the scope may degrade this performance
by virtue of its - 3 db BW adding to the
scope’s - 3 db bandwidth, therefore
intensifying the loss at the limits of their
individual stated bandwidths. Some
manufacturers design probes to match
their line of scopes to preserve their rated
bandwidth or even extend it slightly. Now
let’s look at how a probe gets its rated
bandwidth specs. This is done in a lab
environment using sine wave generators
with a 50 ohm source impedance
immediately terminated with a 50 ohm
load that has a BNC jack built into it. The
ground lead and clip tip (witch’s hat) are
not used in this setup, but rather the tip
has a special tip-to-BNC adapter that
matches up to the termination jack. This
effectively removes most problems
associated with ground leads (as will be
discussed later). Also, the probe tip is
looking into an equivalent 25 ohm
impedance by virtue of the generator
source impedance ( 50) in parallel with the termination
impedance ( 50). This constitutes an almost ideal condition
for measurement, but rarely is the normal setup for probing
in the real world.
FIGURE 1A
Figure 1A shows a typical 10x passive probe connected
to the scope’s vertical input and the standard 10 megohm
input resistance that it presents to all circuits probed. It also
presents a capacitive load on those circuits. This is the series
combination of C and 9C. The typical capacitance the
circuits under test will see at the tip will range from 10 to 14
pF, so I will use 12 pF as a middle of the road spec. Scopes
have a certain amount of inerrant input capacitance, along
with the probe’s cable, etc., connected to it. To obtain a flat
frequency response in a resistive divider network, the time
constant of the probe’s attenuator network (9R in parallel
with C) must exactly equal the time constant of the probe
cable and scope input values (R paralleled by 9C). Since the
probe’s tips are constructed with a fixed value of resistance
and tightly controlled parasitic capacitance C, the RC values
on the scope end must be adjusted to exactly match this.
This is done with a capacitive trimmer mounted in a
termination box which is attached to the cable’s BNC
connector for the scope.
As mentioned, 9C is the combination of scope, cable,
and trimmer capacitance. In the circuit shown, R probe = 9
megohm; C probe = 13 pF, and R scope = 1 megohm.
Therefore, 9C must equal 9x13 or 117 pF to match the two
time constants involved. With the aid of the scope’s
calibrator, 9C is adjusted to present the traditional “perfect
square wave” on the display.
With an input signal of DC to very low frequency sine
waves (<100 Hz), the probe’s input impedance is basically
all resistive and presents a 10 megohm load on the circuit
being probed. At approximately 1,500 Hz, the probe’s input
R and Xc will be equal in magnitude, and probe input Z will
be reduced to about 70% of that 10 megohm value. Input Z
will continue to reduce with increasing frequency injected
FIGURE 1B
into the probe tip. However, due to the matched RC time
constants in the probe’s divider network, the 10:1 division
ratio will be maintained. The probe will continue to give
good performance up to 10 MHz and beyond, but as it
enters the VHF range (30-300 MHz), problems start to
occur. Referring to Figure 1B, the probe appears to the
circuit as shown, when probing these higher frequencies. In
this region, the probe’s resistance no longer enters the
picture as the magnitude of impedance is almost totally
capacitive (when using a very short direct ground
connection). Also, the XL of the clip tip begins to be of
some significance, but even more so is the increasing XL of
the ground lead attached to the probe. These two
components (Cin and L ground) will form a series resonant
circuit at some point which can really mess up
measurements. Even the best of probes have a resonant
point regardless of the grounding system used.
High-end probe designers are constantly trying to move
this resonant point up higher in frequency to push it up and
well outside of the passband of the equipment being used.
Small values of distributed ‘L’ and ‘C’ may be inserted at key
points to flatten out response and raise the input
impedance, but even with the best efforts, this will only
show a certain degree of improvement.
So, in light of these facts, what do we end up with
here? It is evident that in high frequency circuits, the probes
Xc is decreasing and loading the test circuit more. At the
same time, the XL of the ground lead is increasing and, at
some point, will resonate with Cin. Also, the cable ground
potential at the scope end is not the same potential as the
probed circuit ground at the other end due to the ground
lead’s (typically 5-6”) impedance. The scope faithfully
displays its input signal, however that is only the signal that
appears across Cin (in Figure 1B). The probed circuit is
applying its signal across both Cin and Lgrd, thus producing
a volt divider action (along with some other nasty things) at
the actual scope input. At some point, the signal resonates
November 2010 45
