Regarding the Halcon series targets:
The official Halcon documentation suggests that the effective pattern size of the calibration target should be more than 1/4 of the whole field of view but less than the whole field of view (the Halcon software will fail to recognize the plate if the effective area is bigger than the whole field of view). 15 to 20 photos are preferred to archive good result. 

Based on our interviews with many customers, we suggest to select a plate with effective area of more than 1/2 of the field of view in 3D application, and plate with effective area covering the whole field of view in 2D application. After the calibration the deviation of the dotspacing will be nanometer level.

Regarding the Halcon 12 series targets:

We suggest to select a target with a effective pattern area size covering the whole field of view. If any group of the characteristic points falls within the field of view Halcon 12 will recognise the plate. The Halcon 12 series plates are using more dots than the once for a previous version of Halcon. For applications using Halcon version 12 or higher you should  select Halcon 12 series calibration target.

The size of the target should generally be between 100% and 200% of the field-of-view. Smaller or larger targets are possible, but special care has be taken during the calibration process in this case.

. If the target is smaller than the FOV, you can compensate for this by acquiring several calibration images in order to cover the FOV entirely.

. When working with low resolution images, make sure the diameter of the individual dots is no less than 20 pixels, Otherwise the mark extraction may fail. For an image with VGA resolution (640×480), for example, the target size must be twice the FOV to meet this requirement.

. If the target is significantly larger than the FOV (e.g. more much than 200%), you can still use it. Since you are able to fit fewer points into the FOV, you can compensate for it by moving the plate around and taking more images. Should you need to perform a 3D calibration, you might be limited by the depth-of-field, since the plate needs to be tilted. For 2D applications, however, this is not a problem.

. If you do decide to print your own plate, note that your office printer may not provide the accuracy you might need. As a rule-of-thumb, printing your own plate is OK for sizes 160 mm and larger. You can print smaller plates for proof-of-concept purposes, of course, but keep in mind that the accuracy will suffer or, in the worst case, the calibration may fail altogether. 

Regarding the OpenCV series targets:

We suggest to select a plate with effective area more than 1/2 of the field of view in 3D application, and plate with effective area covering the whole field of view in 2D application, and to take 15-20 pictures to calculate a calibration.We recommend to take pictures of the target in it’s whole as the calibration accuracy of a dot will be significantly higher as the accuracy based on the corner points only. If you require a high calibration accuracy a dot based pattern is highly recommended. OPENCV also supports calibration targets based on dot patterns.