How to Eliminate Processing Drops

Processing drops are one of the many types of "drops" that are registered on a Gocator Dashboard page. These drops indicate that an image (either profile or surface depending on the scan mode) has been dropped due to an excessive processing load. In many cases, this is caused by a failure of the sensor to process images as fast as (or faster than) they are gathered. 

You can predict that processing drops may start to occur in your Gocator system when your Processing Latency Peak (also on the Dashboard page) continues to slowly increase over time rather than achieving a steady state. 

When you start to experience Processing Drops, you will need to make a change in your system to eliminate them. Try one, or several, of the following improvement tactics. 

1 - Slow down your production line speed. This will give the sensor a greater amount of time between surface generation times to process the previous image

2 - Increase the X_Spacing, the Encoder_Spacing, or the Z Sub-Sampling. With greater values in 1 or more of these parameters, the generated surface will have less data in it to process, and therefore, the processing time will decrease

3 - Decrease the Surface Max Length. In the case of discrete widget manufacturing, you may have extra margin on the leading or trailing end of a surface scan. Removing or reducing that margin can shorten the overall length of the surface. This makes the surface to process smaller with less lines in it. This method will not necessarily work in the case of continuous extrusion inspection as each individual image would shorten, but you would be gathering more images per second back to back.

4 - Reduce/optimize the Field of View and Measurement Range to exactly what you need and no more. This will decrease the amount of time it takes for the imaging chip to "read out", potentially improving the Max Frame Rate capability as well. 

5 - Reduce your Measurement toolset. If there is another way to get the same result with less tools, consider it as a lower number of overall tools will apply a lesser processing load to the sensor's processor.

6 - Remove unneeded measurement toolset Measurement, Feature, and Data type outputs. Some tools are added to the toolset with 1 or more outputs enabled in each of these output types. Check to make sure that only the outputs that are needed as Anchors, input Features or data Streams, or overall Output values to a control system or PLC are enabled. 

7 - Accelerate the sensor system using a GoMax hardware accelerator, GoAccelerator PC software accelerator, or GoSDK application software accelerator. Accelerating a sensor leverages the processing power of the GoMax (with its onboard CPUs, GPUs, and optimized OS environment) or your laptop or desktop to perform the image processing. In many cases, even your engineering laptop may be more powerful than a singular sensor's onboard processing capabilities. This improvement, while it may take more effort or cost more in hardware, may allow you to continue scanning with all the same process and image parameters while decreasing the per-image processing latency.

8 - Check for proper sensor grounding using the methods detailed in the Gocator Grounding Guide APPNOTE from downloads.lmi3d.com

In some cases, multiple of these techniques may be needed to eliminate processing drops in your system.