4 ms·
As I understand it, the poor refresh is because each pixel element doesn't store its state in silicon; the physical orientation/charge/whatever of the display e
by cma 16y ago
As I understand it, the poor refresh is because each pixel element doesn't store its state in silicon; the physical orientation/charge/whatever of the display element itself is the state. So you have to wait on each pixel to change before moving on to another (you can of course do multiple chunks in parallel, but your bus is only so wide). In an active-matrix lcd you write to a memory cell that lives close to the liquid crystal, and that cell continuously drives the crystal to twist the desired amount. You can rapidly write to those memory cells without waiting on the physical twist of the crystals to complete.
You can see on the kindle the loading indicator is able to spin at a decent refresh rate; ditto word selection, menu selection, etc. This is because you are waiting on few elements to make the physical change at a time.
A simple pen based drawing surface would only update a few pixels at a time would and presumably work quickly as well.
(Any corrections welcome; this is just sort of how I assume it works based on using a kindle, not based on much looking into of how eink actually works)
- aamar 16y agoThe response time for individual pixels is quite slow, around 250ms. It takes that long to apply the necessary sequence of voltages to get a clear image without visible ghosting. That slowness is an obstacle even if you had no bus limitations. Something like the loading indicator might be moving faster, but ghosting is acceptable for something like that. I wrote low-level software for E Ink displays years ago, when the practical response time was more like 400ms IIRC. At the time, drawing lagged too much to be pleasant for a consumer device. Maybe now, if you do something clever with a quick 1st pass and a slow 2nd, darkening pass, you could get a reasonable result... Maybe.