[Arm-netbook] RGB/TTL interface selection, why in EOMA-68

[luke.leighton]

On 11/29/12, jm <joem at martindale-electric.co.uk> wrote:
> On Thu, 2012-11-29 at 11:36 +0000, luke.leighton wrote:
>> On 11/29/12, Henrik Nordström <henrik at henriknordstrom.net> wrote:
>> > tor 2012-11-29 klockan 09:06 +0000 skrev luke.leighton:
>> >> On 11/29/12, Tom Cubie <mr.hipboi at gmail.com> wrote:
>> >>
>> >> > The lcd is from BOE. The spec is here:
>> >> > http://tom.linux-sunxi.org/hardware/datasheet/HV070WSA-100%20product%20spec%20rev.0%20%5b%e6%9c%80%e6%96%b0%e7%89%88%5d.pdf
>> >>
>> >>  superb!
>> >
>> > I wonder... maybe EOMA68 should already be revised to use LVDS instead?
>> > Feels kind of stupid to have to do TTL->LVDS convertion when the CPU is
>> > perfectly capable of delivering LVDS with just the flip of some software
>> > bits.
>> >
>> > I kind of doubt there will be need for a CPU module based on a CPU that
>> > can not do LVDS.
>> >
>> > But you know these things better than me.
>>
>>  not necessarily!   but i have been over this, and nobody came up with
>> an objection, the logic went as follows:
>>
>> * EOMA_68 covers from as low as 320x240 all the way up to e.g. 2048x2048.
>> * therefore whatever interface is chosen, it must NOT cause products
>> which have 320x240 or even lower resolution LCDs to have to add 10%
>> onto the BOM just for an LCD converter IC (whatever it is)
>> * $1 for a converter IC when the BOM is $80 is however considered
>> "acceptable"
>> * therefore 24-pin RGB/TTL is pretty much the only
>> lowest-common-denominator, lowest-cost option.
>>
>> if you make dual-channel LVDS part of the standard as an "option", you
>> just screwed absolutely every module provider whose SoCs do not have
>> the capability to put both 24-pin RGB/TTL *and* dual-channel LVDS onto
>> the same pins.
>>
>> so, ridiculous as it seems initially, 24-pin RGB/TTL is the best
>> option that i can see, when taking into account all the factors that
>> i've been able to consider *so far*.
>>
>> if anyone can come up with better reasoning it'd best be done quick
>> because the A10 module is almost ready.
>
> http://en.wikipedia.org/wiki/Low-voltage_differential_signaling
> LVDS is differential signalling.
> Its less noisy and less noise prone than parallel voltage buses.
> Longer cables possible with LVDS.
> My bet is that the future is likely to be LVDS and much lower power than
> parallel buses.

 in the case of products with larger screens, we need to get the
RGB/TTL signals only off the card and into the LVDS IC (dual or
single).  it's not far.

l.