Trigger And Timebase Errors - Agilent Technologies 3458A User Manual

Trigger and
Timebase Errors
Figure 62. Analog-to-dig
ital converters that
exhibit non-linearity
errors cause spurious
responses that
averaging will not
remove. The 3458A is
linear to 16 bits at
100,000 readings/s.
The 3458A offers two input paths. The differences are that the direct ADC path
(DCV) offers up to 160 kHz bandwidth up to a sampling rate of 100,000 samples
per second; the track-and- hold path offers 12 MHz bandwidth at a sampling rate
of 50,000 readings per second. Both paths exhibit single pole roll-off; both are
nominally three dB down (half power) at the bandwidth point. Hence, two errors
can creep into your measurements: aliasing and amplitude roll-off. In the
track-and-hold path aliasing can be eliminated by increasing the effective
sampling rate up to 100 MSamples/s and the track-and-hold circuit can be
characterized for amplitude roll-off over the band of interest to compensate for
the roll-off. In the case of the DCV path, the only real solution to aliasing is to
supply a low pass analog filter. See Figure 63.
Figure 63. Amplitude
roll-off of the 3458A
Multimeter for its two
different measurement
paths.
Finally, the accuracy of the measurement itself, although not often discussed with
digitizers, is related to the reference accuracy of the 3458A. For static and dynamic
measurements, the absolute accuracy actually exceeds the dmm's resolution. And,
in terms of long term drift, the absolute error is less than 7 ppm per year.
The timebase, a precision temperature compensated quartz crystal, has its drift
and jitter which will affect the amplitude measurement of the input signal. But,
these tend to be very small - less than 50 ps. Hence, the clock accuracy and jitter
do not really affect the measurement within the measurement bandwidth of the
3458A. The timebase jitter error is not cumulative; therefore each sample point
has only its own jitter error and not the combined jitters of previous sample points.
The effects of all the time axis errors are shown in Figure 64.
Appendix E High Resolution Digitizing With the 3458A 361