Leginon

Was there a software upgrade from TFS or alignment done by service engineer between when it worked now ? Such problem is often associated with calibration change at that level.

There are two types of problem with image shift targeting: Either the specimen area of interest moves or the beam slip off but the are of interest are as expected. Can you tell me which problem you are having ? Or both ?

Hello Anchi,

No, there was nor any TFS upgrade nor an alignment done by a service engineer. Our problem is that the beam is slipping off and is not going to the target that we selected in the exposure_targeting. The region where the picture is taken is around our selected target but is usually up to 1 um off.

Thank you for your help and fast response.

Best,

Eduardo

What we call image shift in leginon is the "User Image Shift" in TFS terminology, or "Beam-Image Shift" for general optics. It involves a beam shift above the specimen to the area of interest and then a image shift below the specimen to bring the image formed back to the optical axis. TFS software has a calibration in its alignment procedure to determine the amount of beam shift it needs to do for given image shift coil change.

To properly test this, and show it to the service engineer outside leginon, here is what you do:

1. Make sure your optics is otherwise aligned: Stage and eucentric height, Objective Lens at eucentric focus, Rotation Center (i.e. beam tilt) is correct, Pivot point is correctly set at this position.
2. Find an object that you can see well at your en mag, move that to the center of the screen with stage movement, and move the beam to center it.
3. Spread the beam so that you get some beam when you perform the next step.
4. Watching the image of the object in the Flu cam while move the beam with the track ball. If the feature moves, your "Beam-image Shift" calibration is off.

If this calibration is indeed off, you will need to do this "Beam-Image Shift Calibration" in Procedure alignment under HM Imaging. This might not be available unless you are in service mode and you will need help from the service engineer.

This is not a calibration that goes off often, but has been known to happen. You might find some other problems when you are doing step 1.

Thank you for your response Anchi,

Our image shift calibration in the our Talos Artica was off and we already calibrate it. Despite this we still observe that the targeting is not too precise. The targeting is better that before but it shift itself following a clockwise behaviour. Perhaps I have to improve the leginon calibration matrix? if so which are the recommended parameters to do the image shift calibration and what type of grid do you recommend to do it. If you have any thoughts regarding the clockwise shift of the targeting we will be more than happy to receive them.

Thank you,

Eduardo

So far I am assuming that you have used the same defocus at hr and en in your sessions before and after the problem starts to happen. You may want to check if it is indeed the case. The image projection rotates when it is highly defocused, so it could be your problem.

What we normally do is to do Leginon image shift calibration at eucentric focus for hl and then add rotation with ScaleRotate node. This is usually a small rotation if hl preset defocus is not too far. You probably have this done already in the old calibration when it worked. You should check that.

I normally do my calibration with cross-grating grid that has latex beads on, so that the pattern does not repeat itself too much.

Anchi,

Thank you for your quick reply, my defocus values at hl and en are different, maybe that is the problem that I am having. Regarding the calibration there are couple of parameters that I would like to know which are their optimal values i.e., % shift, tolerance, average and interval.

Thank you,

Eduardo.

I did not mean that hl needs to be the same as en, but to ask if hl preset defocus is different from when you had no issue.

en preset is always within 5 um range, so it seldom rotate significantly.

hl preset is usually in 10-100 um. You just have to be consistent with what you decide to use. Just do an experiment by turning defocus knob manually and you will see its effect.

The default image-shift matrix calibration % shift is 25%, I believe, average is 1 and interval does not do anything when average is 1.
These are all good values. Tolerance is the one that maybe different when your hl defocus is very difference since it checks agains the pixel size at eucentric focus value. Give it a bigger value if you are confident that the correlation is detecting the correct peak so that you want your slightly off calibration to be saved.

thank you Anchi,

We haven't changed the hl defocus value which is of 100 um. I will try to make the calibration again using a cross-grating grid with latex beads and the default calibration parameters. I will let you know what comes out of this.

thank you for your help it is very appreciated.

best,

Eduardo

Hi Anchi,

Jumping on this thread - I'm struggling to get our image shift targeting to work accurately on our Glacios. Right now, if we use stage position (presets manager mover) the targeting is good.

If, instead, we switch to image shift then we see that the targeting is off by 30-50% of the hole.

The targeting is off even after doing the following: (at Z-height, eucentric focus, PP aliged)
-stage position matrix calibration using 15 positions at hole-level (M 1250X, microprobe) (interval 1.5e-06), defocus = 0 um
-scale matrix to higher magnifications
-image shift matrix calibration using 1 position at hole level (M 1250X, microprobe) (interval 3e-06), defocus = 0 um
-scale shift matrix to high magnifications

Could it be something between nanoprobe and microprobe since our exposures are collected in nanoprobe? Or just I just repeat this until I get it to work better?

Mike

I see one problem here:

-scale image shift matrix from 1250x microprobe to high magnifications in nanoprobe.

Scaling stage position matrix is a good idea because it is not possible to calibrate stage position matrix well at high magnification. It is not so with image shift. The coil used at different project mode and probe mode are different. It is best to

-Perform image shift matrix calibration at the exact mag and defocus at each preset involved. i.e, hl, en presets.

If you are using leginon after a long-existing bug fix in Issue #11967, make sure you reset additional image scale rotation back to 0,0. We've found that the new method does not need these any more.

Thanks, Anchi - yes you're right, that was an oversight on my part.

Continued testing showed me that the issue appears to be due to the M - SA targeting. If I try to image shift from an M magnification the targeting is off. But if I go up a click into the lowest SA magnification, the targeting works fine.

I'm confused about why there would be a difference and how to address the problem since it would be preferable to image shift from lower magnifications for throughput.

Short term, though, this will work "OK" for now.

Thanks!
Mike