An extension tube moves the lens away from the sensor to increase magnification. The lens was designed to project a cone that fully covers the sensor at its original flange distance. Moving it further out enlarges the projected image circle, but the cone angle stays the same, so the circle grows more slowly than the sensor area it must cover. At some extension the corners of the sensor fall outside the illuminated cone. The result is a hard, roughly circular darkening that follows the image circle rather than the rectangular sensor.
A zoom lens has to keep the image circle acceptable across its whole magnification range, and it does so with a mechanical compensation group that moves internally. At the long end of the zoom the internal groups are already near their travel limit. Adding an external extension pushes the whole assembly beyond the range the compensation group was designed for, so the corner illumination falls off faster than the nominal magnification increase would suggest.
Place a uniform white target in the field and capture at the intended extension. Measure the grey value at the centre and at each corner. A fall-off of about one stop is normally correctable in processing; beyond roughly two stops the corner noise floor rises faster than the signal and the corner data is no longer trustworthy for dimensional measurement. Do not judge from a normal workpiece image, because a dark workpiece hides the fall-off.
First, reduce the extension and gain the magnification optically by choosing a longer focal length or a higher-ratio zoom body. Second, move to a sensor whose diagonal fits inside the illuminated circle at the required extension; this is usually the cheapest fix if the sensor is not already fixed. Third, accept the fall-off and add flat-field correction, which is only valid if the corner fall-off stays below the two-stop limit and the illumination is stable over time.
Flat-field correction divides the image by a stored reference. It corrects a smooth gain difference, but it also amplifies noise in the corners by the same factor. At two stops the corner noise is already four times the centre noise in signal terms, so the corrected corner becomes visibly grainy and edge detection becomes unstable. This is why the correction approach must be validated on the actual measurement, not on a visual check.
Write the requirement as a maximum extension for a stated sensor size, not as a magnification. The magnification can be reached in more than one way, and only some of those ways preserve corner illumination. When the magnification requirement is fixed, state the sensor format explicitly so the supplier can confirm the image circle at that extension.
No. Stopping down reduces the cone further and usually makes vignetting worse, not better.
By the combination. The tube moves the lens beyond the flange distance its cone was designed for, so the image circle no longer covers the sensor corners.
About one stop is normally correctable. Beyond roughly two stops the corner noise rises faster than the signal and measurement data becomes unreliable.
Yes, usually. A sensor whose diagonal fits inside the illuminated circle at the required extension removes the vignetting without changing the optics.
Because it amplifies corner noise by the same factor it corrects. At two stops of fall-off the correction factor is large enough to make the corner visibly noisy.
The maximum extension for a stated sensor size. Magnification can be reached several ways and only some preserve corner illumination.
Simply enter your email to receive the latest news and insights from Pomeas. Stay connected with Pomeas and be the first to discover new innovations in optical excellence.