Audio/Video distribution inquiry

[[email protected]]

I have a bit of experience in hardware design, but am not overly conversant in
analog such as Audio and Video.

I need to come up with a distribution switch for a CATV application involving
broadcast A/V and am looking for some advice or pointers on details on the
video side. I have the buffers/amps picked out, but I am not very good at the
biasing, and input/output loading to know exactly how to bias everything when
it comes to NTSC video.

I want to take a video source and distribute it to four output channels that
will be sent to commercial inserters for broadcast, so it has to be a good
quality design for obvious reasons....

I need a buffer coming on, an output buffer for each channel, and perhaps
something in between to drive the output buffers.

The audio can wait till I get a grip on the video, but I think that will be a
bit easier to work out. I have specific questions and haven't given much
information here, but what I am hoping for is someone who is better schooled
in the finer points of this kind of design, or perhaps url pointers or
books/manuals that I can get that will guide me through the selection of
biasing and loading the input and outputs.

I appreciate any help or comments.

Thanks,

John

 

[Rich Grise]

I have a bit of experience in hardware design, but am not overly conversant in
analog such as Audio and Video.

I need to come up with a distribution switch for a CATV application involving
broadcast A/V and am looking for some advice or pointers on details on the
video side. ....
I appreciate any help or comments.

If you're insisting on DIY, rather than something off the shelf, e.g.:
http://www.google.com/search?hl=en&q=video+distribution+amp

Then I'd say, as long as you have a handle on the design considerations
for those kinds of frequencies, and impedance matching the cables, I'd
say just bias the amps per manufacturer's spec, and AC couple the video.
NTSC video sets its own black level, you know. ;-) And app notes on
whatever amps you're using, and even mfr's reps, can be a good source
of tips. It seems that the grounding rules for systems with coax
grounded at both ends are different from those for an an audiophile
entertainment center, so keep an eye out for gotchas in this department.
;-)

Good Luck!
Rich

 

[[email protected]]

In <[email protected]>, on 10/25/04 at 06:53 PM,
Rich Grise <[email protected]> said:

If you're insisting on DIY, rather than something off the shelf, e.g.:
http://www.google.com/search?hl=en&q=video+distribution+amp

Its not that I insist, its the customer who has physical and feature
requirements that are not found on commercially available switches. Off the
shelf is not an option.

I will continue to search for more details.

Thanks,

John

 

[Mac]

I have a bit of experience in hardware design, but am not overly conversant in
analog such as Audio and Video.

I need to come up with a distribution switch for a CATV application involving
broadcast A/V and am looking for some advice or pointers on details on the
video side. I have the buffers/amps picked out, but I am not very good at the
biasing, and input/output loading to know exactly how to bias everything when
it comes to NTSC video.

I want to take a video source and distribute it to four output channels that
will be sent to commercial inserters for broadcast, so it has to be a good
quality design for obvious reasons....

I need a buffer coming on, an output buffer for each channel, and perhaps
something in between to drive the output buffers.

The audio can wait till I get a grip on the video, but I think that will be a
bit easier to work out. I have specific questions and haven't given much
information here, but what I am hoping for is someone who is better schooled
in the finer points of this kind of design, or perhaps url pointers or
books/manuals that I can get that will guide me through the selection of
biasing and loading the input and outputs.

I appreciate any help or comments.

Thanks,

John

First of all, I don't work in the A/V world. So take what I say with a
grain of salt.

But National Semiconductor (national.com), Maxim Integrated Circuits
(maxim-ic.com), Linear Technologies (linear.com), and Analog Devices
(analog.com) all make op-amps that are specifically designed to drive
75-Ohm video loads.

It seems to me that you could just take four of them and set them all up
with a gain of 2, then put a 75-Ohm resistor in series with each output
and you are done. Put a 75-ohm resistor on the common input to ground for
input termination, of course.

Like this (this is ASCII art, use courier or similar proportional font:


|\
| \ 75
IN-------+----+-----|+ \___________/\/\/\_______OUT 1
| | | / |
/ | +--|-/ |
\ | | |/ |
/ 75 | | |
\ | |_/\/\/\__|
/ | | Rf
| | |
| | /
GND | \
| / Ra
| \
| /
| |
| |
| GND
|
| |\
| | \ 75
+-----|+ \___________/\/\/\_______OUT 2
| / |
+--|-/ |
| |/ |
| |
|_/\/\/\__|
| Rf
|
/
\
/ Ra
\
/
|
|
GND

You would have two more op-amps connected exactly the same way.

You can get values for Rf and Ra from the datasheet. For some op-amps
(current feedback types) you basically have to use the recommended value
for Rf. And in your case, you want a gain of exactly 2, so Rf and Ra have
to be equal.

One example of a high performance op-amp (it may be overkill for your
application) is the lmh6702 from national.

I suspect there is something about this problem that I am not getting,
because it seems too easy to me. ;-)

--Mac

 

[Glenn Gundlach]

I have a bit of experience in hardware design, but am not overly conversant in
analog such as Audio and Video.

I need to come up with a distribution switch for a CATV application involving
broadcast A/V and am looking for some advice or pointers on details on the
video side. I have the buffers/amps picked out, but I am not very good at the
biasing, and input/output loading to know exactly how to bias everything when
it comes to NTSC video.

I want to take a video source and distribute it to four output channels that
will be sent to commercial inserters for broadcast, so it has to be a good
quality design for obvious reasons....

I need a buffer coming on, an output buffer for each channel, and perhaps
something in between to drive the output buffers.

The audio can wait till I get a grip on the video, but I think that will be a
bit easier to work out. I have specific questions and haven't given much
information here, but what I am hoping for is someone who is better schooled
in the finer points of this kind of design, or perhaps url pointers or
books/manuals that I can get that will guide me through the selection of
biasing and loading the input and outputs.

I appreciate any help or comments.

Thanks,

John

The unit you're looking for is a routing switcher. If you're intent on
DIY, my advice is to not bias it at all. Use bipolar power supplies
and DC couple the signals as it will cause you the least grief. Maxim
and ADI (and others) make crosspoint chips. Some even include the line
drive amps in the chip. The audio doesn't require the frequency
response but does need better crosstalk specs than the video. 60db on
the audio would probably be barely usable but would be quite
acceptable for the video. Its definitely not a trivial project but is
certainly possible. The control system alone is a decent project by
itself.

GG

 

[[email protected]]

The unit you're looking for is a routing switcher. If you're intent on DIY,

Its not a matter of being intent on DIY. Surely the engineers here understand
the need for physical specs, as well as other requirements that are not met by
standard off-the-shelf components. I also need DTMF detection and switching,
as well as capability to expand the number of channels, fit in a one rack
unit, and a handful of other features that are only found on specific devices.
Apparently the customer would rather build his own products than pay
exhorbitant prices to a third party OEM who he believes is ripping him off.
Lots of issues to consider beyond going to the net and just buying anA/V
switch.

my advice is to not bias it at all.

Would you mind expanding on that concept just a bit? I am not sure why that
would be a good idea, but I am also not certain I understand exactly what you
are implying.

Thanks for your time,

John

 

[[email protected]]

It seems to me that you could just take four of them and set them all up with
a gain of 2, then put a 75-Ohm resistor in series with each output and you
are done. Put a 75-ohm resistor on the common input to ground for input
termination, of course.

I am loathe to admit my ignorance in a public forum, but I do want to learn
something so I have to ask why I want a gain of two? I am not trying to fool
anyone here, its not something I know a lot about and would rather not be
sitting in the back room of this place in a month, trying to figure out why
what I did does not work. I would prefer to look silly up front and get it
right, if you know what I mean

I suspect there is something about this problem that I am not getting,
because it seems too easy to me. ;-)

No, its just that I have never video distribution before, and will listen to
any and all advice before commiting to a design that is flawed from the
begining. Some would probably tell me not to do it at all, but there is enough
help and wisdom in the world that I will give it a try. I need the money, and
I am sure I can figure it out as I go.

Thank you for the artwork. It came across fine once I switched the fonts. I
appreciate you taking the time to draw it up and explain it.

Much obliged to you,

John

 

[Rich Grise]

I am loathe to admit my ignorance in a public forum, but I do want to learn
something so I have to ask why I want a gain of two?

The 75 ohm terminator resistor appears in series with the 75 ohm coax
impedance, making a 2:1 voltage divider at the output.

I am not trying to fool
anyone here, its not something I know a lot about and would rather not
be sitting in the back room of this place in a month, trying to figure
out why what I did does not work. I would prefer to look silly up front
and get it right, if you know what I mean

This is a rare quality, and is to be commended, and nurtured.

No, its just that I have never video distribution before, and will
listen to any and all advice before commiting to a design that is flawed
from the begining. Some would probably tell me not to do it at all, but
there is enough help and wisdom in the world that I will give it a try.
I need the money, and I am sure I can figure it out as I go.

Well, don't give yourself analysis paralysis. ;-) Bottom line is, it's
not going to be as hard as it appears. It's only video, after all.

Thank you for the artwork. It came across fine once I switched the
fonts. I appreciate you taking the time to draw it up and explain it.

Cheers!
Rich

 

[Mac]

I am loathe to admit my ignorance in a public forum, but I do want to learn
something so I have to ask why I want a gain of two? I am not trying to fool
anyone here, its not something I know a lot about and would rather not be
sitting in the back room of this place in a month, trying to figure out why
what I did does not work. I would prefer to look silly up front and get it
right, if you know what I mean


No, its just that I have never video distribution before, and will listen to
any and all advice before commiting to a design that is flawed from the
begining. Some would probably tell me not to do it at all, but there is enough
help and wisdom in the world that I will give it a try. I need the money, and
I am sure I can figure it out as I go.

Thank you for the artwork. It came across fine once I switched the fonts. I
appreciate you taking the time to draw it up and explain it.

Much obliged to you,

John

Well, I have to repeat, I'm not a video guy. Most of what I know about
video I learned from datasheets. ;-)

Anyway, I guess I'll address this by giving a quick review of transmission
line basics.

Typical transmission lines have a characteristic impedance. For radio and
microwave equipment this is usually 50 Ohms. For video it is usually 75
Ohms. Other values are also possible. Twisted pair might have around 80
or 100 Ohms or 150 Ohms.

But lets stick with video cable which is 75 Ohms.

What does the characteristic impedance mean? It means that if the 75
Ohm cable were infinitely long, it would look exactly like a 75 Ohm
resistor as far as the driving buffer (your buffer) is concerned.

But with finite length cables, the signal eventually reaches the end of
the cable and might reflect back towards the source. This can lead to some
messy interference on the cable, but there is a simple way to avoid it.
You just put a 75 Ohm resistor at the far end of the cable. As far as I
know, all video equipment will be designed this way. That is, it will look
like a 75 Ohm resistor to whatever drives it. Because of the way
transmission lines work, a cable terminated with a resistor of the same
characteristic impedance looks just like the resistor, as far as the
driver is concerned. It doesn't matter if the cable is 1 foot or 1000
feet. It will present a 75 Ohm load to the driver.

So, now you know that your buffer needs to drive a 75 Ohm load. That
still doesn't explain why you need a gain of two instead of one.

Sometimes the resistor (or whatever) at the end of the cable is not
exactly 75 Ohms. When this happens, a little bit of energy will reflect
back towards the source. When it gets to the source, depending on the
impedance presented by the source, some of that energy will reflect back
towards the load again. If you don't have the series 75 Ohm resistor,
virtually all of the energy will be reflected back toward the load, and
this can lead to interference. But WITH the 75 Ohm resistor, almost all of
the energy will be absorbed. This will lead to the cleanest signals,
especially if long cables are involved.

There is another reason to have the 75 Ohm resistor, too. When the buffer
is not attached to a load, the op-amp will still see a small capacitance
as a load. This capacitance can cause the op-amp to oscillate and possibly
overheat. The series resistor will isolate this tiny capacitance and keep
that from happening. The series resistor will also provide pretty
bullet-proof short circuit protection.

So now that we have this series 75 Ohm resistor, we need a gain of two
because the load at the end of the cable and the series resistor will
share the voltage equally. That is why you "need" a gain of 2. If you
could be totally sure that no one would ever short circuit it, leave it
open, or connect it to a load that was way off of 75 Ohms, you could drop
the series resistor, and stick with a gain of 1. But I would strongly
recommend that you include the series 75 Ohm resistor.

As a general rule, whenever you have a long cable, you will have the best
signal integrity if both ends of the cable are terminated in the
characteristic impedance of the cable.

Oh, since the op-amp is a voltage source, we want a series resistor to
terminate it (just like I showed in my ascii art diagram). If we had a
current source (this is rare, AFAIK, in these types of circuits), we would
want a shunt termination, just like we use at the load end.

So there are a couple of main things you should take away from this:

1) You want a series termination for each 75 Ohm load you drive.
2) You want to make sure that whatever drives your board from the outside
world sees your board as a 75 Ohm load.
3) You will need to keep your input and output signal traces as short as
practical, and, if possible, choose a trace width that will give you 75
Ohm impedance for the trace. (There are equations for this.)

If you have some video signals on the board which originate on the board
and do not leave it, you probably don't need to worry about impedance
as much for those signals.

I hope this helps!

--Mac

 

[[email protected]]

In <[email protected]>, on 10/27/04 at 02:58 AM,
Mac <[email protected]> said:

1) You want a series termination for each 75 Ohm load you drive. 2) You want
to make sure that whatever drives your board from the outside world sees your
board as a 75 Ohm load.
3) You will need to keep your input and output signal traces as short as
practical, and, if possible, choose a trace width that will give you 75 Ohm
impedance for the trace. (There are equations for this.)

I guess kind of knew much of that, but I never really had it explained so well
to verify it for me.

To find such a useful, well stated explanation on a usenet group has to be a
giant stroke of good fortune, making it worth the trouble of wading through
the postings of all the political losers in the group.

Simply put, I thank you very much for the explanation, and for the time it
took to write it up. I reckon I can handle doing the design, and what you have
described will be a very big help to me.

Much obliged.

John

 

[Rich Grise]

To find such a useful, well stated explanation on a usenet group has to be a
giant stroke of good fortune, making it worth the trouble of wading through
the postings of all the political losers in the group.

Something that's also rare, and to be encouraged, the way I see it, is
folks who come back and not only acknowledge that there were answers,
but actually express gratitude for them. (how do you do a "good job"
smiley without getting it confused with a "haha just kidding" smilie?
'cause I'm not kidding.)

This is way cool, and is one of the inducements for the folks who answer
stuff to keep answering stuff. <- good smilie

Thanks for paying attention!

And thanks for the progress report!

Cheers!
Rich