[Arm-netbook] Questions about display resolution

[luke.leighton]

On Fri, Oct 12, 2012 at 7:42 PM, Derek LaHousse <dlahouss at mtu.edu> wrote:
> Arm-netbooks,
>
> So, I've been thinking on the issue of a laptop using EOMA-68, as well as the
> mini-eng/monster-eng boards.  Naturally, these things will want to run a
> display.  According to the EOMA-68 spec, we've got 8-bit color on a ttl
> interface.  This interface is expected to be converted to LVDS to run a screen,
> or DVI for an eng-board.
>
> TI says displays typically use 1 channel LVDS (that's 4D+C for 24-bit) up to
> 1400x1050, and 2 channel LVDS up to 1920x1200 [1].  TI says 70-75 MHz pixel
> clocks are common.
>
> Some math:  1 channel at 75MHz / 1400 / 1050 = 51 Hz.  This seems reasonable for
> a display (1400x1050 at 50Hz).  2 channels at 75MHz / 1920 / 1200 = 65Hz.  Okay.
> But to run a 1920x1200 at 60Hz display on one channel (Like the TF701), we need the
> RGB-TTL to have a rate of 138MHz.  TI's fastest chip tops out at 135MHz.  Oops?

 many LCD panels can actually go down well below their "rated" refresh
rate - 40hz is not uncommon.

> So, a few questions:
> - What's the fastest rate the A10 can output RGB-TTL?
> - What screens take a single channel LVDS at more than 1400x1050?  What're their
> refresh rates?

 none that i know of.  above 1440x900 they start to only do dual LVDS.

> TI does have a chip that can take a single RGB-TTL interface and output to two
> LVDS channels.

 yes.  i don't know if you saw on irc yesterday, you'd gone "away"
just as i responded about looking at the datasheet: it's possible to
set that IC you found into "dual" mode by setting the voltage to 1/2
of Vcc.

 when you do that, it reads only *one* of its 24-pin RGB/TTL
interfaces and outputs odd/even pixels onto alternate LVDS channels.

> I thought this was only good for pixel doubling.  Is it
> reasonable to expect a single RGB-TTL to run at > 100 MHz in other chips?

 i don't honestly know.  we'll find out, eh? :)  but i'm going to
require that the Cardbus (shielded) interface be used.  i was hoping
not to have to do that, but.... the speeds are a bit pushing our luck.

 the balanced-line pairs it should be fine.

l.