Setting up metal detectors can be difficult, but Sentinel Autolearn makes things simple, fast, and accurate. Simply pass your products through the Thermo Scientific Sentinel Multiscan Metal Detector and check the Autolearn dashboard. You need not be concerned with the underlying details in most cases. Most importantly, you are up and running with a configuration that both detects your metal sizes/types and does not false reject. Sentinel Autolearn is up to 20 times faster than previous manual learn/adjust processes (including in Sentinel), and requires almost no knowledge of the underlying metal detection technology.
Prior to Autolearn, Sentinel required you to experiment with settings in many cases. Sentinel Autolearn captures signals from the product and contaminants, so settings are optimized automatically. Sentinel Autolearn also lets you know how detectable your metal samples are prior to doing any testing, saving a lot of time. The complex analysis and scoring system underlying Autolearn yields better results than the experienced human in the vast majority of cases, and does it much, much faster.
|Traditional Learn||Sentinel Autolearn|
|Uses only clean packs||Uses multiple clean AND contaminated packs|
|Detection capability must be manually tested||Shows scores and detectability, easy to "try again" with new metal sizes|
|Time consuming adjustment to optimize operation for the product AND the metal||Takes a few minutes, detection parameters set automatically for each frequency|
|RESULT: Only expert users can achieve an optimal result||RESULT: Easy to use by novices, set-up is determined to balance false rejection and metal detection|
Like the previous Sentinel learn function, the first step in Autolearn is the same. The next two Autolearn steps are new and yield an optimized set-up in a matter of minutes.
Step 1: Here you pass a minimum of five clean products through the metal detector. Just like legacy learn functions.
Step 2: Next you enter your types and sizes of metal contaminants and then pass five of each along with the product. This way the system “learns” what the contaminated product might look like. By knowing the clean AND contaminated signal levels the Sentinel can optimized set-up.
Step 3: In this step the system predicts how detectable the metals you want to find are. Note the needle and blurred regions below. These indicate average detection and the worst/best case detection based on the data captured.
At this point, if all metals are detectable you are done and can run production. False rejects versus detections are automatically balanced for you. There is no need to look at or adjust any settings. You do not need to understand detailed settings like attenuation, phase angle or in/our phase sensitivity.
If you would like to see how detectable smaller (Ferrous example above) or larger (Non-Ferrous example above) metals are you can pass more samples, capture that data and look at the new “scoreboard.” Once you get results you like you are done.
Time and accuracy. The traditional learn method may appear simple, but it can take many hours to a full day to optimize in more difficult applications. With Sentinel Autolearn, it takes minutes to enable the system to automatically determine the best settings. These settings are typically much better than a moderately experienced user could achieve. This is very significant in a production setting with time pressure, turnover and limited hands-on training.