User`s guide

Keep Out Enable - By turning this feature on, the user is able to define an exclusion region for Tail Fit. When a keep
out region is defined, the Tailfit or Gaussian distributions will only be fit to the left and right of the keep out region.
To define a keep out region the Current View must be either on Left or Right Histogram. This feature can be useful
for capturing low probability events.
Keep Out Left Limit (ns) - A user definable time for the left side of the distribution for exclusion.
Keep Out Region - For defining a "keep out" region of the Right and Left histograms.
When a keep out region is defined, the Tailfit or Gaussian distributions will only be fit to the left and right of the
keep out region. To define a keep out region the Current View must be either on Left or Right Histogram. The keep
out region is useful because the mean position of the Gaussian tails is a loosely constrained variable and defining a
keep out region can reduce the time required to converge to a stable location.
Keep Out Right Limit (ns) - A user definable time for the right side of the distribution for exclusion.
Known Pattern with Marker Arm Violation - There are 3 potential arm violation problems in this tool:
The first is when the dataCOM software gives an error "Failed DDJ measure – Bad pattern or Signal". If the setup is
correct (see Known Pattern with Marker setup), this may be an arm violation. After the arm event, the tool measures
edge transitions (see Known Pattern with Marker Measurement Theory). If the tool gets many arm violations, the
measured edges are not in the expected position, which results in an error.
The second example is when there are only occasional violations during the 1-sigma measurement section of the tool.
The result here can be seen in the figure below. This will be viewable from the 1-sigma plot of this dataCOM tool.
Exam
p
le 3
Exam
p
le 2
The third situation is when the tool is in the DCD&ISI section of the measurement and it gets only one or two arm
violations. When this happens the DCD&ISI vs. Edge plot will look like the figure above.
As it is trying to measure a given edge of the pattern (see KPWM measurement technique) it gets many
measurements that are near the ideal position but it gets a couple that are 1 UI or 2 UI off (depending on the pattern).
This happens because the arm plus the user input delay (19-21ns) is very close to the first edge of the pattern. The
tool usually sees the first edge and then makes the correct measurement. But, due to jitter on the arm and/or the
data, during a couple of the measurements it doesn't see it and thus measures the next edge to the 5
th
edge after that.
This is the wrong transition and will be 1UI or more off.
© WAVECREST Corporation 2005 Glossary
321