filters with measurement data in the time- and frequency- domain.
Unfortunately, the filter data is performed with an entirely resistive
load and therefore the data looks very good. However, in practice
an entirely resistive circuit seldom exists. Therefore, the filters
must be evaluated when installed in a practical circuit. It is then
observed that the performance of the filters is not as promised in
the catalog. This shall be demonstrated with a series of illustrative
examples which are measured in circuits of the 74 ACT family. The
gates are always operated with a 5MHz frequency.
Figure 1 shows the time and frequency domain outputs of such a
gate which is mounted on a printed circuit and is not loaded. The
frequency spectrum is measurable to 1,000MHz. In fact, it extends
even above 1,000MHz, but for comparison purposes all measu-
rements are scaled only 1,000MHz. In the time domain relatively
strong over and under shoot and fast risetimes are observable.
This signal is very poor relative to the EMC characteristics. The
excessive bandwidth permits radiation to take place on relatively
small printed circuit boards. When this signal is conducted to
other parts, it is especially important to limit the spectrum to avoid
excessive shielding structures.
As a first measure to limit the spectrum, a resistor is recommended
between the gate output and the conductor connection. The con-
ductor is simulated by an individual gate input to obtain a realistic
circuit. The connection and the conductor length must correspond
to the actual relationship to make the measurements of signal
line filter evaluation meaningful. The effectiveness of line filters is
strongly influenced by their termination.
37
Änderungen vorbehalten / Subject to change without notice