Troubleshooting

This documentation is intended to list known/common issues that may happen while configuring the MVS system.

3rd Party Applications

IPM

  • Cognex Deep Learning 1.0.1
  • Cognex Vision Pro 9.7

MVS Server

  • Kepware v6 (if using PLC Mode)

Fixed Camera

  • Basler Pylon 6.3 (Basler)
  • Vimba (Allied Vision)

System Logs

The MVS Application server will save all logs to a logs folder in the same directory as the executable. These files will contain all errors raised in the application.

To enable more verbose logging the application can be started with the command line argument --debug. This will log all PLC I/O along with additional information about the current state of the application.

Server

Offline Activation Response Does Not Activate

It has been observed that the offline activation will fail if the application has been restarted between generating the request file and uploading the response. Generate a new activation request and do not restart the application during the process. This will require support from AIS as the license may no longer be able to activate.

Can't connect to the OPC server

KEPServer is either misconfigured or KEPServer is unable to connect to the PLC

See PLC Instances for further details.

Mobile App

Installing the Mobile App Fails

Sometimes android will not allow the app installation to complete if a older version of the app already exists. Try uninstalling the old version by navigating to device settings -> Apps -> MVS Remote Camera -> Uninstall and attempt the installation again.

Long Image Processing Times (Mobile) / Poor Network Connections

Most workspaces can process the image in under 500ms, but sometimes when acquiring a image it may take several seconds for a response to get back to the device. This is due to a poor network connection. The images acquired by the app at a normal resolution of 1600x1200 will be about 1-2MB in size by default. Assuming your wifi provides a 10MBit connection to the device, a 1.5MB image will take 1.2 seconds to upload.

To speed up the transfer time the app provides control over the quality of the JPG image being uploaded. The JPG Quality setting can be found in the app settings page accessed from the login screen. Adjusting this setting to a value under 100 will increase the compression of the JPG prior to transferring to the server.

PLC mode push messages not prompting

Verify that the app and server are running compatible versions. App version 1.6.1 requires server version 1.6.0 to work with push messages.

PLC mode inspection not triggering

Ensure that the PLC scan rate is at a level that can be caught by KEPServerEX (our default scan rate for PLC outputs is 20 ms). We only read tag changes if they change. Trigger for example only works when the tag goes from low to high (the rising edge).

IPM

Crashing - Access Violation Exception

The IPM has been seen to hard crash to desktop with no exception logged and no error message. Troubleshooting has shown that is was due to a Access Violation that occurs within the Cognex Deep Learning library, and was likely due to a Cognex Deep Learning update that was recently installed on the PC. It would only crash when specific workspaces were loaded into the system.

The resolution is to ensure that the PC has the latest NVidia drivers installed.

Another possible resolution was to update the local dlls used for Cognex Deep Learning (ensure that the installed version of Cognex Deep Learning matches the version built into the IPM - Contact AISTG for details on this)

(Default allocator) failed to allocate array, not enough memeory on device ArrayAllocator Allocate

Restart the IPM. This generally occurs when a system does not have enough memory to run all the workspaces on the IPM and train new workspaces in the Deep Learning Studio. It may also occur if the Cognex license key is removed from the computer and put back.

Specified argument was out of the range of valid values. Parameter name: The Guid '...' can't be found in the list of workspace services. Is it loaded?

The IPM does not have enough GPU memory to properly allocate to all the workspaces on the machine. Check to make sure the green circle is lit over the workspace that was ran. If it is not lit, then the workspace is not loaded and either the IPM must be restarted or the workspace is broken. If the problem persists and the workspace can be loaded on its own (the workspace is not broken), then this indicates a driver or hardware issue with the graphics card.

Fixed Camera

Cameras do not appear in the list of Add New Camera

Ensure that the vendor supplied camera software (Pylon for Basler, Vimba for Allied Vision) can see the camera and make sure that static IPs are set for each camera.

Could not acquire, timeout

The camera took too long to acquire an image. Ensure that jumbo frames is enabled and that the network adapter has been set to allow large packets. Ensure that each camera is on its own subnet as they behave strangely when they can talk to other devices besides what they are supplying frames to.

Could not connect to camera after it was connected

The camera configuration changed or the camera hardware broke. Reconfigure the camera using the vendor supplied software and restart the computer and application. If the problem persists, delete the camera from the application and add it again making sure to unregister the old camera from the server (see here).

The camera is open in another application

Restart the fixed camera app after ensuring that the vendor supplied camera software is not running. If the error persists, restart the computer.

Error while setting the hardware camera settings for camera ...

Caused when the camera configuration has invalid values. The invalid values will not be set on the camera, but all other settings in the configuration will be, so sometimes it does not affect performance.

Camera acquisition is slow.

Ensure that the camera has jumbo frames enabled and that it is connected to its own NIC.