In message , Bill Wright
writes
Michael A. Terrell wrote:


Do you know that the channel combiners in a CATV head end
were wired in odd and even banks, on separate groups to prevent IMD
caused in the passive mixing?

Even now if we use passive filters to combine channels we try to
arrange the filters as follows (for example)
Filter 1: 21, 25, 29, 33 etc
Filter 2: 22, 26, 30, 34 etc
Filter 3: 23, 27, 31, 35 etc
Filter 4: 24, 28, 32, 36 etc
The filter outputs are combined using passive combiners. The idea is to
prevent interactivity between adjacent channel filters.

Ah but, unless they really have to, most large cable TV networks don't
use filter combiners. This keeps the whole of the headend combining
system wideband. Of course, you can only do this if the modulators are
themselves sufficiently clean in respect of out-of-channel unwanted
signals and noise.

Normally, the losses associated with wideband combining are not a
problem. At least one manufacturer made a purpose-made 8-port (and later
a 16-port) combiner, with insertion losses of around 20 and 24dB
respectively. They were designed specifically to provide high isolation
between input ports. A 48-channel headend could be made using four
16-port combiners followed by a standard 4-way turned-around splitter,
so the total loss was around 28dB. As the modulator outputs were around
60dBmV, the combined spectrum level was 32dB - which was more than
enough to allow splitting to provide multiple feeds for the inputs of
individual trunk-line launch amplifiers.

Of course, there are also losses in the interconnecting cables, in-line
injection points for test signals, monitor points etc, so if required, a
special low-distortion level-raising post-amplifier might be used to
ensure that plenty of signal was available. The last thing you want in
large cable TV headends is to be scraping the barrel for signal level.
--
Ian

Ian Jackson wrote:
In message , Bill Wright
writes
Michael A. Terrell wrote:


Do you know that the channel combiners in a CATV head end
were wired in odd and even banks, on separate groups to prevent IMD
caused in the passive mixing?

Even now if we use passive filters to combine channels we try to
arrange the filters as follows (for example)
Filter 1: 21, 25, 29, 33 etc
Filter 2: 22, 26, 30, 34 etc
Filter 3: 23, 27, 31, 35 etc
Filter 4: 24, 28, 32, 36 etc
The filter outputs are combined using passive combiners. The idea is
to prevent interactivity between adjacent channel filters.

Ah but, unless they really have to, most large cable TV networks don't
use filter combiners. This keeps the whole of the headend combining
system wideband. Of course, you can only do this if the modulators are
themselves sufficiently clean in respect of out-of-channel unwanted
signals and noise.
The manufacturers of many of the modulators sold for communal TV system
use claim that the output is so clean that filter combiners are
unnecessary. But I use filters anyway. The fact is that I seem to get
better results.


Normally, the losses associated with wideband combining are not a
problem. At least one manufacturer made a purpose-made 8-port (and later
a 16-port) combiner, with insertion losses of around 20 and 24dB
respectively. They were designed specifically to provide high isolation
between input ports. A 48-channel headend could be made using four
16-port combiners followed by a standard 4-way turned-around splitter,
so the total loss was around 28dB. As the modulator outputs were around
60dBmV, the combined spectrum level was 32dB - which was more than
enough to allow splitting to provide multiple feeds for the inputs of
individual trunk-line launch amplifiers.

Typically the loss using four six-way filters and a four way combiner
would be about 15dB. Modulator O/P is usually 25dBmV (analogue mods) so
the combined feed is 10dBmV. That would usually be amplified to
something between 40dBmV and 55dbmV before feeding out.

One advantage of using filters is that if some of the RF sources only
provide low level signals it doesn't hurt, noise-wise, to amplify before
combining.

Bill

In article , says...

The manufacturers of many of the modulators sold for communal TV system
use claim that the output is so clean that filter combiners are
unnecessary. But I use filters anyway. The fact is that I seem to get
better results.


I wonder if they have any active stages AFTER the filter, possibly
because they might be used in installations with less than perfect
matching?

We used to use some tunable modulators, although most of them were
fixed. The idea of the tuneable one was that you had a near instant fix
if one of the modulators developed a fault. The downside was that the
broadband noise from the unfiltered output degraded the headend by 1dB
per modulator ...

The number of tunable modulators was reduced to one - we used it as the
reference for all of our distortion tests which, of course, used a spare
channel, hence no (fixed) modulator ...


One advantage of using filters is that if some of the RF sources only
provide low level signals it doesn't hurt, noise-wise, to amplify before
combining.


Obviously you will add noise to the amplified source but not to the rest
of the system.

--

Terry

In message ,
Terry Casey writes
In article , says...

The manufacturers of many of the modulators sold for communal TV system
use claim that the output is so clean that filter combiners are
unnecessary. But I use filters anyway. The fact is that I seem to get
better results.


I wonder if they have any active stages AFTER the filter, possibly
because they might be used in installations with less than perfect
matching?

We used to use some tunable modulators, although most of them were
fixed. The idea of the tuneable one was that you had a near instant fix
if one of the modulators developed a fault. The downside was that the
broadband noise from the unfiltered output degraded the headend by 1dB
per modulator ...

The number of tunable modulators was reduced to one - we used it as the
reference for all of our distortion tests which, of course, used a spare
channel, hence no (fixed) modulator ...


One advantage of using filters is that if some of the RF sources only
provide low level signals it doesn't hurt, noise-wise, to amplify before
combining.


Obviously you will add noise to the amplified source but not to the rest
of the system.

Using unfiltered combiners, the tuneable ('frequency agile' in American
parlance) modulators which I worked with were specced to give a CNR of
better than 60dB after 48 headend channels had been combined. Typically,
it was 63dB.

I recall visiting one particular headend in order to prove to that we
met the spec, and to explain why 'Brand X's' modulators gave more like
70dB or better. I spent the day making lots of measurements on a goodly
selection of the 48 channels, and in the end, everyone was satisfied
that the usual 63dB was being obtained.

The reason that the Brand X units were so much better was quite simply
that they incorporated individual channel filters. The disadvantage was,
of course, that they were not instantly tuneable to any network channel.
But, as Terry says, what a headend operator could do was to have a few
fully tuneable modulators to do system measurements with, and to use as
backup units.

Although the competitor's equipment undoubtedly gave around 10dB better
SNR than ours, individual channel filtering was something we had ditched
two generations of equipment earlier, as it had been decided that there
was absolutely no advantage in getting a 70dB CNR at the headend when
the rest of the network was going to degrade this to something like 48dB
(or even less). The overall impact of changing the headend launch CNR
from better than 70dB to our typical 63dB would have been negligible.
Nevertheless, I believe that later on, all our gear was replaced with
the Brand X units.
--
Ian

On Fri, 10 Feb 2012 16:04:22 +0000, Ian Jackson
wrote:

In message ,
Terry Casey writes
In article , says...

The manufacturers of many of the modulators sold for communal TV system
use claim that the output is so clean that filter combiners are
unnecessary. But I use filters anyway. The fact is that I seem to get
better results.


I wonder if they have any active stages AFTER the filter, possibly
because they might be used in installations with less than perfect
matching?

We used to use some tunable modulators, although most of them were
fixed. The idea of the tuneable one was that you had a near instant fix
if one of the modulators developed a fault. The downside was that the
broadband noise from the unfiltered output degraded the headend by 1dB
per modulator ...

The number of tunable modulators was reduced to one - we used it as the
reference for all of our distortion tests which, of course, used a spare
channel, hence no (fixed) modulator ...


One advantage of using filters is that if some of the RF sources only
provide low level signals it doesn't hurt, noise-wise, to amplify before
combining.


Obviously you will add noise to the amplified source but not to the rest
of the system.

Using unfiltered combiners, the tuneable ('frequency agile' in American
parlance) modulators which I worked with were specced to give a CNR of
better than 60dB after 48 headend channels had been combined. Typically,
it was 63dB.

I recall visiting one particular headend in order to prove to that we
met the spec, and to explain why 'Brand X's' modulators gave more like
70dB or better. I spent the day making lots of measurements on a goodly
selection of the 48 channels, and in the end, everyone was satisfied
that the usual 63dB was being obtained.

The reason that the Brand X units were so much better was quite simply
that they incorporated individual channel filters. The disadvantage was,
of course, that they were not instantly tuneable to any network channel.
But, as Terry says, what a headend operator could do was to have a few
fully tuneable modulators to do system measurements with, and to use as
backup units.

Although the competitor's equipment undoubtedly gave around 10dB better
SNR than ours, individual channel filtering was something we had ditched
two generations of equipment earlier, as it had been decided that there
was absolutely no advantage in getting a 70dB CNR at the headend when
the rest of the network was going to degrade this to something like 48dB
(or even less). The overall impact of changing the headend launch CNR
from better than 70dB to our typical 63dB would have been negligible.
Nevertheless, I believe that later on, all our gear was replaced with
the Brand X units.

There is always a downside to individual channel filtering, and that
is the group delay that gets screwed at the edges, unless you also
have all-pass equalizers. It all tends to get a bit messy.

d

In message , Don Pearce
writes



There is always a downside to individual channel filtering, and that
is the group delay that gets screwed at the edges, unless you also
have all-pass equalizers. It all tends to get a bit messy.

Indeed. Some of the analogue modulators etc used on many large cable TV
systems were designed to provide signals at least equalling broadcast
quality. Anything inserted thereafter could certainly screw things up to
some degree - effectively rendering the manufacturers' painstaking work
rather a waste of time!
--
Ian

Terry Casey wrote:
In article , says...
The manufacturers of many of the modulators sold for communal TV system
use claim that the output is so clean that filter combiners are
unnecessary. But I use filters anyway. The fact is that I seem to get
better results.


I wonder if they have any active stages AFTER the filter, possibly
because they might be used in installations with less than perfect
matching?
Mebbe so, I hadn't thought of it. The out-of-channel noise from them is
impressively low, but with twenty of them it adds up.


We used to use some tunable modulators, although most of them were
fixed. The idea of the tuneable one was that you had a near instant fix
if one of the modulators developed a fault. The downside was that the
broadband noise from the unfiltered output degraded the headend by 1dB
per modulator ...
That sounds about right! The most amazing thing I ever saw like that was
at the old Visnews place at Park Royal. They had a system done by
'experts' who I dare not name even now. The modulators were terrible. No
kidding they drifted 1/4 of a channel if you put your hand on the case.
Their outputs were combined with a 20-way resistive star network. The
result was the most horrendous garbage you've ever seen. The modulator
outputs after the combiner were -10dBmV and the noise was about 6dB
below that! Of course there was a lot of amplification following, and a
lot of cross mod because in attempts to get rid of the snow and
patterning people had turned everything up. Hilariously, before I went
to it they had the local aerial firm round and they'd metered the C Pal
signal at one outlet and diagnosed 'weak signal' so they'd fitted a
massive high gain aerial! This was looking at C Pal which was about five
miles away. Happy days!


Obviously you will add noise to the amplified source but not to the rest
of the system.

Well yes, but better to add a little noise there than let the channel go
through the system 10dB down.

Bill

In article , says...

Terry Casey wrote:

Obviously you will add noise to the amplified source but not to the rest
of the system.

Well yes, but better to add a little noise there than let the channel go
through the system 10dB down.


I wasn't disagreeing with you Bill!

Merely saying that noise introduced here wouldn't affect the overall
system SNR - whereas, if it was not filtered, it would ...

--

Terry

Terry Casey wrote:
In article , says...
Terry Casey wrote:
Obviously you will add noise to the amplified source but not to the rest
of the system.
Well yes, but better to add a little noise there than let the channel go
through the system 10dB down.


I wasn't disagreeing with you Bill!

Merely saying that noise introduced here wouldn't affect the overall
system SNR - whereas, if it was not filtered, it would ...

I know you weren't disagreeing with me. I was merely appending to your
comment!

Bill