Inconsistent geometry measurements

[asarnow]

Hi Fabian,
I had a problem over the weekend with inconsistent geometry measurements, that I couldn't quite figure out. The milling angle was 11˚ with 1˚ overtilt after rough milling. PACEtomo was giving me impossible pretilts like 80˚ or unlikely ones like -4˚, depending on the lamella which looked similar otherwise. Do you think that could be caused by uneven lamella? Would using small PACE groups only spaced out along the tilt axis make it more robust to bad pretilt estimation (at the cost of speed)? Thanks for your advice!

[eisfabian]

Hi Daniel,

Yeah, that sounds wrong. Do you have the log output?
If you get a chance, could you also set debug = True in the script, run it and send me the log output?

The only time I see values that are off by that much is if all your geo points are pretty much lined up. If you do have a large lamella (>20 microns maybe) it could be that some geo points are outside the image shift limit and could not be considered.
It should say in the log how many points it used for the fitting.

Thanks and best wishes,
Fabian

[asarnow]

Unfortunately I didn't save the logs, and I haven't encountered the issue again on other grids. If it does I'll use debug mode and keep the logs. As a point of clarification when I measure the geometry is it ok to move the most central point of the group to the top or do I need to pick them in the right order initially?

[eisfabian]

Okay, thank you. It should be fine to move it up in the navigator.

Let me know if you run into any other issues!
Best,
Fabian