# Directional couplers for cable tv

Tags:
1. Nov 25, 2015

### manzana

These things are pretty amazing. Very cheap, they are a 3 port device which basically works like a diode, but for AC. For example, it lets a signal pass from coaxial connector A to coaxial connector B but not B to A. There is also a third coaxial port C which passes C to B but not C>A or B>C. So they can protect a main cable from faults on a drop C. They are small passive things maybe an inch square. They are very broad band but do not pass DC. This is powerful magic. thanks

2. Nov 25, 2015

### davenn

do you have a question ?

3. Nov 26, 2015

### manzana

Sorry! Yes, for 35 years I have wondered how these things work. I asked a pHD but he said it was too complicated to explain. thanks

4. Nov 26, 2015

### sophiecentaur

Directional couplers take many forms but the easiest to describe and understand (my personal view) is the coupled line variety. You take two coax lines and open up the side of the screen so that a quarter wave length of the two lines run side by side, inside a common outer. That;s the sort of thing you can make yourself with a knife and a soldering iron. Also, a strip line construction can work well. That's a popular form of construction these days. The devil is in the details of construction but you arrange the spacing so that the coupling is what you need and the lines need to be correctly terminated. The length has to be λ/4. A signal applied to one port will be coupled to the other line and the E and H fields due to the power flow in the input line induce a current and volts in the parasitic line. The phases of the induced signals at the fourth port, due to the E and H fields will cancel (when it's λ/4 long) so all the power must exit through the other end of the line (third port).
If you terminate the fourth port perfectly then all the power coupled by a signal flowing in the reverse direction is dissipated in the termination and nothing exits from the third port. The characteristic impedance of the parasitic line governs the ratio of the E and H fields.
I can see the appeal of the 'diode' analogy but it is not really a good one because a diode is non linear but a coupler is a linear device.
I found this link which could possibly help. It shows how the coupling coefficient relates to the impedances of the lines and the impedance between them.

5. Nov 26, 2015

### manzana

Wow, thanks for the insightful reply. I think I get it at a particular frequency, but these devices are amazingly broadband. For example, there is one for \$9 on Ebay which works from 5 Mhz to a Gig. Thanks Sophiecentaur.

6. Nov 27, 2015

### meBigGuy

7. Nov 27, 2015

### sophiecentaur

That's a good link. Again, the couplers in it are based on λ/4 lines. The bandwidth can be improved by having more sections with varying separations to give around an octave but 5 Mhz to a Gig is a tall order to do with lines. I have a tatty VSWR meter (VHF) which covers a good frequency range and that seems to do it with very few coils and capacitors. Directivity is fair but frequency response is not good. I have no test gear to measure its actual performance.
I remember a colleague of mine designing directional couplers for power flow measurements on HF feeders. Their directivity was excellent and they covered 3-30MHz, with a big response slope. An equalising cicuit took care of that - there was plenty of signal available! Thinking about it, those lines were only less than 1m long , let into the side of the balanced feeder trunking, so maybe the λ/4 is not so relevant for directivity - just for a good frequency response. The ultra wide band coupler that Manzana has seen may have a considerable slope across its bandwidth, which doesn't always matter. It's a good idea to read the spec carefully.

8. Nov 27, 2015

### manzana

You guys are awesome. Thanks

9. Nov 27, 2015

### Averagesupernova

Most of the more recent directional couplers I have seen for HF use a current transformer in conjunction with a sampling the voltage.

10. Nov 28, 2015

### sophiecentaur

Ah yes. If the emfs the output circuit can be made to cancel at one port and enhance at the other, it will be directional. Making the coupling flat would be an issue , though and would require EQ, . It's a myuch more compact solution than one involving lines. But, at short enough wavelengths, the printed line method avoids the dreaded 'wound components' which pile on the ££s and can be incorporated in the circuit board of a multi transistor amplifier etc.