Using simple wires to measure signals with the oscilloscope would result in unreadable plots on the scope, the main reason being the noise coupled onto the “probe” itself.
The first line of defense against that would be to use a coaxial cable as a probe, which would prevent external noise coupling. Unfortunately, an unwanted deterioration of the measured signal is achieved in this case, due to the capacitive loading that such a piece of cable adds to the signal. An equivalent schematic of an IC to IC signal is illustrated below:
The non-zero output resistance of the signal source is modeled, as well as the input capacitance of the load. As it can be observed in the plot below, for a 2MHz square waveforms, both the signal at the output of the source and at the input of the load are practically identical:
Things are different however, if you connect a 1x probe of an oscilloscope (which is basically a piece of coaxial cable) in an attempt to measure the signal. Due to the relatively large capacitance of the coaxial cable, an additional capacitive load is placed on the signal line. Assuming a 1 meter long probe, and taking into account that the capacitance of a coaxial cable is about 90pF/meter, the following equivalent schematic can show us the degradation of the signal (both at the load and measured on the scope):
As it can be seen in the plot below, the non-zero output resistance of the signal source, together with the additional capacitance of the coaxial cable and that of the oscilloscope itself, form a severe RC filter which seriously deteriorates the signal on the line:
Things become even worse if higher frequency signals are to be measured. For instance, trying to measure a 10MHz signal (so just 5 times higher frequency than in the previous example) in this manner will lead to seeing almost a triangle waveform on the oscilloscope:
Attempts to reduce this effect have been directed towards decreasing the capacitance of the coaxial cable. Hence, actual oscilloscope probes are actually made of special coaxial cable, with very thin inner conductor, so that the equivalent overlapping surface between the inner conductor and the shield itself is decreased. In addition also the distance between the inner conductor and the shield itself increases, leading to smaller parasitic capacitance of the probe:
This improvement alone does make things better, but it is not enough to make signals in the 10MHz range measurable. Compensated 10x probes must be used in order to accomplish reasonable measurements in this frequency.
The idea behind the 10x probe is to place a 9MegOhm series resistance in series with the tip of the probe. This configuration presents much higher impedance to the measured circuit, severely reducing the unwanted effects of the oscilloscope on the circuit itself. The entire effect of the oscilloscope input capacitance and that of the coaxial cable are virtually eliminated. Of course, the signal that reaches the oscilloscope itself is divided by a factor of 10, due to the resistive divider formed by this series resistor and the 1MegOhm input resistance of the capacitor; the scope itself however is capable of multiplying the actual measured signal by the same factor in order to display it correctly.
All these are true, however, for very low frequencies only. For higher frequencies, the 9MegOhm resistor together with the capacitance of the cable and that of the oscilloscope itself forms a real RC filter which literally kills off everything but the DC component of the measured signal. In order to prevent that, older designs of the 10x probes used to bypass the 9MegOhm series resistor with an adjustable capacitor (Cvar in the schematic below). This capacitor, together with the capacitance of the cable and that of the input of the scope would form a capacitive divider with the same 10:1.
As it may be seen from the plots below, the measurement of a 10MHz signal is greatly improved when using such a probe (when compared to the previous situation when the same signal was taken to the input of the scope with a 1x probe):
The actual signal reaching the scope itself is illustrated in blue in the plot above, and it may be seen that it is attenuated by a factor of 10. The oscilloscope can multiply this (either by internal amplification or through internal digital processing of the acquired samples) and display the correct result that would be identical to the green trace (the one the load of the signal sees too).
It is necessary for the bypass capacitor to be adjustable to allow for the matching of the probe to the various oscilloscope inputs. Sometimes, even for the same multichannel oscilloscopes, the input capacitance and resistance can be slightly different from channel to channel, so moving a probe from one channel to the other would involve slight adjustment of the bypass capacitor in order to reach the necessary capacitive divider ratio similar to that of the resistive divider. This adjustment is done by probing a known square wave signal and adjusting the Cvar capacitance until the signal displayed on the oscilloscope represents the expected shape of the signal. This is the process known as probe compensation.
Newer 10X probe designs have replaced the adjustable capacitor bypassing the 9MegOhm series resistor with a fixed one. This circuit is laser trimmed and has relatively low tolerances. For these probes, the compensation is done by adjusting a different variable capacitor which is placed in parallel to the oscilloscope input (Cvar in the schematic below) which is physically located at the end of the probe which connects to the scope. The purpose is the same: to obtain identical capacitive divider ratio to that of the resistive divider.
The advantage of such a construction is that you can still compensate your probe even if the tip of the probe is not accessible. For instance if you measure a DUT placed in a climatic chamber which keeps an inside temperature of +85Celsius, this ambient temperature will significantly alter the bypass capacitor of the 9MegOhm resistor. However, the tip of the probe being in the oven, the user cannot adjust that capacitor even if the probe would allow for that. However, the user can still adjust the variable capacitor at the other end of the probe so that it would match the new value of the bypass capacitor. As it can be seen from the plots below, the results are the same as with the previous type of 10x probe: