How to use a power combiner efficiently?

[g86]

phase matching inputs power combiner

I have seen in many places people use a power combiner to combine two signals with arbitrary phases. But we know it is also possible that no power may come at the output when two signals are out of phase. So my question is ......

:!: :idea: :?:

 

[flatulent]

required phasing

The reduction in total power is a desired situation some times when there is a signal processing function.

 

[g86]

That is true for processing purposes. But say when we need to add two 1 watt signal to get 2 watt we cannot use a power combiner at arbitrary places. But thats what i have seen in many places whre people use arbitrary length transmission line to connect 2 sources. Now when these power appears with -180 degree phase shift, there won't be any power at the output. So to obtain the 2 watt output we should have two equal amplitude and equal phase inputs. Which can be achieved by changeing the transmission line length at the input. But what can be the efficient way to achieve it?

:!: :idea: :?:

 

[flatulent]

right

your observations are correct. The phasing (line lengths) must be kept in mind. This is why in power amplifier design extreme effort is given to matching the phases between the two paths which includes the phases before the amplifiers (ballance of splitter and equal line lengths) and the phase shift through the amplifiers and equal gains in them.

 

[Gemini1706]

"whre people use arbitrary length transmission line to connect 2 sources."

Just wanted to mention that if you are combining to 'different sources' as I understood from your comments above, then there is no way to combine them in phase at all, since the sources are different and carry different phase info...etc.

The only way to acheive max combining gain is if the two signals combined are 100% identical in phase, shape, transmitted info..etc. This usually is true if you have just ONE source of signal, then you SPLIT this source with a splitter, then try to combine them back again (maybe after amplification) with a combiner. Only then you need to worry about phase matching the two signals in the combiner.

If you have two different sources (as an example two HP signal generators) there is no way to combine them with combining gain unless they are transimitting the same info, and they PHASE-LOCKED together, not just frequency locked.

If you are talking about just pure sine signals (CW singals, or signals without any info in their phase), then yes, matching their phases with appropriate length of transmition lines is important to get combining gain.

I hope I understood you correctl...

 

[flatulent]

antennas

The above post describes the classical 6 dB loss for nonidentical signals. From one input of the two port side to the one port side is a 6 dB loss if the other of the two port side inputs is not identical to the first one.

In some cases, like antenna arrays, the unequal line lengths shift the angle of the main beam. In other words, signals from another direction now have the right combination of phase shift between the antenna elements (due to path length differences) plus line length differences to be in phase at the input of the combiner.

 

[g86]

"whre people use arbitrary length transmission line to connect 2 sources."

At WLL, GSM application networks at different buildings. Many other cases too.

Just wanted to mention that if you are combining to 'different sources' as I understood from your comments above, then there is no way to combine them in phase at all, since the sources are different and carry different phase info...etc.

Yes there is no way.

The only way to acheive max combining gain is if the two signals combined are 100% identical in phase, shape, transmitted info..etc. This usually is true if you have just ONE source of signal, then you SPLIT this source with a splitter, then try to combine them back again (maybe after amplification) with a combiner. Only then you need to worry about phase matching the two signals in the combiner.
If you have two different sources (as an example two HP signal generators) there is no way to combine them with combining gain unless they are transimitting the same info, and they PHASE-LOCKED together, not just frequency locked.

And exactly there lies my question. In such situation how to combine such arbitray signals efficiently?

 

[flatulent]

no way

There is no way to combine them without loss of some form.

Semiconductor gain is cheap. Take your 6 dB loss in the passive combiner.

In some situations isolation is the main concern. For instance, in combining two signals to measure IMD. The leveling loops of a signal generator will remodulate spurious sidebands onto its signal if the other signal is put into the output by the nonideal isolation of the combiner.

 

[Gemini1706]

"And exactly there lies my question. In such situation how to combine such arbitray signals efficiently?"...

As flatulent said, there is no way to combine without the theoritical loss of (3dB per signal)+added loss due to non-ideal combiner.

The later, added loss due to non-ideal combiner, can be minimized using short lines into/outfrom the combiner to minimize dielectric/copper losses. Other than that, nothing much can do..

Specifically, nothing much that has anything to do with PHASE can minimize loss. Phase is outside the loss causes, so nothing much about it you can do to minimize losses...

Hope I am understanding your point correctly...

 

[g86]

Re: no way

I was also thinking simillar things. Means there is no way to avoid that loss of power and not only power, the information contained at two input may also can be lossed or distorted during these type of addition.

RF is [Really Funny]

:!: :idea: :?:

There is no way to combine them without loss of some form.

Semiconductor gain is cheap. Take your 6 dB loss in the passive combiner.

In some situations isolation is the main concern. For instance, in combining two signals to measure IMD. The leveling loops of a signal generator will remodulate spurious sidebands onto its signal if the other signal is put into the output by the nonideal isolation of the combiner.

 

[Gemini1706]

Re: no way

the information contained at two input may also can be lossed or distorted during these type of addition.

RF is [Really Funny]

No. Information will not be distorted, unless you combine two different signals at the SAME frequency, which is very unlikely you wanna do that..

But true, RF is funny

 

[flatulent]

linear system

The passive combiner is a linear system. This means that each input will appear at the output with an amplitude loss factor and a phase shift factor. There will be no IMD type added signals at the output.

 

[dowjones]

one possibility to improve combining efficiency is to add a phase shifter at the one input of the combiner, and to adjust the phase until maximum power is obtained at the common output.

Obviously this works if both inputs are at the same frequency.

 

[Gemini1706]

There is a way to get a combing loss much less than the 3dB typical loss per signal for combining two signals with different frequencies:
Use of cavity filters, where you tune one filter to signal1, and the other filter to signal2, then combine them with a regular simple TEE coax connection or something. This method was used in the old days of cellualr to combine tens of signals through a massive expensive cavity structure. It is not used anymore. Also they used to have cavity filters controled by motors to automatically tune the filters to the desired frequencies of the signals, but you know, the cost of this sturucture+motors+control is not justified compared to the cost of just adding more power devices to your stations and live with the 3dB loss for two signals (ofcourse you cannot combine more than two without even having a much greater loss)...

Let us know if you need more info.

Cheers.

 

[dowjones]

gemini, if i understand well, the cavity combiner you mention is a diplexer, a passive device used to separate Rx/Tx in frequency duplex systems. Certainly this works when you use signals at different frequencies, and the loss from the common port to any of the other two is(or must be) very low.

however it cannot be called properly combiner, since I think that it cannot work with signals working at the same frequency, and even with that, the problem of the phase is still present.

in fact, if i remember well, in a diplexer junction, if you are using the path at f1 between ports 1 and 2, then port 3 (designed at f2) presents an open circuit impedance at f1 to the junction, so that it is not "seen". The same happens between ports 1 and 3 at f2.

regards

 

[flatulent]

diplexr

Right. The two passbands must be far apart so that the filters can have a decent transition band. This is not technically a combiner, but is useful for putting two signals of widely different frequencies on the same wire.

Economically, it is much more expensive than an extra 6 dB of semiconductor gain for small signals. For power signals, such as broadcast stations, it is the cheap solution for putting two transmitters on the same antenna.

 

[tungchang]

Hoe abput using High temperature superconductor

Hoe abput using High temperature superconductor material as the phase shifter? It will reduce the loss of signal, right?

 

[mwpro]

1. Are you guys talking about same frequency or different frequency? If you are combining different channels, there is no phase match problem. If you are talking about same frequency, a phase shifter may help.

2. The 3dB loss is due to the wilkson divider. There is no way to get away with this design.

3. IMD result from combining is due to the fact the output stage of the signal sources are nonlinear and the isolation in the combiner designing is insufficient. Isolators may be used in this case.

my two cents.

regards,
mwpro

 

[fcobarr]

Re: How to use a power combiner efficiently? - for flatulent

This thread has been inactive for a while...actually a long time... but I'm hoping that I can refresh it and create some interest in it again. Specifically my question is about a statement flatulent made

You mention there is a "classical 6 dB loss per side when combining non identical signals..."

Further down in the thread Gemini1706 references the same comment, but calls out 3 dB per signal.

Which of the following statements are correct for a 2 way combiner of non-identical signals:

1. If the loss is 6 dB per signal, then my isolation resistor needs to dissipate 75% of each signal. For example, combining two 1 watt non-identical signals into the combiner, the resistor will need to dissipate .75 W minimum (per input) in addition to conductor and dielectric losses.

2. If the loss is 3 dB per signal, then for the same scenario, .5W minimum (per input) will be lost in addition to other losses.

Thanks for clarification and helping in my understanding of power combining.

 

[zeshan102]

in combining of two signal @ same frequencies let suppose the input power of 6 watt first split and than combine with two 12 watt devices so pratcically how much power i can get from ideal case of 24 watt , supposing devices gain of 5 dB in any idea about how to implement tunable phase shifter at input to reduce the mismatch.

Regards