Power combiner for non-coherent signals

[Khashia]

An 8-1 power combiner is to be used in transceiver circuit,the 8 channels carrying different frequency signals. Usually power combiners add zero-phase difference signals, but what if these eight signals are differing in phase ?Want to study phase imbalance effects in power combiner circuits for different frequency signals and also phase imbalance compensation techniques if any?

What I found on this topis is power handling capability and excellent phase match of power splitters/combiners.
if anyone could refer any useful material that could be of great help.

 

[Darktrax]

Er.. the whole point of combiners is to combine signals, that is, to add signals to the spectrum of those already there, so long as they are all within the operational passband of the combiner. They do not have to be related in any other way.

There are uses for combiners where the inputs are deliberately coherent, though possibly phase shifted, for example in driving beam-steering arrays, or contriving circular polarization, but combiners don't care. They can do coherent as well.

If you want to add another modulated channel, or a beacon pilot signal, you can, so long as they are not so on top of each other that they cannot be separated later.

Thinking on it, a combiner would not be much use if the only signals it could combine were so phase coherent they cannot represent new information.

 

[Khashia]

Thanks for replying to the querry,As you mentioned:

Er.. the whole point of combiners is to combine signals, that is, to add signals to the spectrum of those already there, so long as they are all within the operational passband of the combiner. They do not have to be related in any other way.

Thinking on it, a combiner would not be much use if the only signals it could combine were so phase coherent they cannot represent new information.

Do you say that as long signals are in frequency band of combiner their phase coherence is not that essential and we could continue without adding any phase compensation or adjustment? Wanted to be sure that whether we need to add any phase compensation before combining our signals or not so that extra effort shouldn't be wasted on prototyping circuit multiple times.

There are uses for combiners where the inputs are deliberately coherent, though possibly phase shifted, for example in driving beam-steering arrays, or contriving circular polarization, but combiners don't care. They can do coherent as well.

If you want to add another modulated channel, or a beacon pilot signal, you can, so long as they are not so on top of each other that they cannot be separated later.

I could not understand your statement, and I want to explore the same effects but could not find any relevant material,could you please suggest some site where the same information could be found?I wanted to study these phase in-coherence effects on power combiners.

Further how could we use variable capacitors in such a situation if we want phase coherence?

 

[G4BCH]

If the signals are on different frequencies then they will have a time varying phase relationship with on another, so making any phase adjustments will have no effect.
As Darktrax said, phasing is important in a combiner if the signals being combined are coherent, such as the signals received in each element of a beam stearable antenna.
Look up antenna combiner and Butler matrix for more information on beam forming.
peter

 

[Darktrax]

OK - we need to be clearer.

Do you say that as long signals are in frequency band of combiner their phase coherence is not that essential and we could continue without adding any phase compensation or adjustment?

Yes indeed - and more! Signals in a combiner are like different folk sharing the same street. Taking the analogy further, some of them may be marching in step (coherent in phase), or maybe a bit out of step and with the wrong stride length (messes up the ranks).

You can have a already combined set of signals offered up one input, and two more offered up other inputs where the two are specially phase related. For example, one might be deliberately delayed by 90 degrees, so as to be used for circular polarization from a pair of linear antennas set at right angles to each other. All these can live with each other through a combiner, and do what is expected between them.

You can even, simultaneously, send a signals the other way, up the output, and have them split and go out the inputs. The parts that encountered the phase delay intended for those example incoming circularly polarized, will end up being transmitted as circularly polarized, not interfering, and travelling the opposite direction.

My point is, combiners are not inherently bound by whatever phase-related stuff the user is doing. If the inputs are phase-related, the combination will arithmetically add. That there may be a bunch of other stuff going through makes no difference. It's a good thing. Until they abolish slavery in combiners, I am happy to keep stacking more channels in (so long as they fit in the passband)!

Except for combiners implemented as summing op-amps, combiners are passive, linear, bilateral, and quite friendly.

 

[mtwieg]

If the signals being combined aren't identical (phase, frequency, and amplitude) as referred to the combiner output, then there will be power dissipation and attenuation, which may be critical for some applications. This is necessary in order to maintain good isolation between combiner ports. So if you have incoherent signals at the inputs, expect significant attenuation at the output.

 

[Darktrax]

If the signals being combined aren't identical (phase, frequency, and amplitude) as referred to the combiner output, then there will be power dissipation and attenuation, which may be critical for some applications. This is necessary in order to maintain good isolation between combiner ports. So if you have incoherent signals at the inputs, expect significant attenuation at the output.

This flies in the face of everything I experience with combiners. Out of the of available bandwidth, I would fit channels of phase modulated carriers that sit separately in the spectrum. These do not get attenuated any extra just because a signal in one of the channels starts up. In this application, there is frequency separation. They live independently.

I think the case you refer to might be when combiners are used to split a signal into several identical phase-coherent paths, to enable each to be amplified separately, and then re-combined carefully, with identical phase delays in each path, and featuring isolation, so that not much of any input can go back out another input. This allows a set of lower power devices to deliver a large output, and provide redundancy should one of the amplifiers fail. In this case, the parts had better arrive at the output in phase. I agree that then, getting it wrong might get uncomfortable!

But that is a particular case where the signals being combined originated from a single source (necessary if you want them to be coherent). To combine several unrelated signals, there is no phase coherence, but they combine into an expanded spectrum anyway.

There are also many applications where one can combine coherent signals, identical in frequency, but definitely and deliberately not identical in amplitude or phase. We mess with these two, to do things like beam steering, or placing nulls in a particular direction to lose co-channel interference. I cannot agree that the price of such games is any significant attenuation beyond that inherent to the combiners.

 

[mtwieg]

I'm just referring to combiners like wilkinson types, where there are resistors between the ports which provide high isolation. In such cases, any difference in voltage between the ports, due to phase, frequency, or amplitude difference, will lead to extra dissipation in those resistors beyond the normal loss of the combiner. Other combiner types may not suffer this problem, but will usually have poorer isolation, to my knowledge.

 

[Darktrax]

OK - I understand your point. It makes the Wilkinson type even more interesting.

The "resistor" feature was a novel addition contrary to most expectation at the time. Built for splitting, the surprise came when discovering the performance when combining. I am sure the Wilkinson has been analysed to extremes by now, but I have always simply used them from looking at the vendor performance data.

The resistor, being actually between two ports, manages not to dissipate anything of a incoming signal in Port1 because the voltages at Ports 2 and 3 are the same, and in phase. The rest of the explanation for what happens to a signal offered up Port 2 for combining, from one source (Microwaves101) then starts to go wrong for me.



**broken link removed**

I will have the truth of it - in the end..

Append: A quick look indicates the superposition theorem should apply. Voltages at the resistor ends from one signal, if destined to cancel, should not change the destiny of voltages from a separate signal, even if that one is a non-coherent, separate, new frequency signal. The components from the second will cancel also. The complicated mess of both signals together at (say) Port 2 is matched by an similar complicated mess 180 deg out of phase at Port3.

 

[zorro]

These two types of combiner are very different things and must not be confused:

- devices that combine signals at different frequency bands without power loss (ideally)
- splitters/combiners like Wilkinson, 180º hybrids or 90º hybrids

In all cases superposition applies (they are linear).
In the two-way Wilkinson combiner, even mode is perfectly combined and odd mode is completely dissipated.

Regards
Z

 

[tony_lth]

The combiner has a OIP3, maybe about +86dBm. That can lead to IM3 spurs when input different freq signals.

 

[naderff]

the best power divider for removing any distortion in output phase is lange coupler