Acquisition slows down after a while when using autofocus with PFS status


I have a problem regrading multi-dimensional acquisition for 24 hour imaging. Acquisition starts to slow down after a while. There is a pause between one position to another position. The problem occurs when the autofocus is enabled which uses PFS status. I am using MM 2.0. The computer has 32 GB of memory. Imagej is allocated 24GB of memory. I am attaching the log report here. I really appreciate any thoughts on this issue.
Thank you.
Problem Report.txt (1.0 MB)

Just to make sure that I understand, you only see this problem when you use the PFS as the autofocus device during the acquisition? Does the PFS lock at each position (i.e. when you look at the scope, does it make the happy sound;) during the experiment? If not, there likely is a certain timeout before MM gives up and continues anyways.

PFS locks for all the positions. Problem occurred when PFS is used as autofocus. The delay between the positions keeps increasing and sometimes it is about half an hour.

There are problems with the PFS in your logs. Specifically, the PFS often fails to lock, and MM has a time out of 5 sec. However, I do not see extended delays anywhere in the log. In fact, the log starts at 09:38:45 and ends at 09:40:50, so covers about 2 minutes. The first problem seems to be that the PFS does not lock reliably. That could be caused by too much sample movement between setting up the acquisition and actually running it, or by settings up the wrong z-height, or something along those lines. It is always tricky to get everything right. If the problem is extended delays, please send me a log of such a session with long delays.

In your channel definitions, you set the Arduinos both by Label and by State. That is unnecessary and causes overhead (the computer will send the same command twice to the Arduino). Not a big deal, but takes an extra 20ms or so everytime for no good reason.

Thank you very much for your input. I mentioned about the PFS engages because I didn’t see any out of focus images. I had 12 positions in this acquisition and it was taking less than 2 minutes(~1.5 minutes) at the start of acquisition. In the log I can see only four positions with in 2 minutes. The longer delay happened later stage of acquisition. Unfortunately I don’t have the log for that. I will definitely record next time.

I will look into the channel definition, and try to modify that. It is really useful information and will be important for fast imaging.

Since I am new to micromanager, I am not sure whether I understand the log report correctly. Sorry for any mistakes I make which are obvious.

Thank you very much.