We’ve often seen people put together test stations to collect data, but have no idea what to do with those test results. It’s important to think about what sort of information you want to derive from the data. Do you want to look at trends for power out on a unit, or missed or corrupt bits in gigabytes of digital data? Make sure you consider your data analysis goals before you finalize your test so that the data can be collected in a manner that facilitates the type of analysis you want to perform.
Not knowing what you are doing with your data after you collect it can cause multiple issues. If you don’t know what information you are targeting you may not construct your test in a way that collects that data. Determine the key metrics before you write the test to make sure that the information isn’t missed.
Organize your data in a scheme that makes it easy to retrieve later on. If you don’t have identifiers on the type of data being collected it could get mixed in with the data from unrelated tests. Using a common prefix when storing your data can help with this. For example, you may have two tests that record BERT values. If you name the test data BERT test for both it will be impossible to differentiate between the two. Instead you may want to add prefixes or unique identifiers. For example, 1.1.1 Uplink BERT = [measured BERT] and 2.1.1 Downlink BERT = [measured BERT].
If your goal is to improve manufacturing efficiency by identifying your most frequent failures, you may want to create a pareto chart from your data.
This charting can only be done well if you store the data in a way that is specific enough to identify the failures. For example, saying serial communications failure isn’t as useful as providing the specifics of the failure such as unable to communicate due to no response, frame error, incorrect response, response timeout, etc. Having this level of resolution can really help when trying to slay these systematic issues with your test.
Using your data to drive manufacturing efficiency is important. We’ve worked with customers to reduce their production costs as much as 30% across multiple product lines. Check out this report for more information on how we did this.