Click here to go to our main page on power combining
Click here to go to our main page on couplers and combiners
Click here to go to our main page on Microwave Figures of Merit
New for May 2020. This content came from Mohamed. Many thanks!
First, some comments from the Unknown Editor... Combiner loss is an active area of study as solid-state power amplifiers roll out and replace TWTs. A figure of merit can help distinguish between different approaches early in a design. Following up on combiner loss, it is useful to further quantify where the loss burglars are... dielectric, metal, radiation, isolation and mismatch each play a part. Maybe someday we can follow up with an example where the loss mechanisms are separated.
Throughout many designs of power combiners (whether simple hybrid couplers, multi-branch couplers, Gysel couplers or radial combiners) I was personally faced with the question of how to quantify how good a design is. What is its bandwidth? How to compare two designs? In short, is there a figure of merit that can be simply referred to by designers?
When I worked in industry, it was a varied approach and it really depended on whom you are talking to so here I’m trying to make some sense of all of it.
The major three directly measurable quantities for the combiner are return loss, insertion loss and isolation. One can argue that good return loss is needed for the amplifier at the input to see a good match and produce its best output, With the same token, isolation provides “well…..isolation” between the two or more amplifiers at the inputs, and insertion loss tells you how much of your signal is used to worm the environment.
But again if you have design A and design B how do you chose? 15 dB RL and 15 dB isolation or 20 dB RL and 10 dB isolation?
I personally came to believe that combining loss (CL) is the overall figure of merit that should be used (others may disagree this is why I put the idea here for others to pick at it).
What I came to see is that, and particularly if we are comparing designs, CL (or combining efficiency whichever way you prefer) tells you how good a design really is, if the return loss is bad, then for sure that will reflect on the CL, and the same goes for isolation and insertion loss. For example when one wants to quantify the BW of a combiner, I find it more useful to say well it has a CL of 0.5 dB over 50% fractional bandwidth , rather than saying well it has 20 dB RL over 40% and 15 dB isolation over 30% and an IL of 0.5 dB across 50% etc etc.
Also this approach can tell the designer, what is the effect of all losses stemming from imperfections in the electrical design excluding the ohmic loss, so the designer can simulate the design assuming perfect conductors and this will tell him right off the bat how much energy will be lost , and then ohmic loss effects can be quantified. It would not make sense to silver or gold plate a badly designed combiner as it won’t help much.
Mathematically combining efficiency is defined as the ratio of the combined output power to the summation of the input powers at all ports
For example for an in-phase 2-way combiners the efficiency and hence loss can be defined as:
The magnitudes must be separated.... it should say |S12|^2+|S13|^2.... UE
For quadrature combiners , similar formulas can be calculated.
To generalize the formula and use it for radial combiners, one can write that for an N-way (with N+1 ports with port 1 as the combined port) combiner the loss can be expressed as:
To show an example I show here a design I presented at IMS2018 of a Gysel power combiner, the response is shown below. Port 1 is the combined port, Ports 2 and 3 are the ports to be combined, and Ports 4 and 5 are isolation ports and are terminated.
(From M. M. Fahmi, and R. R. Mansour, “Compact Ridge Waveguide Gysel Combiner,” in Proc. 2018 IEEE MTT-S International Microwave Symposum Digest, Philadelphia, PA, USA June 2018”)
I’d say a 0.5 CL which is about 90% combining efficiency is a good benchmark to define the bandwidth of the combiner rather than the RL or isolation levels.
I hope this small thought will make the life of some designer somewhere a tad easier J