The Secret Life of Splitters

Everyone uses splitters. In the 1980s when you bought a second TV (before those damn cable boxes) Dad popped off to the radio shack bought a coaxial splitter and just plugged in the new TV.  Same thing went for the rotor-antenna on the roof. This usually worked just fine. But maybe there was a third TV, or a line out to the garage. Then eventually a line for the computer. Strangely some of these underperformed and without an engineer in the house, the mysterious problem persisted. More on coaxial cable here.

I'll start with the Power Tree. It's a type of diagram that explains the relationships of numbers. Knuth used one in his semi-numerical algorithms.  This will be much simpler.  Though for the record I spent entirely too much time on this diagram.But what you start with is full power, or full power minus loss over distance. Over 150 feet you probably lost about half a dB so I'll call that loss before the splitter "trivial."  This tree illustrates equal power loss across the system. Sometimes it's much worse.

What you get is that after the first split the power is reduced by half, "P2" or approximately 3dB. Decibels are measured logarithmically and so they don't divide in even increments like you might otherwise expect. I'll use round number here for simplicities sake. If you want the exact fractions you can use the calculator here.

The dB is a logarithmic unit used to describe a ratio.  It could be measuring voltage or pressure, loudness, or half a dozen other things which are all related but different. In this case it's voltage. So at the next split we divide the original power by 4 and reduced power by 75%  which is about 6dB that's only "P4."  At "P8" there is almost bupkis left, a reduction of about 9dB. At 10dB there would be about 1% of the initial voltage remaining. Amazingly, frequency modulated signals can survive beyond this point depending on their strength of course.  But this is only if it's an even division in the tree. See the image below.

This is a 8-way splitter. It exactly mirrors the tree. you can actually see the copper threads loop into ferrite beads then in turn be looped into other ferrite beads. The ferrite bead (aka ferrite choke) is not wound with copper thread you see. So it is not a toroid. The reason they use those and not just solder copper to copper is that it effects the impedance matching.  impedance mismatches create loss and as you can see in the power tree.. there is already a lot of loss. More here.

Inside this splitter (above) you can see that the tree isn't always even. Here the output on the right is "P2" but both the center output and the one to the left are both at "P4."  That means that one of the outputs is 25% stronger than the other two. This is significant enough to effect the receiving device, lower volume, color distortion etc. In analog TV it meant snow. Digital TV being more binary would could mean that some channels are entirely missing. These situations are worsened by connecting splitters in a daisy chain, i.e. the out put of one splitter connects to another in turn. the result is loss compounding upon loss. Some signals are very tolerant. You can attenuate FM radio 20dB will very little appreciable loss. Some attenuation is even desirable as it reduces bed noise. But you should not do so unknowingly.  So choose wisely and cable carefully. I exclusively buy Toner splitters myself.