 0.65% of the time. Aug 11 at 21:31
  • 1
    There was never any monochrome "MacBook". You may be referring to the "Macintosh Portable" or an early PowerBook.
    – nobody
    Aug 12 at 0:53
  • @nobody: Maybe a Powerbook. I think it had "book" in the name, but I know it wasn't the original portable (which could survive being run over by a truck), but it wasn't my machine so I didn't pay much attention to which model it was.
    – supercat
    Aug 12 at 14:43
  • @nobody: I'd normally like to have more positive knowledge when answering, but part of the question which I haven't seen anyone else address is why the line is blue rather than black, and I think the fact that my dad's portable had a bright yellow rectangle on an otherwise-monochrome screen is almost certainly relevant. On a related note, I wonder why the displays on Casio graphing calculators seem to be of a unique type which I haven't seen used anywhere else, since I would expect they're lower power than any other kind of LCD that's clearly visible in low ambient light.
    – supercat
    Aug 12 at 14:50
11

LCD display segments (pixels) are electrodes that are used to control the electric field applied to the liquid crystals that are between the electrodes. So in essence they are tiny capacitors, and to turn the segment on, a voltage is applied to electrodes to charge an electric field which quickly aligns the crystals to polarize the light so that it gets blocked in the polarizers so the segment looks black. To turn off the segment, the voltage between the electrodes is set to zero to discharge the electric field, which allows the liquid crystals to quickly unalign to remove light polarisation so the segment looks transparent again.

When the power is turned off, at some point the active display driving stops and the operation of the display driver might be undetermined due to the decaying supply voltage. It is possible that it allows the residual voltages to be applied to all segments of the column briefly or constantly to charge up the electric fields. Either way, the driver clearly does not discharge the fields to turn the segments actively off.

Therefore the segments remain aligned and will block the light, until the electrode voltage and thus the electric field starts to decay and the segment will slowly turn off. When only partial amount of the crystals are polarizing and blocking the light, the hue can be different. It is basically like looking at the display while slowly turning the contrast setting to zero.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.