RF "hub"

[NiHaoMike]

Is there some passive circuit with N ports (where N is 3 or more) such that if a RF signal enters any port, it is distributed among the remaining N-1 ports? The most obvious answer is a resistive splitter, but is there an alternative that is lower loss and still passive and symmetrical? (The latter excludes common RF splitters.)

[AF6LJ]

What frequency range are you planning on using said device in??

[KJDS]

There are many possibilities, depending on frequency, bandwidth and other requirements.

[NiHaoMike]

Something like 500-600MHz for those DirecTV Ethernet over coax gadgets or 2.4GHz for running Wifi over coax. But I'm actually more interested in how such a circuit would be built.

[T3sl4co1l]

If you can handle lots of reflections, you can just short a bunch of connectors together.  The return loss will be about as bad as the insertion loss if you had used a resistor divider.

Otherwise, I don't think there exists a solution for N > 2 ports where all share evenly, with no reflection.  And the case for N = 2 is trivial (just a union connector).

After all, in a simple one-to-all connection, "all" includes "self"!

If you want it for "all besides self", on all ports, you'll have a harder time.  I suspect one isn't possible without circulators (which means it would have to be done in the GHz, where those are available).  It wouldn't be cheap.

Remember that an RF network requires matched loads on each port, otherwise the load seen at other ports will be different (reflection from one port causes reflection on the others).  So it still won't be like a hub in the network sense, where up to N ports can be in use.

I think hubs with rx/tx integrated are more common, which also has the benefit of decoding, cleaning up and encoding the signal anew, and as long as you're in there looking at the data, you can put routing, inspection, etc. on it too, saving bandwidth as well.

Tim

[AF6LJ]

For what you are trying to do; you need a preamp and a resistive splitter.
Otherwise the reflections will degrade your signal (reflections as in VSWR).

[KE5FX]

Charles Wenzel published a nifty active circuit to do this sort of thing, but I don't know of any way to do it passively with less loss than you'd get with a resistive splitter and/or loss of reciprocity.

[tekbasse]

If you want to experiment, maybe reflection can be minimized by tapping just the shield (outer conductor) of the coaxial cable for N-2 connections. 

Ideally, signal would be polar opposite on the shield, but reversed again upon induction of the N-2 inner conductors.

cheers,

[ConKbot]

N-way divider (non resisitive) with the sum port open or shorted so any power coming in goes to the sum port, gets reflected and divided between all the ports? Not sure how'd that would compare to a resistive divider. The swr would still be crap though, but that may or may not be a problem.  If you have link budget/margin to spare an attenuator on each port would help with return loss.

[rfeecs]

Is there some passive circuit with N ports (where N is 3 or more) such that if a RF signal enters any port, it is distributed among the remaining N-1 ports? The most obvious answer is a resistive splitter, but is there an alternative that is lower loss and still passive and symmetrical? (The latter excludes common RF splitters.)

Write out the S-matrix for what you are describing:  symmetrical and assuming matched ports, evenly distributing power in to all the other ports.

So S₁₁, S₂₂, etc. = 0 for matched ports.  All the other elements, S_nm are the same number to evenly split between ports.

Look up the rules for S-matrices for passive lossless networks.

For N more than 2, it can't be lossless.

For matched ports, it seems likely that the resistive divider is the best you can do?

[KJDS]

Now I've read the requirements when I'm awake, it can't be done whilst maintaining a respectable input return loss.

[LukeW]

There are lots of practical options - say, a cascaded stack of Wilkinson dividers - but in the real world none of them meet theoretical, perfect ideals.

Is this a theoretical problem or are you looking for solutions to some real-world problem? And if so, what is it?

[AF6LJ]

There are lots of practical options - say, a cascaded stack of Wilkinson dividers - but in the real world none of them meet theoretical, perfect ideals.

Is this a theoretical problem or are you looking for solutions to some real-world problem? And if so, what is it?

I think he wants to split up his Direct TV signal which he says is in the 500-600MHZ range.
So..
This would be a practical problem and perhaps since it is digital my concerns about reflections are overstated.

EDIT:
I do like the Wilkinson divider idea.
I was thinking about that yesterday when the ideal of the signal being digital has slipped my mind.

[rfeecs]

I'm thinking DirectTV uses something like these:

http://www.hollandelectronics.com/catalog/upload_file/MSAT-Splitters.pdf

They are quite lossy and have lousy port return loss.  They specifically claim "low port to port isolation".

I suspect these use some kind of transformer based or lumped element splitters.  Could even be Wilkinsons, I guess.

[AF6LJ]

If we cut one of those open we may find a resistive splitter inside.
The OP could build one, but this is not an exercise for the casual hobbyist.

[NiHaoMike]

I think he wants to split up his Direct TV signal which he says is in the 500-600MHZ range.
So..
This would be a practical problem and perhaps since it is digital my concerns about reflections are overstated.

EDIT:
I do like the Wilkinson divider idea.
I was thinking about that yesterday when the ideal of the signal being digital has slipped my mind.

To be more precise, 500-600MHz is roughly the range DirecTV uses to run IP over coax. The adapters they use to connect Ethernet are going for super cheap nowadays (as in 24 for $30!) and are usable as general purpose Ethernet over coax adapters, hence the interest in repurposing those adapters. Thus far, it looks like the best option is to just use an off the shelf splitter with the input disconnected so that the signal would travel backwards through the splitter, bounce off the input, and then distribute to all the ports. Or if some other signals are to be multiplexed on top, add an adapter that appears as an open or short at 500-600MHz and a pass through for the other signals of interest.

[hendorog]

If we cut one of those open we may find a resistive splitter inside.
The OP could build one, but this is not an exercise for the casual hobbyist.

Huh? Why not? Nothing about a splitter is rocket science.

Good info here on various options for the OP.
http://www.microwaves101.com/encyclopedias/couplers-and-splitters

[AF6LJ]

If we cut one of those open we may find a resistive splitter inside.
The OP could build one, but this is not an exercise for the casual hobbyist.

Huh? Why not? Nothing about a splitter is rocket science.

Good info here on various options for the OP.
http://www.microwaves101.com/encyclopedias/couplers-and-splitters

I agree for those of us who have experience in RF up through UHF and beyond this wouldn't be a big deal.