Highland T560, Technical Manual, Page: 38 / 49

38 T560MANK5.DOCX
7. Jitter Notes
Jitter is defined as the 1-sigma standard deviation of delay. It is the shot-to-shot time
uncertainty from the external trigger to any output's rising or falling edge, or the uncertainty
between edges of channel outputs. Jitter is measured in RMS picoseconds. Visual peak-to-
peak jitter is roughly 5 times that of RMS.
"Jitter" is usually accepted to indicate time variance as observed over an interval of 0.1
seconds, with the term "wander" used to describe slower changes of delay. Wander thus
encompasses changes in delay driven by temperature changes and other slow effects. The
T560 jitter specs are valid for observation periods up to 10 seconds in the absence of
radical temperature changes. Note that coaxial cable propagation delay can change
considerably with temperature and can contribute to observed timing variance.
Uncorrelated jitters add trigonometrically, as the square root of the sum of the squares of all
jitter contributors.
Jitter can be difficult to measure. The trigger input to the T560 must be clean and fast (< 2
ns risetime) and the measuring instrument must have a jitter noise floor well below that of
the T560. Most oscilloscopes and counters are not capable of resolving T560 jitter
performance, especially so for longer delays. For example, a Tektronix 11801C sampling
oscilloscope (or the newer DSA8200 without the optional phaselock module) has a short-
delay jitter well below that of the T560, but has added jitter on the order of 10
microseconds per second of delay, whereas the T560 starts with a greater basic jitter but
typically adds about 4 ns of jitter per second of delay.
For lowest jitter from an external trigger, the T560 trigger level should be set to the steepest
part of the input edge, typically 1/3 to 1/2 of the peak amplitude.
Jitter is a function of the generated time delays. Very short delays have a baseline jitter that
depends on fundamental triggered-oscillator phase noise. After about 500 ns, the DSP
stabilization loop becomes active and disciplines the triggered oscillator, limiting its jitter
accumulation.
Long delays, in the milliseconds range, become dominated by the phase noise of the
internal crystal oscillator, typically about 4 ns per second of delay. Long-delay effects are
zero relative to a user-provided 10 MHz reference clock.
Jitter between successive triggers, referred to as "period jitter", depends on the quality of
the trigger source. The internal DDS trigger synthesizer has jitter typically about one part in
20,000 of the trigger period. DDS jitter is best if its frequency is in the 2-10 MHz range,
where the period jitter, measured at a channel output, is typically about 25 ps RMS. For
lowest DDS jitter at lower rates, keep the DDS synthesizer frequency in this range and use
a trigger divisor to get lower trigger rates.
Dividing down the internal 80 MHz clock results in period jitter in the neighborhood of 40 ps
RMS, until millisecond-range periods when crystal oscillator phase noise again becomes
important.