Modeled stage (mag only) not saved
Added by Bob Grassucci over 10 years ago
Modeled stage (mag only) calibration does not seem to be seen by Leginon for our en mag. The result is the target from hl preset are never centered and we can no longer get more than one image per 2 micron hole. For some reason modeled stage is saved in en preset but not (mag only). When trying it with navigator it is way off. Any suggestions?
The reason for the bad en targetting was bad image shift calibration. Somehow it changed. I am still curious about the missing mag only calibration.
Replies (1)
RE: Modeled stage (mag only) not saved - Added by Anchi Cheng over 10 years ago
Model stage calibration data has two parts. First is the harmonic function that describes the periodic deviation from linear motion, and the second is the scale and rotation of the linear model like a 2-D transformation matrix.
The harmonic function is mechanical and can be applied to all mags, while the scale and rotation need to be determined at each mag at approximately the defocus used for data collection since the rotation/scale does change with strong defocus. The calibration table is telling you that you only have the former (for all mags once it was determined at one of the mag) but not scale-rotation calibration at the en mag.
Two things will happen here:
1. You may find that it is impossible to calibrate properly stage movement at your en preset. The move to be a faction of the image that allows correclation is too small for accurate movement for many scopes. (Our F20 en mag often covers <= 0.4 um). In most cases, the direction of the move is reasonable but it does not give consistent scale. There is little can be done typically. However, there is a possibility that your scope need a different backlash correction. In pyscope/tecnai.py and the function setStagePosition, you will find the number we set and work for all our scopes: 2e-6 m. You can change that for your scope. We just found that we need to put 0 for Krios stages.
2. Because the way Leginon is designed, movement is determined by the image mag at which the target is obtained. When you test your calibration results with navigator, this is that of en preset. During experiment, this is hl preset. Therefore, to improve your stage movement during experiment, you really should be re-calibrate scale-rotation at hl mag, not en mag.
There are two reasons for your en targeting with image shift calibration to be bad:
1. You have changed the defocus significantly since your good calibration.
2. You chose the targets too far away from the center which demands large image shift. Image shift coils are not strictly linear but spiral out a bit. Therefore large image shift with a linear model as we do is not good enough. I once asked whether the non-linearity can be modeled, but got a negative answer.
3. Because image shift rotation is not always the same between mags (most obvious between LM and M), we try to use a combination of hl and en image shift calibration for image shift movement rather than the parent-only approach I described for the stage. You may try to calibrate the image shift movement in en (but bin it down) to improve your movement accuracy.
Hope this covers them all.