RE: Andor3 IOC detects and connects to Marana, can get frames, but frames appear noisy/corrupted

[Mark Rivers via Tech-talk <tech-talk at aps.anl.gov>]

Which BitFlow card is this?  I have seen stripes on their Claxon CoaxPress card when it was plugged into an x4 rather than x8 PCIe slot.  They are due to dropped packets.

 

Mark

 

 

From: Tech-talk <tech-talk-bounces at aps.anl.gov> On Behalf Of Wlodek, Jakub via Tech-talk
Sent: Wednesday, November 29, 2023 9:17 AM
To: tech-talk at aps.anl.gov
Subject: Andor3 IOC detects and connects to Marana, can get frames, but frames appear noisy/corrupted

 

Hi all,

 

Recently I configured an Andor Marana detector utilizing the Andor3 areaDetector driver and the bitflow kernel driver included with the andor3 SDK. This has been working fine and collecting data for ~6 months. At one point, after the detector had been running for a long time, the beamline reported that the IOC was displaying images that appeared corrupted/didn't make sense, but stopping/starting acquisition got back the expected readout, and they ran OK for another several months.

 

Now, however, this same corrupted image issue has re-emerged, but this time a stop/start does not fix it. In fact, a camera power cycle, IOC reboot, kernel driver rebuild and reload, and IOC server reboot do not appear to have solved the problem. The IOC can detect and connect to the camera, all firmware/serial number information is correct, and it gets images of the correct size, but they do not appear to be valid images. The different PreAmp Gain modes produce different issues, please take a look at the linked github gist for screenshots:

 

https://gist.github.com/jwlodek/b50af1d823f159bb78ecd088b71298b3

 

The 12 bit mode exhibits vertical stripe artifacts, while the 16 bit HDR mode exhibits horizontal lines that cross the image. The overall average pixel value in the 16 bit mode with no beam used to be ~100 counts, it is now ~35 counts, and spikes to 200 when the artifacts appear. I don't see any error messages that would explain this in the IOC shell, and during acquisition the detector outputs the correct framerate and image dimensions. The image also does not appear to be influenced by outside stimuli i.e. flashlight/beam.

 

Has anyone seen this on an Andor Marana, or Andor camera in general before? If so, is there a software based fix for this or is this a hardware related issue? My next troubleshooting steps will probably be to pull out the PCI card and test the setup on a Windows machine with a shorter cable (the bitflow software on windows lets you collect images outside of the Andor SDK), but if this is some kind of configuration issue I'm missing I'd rather ask first before I pull the card out of the server in the rack.

 

Thanks in advance for the advice!

 

Jakub