On Fri, Mar 17, 2006 at 01:11:55AM -0700, James Richard Tyrer wrote:
> Dieter wrote:
> >>If it is going to have BNC connectors, they it probably should have 50
> >>ohm termination resistors.
> >>
> >>But, if you are going to use a VGA "D" connector, the I would use 75 ohm.
> >>
> >>This is a constant issue with many things -- 50 ohm vs 75 ohm.
> >>
> >>If you are going to use BNC connectors and 50 ohm, you really need to
> >>design the circuit board so the three traces also have 50 ohm impedance.
> >> Otherwise with ps rise times the trace will cause ringing if the
> >>propagation time for the trace is more than half the rise time.
> >
> >Uhm... are you saying that displays with BNCs are 50 Ohm?
>
> We would hope so. What we do know is that the VGA standard is 75 ohm
> and the standard impedance for BNC connectors is 50 ohms. Although
> people kept putting BNCs on 75 ohm coax so now you can buy 75 ohm BNCs
> but they don't have as high a frequency rating.
>
> The monitor issue should be researched.
I can answer this one. Both of my BNC monitors are 75 ohm, and they
are equipped with 75 ohm BNC connectors. Whatever the rise time limit is,
the result is razor-sharp pixel edges with no ghosting, at a 135 MHz
dotclock rate. 75 ohm BNC connectors are a little hard to find; you have to
look in major manufacturer's catalogs and maybe in high-end video shops.
The reason? Way back in history the radio industry standardized on
50 ohm coax because it was easier to match that to the most common antennas,
but the video industry standardized on 75 ohms right at the beginning,
because the loss is lower for the same diameter cable. In consequence,
several of the major coax connector families have separate 50 and 75 ohm
series, which in most cases are intermateable without damage, but not of
course, without reflections.
If any BNC-equipped monitors have 50 ohms impedance, that's news to
me, but I'm always willing to learn new things.
Recommendation: make 75 ohms standard. If there is a requirement to
drive 50 ohm cables, offer 50 ohm source terminators and 50 ohm BNC
connectors as a factory option with a specific code in the assembly part
number. But then you have to bypass the 75 ohm traces on the PC board.
However, it wouldn't be possible to route traces to both a VGA connector and
a set of BNCs anyway, because that would create stubs, which would cause
reflections. I can think of two solutions: either a separate artwork for a
board with coax connectors, or run miniature coax from the BNC rear-panel
connectors to the pads on the board. In that case, it would be best to cut
the runs at the source end, and solder the miniature coax to the board at
that point.
However, there are other considerations. If the project wants to
offer an option for BNC monitors, that doesn't necessarily mean the board
has to have BNC connectors. Software Integrators sells high-quality video
cables with a VGA or DVI connector on one and and BNCs on the other. For
higher performance, there are D-sub board connectors with 3 or more true
coaxial contacts. My Apollo video board has 3 75-ohm coax contacts in a
DB-size shell. The Sun standard is probably more common; that connector is
called a 13W3, I think. It's a smaller D shell, a DA-size, I think, with 3
coaxial and several pin-and-socket contacts. I don't know the standard;
perhaps the plain pins are for interrogating the monitor type. Also, the
DVI connector uses something like a stripline contact pair for the analog
signals, instead of plain pin-and-socket; it looks like it might offer
better analog signal integrity than the VGA type. If the project wants to
standardize on just one rear-panel connector and let the cable do the
interfacing, we could try to find out more about the DVI.
Speaking of analog black magic, I have _never_ been able to figure
out how it's possible to run these high-speed analog signals through a
pin-and-socket VGA connector without causing disastrous inductive
discontinuities. Apparently, high-end graphics board makers have this
problem solved. We need to know how they do it. This is one wheel we don't
want to find ourselves re-inventing.
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
