Project

General

Profile

stage accuracy

Added by Anonymous over 14 years ago

Hello,

Does somebody have experiences with a rather inaccurate stage?

We plan to use Leginon for generating large montages (up to 8 x <img src="{SMILIES_PATH}/icon_cool.gif" alt="8)" title="Cool" /> of 16MP images at 10k or 5k magnificaton across complete series of sections on a Jeol 1230. Similar as it has been done by Hartenstein et al. with the fly brain.

We now tested our stage with a 1k x 1k camera at 10k magnification: We took a picture A at 10k magnification at stage position A, moved to anohter position B, moved back to stage position A and took a second picture A'. Image A' is shifted up to 700 nm compared to image A either in x or in y direction. This is up to 6% of the image dimension.

Will this cause insurmountable difficulties for our plannings? Can this be corrected by stage calibration?

Harald


Replies (1)

Re: stage accuracy - Added by Anchi Cheng over 14 years ago

Harald,

If you have 6 % returning error, you will need 6% overlap on each side of your tiles. Therefore, the distance between tile target is 12% less than ideal. In addition, if you use software to stitch the tiles together, you will need to allocate overlap for it. More overlap reduces the efficiency, that is all its disadvantage. It should not stop you from doing it.

As far as correction, it depends on the nature of the inaccuracy:
(1) Your test is about returning to the same position, the motor is driven in a different direction when you return. Therefore, it is possible that you can improve that by a back-lash correction, i.e., always move to a position from the same direction. Leginon and other automated image acquisition software all (or should) have this correction built-in.

(2) Some goniometer has periodical movement, see http://ami/documentation/leginon/bk01ch11s11.php for an example. That was on an older FEI tecnai. Most newer FEI scope has not shown such strong oscillation. I don't think JEOL scopes do the same, at least not at the same scale (Our oscillation gives more than 20% error at some location). This is the type of error that can be corrected by calibration.

(3) For random error, calibration will not help. On FEI scope, the further distance you want to drive to, the less likely you will get there accurately, therefore, there is an iterative move option.
see http://ami/documentation/leginon/bk01ch12s07.php on an example setup. However, this type of movement come at the price of time because it needs to try, check, and retry until it is good enough. I think the same strategy can be applied to JEOL scope.

Anchi

    (1-1/1)