Cable Splitter Loss and Impedance Matching

Question

From my memory, cable splitters (such as for splitting a cable tv coax cable into two televisions) provide impedance matching to prevent signal reflections, etc.

If one 50 Ohm line is split into two 50 Ohm lines, the two will appear as a single 25 Ohm line (two 50 Ohm lines in parallel). The splitter, therefor, will include 25 Ohm impedance in series with the incoming line. Half the incoming power is dissapated in this impedance and the other half is split between the two cables, so there is -6dB loss on each line.

Yet... recently looking at splitters I've noticed some for sale advertising -3.5 dB attenuation on each line.

How is this possible without impedance matching issues?

Answer 1

Resistors are about the least efficient way to build a splitter. A resistive splitter will split the current and waste the excess voltage. So for a 2-port resistive splitter you lose 6dB. They do have some advantages in specialist situations but practically they aren't used much (I found when I needed one I had to build it myself).

Afaict low loss TV splitters are based around transformers. The transformers can match the impedances without too much loss and can maintain reasonably consistent behaviour over a very wide frequency band. A transformer based splitter built out of ideal components would have a 3dB loss but real transformers are not ideal, so 3.5dB is about as good as you can get.

Mini circuits (a maker of high quality RF splitters) have an appnote describing how transformer based splittlers work. https://www.minicircuits.com/app/AN10-006.pdf

Transmission line or inductor/capacitor based splitters (like the Wilkinson one mentioned in another answer) can be even more efficient (real transmission lines and simple components have less loss than real transformers) but they only work over a narrow frequency band so are not suitable for TV (there is approximately a factor of 2 frequency difference between the top and bottom of the UHF TV band, far more if you include VHF).

Answer 2

I don't know where does 6dB loss on each line is coming from.

If you have an RF splitter, then each path is attenuated ideally by 3dB (so -3dB). For example, the Wilkinson Power Divider ideally splits the signal and so each output is lower by 3dB.

However, because of imperfections in the construction of the splitter or divider, such as impedance mismatches, you will have extra attenuation. So, in your case an extra 0.5dB attenuation.

Depending on the frequency and requirements, 0.5dB can be a bit bad or completely acceptable (given the cost).

Answer 3

I think OP is confusing power with amplitude.

Half power calculation: 10*log10(1/2) ~= -3 dB

Half amplitude calculation: 20*log10(1/2) ~= -6 dB