From the last two traces we can see that your cables are unterminated, and the driver cannot supply the necessary current to drive the cable. The 6ns and 12ns numbers aren't helpful and should be ignored.
The first 2 measurements probed with the setting on 10x. The BNC cables lack such a mode so would’nt those pull more current?
You need to look up the theory of digital signals and transmission lines.
Your "high impedance *10" scope probe's input impedance is lower than a "low impedance Z0 resistive divider" scope probe. Work out the impedance of a 15pF capacitor at 100MHz.
Some scopes have a 50 ohm input capability would that be in this artificial situation better? Is it possible to add a normal 50 ohm resistor, or does it need to be combined with a capacitor/inductor, or is a good old fashioned bnc terminator on a T connector a possibility?
A proper 50ohm input is ideal, but they are only found in high end scopes (and Tek 485s). In other cases the 50ohms is simply a resistor added in parallel with the scope's high impedance input (i.e. 1Mohm//20pF).
Any L/C would only be 50ohms at one frequency; at all other frequencies there would be a mismatch.
The BNC terminator plus T connector will be 50ohms//20pF, where 20pF is the scope's input capacitance. It isn't too bad at these risetimes, but becomes more problematic at higher speeds.
What would be the verdict on the probing method which started this thread? Any good?
Look at your second graph.
Okay my judgment would than be, if the transitions are faster than 3V / 50 nS ringing becomes more and more apparent on high impediance loads.
The problem is you
don't have a high impedance load: the circuit "sees" a 15pF load - the probe tip capacitance.
If we use the rule-of-thumb BW=
0.35/
tr, then with t
r=10ns BW=35MHz. I chose 10ns because that is typical of fast logic in the 1970s
Today's jellybean logic is <1ns
Now consider the impedance of a capacitor, Z=
1/
2*pi*f*C, so 15pF ->
300ohms, and it gets lower at higher frequencies (e.g. 80ohms at 100MHz).
The "ringing" frequency will be f=
1/
2*pi*sqrt(LC), where C is the tip capacitance and L is the inductance of the ground connection, typically 90MHz for a 150mm ground lead.
Otherwise it will do just fine. I came to these values because at the rise an fall times of my sig. gen. it was stil (only) a bit noticable. In the analog domain thats is very usefull I guess.
In faster cases, normal probing with the clip lead is not advisable as well.
Whether or not it will "do fine" depends on the signal and output that you are probing. What might be fine for an AWG might not be "fine" for a logic signal.