Hi Sarasir,
Trying to visualize exactly what's happening is not easy, I have some ideas, but that's all they are, no concrete proof and only scant evidence that they match reality!
So here is what I currently think, however at this point I'm not sure how to test or verify it!
The following is just a little background of things I think are pretty much of general knowledge:
As we know, most of this originates at the switchers. Due to stray capacitance, inductance, or due to radiation, some of the fast switching current strays from its main path and flows through an alternate unintended path to reach the return. These stray currents can be differential or common mode from the scope's input point of view. Owon has managed to clean up the differential signals with filtering to the point that there is a reasonably low amount of baseline noise, and a reasonably low amount of distortion on the signal path. However, some common mode currents are still present that are invisible from the scope's input point of view.
The following is what I think is happening inside the Owon in more detail but can't prove it:
So I believe that what remains are: a) Currents that are nearly identical in both phase and amplitude at both the sleeve and center conductor of the BNCs. b) In addition, as long as there is nothing connected to the BNCs, I believe there is an even larger number of current paths that are only present on the ground plane but are not referenced to the BNC's center conductor. Therefore these currents are also invisible from the scope's input perspective.
When you connect something to the BNC, for example, a probe cable or plain coax, I visualize a couple of long physically close parallel conductors analogous to long adjacent parallel traces on a circuit board. Since the shield is part of the ground plane, any currents on it that did not originally have a counterpart in the BNC's center conductor will be coupled to it. Assuming a perfect cable, the result will be common mode currents from the scope's input perspective, and as a result, these currents will still remain invisible.
The following tests out experimentally, but the reason why it happens is just my theory:
However, in reality, due to cable imperfections, I would expect a moderate but gradual increase in differential signals as the length of the cable increases. In other words, you would see a little more noise displayed on the scope as the cable length increases.
The following is just a little background of things I think are pretty much of general knowledge:
As the cable length increases propagation delays become more of a factor. This effect prevents distortion free transmission of differential signals, particularly HF signals, from one end to the other unless the cable is properly terminated to match its characteristic impedance. Improper termination will result in reflections that end up being added to the original signal. The resulting signal is a distorted version of the original signal.
The following seems to test out experimentally, but the reason why it happens is just my theory:
However, the signals of interest in this case are common mode, and the concern isn't signal distortion, instead, we'd like to know the mechanism that converts some of these common mode signals to differential mode signals. First consider a coax cable cut at the end with a sharp razor blade. To a signal this cable will exhibit an abrupt transition from 50 ohms characteristic impedance to nearly infinite impedance. However, because of the clean and nearly perfectly symmetrical cut, there will be little if any impedance imbalance in respect to the shield and inner conductor, so the reflections caused by the abrupt change will result in nearly identical currents, in both phase and amplitude, on both conductors. If the scope's input circuit is properly designed and laid out, an analogous effect will take place when the reflections encounter the 1 meg ohm input impedance. So any increase in visible noise given these circumstances should be small. In contrast, if for example, the coax ends in 1 or 2 inch pigtails, the increase in visible noise is quite a bit larger. There is still an abrupt change in characteristic impedance, but in addition, there is also an impedance imbalance because of the rather raw unequal imperfections at the end of the cable. I believe in this case reflections multiply the effect of the impedance imbalance at the end of the cable. Outside interference probably also plays a role in this scenario. Also, in my opinion, any equipment attached to the pigtails will also have an influence on the impedance imbalance and/or the characteristic impedance at the end of the cable.
I also believe that using proper termination, which we know produces the illusion that the cable has an infinite length and as a result minimizes reflections, will also minimize the multiplying effect that I believe an improperly terminated cable will have on impedance imbalances at its ends.
Anyway, that's the way I view the effect of proper termination as it relates to the increase in visible noise. But again, the main reason to use proper termination is to minimize distortion of the differential signals that traverse the cable on their way to another device.
How much of the above reflects the way that things actually work? I'm not sure, it sounds good to me at the moment. The only thing that I know for sure is that the visible noise is reduced when proper termination is used.