Leginon

Hi, Mandy,

Without knowing your specific imaging conditions, here are my best guesses:

1. It is not uncommon that the beam moves by some amount as MSI routine settles. The preset cycling is like lens normalization. It creates apparent stability by repeating the same process over and over again. The first one or two squares of minimal user interaction is normally enough to settle it. You would then quickly realign beam shift at high mag presets and then continue with more MSI image collection. The higher the magnification, the more obvious the beam shift is to human and detector eyes. If you have been doing so but still sees this, then it may be more a TEM alignment issue. Check you mean intensity graph in the webviewer experiment summary page to see if that is stable. If not, it might mean that upper part of your scope alignment (gun, condenser lens) are bad or simply needing the time to settle.

1. Visit from your web browser to any image viewer of the session.
2. Find the summary tab at the top left corner and click to reach summary.php page
3. Scroll down to Image stats and click on "report >>" tab to see the full report on the image stats.
4. The graphs on the right column are the time dependency of the image mean/stdev for each preset.

2. If your fa preset beam diameter is much larger than fc preset, then it makes sense that, even if the beam shift is off, the fa preset used in autofocusing part still covers the detector while fc is off the detector.

3. The beam can also be shifted if eucentric height is not achieved at EVERY grid square. If the "Z Focus" node does not do its job, the "Focus" node will need to chase the grid height at each square. On a bent grid, this will result in drift of beam shift as the proper defocus relative the grid changes over time as you move to grid square far away from the original position where you aligned the beam shift. Check the autofocus result graph in the webviewer session summary page (up to step 2 in the above instruction).

4. If you are using the default image shift movement method to reach the focus target, and sending fc preset from preset manager gives you good beam shift but not during focusing, then you have an alignment issue. You can correct that using HM_Beam_Shift_Image_Shift_Calibration. This problem should, however, not appears only after 250 images but at the very beginning.

Thanks Anchi! I don't think it's a problem of MSI routine settlement, since it happens every ~200-300 images. I adjusted the beam after the 1st 250 images, but had to come in again after another 200-300 images. Below is a screen-shot of the image stats (ed/fc) and you can see how every so often this happens.
We followed the z-focus correction and it seemed to be doing a good job. Correlation was good and peak was found in corner. I only have this problem when using different mags for ed and fc/fa. The focus adjustment itself went wrong several times (see attached image), but since we have a second focus node in there, most of the time it could find the 2nd one better.
I do think that it is a microscope alignment and we re-did Beam and Image HM-TEM alignments just before starting this session, because before the eucentric height was even different before and after collecting the atlas! Now we are left with this beam shift problem though...