Hi Jinwen,

I'm not sure I understand the circumstances during taking of the signals - are the first and second one taken without the gas cell in place? Obviously, the third one was taken with a cell.

However generally, you have to take into account the different impedances at the tips: The impedance of the "hot" tip depends on the probe used, e. g. 1 MOhm parallel to 20 pF to the reference tip.  The reference tip has the same impedance connected to the hot tip, of course, but additionally 37 pF to ground. I'm not sure about the details of the probe and of the input circuit; it is even possible that an additional current flows at the outside of the coax cable into these 37 pF.

In order to check the latter, you could clamp ferrite cores around the cable of the probe, and repeat the measurements shown in Fig. 2 and Fig. 3.  I'm afraid there is no way to avoid the 37 pF between the reference tip and GND with this oscilloscope.

You could use a differential probe but I guess most of them are designed for lower voltages.

The answer to your general question: If you are able to measure other parameters like supplied power, signal beyond the point of measurement, etc., and you detect changes of these quantities when connecting the probe you can be quite sure that your measurement modifies the original signal.

Otherwise, the only (not very satisfactory) way I can think of is to use different probes or even different scopes, and to try the above mentioned ferrite clamps. If the results differ, sometimes it might be hard to decide which one is nearer to the truth. Certainly, this is no case for a majority decision :-) . Usually, I would believe those results which come closest to expectation based on theory. But even this can be misleading.

May be this is a trivia , the BNC/SMA connector (body) causes ground loop check the datasheet or if it is floating !!!

Thank you for those helpful suggestions @ Joerg @ Victor. In Fig.1 and Fig.2, the circuit contained the cell, but the impedance was not matched well so as to fail the ignition. The output voltage was also detected by the oscilloscope with the yellow line in Fig.2~Fig.4.

I believe that the additional 37 pF to ground ,as pointed by Joerg, is the main reason, since it was the only thing changed.

Dear Joerg, what does a ferrite cores used to? Increase the capacitance between the reference tip and ground to check the path flow? I will try it.

Dear Jinwen, the signal inside the coax cable is accompanied by currents on the inner conductor and on the inside of the outer conductor, and these currents are of the same amount but flow into opposite directions; so, they sum up to zero, and are not accompanied by a magnetic field on the outside of the cable. Consequently, the signal has no effect on the ferrite core, and the ferrite core has no effect on the signal.

Additionally, there might be a current on the outside of the outer conductor; this is easily possible if the shield of the cable is grounded on both ends, but at high frequencies a capacitance at each of both ends is enough to provide a current path. Since this current has no counterpart on the inside of the cable, it is accompanied by an electromagnetic field; consequently, the current "sees" capacitance and inductance on the cable which adds up to a certain impedance. By adding ferrite cores, the inductance is increased, and so is the impedance, resulting in less current on the outside of the cable. Additionally, the ferrite material introduces some loss, thus part of the electromagnetic energy is turned into heat.