Bi-directional diplexer

On a sunny day (Wed, 28 Jan 2009 21:33:11 +0000 (UTC)) it happened Paszczak wrote in :

I dunno, but in the very long ago past I did something with serial and parallel filters to have different transmitters at different frequencies working into the same load:

TX 100 MHz ----------C1--- L1 ------

-------------------- RX 100 MHz |-----------------------------| RX 300 MHz ------------------------- ------C2 --- L2----- TX 300 MHz

C1 L1 tuned to 100 MHz, and C2 L2 tuned to 300 MHz

You can use parallel tuned circuits too. The Q of both LCs must be so that the required bandwidth is transmitted. Add some resistors perhaps. Losses? Dunno what power you have.

Go way back to the beginning and tell us what exactly it is you are trying to do, purpose, specs etc..

--
Joe Leikhim K4SAT
"The RFI-EMI-GUY"©

"Use only Genuine Interocitor Parts" Tom Servo  ;-P

lly

wo

, e

om

ood

ts

o

ld

Seems like you're making it too complicated...if you design a diplexer to split between 100MHz and 300MHz ports, that should work fine at each end; the diplexer doesn't care whether you hook a transmitter or a receiver or a transmission line to each port, though the return loss of what you hook there will affect the behaviour of the system of course. But the point is that if you have a diplexer which looks like

50 ohms (broadband) on the common port when the other two ports are properly terminated, you don't have to worry about what's going on at the other end; it, similarly, will take care of itself.

ELSIE can easily do the design for you--at least the version you pay for can. Not sure if the free version does though.

Where I wanted especially good isolation between ports, I've added transmission zeros to kill the unwanted signal--for example adding a

100MHz zero in the 300MHz branch. The zeros are generally easy to add given that you already have some Ls and Cs in there anyway; with ELSIE you can let the program optimize it for transmission and/or return loss.

What I'm trying to design is a microwave data transmission device. The indoor unit does all the modulation, digital processing, etc., the outdoor unit converts signal to/from microwave band.

The connection between two units is a 50 ohm coax. I need to send transmitted signal (say 300 MHz) over that coax to the outdoor unit and receive signal (say 100 MHz) from the outdoor. Moreover I need to send DC power supply and have some low-speed data communications (like 1000-2000 Hz FSK) to communicate with a CPU in the outdoor unit, but let's leave that for now. So the application looks like (fixed font :) ):

300MHz Tx -->--\\ /-->-- 300MHz Rx +----coax----+ 100MHz Rx --

Ok;

1) I don't see your application for multiple transmitters at one end of the link, are you sure that is the requirement or are you simply calling an IF (intermediate frequency) a "transmitter"?

2) If you have multiple transmitters in same band ie: 300 MHz, at same end of circuit, you would normally use a device called a ferrite isolator to provide isolation between the transmitters. These will be pricy and large at 100 or 300 MHz.

3) I have seen many microwave terminals that are split between indoor signal processor and outdoor RF unit. The ones I have seen that are duplex, generally have separate IF cables for "uplink and downlink", thus simplifying the design.

4) If you wish to duplex the uplink and downlink as you describe, the device you need is a diplexer, which as you have recognized is a LP and HP filter arrangement.

1. DC injection and low freq data can be done. it will be a LPF and DC blocking capacitor arrangement at each end.

--
Joe Leikhim K4SAT
"The RFI-EMI-GUY"©

"Use only Genuine Interocitor Parts" Tom Servo  ;-P

If you post an e-mail address that I can contact you, I can help you..

Mark

I meant IF. There would be 4 mixers on the design above - one on each branch.

No, there's only one signal source on each frequency - so there's 100 MHz IF generated on the outdoor side and 300 MHz IF generated on the indoor side.

Yes, that's what I already found out. I found that the most common way to design such filters is to use 'singly-terminated' ones, as they were called. And that works in following scenario:

/---[ LP filter, cutoff 200 MHz ]--- 100MHz 50 Ohm load Tx 100-300 MHz ---+ \\---[ HP filter, cutoff 200 MHz ]--- 300MHz 50 Ohm load

Signal source 'sees' 50 Ohm over whole band - the 'signly-terminated' LP filter has high impedance outside its passband and HP has high impedance also. So when connected in parallel the resultant impedance is the same for every frequency and there are no reflections.

But that's not the same when replacing load on the diagram above by another signal source - the filter isn't 'symmetrical' and behaves differently depending on which 'direction' you use.

Other filters (that were called 'doubly-terminated' - which were 'symmetrical') have low impedance in the stopband, so they cannot be connected in parallel without additional circuits. But I think that's the way I have to do it - use such filters and somehow change their impedance in the stopband. The question is - how? Everything I found on the web concerns the first type of filters.

P.

Try:

paszczakojad gmail com

P.

On Jan 30, 9:09=A0am, Paszczak wrote: ...

ed.

100MHz 50 Ohm load 300MHz 50 Ohm load

or

er

...

So the only reason that it's "non-symmetrical" is that when you put a source in place of one of the loads, it sees a 50 ohm load only in the passband of the filter it's connected to, no? First, why do you CARE what it sees outside the range where it's generating energy? (Are your sources so poor that they can't work into a load that's not 50 ohms when you get well away from the frequencies of the power they generate??) Second, if you really think you need that, then add components at the end where you have that source which yield a 50 ohm (or whatever you want) impedance at that point. Basically, it would end up looking like it does looking into the "source" end of the diplexer you diagrammed above; there would be two branches in parallel, and one would go to a termination.

I'd be happy to design such a system for you, for a fee, but from what you've described, I'd say it would be a serious waste of money.

Cheers, Tom

There is an easy solution to the low-z in the stop bands, insert a parallel resonant trap at the other frequency.

100 MHz tx --[LPF]--\\ /--100 MHz trap--[HPF]-- 300 MHz rx >----< 300 MHz rx --[HPF]--/ \\--300 MHz trap--[LPF]-- 100 MHz tx ^ additional traps here if needed.

You can also use 4 port splitter/combiners at each end of the line with 2 ports terminated with 50 Ohms. There are also things called circulators but ones for 100/300 MHz are kind of rare, 1 GHz and up are more common.