Fluke DDS Function Generator 271, User Manual

Page: 120 / 132

271
Users Manual
C-2
t = 1.25e
-4
(125 µs) for ramp times up to 200 ms.
Note 2: Phase Error and Phase Jitter
Main Out to Aux Out
The level of phase error or jitter depends on frequency and function. Below 30 kHz the
two outputs are generated in the same way and any jitter is due only to delays and/or
phase shifts in the paths taken by the signals to reach the BNC outputs. As these delays or
phase shifts are constant the jitter results only from the 27 MHz crystal oscillator and is
typically less than -110 dBc/Hz at 10 kHz offset from the carrier.
Above 30 kHz (and becoming more significant as the frequency rises) three things must
be taken into account:
1. The phase is fixed at 0 ° unless the
AUX=
field on the options menu is set to
LOFRQ
.
2. The block diagram and explanation in appendix B show the changes in paths of the
signals as the user makes changes to the various settings. The path changes cause
variations in the absolute value of the phase angle and path variations with frequency
and waveform cause absolute and/or jitter variation in phase. For example with a sine
wave at 1 MHz the delay through the filter is much longer than that at 50 kHz so the
phase error will be greater at 1 MHz than at 50 kHz if LF AUX is used. Using HF
AUX the errors will be much smaller (but fixed at 0 °).
3. Above 30 kHz the DDS waveform generation begins to down-sample the data stored
in RAM, which means that not every point is played back on every cycle. Consider a
one cycle square wave in RAM, which consists of 512 points low followed by 512
points high. At low frequency all points are sampled for every cycle of the waveform
but as the frequency rises above 30 kHz some points get missed. If either of the two
points which define the edge of the square wave are missed then the edge will shift in
time. However, due to the apparently random sampling phase, which changes every
cycle, the points are not always missed in the same way so the edge appears to move
on a cycle by cycle basis, i.e. it jitters. The total jitter from one cycle to the next is
never more than one 27 MHz clock period (36 ns) which at 50 kHz represents 0.65 °
of phase jitter but at 5 MHz the jitter is 65 °. This may be demonstrated by setting
SQWAVE GEN=LO FREQ
on the options menu and observing a square wave at the
two frequencies on an oscilloscope. The above will apply to the
AUX OUT
square
wave above 30 kHz when
AUX=LOFRQ
is selected on the options menu. If this is
combined with a main output square wave generated with
SQWAVEGEN=AUTO
on
the options menu then the phase jitter will become very large at high frequencies.
The above gives a brief outline of phase variation and jitter under different conditions,
but it should be remembered that the best phase accuracy is only 360 °/1024 or 0.35 °,
there being at best only 1024 points each cycle of the waveform.
Between Phase-locked Generators
When phase locking two generators all the above factors must be considered in case they
have an effect on the result. In general, however, when two generators are used to
generate the same waveform at the same frequency with some phase shift and the outputs
are taken from the main outputs of the two generators, then the phase accuracy will be
better than 0.5 ° and the jitter will be determined by the crystal oscillator in the master
generator. Take care to keep the clock and synchronization cables short and of good
quality.
1.888.475.5235 info@Fluke-Direct.com
Fluke-Direct
.com