418

have 2

8

= 256 discretisation levels. If themeasurement area of the oscilloscope is set at

100V per unit and the vertical graticule is divided into 10 units, the minimum resolution

of the input signal will be:

V9.3

256

div 10

div

V 100

=

⋅

Because of the limited resolution of a DSO, it is advisable always to use as much as

possible of the entire vertical monitor area in order tominimisemeasuring errors. If, for

example, an input signal of 75V is measured at a setting of

div

V 10

, a minimum

resolution of 0.4V will be achieved. Using a setting of

div

V 20

instead, the minimum

resolution will be 0.8V. The setting

div

V 10

should therefore be chosen in preference to

the setting

div

V 20

in this case, because of its better resolution.

Another point to remember when measuring with a DSO is its bandwidth. Bandwidth is

defined as the frequency at which a sinusoidal input signal is attenuated to 70.7% or -

3dB of the actual signal amplitude. This frequency is known as the cutoff point.

Accordingly, the bandwidth determines the essential capacity of the DSO tomeasure a

signal with a frequency of >0Hz. The capacity of the DSO to map the signal correctly

decreases as the frequency of the signal to be measured increases. The cutoff point

indicates the area of the DSO in which precise measurements can be made within the

specification. If the DSO does not have sufficient bandwidth, the following problems will

occur duringmeasurement:

•

high-frequency changes of the input signal will be recorded incorrectly;

•

the amplitude of the input signal will bedistorted;

•

details of the input signal will get lost;

•

the slopes of the input signal will be difficult to read.

A good rule of thumb is that the bandwidth of theDSO should be five times greater than

the highest relevant frequency component of the input signal. The measuring error in

this case will be ≤2%. It is vital to chose an oscilloscope with the correct bandwidth

whenmeasuring rapid signal changes such as those that occur whenmeasuring turn-on

and turn-off behaviour of IGBTs.

According to the Nyquist-Shannon theore

, a signal must be sampled at least twice

the frequency of its highest frequency component. This theorem assumes that the signal

to bemeasured is continuous and themeasuring instrument possesses infinite storage

depth. As storage depth in DSO is limited and the signal is often not continuous,

sampling at twice the frequency is not sufficient. As a guide value, aminimum sampling

of 10 referencepoints per slope can beassumed in practice.

2

This sampling theoremwas formulated byUnitedStatesmathematicianClaudeElwoodShannon (1916 - 2001),

based on thework of Swedish-Americanphysicist HarryNyquist (1889 - 1976) in1927.

SEO Version