In message , Ralph
Mowery writes

"Ian Jackson" wrote in message

Strange as it may seem, if you use (for example) a TV 2-way '3dB' splitter
to combine two identical in-phase signals, you DON'T lose 3dB. Apart from
the unavoidable slight inherent losses of the two transformers the circuit
uses (a total of around 0.5dB at low VHF, increasing to 1dB at high UHF),
the splitter is lossless. Ignoring the transformer loss, the 3dB loss
occurs simply because the power at each output port is half of that at the
input. You haven't actually lost anything.

If the splitter is now turned around to become a combiner, it doesn't
suddenly become more lossy. If you again ignore the transformer losses,
the two identical in-phase signals you feed into the 'output' ports are
added, and the result is a signal 3dB higher.


Jeff found a url with the specs for the combiner.
http://www.solidsignal.com/pview.asp...cc-7870-2-way-
tv-antenna-joiner-coupler-(cc7870)

$18.99? They're having a laff! $5 max.

It says 3.5 db of loss. I assume they use the simple resistor network
instead of transformers. If so, then the net results not counting feedline
loss would be a minus .5 db.

A typical TV 2-way equal splitter has 3dB splitting loss (because it
splits the signal in two), plus a little loss in the two ferrite-cored
RF transformers inside. The spec is usually something like 3.5dB at
50MHz, increasing to 4 or 4.5dB at 900MHz.

A resistive 2-way star or delta splitter/combiner has a loss of 6dB (3
due to the splitting, and 3 in the resistors), However, whereas the 3dB
transformer splitter theoretically has infinite isolation between the
outputs (when the input is terminated) - and in practice it is 25 to
40dB - the 6dB resistive splitter/combiner has only 6dB (essentially,
all three ports are interchangeable).

That combiner does not seem to be made to add signals from idinitical
antennas for more gain, but just to let you use one feedling for several
differant antennas such as putting a FM antenna up and a TV antenna up, or a
seperate UHF and VHF antenna up and using one feedline to the receiver.

It is extremely bad practice to use a wideband combiner to connect two
antennas pointing in different directions. For analogue signals, that's
a sure recipe for multipath and ghosting - and it can't be good for
digitals either. The combiner needs to be filtered (eg a diplexer), so
that one antenna provides signals (on the appropriate frequencies) from
one direction, and the other antenna provides signals (on the
appropriate frequencies) from the other direction.

I have one for designed for my ham transceivers,and have measured less
than .5 db of loss, but that is for differant frequency ranges and not to
combind two antennas on the same band.

That sounds like a diplexer (frequency selective), and not a wideband
splitter/combiner. These can indeed be low loss, as they consist of
frequency-selective (or lowpass-highpass) filters connected to a common
port, and there is no need for wideband combining/splitting.

Even if he does get the maximum of 3 db of gain, will that acutally get him
anywhere ? Will that be enough gain for the FM broadcast band to be
noticiable ?

Well, every little helps!


--
Ian

"Ian Jackson" wrote in message
Strange as it may seem, if you use (for example) a TV 2-way '3dB'
splitter to combine two identical in-phase signals, you DON'T lose 3dB.
Apart from the unavoidable slight inherent losses of the two transformers
the circuit uses (a total of around 0.5dB at low VHF, increasing to 1dB at
high UHF), the splitter is lossless. Ignoring the transformer loss, the
3dB loss occurs simply because the power at each output port is half of
that at the input. You haven't actually lost anything.

If the splitter is now turned around to become a combiner, it doesn't
suddenly become more lossy. If you again ignore the transformer losses,
the two identical in-phase signals you feed into the 'output' ports are
added, and the result is a signal 3dB higher.



I don't have a TV splitter/combiner to play with. I do have a MiniCircuits
combiner I have been playing with.
http://www.minicircuits.com/pdfs/ZFSC-2-2.pdf

I am not sure what is in the combiner but it must be transformers of some
type as the resistance of the ports are near zero ohms.

Their specs is for a 3 db loss and then an aditional loss of about .2 to
1.2 depending on the frequecy. Are you sure that is not the case where you
are saying you do not loose the 3 db ? That is the ratings is for the extra
..5 or so not counting the already 3 db of loss.

Maybe you can tell me if I am playing with the wrong type of combiner.. Here
are some results of my tests. using a HP 8924C service monitor and another
signal generator.


With one input port of the combiner having a signal and the other port
terminated with a 50 ohm load (the nominal impedance of all devices) there
is a 3 db loss (small values not being included) not the low values you
mention to the output port. When I hook up the other signal generator, I
get from almost a total of 0 db to 6 db of loss. I asume the spectrum
analizer going from 0 to 6 db is the phasing of the two generators.



---
This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com

On 1/4/2014 5:40 PM, Ralph Mowery wrote:


snip

As the specs is for a 3.5 db loss, I assume that is if you hook up two
antennas to it, the antennas will have a gain of 3 db at the most, then you
go to the combiner and loose 3.5 db for an overall loss of .5 db.

That is where I don't see stacking two antennas and using that combiner for
more signal strength.


Jeff is correct. Your error is believing the combiner has 3.5db loss.

When splitting a signal, you do have about a 3.5db loss per output,
because the signal is halved plus a bit of additional loss. However,
when combining the signals, the signal is NOT halved, so you don't have
the 3db loss there. You only have about 0.5db loss (more or less,
depending on the quality of the combiner and other factors).

Let's take an example. Since a combiner is just a splitter turned
around, we'll start with the splitter end.

Let's feed 2mw to the input of the splitter. This means each output
gets 1mw (3db loss) (we could use voltage also, but since power is
E^2/R, it's not so straightforward).

So now each leg has 1mw on it.

Now let's turn the splitter around and make it a combiner and feed two
signals, 1mw ea., same frequency, to the inputs to the combiner. Since
this is a totally passive device, the effects are reversible. If the
signals are 180 degrees out of phase, of course the output is 0.
However, if the two signals are in phase with each other, the putout is 2mw.

Note there is no 3db loss in the combiner. But of course this assumed a
"perfect" combiner, with no losses. In reality, the combiner will have
a bit of loss (typically 0.5db as noted above), so the output from the
splitter will be slightly less than 1mw and the output from the combiner
will be slightly less than 2mw.

Does this help clarify things?

And yes, phasing harnesses work the same way. The can be either
splitters or combiners, depending on how they are used. The advantage
is they have less loss; the disadvantage, as noted, is they have a much
narrower effective bandwidth.

--
==================
Remove the "x" from my email address
Jerry, AI0K

==================

"Jerry Stuckle" wrote in message
...

Jeff is correct. Your error is believing the combiner has 3.5db loss.

When splitting a signal, you do have about a 3.5db loss per output,
because the signal is halved plus a bit of additional loss. However, when
combining the signals, the signal is NOT halved, so you don't have the 3db
loss there. You only have about 0.5db loss (more or less, depending on
the quality of the combiner and other factors).

Let's take an example. Since a combiner is just a splitter turned around,
we'll start with the splitter end.

Let's feed 2mw to the input of the splitter. This means each output gets
1mw (3db loss) (we could use voltage also, but since power is E^2/R, it's
not so straightforward).

So now each leg has 1mw on it.

Now let's turn the splitter around and make it a combiner and feed two
signals, 1mw ea., same frequency, to the inputs to the combiner. Since
this is a totally passive device, the effects are reversible. If the
signals are 180 degrees out of phase, of course the output is 0. However,
if the two signals are in phase with each other, the putout is 2mw.

Note there is no 3db loss in the combiner. But of course this assumed a
"perfect" combiner, with no losses. In reality, the combiner will have a
bit of loss (typically 0.5db as noted above), so the output from the
splitter will be slightly less than 1mw and the output from the combiner
will be slightly less than 2mw.

Does this help clarify things?

And yes, phasing harnesses work the same way. The can be either splitters
or combiners, depending on how they are used. The advantage is they have
less loss; the disadvantage, as noted, is they have a much narrower
effective bandwidth.


What I am having trouble with is the 'perfect' combiner.
The one by Wineguard specs 3.5 db loss and the MiniCircuits I have specs at
3 db plus slightly more depending on frequency. I had forgotten that I
built one years ago out of the ARRL Handbook. They give it a spec of 6 db
of loss per port. The one I built has that not counting minor errors and
loss. Just checked it out.

My problem is where are you going to find a combiner for a broad frequency
that does not have any large (say over 1 db ) of loss ? Are the ones for
the TV frequencies built differant ?

For the splitters, I see the 3 db because the signal is going to two places
(3 db equals half power as we all know). But then the problem I am having
is the extra 3 db that is lossed in the combiner instead of just half of a
db or so.

Has anyone actually put one on accurate test equipment to see about the loss
like I have been trying to do ?

I understand phasing harnesses for antennas. They are almost loseless. Only
a few feet of coax worth. I have used them on antennas before. They are
not usually very broad banded unless the antennas are broad banded and made
so the impedance is not the nominal 50 ohms. That is for comercial 4 or 8
dipole arays for VHF/UHF. Lots of 'tricks' used to do that.




---
This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com

On 1/4/2014 9:23 PM, Ralph Mowery wrote:

"Jerry Stuckle" wrote in message
...

Jeff is correct. Your error is believing the combiner has 3.5db loss.

When splitting a signal, you do have about a 3.5db loss per output,
because the signal is halved plus a bit of additional loss. However, when
combining the signals, the signal is NOT halved, so you don't have the 3db
loss there. You only have about 0.5db loss (more or less, depending on
the quality of the combiner and other factors).

Let's take an example. Since a combiner is just a splitter turned around,
we'll start with the splitter end.

Let's feed 2mw to the input of the splitter. This means each output gets
1mw (3db loss) (we could use voltage also, but since power is E^2/R, it's
not so straightforward).

So now each leg has 1mw on it.

Now let's turn the splitter around and make it a combiner and feed two
signals, 1mw ea., same frequency, to the inputs to the combiner. Since
this is a totally passive device, the effects are reversible. If the
signals are 180 degrees out of phase, of course the output is 0. However,
if the two signals are in phase with each other, the putout is 2mw.

Note there is no 3db loss in the combiner. But of course this assumed a
"perfect" combiner, with no losses. In reality, the combiner will have a
bit of loss (typically 0.5db as noted above), so the output from the
splitter will be slightly less than 1mw and the output from the combiner
will be slightly less than 2mw.

Does this help clarify things?

And yes, phasing harnesses work the same way. The can be either splitters
or combiners, depending on how they are used. The advantage is they have
less loss; the disadvantage, as noted, is they have a much narrower
effective bandwidth.


What I am having trouble with is the 'perfect' combiner.
The one by Wineguard specs 3.5 db loss and the MiniCircuits I have specs at
3 db plus slightly more depending on frequency. I had forgotten that I
built one years ago out of the ARRL Handbook. They give it a spec of 6 db
of loss per port. The one I built has that not counting minor errors and
loss. Just checked it out.


A perfect combiner (like anything else "perfect") doesn't exist. But it
is a very common (and handy) way of specifying how things work. It's
used all over the place in EE degree programs, for instance.

So you start with the perfect item, then add losses, phase shifts, etc.
as they occur to get a "real" part.

My problem is where are you going to find a combiner for a broad frequency
that does not have any large (say over 1 db ) of loss ? Are the ones for
the TV frequencies built differant ?


There are good combiners and bad combiners. The commercial grade ones
we use typically have 1db loss from 50Mhz to 2Ghz. Note that these
are basically splitters which are reversed to form combiners, when
necessary.

For the splitters, I see the 3 db because the signal is going to two places
(3 db equals half power as we all know). But then the problem I am having
is the extra 3 db that is lossed in the combiner instead of just half of a
db or so.


In a good quality combiner, there is no extra 3db of loss.

Has anyone actually put one on accurate test equipment to see about the loss
like I have been trying to do ?


I haven't actually measured it myself, but I do use commercial grade
splitters/combiners (not as much any more because a lot of video has
gone digital). Typical loss as a combiner is around 0.5 - 0.7 db from
50Mhz to 2Ghz.

But you also won't find these at Radio Shack or Best Buy.

And there are testing labs out there who do test these things; if any of
the ratings were off, the manufacture would quickly lose credibility in
commercial circles.

I understand phasing harnesses for antennas. They are almost loseless. Only
a few feet of coax worth. I have used them on antennas before. They are
not usually very broad banded unless the antennas are broad banded and made
so the impedance is not the nominal 50 ohms. That is for comercial 4 or 8
dipole arays for VHF/UHF. Lots of 'tricks' used to do that.


Phasing harnesses are just another form of splitter/combiner. One way
they combine; turn them around and they split. That's why they work for
both transmitting and receiving.




---
This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com


--
==================
Remove the "x" from my email address
Jerry, AI0K

==================

On Sat, 4 Jan 2014 17:40:20 -0500, "Ralph Mowery"
wrote:

I agree that stacking antennas works, the problem I have is the type of
combiner that is used.

Well, one could always use an active combiner. For just the FM band,
that's quite easy.

The 30% was from another web site and I assume it was from someone at the
Winegard factory. Not sure why he would say 30% instead of db.

This may help explain the problem:
http://www.journalistunits.com
It doesn't include most electronic units of measure, but I think you
see the problem.

Even so 30
% is nowhere near the 3.5 db listed in the ad. From the url you gave, the
spec is for 3.5 db which is around what I would think it could be if simple
resistors were used.

Nope. A resistive combiner/splitter is -6dB.
http://www.microwaves101.com/encyclopedia/resistive_splitters.cfm
"Resistive power dividers are easy to understand, can be made
very compact, and are naturally wideband, working down to
zero frequency (DC). Their down side is that a two-way resistive
splitter suffers 10xlog(1/2) or 3.0103 dB of real resistive loss,
as opposed to a lossless splitter like a hybrid. Accounting for
3.0103 dB real loss and 3.0103 dB power split, the net power
transfer loss you will observe from input to one of two outputs
is 6.0206 dB for a two-way resistive splitter, so they are often
called 6 dB splitters. Dig?"

That combiner seems to be made not for stacking antennas for more gain, but
to combine several antennas either pointed at differant directions or so a
single feedline could be used for a TV and FM antenna or where you hae
seperate antennas on the same mast for UHF and VHF.

When one combines two different band antennas, the usual method is a
diplexer. Since the receiver only sees one antenna on each band, the
impedance is constant. A low pass filter can also be made very low
loss if you don't care much about rolloff and ripple. However, if
we're down to the point where small fractions of a dB produce a
noticeable difference, I suspect that additional gain (tower mounted
pre-amp) or less loss (better coax cable) will be more important.
Something like this:
http://www.solidsignal.com/pview.asp?p=uvsj
0.5 dB insertion loss. Oh well.

As the specs is for a 3.5 db loss, I assume that is if you hook up two
antennas to it, the antennas will have a gain of 3 db at the most, then you
go to the combiner and loose 3.5 db for an overall loss of .5 db.

Please re-read what I wrote. From each of the (input) ports to the
receiver port (output), there is only 0.5dB of loss. If two antennas
provide an additional 3dB of gain, and each port gobbles 0.5dB, then
the combined gain is 2dB.

That is where I don't see stacking two antennas and using that combiner for
more signal strength.

Would you rather make the yagi twice as long? Once we get to very
large antennas, 3dB of additional gain can easily become a mechanical
challenge.

I do agree that to get 3 db of gain from the antenna it would need to be
about twice as long. I did not look up to antenna to see that it was about
10 feet long already. A 20 foot long antenna would be large, but so would
two antennas 10 feet long and seperated by around 5 feet.

Note that FM broadcast stations with directional antennas use various
vertically mounted antennas, not Yagis. They're interested in
survivability as well as gain and pattern. A 20ft long antenna is
possible, but I don't think anyone wants to climb the tower and drop
the antenna to fix a broken element. That's much easier with a side
mounted barbeque grill type antenna, stacked dipoles, crossed dipoles,
horizontal loops, etc.

Maybe not too bad as I have several antennas on booms that are close to 15
feet long stacked about 5 feet apart. Not the best, but it was what I could
do for what I had to work with. You can see them on my QRZ.com page under
KU4PT.

You probably don't have overweight birds sitting on your yagi
elements. Yes, it can be made to work but it's so much easier and
neater to do it with a combiner.



--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On Sat, 4 Jan 2014 21:23:14 -0500, "Ralph Mowery"
wrote:

What I am having trouble with is the 'perfect' combiner.

I feel your pain. Many years ago, I made a similar mistake on the NEC
antenna modeling mailing list. I then processed to make a total fool
of myself and had to be corrected by the experts. Even so, I still
didn't believe it so I built a Wilkinson combiner and bench tested it
for loss. I still have the combiner somewhere as a reminder of my
mistake.

Incidentally, a Wilkinson combiner might be a tolerable solution for
combining two FM antenna. The loss is much less than a bifilar wound
toroid. I'm not sure if it will work over the entire FM band. I can
grind the numbers if anyone is interested.

My problem is where are you going to find a combiner for a broad frequency
that does not have any large (say over 1 db ) of loss ? Are the ones for
the TV frequencies built differant ?

There's only so much you can do with passive only designs. The next
step up is an active combiner:
http://www.rldrake.com/product-ac1686.php
0-3dB gain per port. 54 to 860 MHz.

Has anyone actually put one on accurate test equipment to see about the loss
like I have been trying to do ?

Yep. I have. There's very little loss between the combiner input
ports and the "sum" port. However, in the other direction, there's a
bit over 3dB loss due to the power splitting. See the specs on the
MCL splitter/combiner that you have and try it with a service monitor
or generator. Since it works down to 10 MHz, you might be able to do
the test with a function generator, a few dummy loads, some T
connectors, and an oscilloscope.

I understand phasing harnesses for antennas. They are almost loseless. Only
a few feet of coax worth. I have used them on antennas before. They are
not usually very broad banded unless the antennas are broad banded and made
so the impedance is not the nominal 50 ohms. That is for comercial 4 or 8
dipole arays for VHF/UHF. Lots of 'tricks' used to do that.

It's low, but the phasing harness loss for stacked vertical dipoles is
not zero. I've never calculated or measured it, but this might help:
http://www.kg4jjh.com/pdf/2-Meter%20Vertical%20Dipole%20Array.pdf
"The phasing harness loss at 150 MHz is calculated to be 0.67 dB."
Scaled for 100 Mhz, I would guess about 0.5 dB. Might as well use a
combiner/splitter.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

"Jeff Liebermann" wrote in message
...
On Sat, 4 Jan 2014 21:23:14 -0500, "Ralph Mowery"
wrote:

What I am having trouble with is the 'perfect' combiner.

I feel your pain. Many years ago, I made a similar mistake on the NEC
antenna modeling mailing list. I then processed to make a total fool
of myself and had to be corrected by the experts. Even so, I still
didn't believe it so I built a Wilkinson combiner and bench tested it
for loss. I still have the combiner somewhere as a reminder of my
mistake.

Incidentally, a Wilkinson combiner might be a tolerable solution for
combining two FM antenna. The loss is much less than a bifilar wound
toroid. I'm not sure if it will work over the entire FM band. I can
grind the numbers if anyone is interested.

My problem is where are you going to find a combiner for a broad
frequency
that does not have any large (say over 1 db ) of loss ? Are the ones for
the TV frequencies built differant ?

There's only so much you can do with passive only designs. The next
step up is an active combiner:
http://www.rldrake.com/product-ac1686.php
0-3dB gain per port. 54 to 860 MHz.

Has anyone actually put one on accurate test equipment to see about the
loss
like I have been trying to do ?

Yep. I have. There's very little loss between the combiner input
ports and the "sum" port. However, in the other direction, there's a
bit over 3dB loss due to the power splitting. See the specs on the
MCL splitter/combiner that you have and try it with a service monitor
or generator. Since it works down to 10 MHz, you might be able to do
the test with a function generator, a few dummy loads, some T
connectors, and an oscilloscope.

I understand phasing harnesses for antennas. They are almost loseless.
Only
a few feet of coax worth. I have used them on antennas before. They are
not usually very broad banded unless the antennas are broad banded and
made
so the impedance is not the nominal 50 ohms. That is for comercial 4 or 8
dipole arays for VHF/UHF. Lots of 'tricks' used to do that.

It's low, but the phasing harness loss for stacked vertical dipoles is
not zero. I've never calculated or measured it, but this might help:
http://www.kg4jjh.com/pdf/2-Meter%20Vertical%20Dipole%20Array.pdf
"The phasing harness loss at 150 MHz is calculated to be 0.67 dB."
Scaled for 100 Mhz, I would guess about 0.5 dB. Might as well use a
combiner/splitter.

--

I understand the idea of using 'perfect' items in electronics, then going
for more exect calculations if needed. As most electronic items are often a
5 to 10 percent variation anyway. Often you get a close calculation and
build it and trim for the desired results. I did take 2 year course in
electronics engineering about 40 years ago for an associates degree, so know
about perfect vers real components.

The Wilkinson combiner is possiable for relative narrow frequencies. Not
sure if building one out of descrete components or full size transmission
lines would be broad enough for the whole FM band either. Wild guess it
would be about the same if just two pieces of transmission line of the
correct impedance and length were used.

Isn't the Wilkinson combiner just two pieces of transmission line (or
simulated with components) with a resistor across two of the ports to
absorbe the diffeance if the loads/sources are not ballanced ? I know what
they are and have seen equipment with them in it,but never did much of a
study on it.


As the subject is combining, I have not looked into the losses of splitting,
but it would be 6 db for the simple resistor designs not counting the minor
losses. That would be 3 db for the ports and 3 db lost in the resistors.

To combind signals you would get the loss of the resistors of 3 db and a
fraction of other loss.

I am using a HP 8924C for a test set. It has just about everything you can
think of for a service monitor. Calibrated from 30 to 1000 MHz but usuable
uncalibrated to about a half of a mhz.
http://www.amtronix.com/hp8924c60.htm

As mentioned the only combiner I have is a MiniCircuits and they spec it at
3 db plus small losses depending on the frequency. That is what I am
measuring.

The diplexer/combiner will have very low loss. I have checked out 2 of them
in the past just to see and the losses were about half a db or so. However
that is for frequencies seperated by a very large percentage. Usually one
port is a low pass and the other is a high pass filter. Not suited for
signals on the same frequency as the origional poster wanted to do.

Yes, phasing harnesses on antennas are not totally loseless, but will be
mainly whatever the loss of the coax is between the elements.

If were the origional poster and there were not too many transmitters near
me, I would try a good preamp first. Mast mounted if possiable as it is for
receive only.









---
This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com

"Jerry Stuckle" wrote in message
...
A perfect combiner (like anything else "perfect") doesn't exist. But it
is a very common (and handy) way of specifying how things work. It's used
all over the place in EE degree programs, for instance.

So you start with the perfect item, then add losses, phase shifts, etc. as
they occur to get a "real" part.



I am awear of that 'perfect' vers 'real world'. Took a 2 year asociate
degree in electronics engineering about 40 years ago. Most things are
calculated close and then trimmed to take care of the usual 5 to 10 percent
differance in components.

That is why I was not worried about anything under a DB, but just the parts
close to 3 db.



---
This email is free from viruses and malware because avast! Antivirus protection is active.
http://www.avast.com

In message , Ralph
Mowery writes

"Ian Jackson" wrote in message
Strange as it may seem, if you use (for example) a TV 2-way '3dB'
splitter to combine two identical in-phase signals, you DON'T lose 3dB.
Apart from the unavoidable slight inherent losses of the two transformers
the circuit uses (a total of around 0.5dB at low VHF, increasing to 1dB at
high UHF), the splitter is lossless. Ignoring the transformer loss, the
3dB loss occurs simply because the power at each output port is half of
that at the input. You haven't actually lost anything.

If the splitter is now turned around to become a combiner, it doesn't
suddenly become more lossy. If you again ignore the transformer losses,
the two identical in-phase signals you feed into the 'output' ports are
added, and the result is a signal 3dB higher.



I don't have a TV splitter/combiner to play with. I do have a MiniCircuits
combiner I have been playing with.
http://www.minicircuits.com/pdfs/ZFSC-2-2.pdf

The spec says "• low insertion loss, 0.4 dB typ."
This is the real loss - the loss that will make the splitter get warm
when you feed a signal into it.

I am not sure what is in the combiner but it must be transformers of some
type as the resistance of the ports are near zero ohms.

The usual basic circuit is:
Input port to ground - a 2:1 impedance ratio step-down ferrite-cored
autotransformer T1 to ground.
T1 tap is connected to the centre tap of a second 1:1 ratio
ferrite-cored autotransformer T2.
Each end of T2 of connected to each output port.
A 100* ohm resistor R is connected between the ends of T2 (ie between
the output ports).
*R is 2 x Zo, so for a 75 ohm system, it will be 150 ohms. Note the
purpose of T2 and R is to provide isolation between the outputs (for
signals coming back into the output ports).

Their specs is for a 3 db loss and then an aditional loss of about .2 to
1.2 depending on the frequecy. Are you sure that is not the case where you
are saying you do not loose the 3 db ? That is the ratings is for the extra
.5 or so not counting the already 3 db of loss.

For forward-going signals, T2 and R play no part in the operation of
the circuit. The signal current from T1 tap enters T2 at its centre tap,
splits, and flows outwards in opposite directions to the ends of T2. The
magnetic flux created by the currents cancels out, so T2 presents no
impedance whatsoever. In effect, it isn't there, and the two outputs are
connected in parallel. The impedance presented to the centre tap of T1
is therefore 25 ohms, which is exactly what T1 is there for - to match
the 50 ohm input to 25 ohms of the two (effectively) parallel outputs.

So you see that apart from the unavoidable losses in the transformers
and in the copper wire, there are no losses in this circuit. However,
because the power emerging from each output is half the power of the
input, the loss measures 3dB (plus a bit). If you can devise a passive
2-way equal splitter with less than 3dB loss, you will make a fortune!

Maybe you can tell me if I am playing with the wrong type of combiner.. Here
are some results of my tests. using a HP 8924C service monitor and another
signal generator.

Theoretically, this type of circuit has no loss (whether used as a
splitter of an in-phase combiner). It's only the 'extra' losses that
will give you trouble. The same is true of any other type of combiner,
so which one you use may depend on which will have the least extra loss.
If you don't need high (or any) isolation between the 'output' ports
('input' if a combiner), you might do better with one of the
transmission line alternatives.

With one input port of the combiner having a signal and the other port
terminated with a 50 ohm load (the nominal impedance of all devices) there
is a 3 db loss (small values not being included) not the low values you
mention to the output port. When I hook up the other signal generator, I
get from almost a total of 0 db to 6 db of loss. I asume the spectrum
analizer going from 0 to 6 db is the phasing of the two generators.

Not quite sure what you're doing here. However, for this circuit to act
as a lossless combiner, it relies on there being no voltage across the
resistor R. This means that the two input signals must in phase and of
equal amplitude.

Typical measurements for this sort of device would be (with all three
ports correctly terminated):
In to Out: 3.5dB
Out to Out: 30dB
With one Out unterminated (o/c or s/c):
In to Out: 3.5 to 4dB
With In unterminated (o/c or s/c):
Out to Out: 7dB
--
Ian