wilkinson combiner without resistor

[jsp]

wilkinson combiner

i would like to know is it possible to design wilkinson combiner without isolation resistors with changing in characteristic imp. or length and the same performance for return loss
thanks in advance

 

[mr_ghz]

wilkinson combiners

Hi jsp

if you design a wilkinson combiner without resistor, the return loss on
the S11/S22 port will be poor (about -6dB). Also the port-to-port isolation
S21 will be in this range.
The resistor transforms out-of-phase input-signals into heat. This is the
reason for the matching and isolation.
With stubs or different line-impedances you cannot create a part, that
will absorb the energy, so you will have no chance to create a 'good'
power-combiner.
Is your problem the inductance of the resistor? Perhaps you can use
more than one resistor or a carbon-film on your PCB.

Greetings

 

[Gemini1706]

wilkinson combiner design

Yes you can, only in 1 condition:
That you are trying to combine THE SAME SIGNAL that was split into two pathes before combining (as in balanced amplifiers).

If your signals are exactly coherent(i.e. the same signal, specially in phase), then there is no heat dissipated in the resistor, becasue the two signals will always be in-phase, and the voltage across the resistors will ALWAYS be zero, hence no current flows into the resistor===> resistor is floating and it is completely out the circuit, so you can now remove it without affecting anything pretty much.

Notice my emphasis that you use it to combine the same-signal back (so you will have a combining GAIN of about 2.5 to 3 dB) and not the usual 3-3.5 dB LOSS when you combine different signals.

mr_ghz remarks are true and valid, but only applies if you have different signals (in phase, or what is called 'non-coherent signals')

Cheers...

 

[FANT]

wilkinson combiner

Gemini 1706 You are right, but consider that you use two amplifiers mostly for the following reasons:

1) to increase the output power of an amplifier without a mortgage !

2) to have a " failure resistant " system

In the first case it is quite difficult to maintain ( not to tune, this is easy ) gain and phase constant over temperature, aging, bandwidth, different drive level, ecc and in this case you need the balancing resistors;

In the second case, when one of the amplifiers fails the remaining one must see a quite good return loss and this is achieded only with the balancing resistors.

Depending on the output power uf the amplifiers you can use some non inductive resistors ranging from the 1206 size ( usually we use 4 in parallel to increase power ) good up to 5/6 GHz or some non inductive resistors from Rhom ( but until 500/1000 MHz ).

Mandi

 

[Gemini1706]

wilkinson resistor

FANT, true, but some explanation to your points:

1-to increase the output power: I saw many systems that use inphase combining without a resistor, but usually those systems are intelligent (adaptive, that is) so they automatically adjust for temp and aging variations.. though not perfect, but still supply adequate combining gain without being much in-coherent.

2-to have a " failure resistant " system. Here is a good trick (one of my 'grandma' secret recipes :
You have your wilkinson, without the resistor ofcourse, but instead of having the arms of the combiner quarter-wave length, make them HALF-wave length 50Ohms lines. Here, when both amplifiers are operating, they combine as usual, since having half-wave instead of quarter wave does not do any difference (since it is just operating as a matched 50ohm line independent of length). When an amplifier fails (or better, somebody pulls it out in system level), the impedance at the floating arm is open-circuit, and the half-wave arm translates it as is (open circuit also) so the other arm does not see any "loading" due to pulling the other amplfier off.
I know this works best if you pull the amplifier completely (i.e. hot-swapping systems), but can still also provide some help if an amplifier fails, specially if its drain is damaged and thus acts as an open circuit, or having some simple intelligence to open circuit the failing amplifier.
This worked for me without any trouble for a hot-swapping system. I saw other systems trying to solve this using "REALLY" expensive techniques. In our case, it is just a nother piece of extra coax length .


My point is: there are systems out-there that do use wilkinson (or modified wilkinson, as described above) without need for the extra high-power resistor.
Adding a little bit of intelligence is not that epxensive (you can use very simple techniques, 55 timers...PICs..etc), but can add a lot to the system.

 

[FANT]

wilkinson compiner

Sorry Gemini 1706, but I have seen many amplifiers destroyed and I can assure You that the impedence You see at the output ranges from 0 to Thousands ohm ( also, if you are quite lucky, 50 Ohm with the resistence of the burned PCB).
Also, if the Drain is open, doesn't say that the output looks like an open circuit, there is between the drain and the output a trasforming network, usually a pass bass filter, a directional coupler, ecc.
Also, with your network, if the amplifier output is 0 Ohm also at the summing junction you will have 0 Ohm ( and you can imagine the results ).
Only if your system is hot swappable what you say is true, otherwise we all are in the darkness.
I think we should know more about the system.

Mandi

 

[Gemini1706]

wilkinson combiner phase error

Yeah FANT, the system I described is more suited for hot-swapable systems than for fault tolerance. Totally Agree.

My point is: There are systems that DO use in the real world wilkinson combiners without a resistor.

I have a little clarificaton I did not make in my post above (and irritated me till elektroda is up and running again ):
I said:

You have your wilkinson, without the resistor ofcourse, but instead of having the arms of the combiner quarter-wave length, make them HALF-wave length 50Ohms lines

Should have been described more clearly as:
You have your wilkinson, without the resistor ofcourse, but now add quarter-wave length 50Ohms lines to the arms. (thus making each arm length=1/2 wavelength, but keep in mind that only the last 1/4 wave-length of it is actually 50ohms, and the reqular wilkinson 1/4wavelength arm will have its regular wilkinson impedance)..

Anyway, as FANT said, it depends on your system. You can use the above technique if it suits you, and I hope it added something to your toolbox

Cheers..

 

[FANT]

lossless power combiner non-coherent

Gemini, to be honest I have never seen a Wilkinson without a load resistor.
Anyway thank for your help and the suggestion of the 50 Ohm lines.

Mandi

 

[dowjones]

wilkinson combiner amplifier

hi all,
I've reading all these posts and it seems there is some conceptual error on all this, which was addressed by mr.ghz. A wilkinson splitter needs the resistor. if you remove that it becomes a reactive splitter and then the output match (of the output ports) is 6db, as well as the coupling between the output ports (also 6db).

a wilkinson is (or can be seen as) a directional coupler with all ports matched and good isolation between the outputs.

the resistor is the element which absorbs the non-transmitted power: when you inject power from one of the outputs 3db go to the odd port and 3 db go to the resistor. nothing bounces back due to ideal ports match.

on the other hand, if you remove the resistor and make the same reasoning, 3db go to the input, 6 db are reflected back and 6db are coupled to the other output port.

regardz

 

[flatulent]

gysel powr splitter

The no resistor case is for ordinary paralleling of amplifiers. If one goes down the other may be damaged if it does not have a circulator on the output line. This no resistor system only works well with some form of active feedback to examine the two transmitter outputs and adjust their drive levels to make them equal amplitude and phase.

It is much easier to combine the RF power in the field by splitting the antenna array in sections and driving each with an amplifier.

 

[RF_Router]

no resistor power divider

It will be easy for you to know the scattering matrix of a Wilkinson Power Divider.

0 1/sqrt(2) 1/sqrt(2)
1/sqrt(2) 0 0
s= 1/sqrt(2) 0 0

If you can see only column 1 is lossless that mean that only if you enject singal on port 1 in will be lossless. Also when you so lossless in a device it means the power that already entered the device not the reflected power just the one that has already been entered to the device this one will be lossless.

However if you enject on port 2 or 3 as you can see from column 2 & 3. It will be lossy. You may ask yourself. When is the loss gone. Well it went on the resistor. So half will go out but another half will be spent on the resistor.
Barr in mind another thing that it is impossible to match on all three ports a reciprocal lossless decive. You will have to know this statement there are many pepole in the laborotaries trying to match all three ports and it is impossible.

Also if you do not include the resistor it will be working nice only from port 1 otherwise port 2 & 3 will not be insulated.

Enjoy,
RF_Router

 

[nandopg]

wilkinson combiner resistor

Dear Basil,
The datasheet you uploaded describes: "Splitters are used extensively in wireless signal distribution networks, to split the base station transmitted signal, Tx, into many different paths to reach multiple antennas, ...."
It is clear that what is being regarded in the datasheet is a power splitter to be used to feed power to an antenna array.
Antennas are mechanical devices that will never be damaged as the amplifiers do. In this sense a power splitter to feed antennas really do not need resistor isolation. The more import figure is the insertion loss (as described in the datasheet).
When you talk about power splitter/combiners for solid stae amplifiers, you are talking about something totally different. In this application the combiner/splitter will always have an isolation resistor associated. The only exception is if the amplifier array is too big. In this case circulators should be attached in each input and output of the amplifiers that made up the array. I've seen some solid state high power transmitters (higher than 5KW) doing this way.

NandoPG

 

[Sinuous]

wilkinson combiner for shortwave

If you want to design a wilkinson with the resistors going to ground instead you can use a giesel power divider. This divider basically transform the resistor into the equivalent shunt resistors.

 

[molloy]

power combiner phase error

I suppose it is not too difficult to design a wilkinson without resistors, but of course the output ports wont be so isolated. Then it depends upon the load you have and upon the failure you have to face. As to other performances you wont have problems
..eeer maybe i will have to design a geysel combiner. Anyone has some materials about this? Input power approx 1KW
thanks

 

[Gemini1706]

amplifier combiner ghz 50ohm -75ohm -sat -75 -ohm

Ofcourse you cannot connect both PAs ithout isolators at each output. No doubt in that...

It looks like people did not get my point about removing the wilkinson resistor:
1- The 3 dB loss some people described here ONLY applies if you have different signals. Listing math equations describing the combiner will not help, because you have TWO IDENTICAL SIGNALS. I know what I am talking about, because we have such a system that is working perfectly. SO PEOPLE PLS STOP insisting to talk about this 3 dB loss and the basic combiner equations. IT JUST DOES NOT APPLY.
2- you need to have isolars after each PA (as some said above). I guess this is common sense and does not need expalnation (if you removed the resistor, you loss all your isolation.. Duh )
3- you need to have simple form of intelegence (as I described above) to make sure phase drift with time is avoided.
4- It is used mainly for hot-swapable systems, and it works perferctly fine. In fact, we hold patent for it, and others actualy pay us royalities for such simple technique.

Sorry to repeat what I said over and over again, but this topic have been for a while, and some people still list the basic wilkinson combiner and some quotes from pozar book

Cheers