Analyzing CSS with BERT
How can you go about using BERT with CSS? A test scenario might include a test setup whereby CSS is fed with multiple channels of a PRN23 (or other) pattern and the output of the CSS analyzed for any issues using another BERT instance. Let’s take a look at how these setups might work.
Configure CSS
First we need to get CSS set up for its output to be available. Start by clicking More Output Options…
Enable Alternate Pull Output and optional change the Stream UNC Suffix.
Commit the change and we’ve now made the output accessible to be read.
Configure BERT Analyzer
After creating or selecting a BERT, enable the Receive channel to read the CSS output.
We’ll stick with the default PRN23 pattern, but you should select the pattern type relevant to your case.
Start the BERT instance. Until you start feeding the pattern data to the Sources, you won’t see any values changing.
Latency
Once the data start flowing, you’ll start to see CSS report bits flowing. A caveat when starting BERT data on the sources can be that the pattern generators start offset from one another, as might happen in a manually started scenario. You might see something like the following and wonder why only the first Source is meaningfully participating.
In this case you can see a pretty obvious difference in bit counts, so there’s clearly a fairly large offset in the patterns, even though they may all be perfect. You may need to adjust the CSS configuration to account for the offset in the data, here’s how.
Click Configure… to open the configuration dialog.
You’ll need to adjust the Source Time Offsets and Latency Tolerance value. Adjust this up a ways to try to span the gap between sources.
After adjusting and restarting, assuming the value can buffer and span the offsets, you’ll start to see the CSS working after an initial short period of buffering.
You’ll need to restart the BERT after making changes to the CSS as the BERT will have stopped after the CSS restart. Once started you should start to see results.
