I’ve heard that a calibration is refused if it is too far away from the CGM’s value. It’s like the CGM is skeptical that it could really be so wrong.
One approach to deal with this is to calibrate part way, i.e., use a number between the CGM’s value and the meter’s measurement, maybe choose a number within 20 mg/dL of the CGM’s reading. Then wait a while (20 minutes?) and do another modest calibration, and repeat as needed.
I’ve also heard that the CGM becomes more and more resistant to calibrations as it ages; a young sensor may fully accept a calibration value as the truth, whereas an older sensor may only move part way towards the calibration value.