Table of Contents
Advertisement
Jitter Filter
The jitter filter is a band-pass filter that is applied to the TIE-vs.-time data measured on the signal
under test. The filtered waveform of the trend of TIE over time is displayed in the waveform grid
when Filtered Jitter is selected. The filter is implemented as an FIR (Finite Impulse Response)
digital filter with a maximum tap length of 2000. Not all high-pass filters are realizable, owing to
the tap length restriction or the data record size or both. The instrument will apply a low-pass filter
if the band-pass filter cannot be realized. The filter is applied only to the filtered jitter trace and
does not affect the other jitter measurements: Tj, Rj, Dj etc. The peak-to-peak and rms value of
the filtered jitter trace will be displayed below the trace.
The clock recovery PLL results in a jitter spectrum with a high-pass characteristic determined by
the type and loop bandwidth of the PLL (see the Theory section). The combination of the low-
pass jitter filter and the PLL cutoff will result in a band-pass filter for cases where the jitter filter
function cannot realize the high-pass filter selected.
1. Touch the Filtered Jitter button.
2. Touch inside the Lower Limit field and enter a value, using the pop-up keypad. To realize a
low pass-filter, select Set to min. in the pop-up keypad. In this case the lower frequency
cutoff will be set by the PLL cutoff frequency.
3. Touch inside the Upper Limit field and enter a value, using the pop-up keypad.
TIE Histogram
A TIE histogram is often useful to display the raw measured histogram of the measured jitter
values because it can give clues as to the sources of deterministic jitter and it also gives a good
indication of whether the extrapolated total jitter value is likely to be accurate. The bins in the
histogram at the extreme tails (the last 5 to 10 bins on each side) are used to determine the best-
fit curve (see the Theory section for more details). Generally, if the histogram is smooth between
the two tails, that is, there are no deep troughs in the shape, the curve fitting and extrapolation
will give an accurate and stable Tj value.
There are some cases where very large amounts of ISI or periodic jitter cause deep troughs in
the shape of the histogram. In these cases, a very large population is required in the histogram to
ensure a sufficient population in the tail regions to achieve a reliable fit. Generally 1.5 to 2 million
measurements in the histogram are sufficient to resolve the tails, but more data is always helpful.
A histogram that appears smooth is a good indicator that the total jitter will be accurate.
SDA-OM-E Rev H
SDA Operator's Manual
337
Advertisement
Table of Contents
