Rockbox mail archive
Subject: Re: Survey: how fast is your display?
From: Eric Linenberg (elinenbe_at_umich.edu)
As I move to higher values the stripes move to the right and up the
screen and as I move lower they move to the left and down.. after
about 1 minute of being on they are stable at 68.111 FPS and about 2
minutes later it is stable at 67.792 FPS. If you need any more help I
ID> I'd like to make a little survey on the variation of the LCDs.
ID> The background is that for halftoning it's desirable to "flicker" the pixels
ID> with the same speed as the internal LCD scan. Only then every flicker phase
ID> is actually displayed, important for the appearance. If we're too slow, a
ID> part of the screen is shown twice, giving that temporal value additional,
ID> unwanted "weight". If we're too fast, a part gets not shown at all, therefore
ID> losing it.
ID> The grayscales look perfect and stable on a nice match, grainy on mismatch.
ID> Unfortunately the LCD is internally clocked by an RC oscillator, those are
ID> far less stable and reproduceable than a crystal, e.g. depend on temperature.
ID> And there is no vsync, scan position readout or other means of feedback. So
ID> all we can do is blindly do our best shot. I'd like to find out what a good
ID> average value would be.
ID> I made a diagnostic plugin to find out your value, see attached. It should
ID> be good for Recorder (2MB), V2 and FM, requiring you to run a build not older
ID> than 27/01 (when I optimized the LCD code again). Source code is included, if
ID> you build your own. It tells the LCD rate by making a pattern which
ID> interferes with it.
ID> To use it, simply play it. It shows 2 alternating frames, a dark and a light
ID> one. The alternation is done by a precise timer which you can tune now, with
ID> up and down keys. The goal is to get the screen free of flicker, with only a
ID> slow-moving residual artifact remaining that looks a bit like "a school of
ID> fish", small black horizontal stripes that chase each other. The rest of the
ID> screen may look "cloudy" in varying shades, but is stationary. The pattern may
ID> move off-screen for a while, so you may only see the background.
ID> I tried to take a picture of it, but it didn't came out well. When tuning,
ID> if you see a storm of diagonal patterns, you're way off. If the patterns slow
ID> down, you're getting closer. If the patterns move up, you're too fast,
ID> compensate by going down. Vice versa, increase if they move down.
ID> My display does about 67 Hz when freshly started. It decreases as it gets
ID> warmer (charging and prolonged disk activity), can get as low as 61 Hz. How
ID> does yours?
ID> Thanks for participating,
Page was last modified "Jan 10 2012" The Rockbox Crew