On Sun, May 15, 2005 at 09:41:42AM -0400, Timothy Miller wrote:
> We intend to build DACs into the ASIC, and we'll use a DVI-I
> interface, requiring an adaptor for VGA.
>
> Note that DACs built into ASICs can't have the electrical isolation
> that we like to have, so we may have to produce a separate card with
> an external DAC for high-end analog.
Sounds like this is one place where we get into extra-cost BOM
options. I can think of a couple possibilities:
1. Optionally install fast video amps on-board, configured with a gain of
2, so they can be back-matched to absorb minor cable reflections. This
is the cheapest way to get clean-edged pixels on an analog monitor. If this
option isn't ordered, zero-ohm resistors would be stuffed instead.
There are plenty of cheap, fast op amps these days, so my first thought
would be to include them by default, but I suppose we have to shave every
nickel to compete in the consumer category. It's still worth paying close
attention to impedance match and connector discontinuities in the PCB
artwork, because that has zero per-unit cost.
2. Offer an outboard accessory as you said, pure digital in, and the best
state-of-the-art analog out, which mounts directly on the monitor's input
connector. This shortens the prop delay for reflections, so they return in
much less than a dotclock period, making impedance match and dispersion much
less important. The downside is that separate adaptor cables would be
required for VGA, 13W3, and 3/5-BNC monitor inputs; however, Software
Integrators has really good ones in stock. But it would be the king of
graphic arts drivers. Imagine a GDM-FW900 with that kind of signal purity!
Another possible feature for an outboard analog adapter would be a
socket for one of the common flash devices, which could spoof the monitor
identification data, making it easier to configure X for older monitors.
This raises another question. How will analog output be implemented
on the FPGA development board? That can't be programmed into an FPGA, and
it's important to have top-end analog on the development platform so the HDL
can be evaluated fully. An argument could be made here that the expedient
thing to do is design the outboard analog accessory concurrently with the
FPGA board, and don't even try to fit analog on-board. That way, it's done
once and for all.
That brings up another point. How about implementing any field
reprogrammability of FPGAs and BIOSs in the form of plug-in flash devices?
That would offer an ironclad way to undo any mistake or accident, taking the
white-knuckle factor out of updating a board in the field. Anybody on the
list know whether any of the common flash cartridges offers hardware
write-protect?
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
