Power combining myth on patch array

[chunweimin]

People use alot the reactive power splitters to feed microstrip patch arrays. As well known, the conventional reactive power splitters are one-way devices, which means s21/s31 is not equal to s12/s13 with poor output port isolation (s23/32). Then, can we use such splitters for a receiving microstrip patch array? Might we get any power loss due to the backward mismatch of this type of power splitters? Thanks very much for the thoughts.

Chunwei

 

[kspalla]

You can still use this kind of splitters but for one application only as either in Tx mode or in Rx mode.
As S21 is not same as S12 it is inefficient to use for both Tx and Rx modes.

 

[vfone]

A standard power splitter is a two-way device.
Assuming 1 is the common port, and 2 and 3 are the split ports, then S12=S21 and S13=S31. This is valid only if all three ports are well terminated (usually at 50 ohms).
In this case even the isolation between ports is two-way, S23=S32

Only the circulators (mounted as power splitters) are just one-way devices.

 

[chunweimin]

Thanks very much for the thoughts.
The array will be used only in Rx mode. As I am aware, the reactive power splitter can be used only as "splitter", but not as a "combiner" due to S21≠S12, S31≠S13 with poor isolation S23/S32 (~7dB only). Maybe another way of asking the question. The 1-to-2 reactive power splitter has good matching over a very wide bandwidth when the signals flow from the input port 1 to the output port 2 and 3 (as a splitter, 50Ω to 100Ω). However, when such array is used as a receiving device, the signals picked up by all the patches will couple in to the receiver, combining the collected energy from all the patches and forming a reverse direction of energy flow from the port 2 and port 3 merged to port 1, and to the subsequent components, e.g. BPF and LNAs etc. Since S12 and S13 are poor (~5-6dB instead of idealy 3dB in splitting mode) with poor isolation, will such property degrade the performance of the receiver? I am not sure that makes sense, or it is not the way I have explained on how the receiving patch array with a reactive power splitter works, or actually such splitter is working as a poor combiner? Cheers.

 

[biff44]

I am confused by this post.

The reactive power splitters I have used have S21 = S31. And they work fine as power combiners too.

Working as a splitter, a 50 ohm source comes up to a T junction. At that T junction, looking one way, it sees 100 ohms resistive. Looking the other way it sees 100 ohms resistive. When the two paths are combined at the T junction, there is 100 ohms in parallel with 100 ohms, meaning the input to the junction is matched to 50 ohms.

Travelling down ONE of the two paths leading away from the junction, I use some form of impedance transformation, going from 100 ohms at the junction, to 50 ohms at the load.

It works exactly the same way if I use that circuit as a combiner, with two 50 ohms sources, combining into one 50 ohm load.

 

[kspalla]

I agree with vfone and biff44 that normal reactive power splitters will have S21 = S31.

Let imagine after the power division and before the antenna there is phase control elements (ferrite) which are asymmetrical wrt phase. This case the entire channel S21≠S12.
This case same reactive power splitter path can not be used for both Tx and Rx modes.

 

[chunweimin]

Thanks for your comments and apologise for my misunderstanding. The S21 and S31 are indeed equal to S12 and S13, respectively.
Now, another issue is the isolation. Do you gentlemen think the reactive type is suitable for power combining? Will the isolation of this type have influence on post signal processing at IF downwards and the resolution of the signal? How do you think of the use of a Wilkinson Power Splitter/Combiner? It's got good isolation (>20dB), not very wideband though, but adequate for narrowband systems. Is the noise of the resistor tolerable when used in Rx? As experts like you guys, which one will you probably use for Rx? I am doing scanning arrays for DF systems, and very much appreciate your advices. Cheers.

 

[kspalla]

Can you tell us about the architecture used for DF system?

 

[chunweimin]

I've got to design a set of antenna arrays covering the full azimuthal plane for DF detection. Therefore, electronic scanning for each array may be unavoidable I suppose, and a number of subarrays are necessary that connect down to the Rx then combine together to a single RF output.

The main issue I have been working on is to choose the most appropriate feed network. I understand people use loads of reactive power splitter/combiner in Rx of Satellite applications, and the performance is enough good. I've also found some Wilkinson power splitters/combiners in some airbourne detection systems. Both types have advantages and drawbacks. What I am looking for is a general design consideration on the selection of a splitter/combiner in terms of the system point of view; noise (thermal issue of lumped components) and signal resolution (poor isolation between the routes ). Regard to these two candidates used in Rx mode, is the signal route isolation more important than the noise out of the resistors in a Rx or in opposite? Cheers.