The explanation for reflections into an unterminated open is rather obvious and intuitive: a signal reaching the end of an unterminated cable has nowhere to go but back, so that's where it goes.
But apply the same logic to a
short and the same "nowhere to go" argument fails - obviously it has somewhere to go! In fact, it has a totally unrestricted path to ground. So why would it reflect?
TL;DR: in the case of a short the transmission line is effectively twice as long, doubled on itself, and the signal comes back on the ground return. Yes? No? Maybe?
My conjecture is that the short makes the transmission line effectively twice as long, folded in half, and the reflection is the ground return. Observing signal over ground this ground reflection appears as an inverted version of the outbound signal. This would also suggest that a proper load termination perfectly balances the signal and ground returns so they fully cancel. In other words, the signal has no where to go but to get dissipated in the load. If the ground return is also the chassis ground on the transmitter this returning ground signal will continue onto the chassis and down the mains earth ground. Since ground is low impedance and is never really "terminated", this reflection can travel quite a ways, spreading across a potentially large portion of the mains network, and only eventually gets consumed by earth coupling and resistive losses. This, in turn, would suggest under-terminated antenna cables are a Bad Thing as they will radiate
somewhere.
Is my guess anywhere near correct? We're talking
power so technically speaking of course the power out can be said to reflect. So does the voltage, but understanding it comes back on the ground return actually seems kind of important. Naturally, the source has to conjugate match the load so the outgoing I/V phase can be canceled by the load. (At least at a single frequency, as in the case of an antenna tuner.) Well, source plus transmission line if the latter isn't a multiple of a half wavelength.
I got thinking about this as I was measuring an RG-8 direct burial antenna feed cable. I used an AWG to send a 10MHz single cycle burst (just a single sinusoidal cycle) down the cable, and T connected it to my scope, set to 1M input impedance. Using a 20ft unterminated cable I could see the pulse go out, then return 35.5ns later, for an effective V
P of 0.748ft/ns. That's about 76% VF, actually a bit shy of the advertised 84%. (But it does seem very low loss, well under the -0.2dB/100ft spec.) When I ran it down the main feedline (with the 20' section attached) into an open I measured ~154ns round trip for a length of ~57.5ft. However, into a
short the round trip is 10% longer! Not much, but easily measured. This is what led me to the speculate that in the case of a short the reflection is the ground return (foil plus braid shield), which then has to be slower... (I know I can also hook up the open feedline as a stub filter and measure the notch center on my VNA. I just wanted to eyeball it in the time domain, for additional reflections and whatnot.)
KM6VKX