Large overhead for acquisition times using MerlinEM for cRED

[aurorateien]

Hello,

We are using Instamatic v2.0.0 for doing cRED with MerlinEM detector and have discovered an issue related to the acquisition time of each frame.
For example, for an exposure time of 0,5 s for each frame (0,01 s for defocused frames), the acquisition time is typically 0,8 s (high!), which seems a bit peculiar as the readout time for the MerlinEM 24 bit data is only 1,64 ms. The rest of the time included in the definition of the acquisition time is overhead, which typically is only a few ms. Why would these extra 300 ms be added to the acquisition time and can it be corrected for? After what I've understood, this is not an issue for other detectors.

I came across the note on using Instamatic with Merlin that said: "Changing exposure on the fly (for example, in the gui) incurs a small overhead (~300 ms). "
This overhead (of 300 ms) is in fact not small when doing cRED for 3D ED data acquisition, and is seemingly added without us doing any changes on the exposure time in the GUI. If the reason for the added time in fact is overhead, is there any way to avoid it?

[stefsmeets]

Hi @aurorateien , in my testing with the Merlin, the overhead was negligible (<1 ms as you suggested) for both softtrigger and movie mode. Can you tell me which mode you are using or how you are collecting data?

The overhead penalty when changing exposure happens because instamatic has to send a command to change the exposure to Merlin in between collecting images. This was about 300 ms in my testing. If the exposure does not change, this does not get triggered.

[stefsmeets]

I'm going to close this issue now, but feel free to re-open or open a new issue if you have any further questions!

[emichr]

Hi,

Sorry for not following up on this earlier (@aurorateien finished her masters with us last summer and other things got in the way).

I'm not certain, but I believe we are using the softtrigger mode. I must admit I don't know how to check this or how to change it.

In any case, our problem is still that, no matter what our exposure time is, the acquisition time is 0.091-0.095 s. See e.g. these logfiles:

cRED_log_25ms.txt
cRED_log_10ms.txt

Maybe @GearoidM has some experience with this as well?

Any help is greatly appreciated.

[stefsmeets]

No worries. The soft trigger is always going to have some form of overhead, because Instamatic has to retrieve the image from the Merlin software. I imagine Merlin is smart enough, but check if your Merlin instance is set to not save or post-process every frame (i.e. darkfield correction) before sending it. I put some of my notes here: https://instamatic.readthedocs.io/en/latest/merlin/

Another thought is that your exposure time is very short compared to my experiments (usually in the order of 0.1 - 0.5 s). Could you try with 0.3 s and see what happens?