[Open-graphics] Designing a DDC injection box

[James Richard Tyrer]

Dieter wrote:
>> Many details need to be investigated and decided, but they're not
>> on the critical path to OGC1.  That's important.
> 
> If we are depending on a DDC injection box, we should build and test
> a prototype to make sure it works.  This can be mostly done by
> different people, freeing up the people designing the OGC board
> itself.
> 
>> One question is whether there's a PCB-mount mating connector for
>> the DVI connector on the video board, with low reflections.
> 
> I poked around a bit with google, and found DVI connectors with
> solder connections.  They appear to be intended for cables.
> Connector is listed as 1.95 and the hood is 1.50.  I assume that is
> the price in US$.  If so. DVI connectors aren't as expensive as I
> thought they might be. 
> http://www.action-electronics.com/computerconnectors.htm
> 
> You could take a cable with the correct connectors for your monitor 
> (BNC / 13W3 / HD-15) and solder the coax to this DVI connector. 
> Solder a ribbon cable to the DDS pins on the DVI connector and run 
> them to the DDC box.  (or use twisted pair if it needs that.) If we
> put a connector on the ribbon cable at the box, we could use the same
> DDC box with DVI, HD-15 or 13W3.  Just change the coax to one with
> the right connectors.

Perhaps a D-13W3 or DVI and use adapters.

>> On the cable side, probably the latter.  Assembling that connector
>> to the coax lines without fouling up the impedance m atch has to
>> involve some esoteric black magic.
> 
> Any connector will introduce an impedance mismatch.  At least that's
> what the prof said in transmission line class, back in the dark ages.
> The question is how can it be minimized.

I think that he is suffering from a common misconception about this.
The problem comes from connecting the wire to the connector and there
isn't any way to escape this problem except to have the shielding be
continuous (which isn't possible).  So although the connectors
themselves have gotten better to the point that they are almost perfect,
there doesn't seem to be a way to connect flexible cable without a
mismatch.

I think that the DVI analog signal connectors are strip-line
connectors (like USB) so they might come close to a VSWR of 1:1 but
there is still that change from coax to strip-line to worry about.

OTOH: HD-15 connectors are not good at all for the video signal because 
they are neither coaxial or strip-line connectors.  There is no ground 
inside of the connector to set an impedance.  Using D-13W3 connectors is 
a much better design (but more expensive)

-- 
JRT