We have three Basler USB3 cameras connected directly (not through a hub) to the USB3 ports of a high-end workstation (Intel Xeon Gold 6130 2.1GHz, 3.7GHz Turbo, 16C processor, 64GB RAM, Nvidia Quadro RTX 5000, 16 GB GPU). LabVIEW 2019 and Vision acquisition software 2019 along with Vision Development Module 2019 have been installed on the computer. The cameras configuration has been modified in the MAX, primarily to change the resolution to 1920 x 640 and the exposure time to 1000. note that these changes have been made in MAX once during the camera installation and saved to the camera files. These settings are not accessed or modified during the runtime application. The runtime application uses a loop to continuously grab images from the three cameras. This module is identical to the HL Grab.vi example in the LabVIEW examples which shows how to grab continuously from a single camera. In order to grab from three cameras, we just used a for loop to grab from three cameras. Everything seems to work fine until one or more cameras randomly freezes, most likely due to a timeout error, and never recovers from it. This happens randomly in the sense that sometimes it occurs within a few minutes and some other time after a few hours. Note that these cameras are required grab continuously for at least eight hours without any interruption. When the camera freeze issue occurs, the only way to recover from it is to restart the computer and restart the application, which is undesirable. On the NI forum, we found a discussion which seems to be very similar to our issue here: https://forums.ni.com/t5/Machine-Vision/IMAQdx-Timeout-0XBFF6901B-after-running-camera-acquisition-f... . We have tried almost everything possible including changing the USB3 cables, USB3 ports, deleting camera icd files, cleaning the cameras and so on, however, the issue seems to reappear randomly every now and then. Please advise any resolution for this issue. Also, is there a way to handle the exception programmatically with something like the following:
Pseudo Code:
if (frame_acquisition_time > 3000 or frame_corrupted)
{
acquire a new frame for that specific camera;
continue;
}
What appears to be causing the issue is one or two corrupted frames in thousands of frames and the above approach might work. Does LabVIEW do something like this already? If not, is there a way to code this into the acquisition module?
