achieving more gain by stacking/baying two large UHF aerials

It used to work OK in the old days, but here where I am the great reception
I got from Hannington by using two of the older amps I used was ruined when
some nasty out of band but very large signal got in and gave intermittent
herring bone patterns all over the picture. This was the 70s though,
headroom and passbands may well be much better these days!

Brian

--
Brian Gaff -
Note:- In order to reduce spam, any email without 'Brian Gaff'
in the display name may be lost.
Blind user, so no pictures please!
"Andy Wade" wrote in message
...
wrote:

No it isn't. In this particular case my mind was partly occupied by
the need to watch the van in case someone robbed it. I shall do it
again at a calmer place.

Interesting thread. I agree that it's almost certainly mismatch effects
that are the cause of the apparent inconsistent results. The o/p match of
most commercial TV aerials is not particularly good - the aerial
benchmarking standard only requires a return loss of 6 dB (equivalent to
about 3:1 SWR). Combine that with 9.5 dB RL for the simple parallel
combiner, or around 6-7 dB for a typical low-noise preamp and the scope
for 'SWR ripple' is considerable.

For test/comparison purposes a few strategic 6 or 10 dB pads (fixed
attenuators) will calm things down a lot and should make the results more
consistent - but obviously you won't want to leave those in place in a
real installation.

Using ordinary 2-way hybrid combiner/splitters for combining aerials isn't
an optimum solution - (a) there's usually 1 dB or so excess loss in the
ferrites at UHF and (b) it makes the phasing more critical than a simple
parallel connection. Any phase error between the incoming signals means
that some power will be dissipated in the balancing resistor. Two equal
75 ohm cables teed into a 50 ohm quarter-wave line section (as per the old
J-Beam 2-way phasing harness) is best. The quarter-wave transformer will
work well-enough over the near-octave band, just size it for the geometric
centre frequency as Jim said. (And the band will probably end at ch 60
before too much longer, easing things a bit more.)

Noting use of the 9 dB Proception preamp, something you could try
(experimentally) is to use two preamps - one on each aerial with a
combiner/splitter on the outputs. That type of splitter usually passes
power to both legs, so there's no powering problem (but twice the DC
current load, obviously). Advantages are that you'll gain a couple of dB
on signal handling and that the combiner loss becomes much less
significant, being post-amplifier, so the overall G/T should be maximised.
Disadvantages are a bit of extra cost and the hassle of having to
phase-match two extra bits of cable. I haven't tried this - its just a
suggestion offered with no guarantee of anything :~)

--
Andy

On Sun, 9 May 2010 10:00:35 -0700 (PDT), "
wrote:


Yes, I'll definitely try this with the two aerials at different
spacings next time.

I feel a Russ Andrews moment coming on, how about an electro-hydraulic
aerial separator?

On May 11, 12:45*am, Andy Wade wrote:
wrote:
Using ordinary 2-way hybrid combiner/splitters for combining aerials
isn't an optimum solution - (a) there's usually 1 dB or so excess loss
in the ferrites at UHF and (b) it makes the phasing more critical than a
simple parallel connection. *Any phase error between the incoming
signals means that some power will be dissipated in the balancing
resistor. *Two equal 75 ohm cables teed into a 50 ohm quarter-wave line
section (as per the old J-Beam 2-way phasing harness) is best. *The
quarter-wave transformer will work well-enough over the near-octave
band, just size it for the geometric centre frequency as Jim said.
This needs some experimentation, but do you really think the bandwidth
of such a matcher would be adequate? Instinctively I find the idea
worrying! How do you think such a matcher would perform at the ends of
the band compared to a standard inductive splitter? That, I suppose,
is what matters.
Possibly it would be a good idea to bias it towards the top end a bit,
since that's usually where you need the gain.

Noting use of the 9 dB Proception preamp, something you could try
(experimentally) is to use two preamps - one on each aerial with a
combiner/splitter on the outputs. *That type of splitter usually passes
power to both legs, so there's no powering problem (but twice the DC
current load, obviously). *Advantages are that you'll gain a couple of
dB on signal handling and that the combiner loss becomes much less
significant, being post-amplifier, so the overall G/T should be
maximised. *Disadvantages are a bit of extra cost and the hassle of
having to phase-match two extra bits of cable. *I haven't tried this -
its just a suggestion offered with no guarantee of anything :~)
But wouldn't the s/n ratio be set at each amp, and thus would not
improve when the signals were combined? Surely the result would be a
signal/noise scenario identical to that from one aerial, but plus a
couple of dB? (Otherwise, we could combine the outputs from LNBs!)

By the way I can confirm that the input filters on those 9dB amps work
well. T'other day I went to a communal system absolutely wrecked by
VHF and low UHF ****e. There were two masts and a multistorey car park
within 75m, all of them festooned with God-knows-what tx aerials. I
was expecting a fair bit of bother but I removed the old Labgear
masthead and fitted a Proception one, and turned the gain up a bit on
the dist amp, and there was no problem at all. It were a proper
anticlimax. I had a load of bandpass filters and channel filters and
notch filters in the van, and they all went back to base with tears in
their eyes.
Bill

wrote:

[lambda/4 transformer]
This needs some experimentation, but do you really think the bandwidth
of such a matcher would be adequate?

Yes, and it's probably the best you can do using standard coax (50 ohm).
The twin-preamp idea avoids the issue of course.

Instinctively I find the idea worrying! How do you think such a
matcher would perform at the ends of the band compared to a standard
inductive splitter? That, I suppose, is what matters.

I'll work out some figures tomorrow.

Possibly it would be a good idea to bias it towards the top end a bit,
since that's usually where you need the gain.

But bear in mind that the top end will likely be ch60 or 62 before too
much longer.

[twin preamps]
But wouldn't the s/n ratio be set at each amp, and thus would not
improve when the signals were combined? Surely the result would be a
signal/noise scenario identical to that from one aerial, but plus a
couple of dB?

Definitely not. You're collecting twice the amount of signal and the
same amount of thermal noise. (The individual noise contributions from
each antenna+preamp are uncorrelated and don't double when combined.)

One way to look at it is this: to take a concrete example let's assume
that the signal collected from each aerial is 40 dBuV, preamp gain is 10
dB with a noise figure of 2 dB and the antenna noise temperature is
around 290 K, giving about 4 dBuV antenna noise floor in 8 MHz bandwidth
(75 ohm system). At each preamp o/p the signals are 50 dBuV and the
noise floor will be ~16 dBuV. The individual C/N's are 34 dB. Now
combine in a hybrid combiner/splitter - the signals are (ideally) 100%
correlated - equal amplitude and exactly in phase - so no power is lost
in the balancing resistor. With a loss-free combiner the o/p signal
will be 53 dBuV. Now the noise inputs to the combiner are completely
uncorrelated and each will experience 3 dB insertion loss (half the
noise power to each input is dissipated in the balancing resistor). By
superposition the combined noise o/p is the same as at each input (two
halves are adding up to make a whole), i.e. ~16 dBuV. The output C/N is
37 dB and we have gained 3 dB relative to using one aerial and preamp.

(Otherwise, we could combine the outputs from LNBs!)

'Fraid not. There the noise contributions originate from the same
antenna and front-end. Both signal and noise are correlated and all
you'll gain by combining is an increase in the amplitude of both but no
more C/N. You would gain with separate LNBs on separate dishes, but
getting the signals in phase at 12 GHz might be a little tricky. A
bigger dish is much easier :-)

By the way I can confirm that the input filters on those 9dB amps work
well. T'other day I went to a communal system absolutely wrecked by
VHF and low UHF ****e. There were two masts and a multistorey car park
within 75m, all of them festooned with God-knows-what tx aerials. I
was expecting a fair bit of bother but I removed the old Labgear
masthead and fitted a Proception one, and turned the gain up a bit on
the dist amp, and there was no problem at all. It were a proper
anticlimax. I had a load of bandpass filters and channel filters and
notch filters in the van, and they all went back to base with tears in
their eyes.

That's good to hear. Essentially the same input filter is used on the
UHF inputs of all products. It's around 26 to 30 dB down at 400 MHz -
far more in the VHF and below. If you've got high power TXs between
400 and 470 you might sometimes find you need some external filtering.

--
Andy

On May 14, 1:10*am, Andy Wade wrote:
wrote:
Definitely not. *You're collecting twice the amount of signal and the
same amount of thermal noise. *(The individual noise contributions from
each antenna+preamp are uncorrelated and don't double when combined.)

This idea is a revelation to me, and I do so hope you are right. I'm
definitely going to experiment with this. It would make combiner loss
irrelevant.

Bill

In article , Andy Wade
wrote:
wrote:

[twin preamps]
But wouldn't the s/n ratio be set at each amp, and thus would not
improve when the signals were combined? Surely the result would be a
signal/noise scenario identical to that from one aerial, but plus a
couple of dB?

Definitely not. You're collecting twice the amount of signal and the
same amount of thermal noise. (The individual noise contributions from
each antenna+preamp are uncorrelated and don't double when combined.)

Purely for the sake of engaging in academic quibbling, I'd like to point
out that the above isn't totally correct. :-)

The noise from the 'antennas' will mainly come by radiation from the
surrounding environment, weighted in directional terms by their antenna
patterns. As such it comes from the same sources to both of them so will
correlate to some extent. (Otherwise much of radioastronomy would not
work!) So there will be some amount of correlation in the noise.

Similarly, some noise from the amps may come from a common power supply
injecting the noise into them.

To what extent that matters will depend on details like the individual
antenna patterns, and the relative environmental and amp noise levels, etc.

But I'd agree that in practice you can usually expect an improvement that
should approach 3dB. Indeed, if you are 'lucky' and there is a noise
hotspot that you can get antiphase between the two patterns you might do
better than 3dB. :-)

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

In article
,
wrote:
On May 14, 1:10 am, Andy Wade wrote:
wrote: Definitely not. You're collecting twice
the amount of signal and the same amount of thermal noise. (The
individual noise contributions from each antenna+preamp are
uncorrelated and don't double when combined.)

This idea is a revelation to me, and I do so hope you are right. I'm
definitely going to experiment with this. It would make combiner loss
irrelevant.

I'd have expressed it slightly differently and said you get 'twice the
signal' and 'twice the noise' presented for combination. *But* that the
signals correlate (are in phase) so add up nicely, but the noise is
(mainly) uncorrelated so adds up less effectively. Thus giving a SNR
advantage of about 3dB.

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

"Andy Wade" wrote in message
...

Definitely not. You're collecting twice the amount of signal and the same
amount of thermal noise. (The individual noise contributions from each
antenna+preamp are uncorrelated and don't double when combined.)

I no nothing about this topic but am keen to learn, but I'm having trouble
getting my head around this bit.

Let's just put aside SNRs for now and talk about the basics. Are you saying
that if you mix two independent white noise sources, each generating white
noise at the same level, the resultant mixed signal is at the same level,
too? What about mixing three or four, or ten, such sources?

My brain is struggling with this! :-)

SteveT

PS: Also, maybe I'm struggling with the difference between "mixing" and
"adding" signals. Is there a difference?

In article , Steve Thackery
scribeth thus

"Andy Wade" wrote in message
...

Definitely not. You're collecting twice the amount of signal and the same
amount of thermal noise. (The individual noise contributions from each
antenna+preamp are uncorrelated and don't double when combined.)

I no nothing about this topic but am keen to learn, but I'm having trouble
getting my head around this bit.

Let's just put aside SNRs for now and talk about the basics. Are you saying
that if you mix two independent white noise sources, each generating white
noise at the same level, the resultant mixed signal is at the same level,
too? What about mixing three or four, or ten, such sources?

My brain is struggling with this! :-)

SteveT

PS: Also, maybe I'm struggling with the difference between "mixing" and
"adding" signals. Is there a difference?

Could see that either way but white or random noise will as much add as
it will cancel or subtract so it tends to even out. Remember this is the
signal you do not want. Whereas the wanted signal is the same i.e. the
signal from each aerial is "together" as it where, and hence will add...
--
Tony Sayer

In article , Steve Thackery
wrote:

"Andy Wade" wrote in message
...

Definitely not. You're collecting twice the amount of signal and the
same amount of thermal noise. (The individual noise contributions
from each antenna+preamp are uncorrelated and don't double when
combined.)

I no nothing about this topic but am keen to learn, but I'm having
trouble getting my head around this bit.

Let's just put aside SNRs for now and talk about the basics. Are you
saying that if you mix two independent white noise sources, each
generating white noise at the same level, the resultant mixed signal is
at the same level, too?

For the sake of clarity/simplicity, ignore the question of impedance
matching. And ignore questions about what we mean by 'random'. :-)

When combining the 'signals' we find that they are essentially copies of
each other. Thus when combined you can imagine simply adding together their
voltages instant by instant to *always* get double the voltage. So an
increase of 6dB in power. This is because the two copies are perfectly
'coherent' - i.e. identical in their voltage pattern with time.

Now consider two independent noise sources (e.g. the noise being generated
in two amplifiers). These may be identical amplifiers, but the noise in
each case is 'random'. That means that although the time-averaged noise
powers (and hence rms noise voltages) from each are the same, their
time-patterns are different, and you can't specify the voltage of one at
any instant from knowing the voltage of the other at any instant.

Statistically the two noise voltage patterns have no correlated or
predictable relationship. They are in fact quite 'different' beyond having
similar statistics. On time-average they add together to produce twice the
power (3dB) i.e. the resulting combined time-averaged rms voltage is
increased by root two.

Hence the noises combine and rise by two in power. But the signals rise by
two in voltage - i.e. four in power. The SNR therefore rises by 3dB.

When you take impedance matching into account there will be a change in the
overall effective 'gain' seen in voltage terms. But this affects both
signals and noise in the same way. So although the output levels then
depend on the matching, you still get a nominal 3dB improvement in SNR.

The essential difference is that the signals are perfectly coherent (e.g.
the same) but the noises are not. The noises have time-averaged levels that
are similar, but their instant-by-instant levels vary in randomly different
ways. So as Tony has pointed out, they can easily tend to cancel rather
than combine.

If you wish you can think of this as being: "Half the time the noise
voltages cancel to give nothing. Half the time they combine to double the
voltage - giving *four* times the power. So you get four times the power,
but only half of the time. Whereas with the (correlated) signals you are
*always* getting four times the power all the time." So you 'win' by an
overall factor of two in power. 8-] This isn't the correct explanation,
but hints at the reason for the distinction.

Alternatively, you can use a coins analogy. If you have two coins and throw
them, how often do they agree? This lack of complete agreement is due to
the uncorrelated random behaviour. [noise]

Now weld the two coins together at their edge so they both show the same
side. If you now throw the (correlated by welding) coins you get them
agreeing all the time. [signal] Hence the outcome taken over many throws is
quite different.

What about mixing three or four, or ten, such sources?

The effect on SNR scales with the number of sources. Assuming their noise
levels are all similar and independent (uncorrelated) whilst the signals
are all identical (perfectly correlated).

A method based on this is used to make low-noise transistors where many
junctions are used in parallel to ensure their individual noise
contributions do not correlate. What you buy still has three legs and is
sold as a 'transistor' but actually has many devices inside it.

The snag is that some noise may actually be correlated, and that sometimes
the signals are not perfectly correlated.

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html