RF combiner

[SilverFoil]

I'm trying to figure out the best way to get the vector sum of 16 different signals of different phases (and probably multiple frequencies as well). My primary concern is minimising changes to the signal phase, followed by noise figure and then gain. Is there a accepted way to do this? Either active or passive is fine.

The signals in question would be in the range 100MHz to 400MHz, going up to about 30dBm (because other parts of the circuit let out the magic smoke above that). This is for a pseudo-doppler direction finding system if anyone's curious. I'm aware that switches can be used to reduce it down to a two-way but I'm hoping to find a solution that doesn't involve active control.

So far, the best solution I think I've found is stacking combiners like the ADP-2-1W+ https://www.minicircuits.com/pdfs/ADP-2-1W+.pdf. Downside is I'd need 15 of them. A resistive combiner/s would spend too much gain for my peace of mind (for a single-stage radial: -12dB? -24dB? I'm having some trouble finding information on combiners that isn't just a bit at the end saying 'they're like splitters but backwards').

My thanks for any help you can provide.

 

[G4BCH]

they're like splitters but backwards

Yes that is right they work in both directions, split one way combine the other.

Assuming the amplitude and phase errors are within your budget why not just get a MiniCircuits 16 way splitter combiner? You will not get much lower loss by doing it yourself, but you may get slightly better phase and amplitude performance because you can optimise for you frequency range.

 

[SilverFoil]

Yes that is right they work in both directions, split one way combine the other.

Assuming the amplitude and phase errors are within your budget why not just get a MiniCircuits 16 way splitter combiner? You will not get much lower loss by doing it yourself, but you may get slightly better phase and amplitude performance because you can optimise for you frequency range.

Aaaand I just realised that I forgot to mention that this is a PCB mounted circuit. The only 16 way splitter/combiners they make are connectorised.

Also, by 'doing it yourself', do you mean cascaded ADP-2-1W+ that I mentioned in the original post? There's no weird gotcha for why that wouldn't work?

Also, thanks for answering.

 

[G4BCH]

Cascading splitters works fine, that is probably what is inside the 16 way parts, the connectorised 4 way parts I have used are made that way.
By doing it yourself I mean designing your own, a colleague of mine, some years ago now needed a 32 way combiner and to get the performance he needed he had to 'grow his own', cascaded minicircuits or similar parts had too high loss and inadequate phase match,
For your frequency range I would look at a part with a higher maximum frequency, the performance degrades faster at the higher frequencies than low, and your low end is 100MHz so the ADP2-4 or ADP2-20 may be a better choice.

 

[SilverFoil]

Cascading splitters works fine, that is probably what is inside the 16 way parts, the connectorised 4 way parts I have used are made that way.
By doing it yourself I mean designing your own, a colleague of mine, some years ago now needed a 32 way combiner and to get the performance he needed he had to 'grow his own', cascaded minicircuits or similar parts had too high loss and inadequate phase match,
For your frequency range I would look at a part with a higher maximum frequency, the performance degrades faster at the higher frequencies than low, and your low end is 100MHz so the ADP2-4 or ADP2-20 may be a better choice.

Alright, thanks for the advice. I have to say that I'm not confident I could design one from scratch and expect it to work properly with my current level of understanding. Don't suppose you'd know of a comprehensive theory document or simulation software or something like that?

 

[biff44]

i would highly recommend you breadboard this subsystem with connectorized parts, say 1:4 or more power combiners with sma connectors and semirigid cable connecting it all up.
Power combiners are not strictly "vector signal combiners", and you can get some interesting nulling and phase/amplitude distortions of two adjacent antenna sectors.

it would be best to breadboard it, and make SURE it is working like you expect before you layout out and fabricate a PCB board version of it.

 

[SilverFoil]

i would highly recommend you breadboard this subsystem with connectorized parts, say 1:4 or more power combiners with sma connectors and semirigid cable connecting it all up.
Power combiners are not strictly "vector signal combiners", and you can get some interesting nulling and phase/amplitude distortions of two adjacent antenna sectors.

it would be best to breadboard it, and make SURE it is working like you expect before you layout out and fabricate a PCB board version of it.

Alright, I'll look into that. Thanks.

 

[FvM]

Power combiners are not strictly "vector signal combiners", and you can get some interesting nulling and phase/amplitude distortions of two adjacent antenna sectors.

I'd say they are strictly working as vectorial signal combiner, but the effect of vectorial signal addition may be unexpected in some cases. The ideal 16:1/1:16 combiner/divider has a gain of -12 dB, not sure if you're aware of this.

 

[SilverFoil]

I'd say they are strictly working as vectorial signal combiner, but the effect of vectorial signal addition may be unexpected in some cases. The ideal 16:1/1:16 combiner/divider has a gain of -12 dB, not sure if you're aware of this.

Dimly, yes I am aware. Thanks for checking, though. I'm given to understand that adding an amplifier after it would restore the gain but not the signal to noise ratio? Or is 'attenuation equals loss of noise figure' rule only for 'real' (resistive) loss?

 

[G4BCH]

Alright, thanks for the advice. I have to say that I'm not confident I could design one from scratch and expect it to work properly with my current level of understanding. Don't suppose you'd know of a comprehensive theory document or simulation software or something like that?

You could look at :

http://mwl.diet.uniroma1.it/people/pisa/SISTEMI_RF/MATERIALE%20INTEGRATIVO/Kikkert_RF_Electronics_Course/05-RF_Electronics_Kikkert_Ch3_RFTransformers.pdf

https://www.highfrequencyelectronics.com/Dec07/HFE1207_GrebennikovPart1.pdf

Power Dividers and Combiners

not comprehensive design data but will give you some idea of what is involved.
To get the best out of the wound components is part art part manual skill and part science., they are 3D structures that are difficult to model and the placement of windings has a significant effect on HF performance as does the ferrite used. I have worked with people who were good at designing them but never done anything but simple devices myself.

Dimly, yes I am aware. Thanks for checking, though. I'm given to understand that adding an amplifier after it would restore the gain but not the signal to noise ratio? Or is 'attenuation equals loss of noise figure' rule only for 'real' (resistive) loss?

Signal loss is loss and it does affect the system noise figure.

 

[SilverFoil]

You could look at :

http://mwl.diet.uniroma1.it/people/pisa/SISTEMI_RF/MATERIALE%20INTEGRATIVO/Kikkert_RF_Electronics_Course/05-RF_Electronics_Kikkert_Ch3_RFTransformers.pdf

https://www.highfrequencyelectronics.com/Dec07/HFE1207_GrebennikovPart1.pdf

Power Dividers and Combiners

not comprehensive design data but will give you some idea of what is involved.
To get the best out of the wound components is part art part manual skill and part science., they are 3D structures that are difficult to model and the placement of windings has a significant effect on HF performance as does the ferrite used. I have worked with people who were good at designing them but never done anything but simple devices myself.


Signal loss is loss and it does affect the system noise figure.

Alright, that looks like plenty to get my teeth into . Thanks. One last question, if I may. I've seen it suggested that a op-amp in inverting summing configuration could be used as a active combiner. I haven't seen it in any literature, which implies it's a really stupid idea. Is this just because op-amps tend not to be sufficiently high-bandwidth, or is there some reason beyond my understanding of the theory?

 

[G4BCH]

An op amp would work as a combiner if it were near perfect. You don't see it used at much greater than audio or video frequencies because the imperfections become too serious; gain bandwidth product, noise figure & distortion etc.

 

[SilverFoil]

An op amp would work as a combiner if it were near perfect. You don't see it used at much greater than audio or video frequencies because the imperfections become too serious; gain bandwidth product, noise figure & distortion etc.

Makes sense, thanks.