Requalifying a Station for 850 nm: The Lens Data Behind a Focus Shift

The focus moved because the lens is corrected for the visible band and 850 nm does not come to the same image plane, which is longitudinal chromatic aberration rather than a fault. Requalifying the station is not a matter of moving the focus ring and running the same recipe. It needs four things you can file: the maker’s measured focal shift at your working distance, your aperture and your actual window stack; MTF at 850 nm at named image heights; a per unit offset figure, because two stations on the same lens model will not need the same correction; and a bench record from your own station that shows what you applied and what it did. Without the fourth, the first three are someone else’s drawing of your problem.

Why the Focus Moves and Why “IR Compatible” Says Nothing

A lens bends light by refraction, and the amount of bending depends on wavelength. Glass is dispersive, so a design that brings the visible band to a common focus at the sensor will leave 850 nm at a different plane. Nothing is broken and nothing is dirty. The image is simply being judged at a plane where the 850 nm light is not in focus. The Schneider Kreuznach note on how different illumination sources change lens focus works through the consequence: if the focus stays where visible light was best, the 850 nm MTF can collapse at spatial frequencies well below the sensor’s capability, and the apparent recovery at higher frequencies is spurious resolution with a phase shift, which is not data you can use. Move the focus by the amount and in the direction the chromatic focus shift curve gives, and the performance comes back.

That is also why the phrase IR compatible should end the conversation rather than start it. A lens can transmit near infrared perfectly well and still form a poor 850 nm image, because transmission and correction are different properties. The datasheet line usually describes the coating’s pass band or a vague claim about NIR use. What you need is measured behaviour, at your geometry, on the axis and off it. Ask for the measurement, not the adjective.

The Four Numbers That Turn a Datasheet Into a Requalification

The first number is the chromatic focal shift itself, with a sign and a magnitude, between the visible reference condition and 850 nm. A magnitude without a direction tells you how far to move and not which way, and the direction depends on the glass and the design, so it cannot be assumed. It must be quoted at your object distance, your aperture and your sensor position, because the shift in image space converts to a different amount of focus travel at a different magnification.

The second is the available focus travel in the mount. If the shift needs more travel than the lens barrel or the mount allows, the answer is a different lens, not a longer adjustment. Check this before you plan the rework, not after.

The third is MTF at 850 nm, at the spatial frequencies your inspection actually depends on, at centre and at named image heights, with sagittal and tangential curves separated, at the f number you run. A center only figure hides the field behaviour, and a lens that is acceptable on axis at 850 nm can be much worse at the edge of a large sensor. Reading those curves is its own skill: how to read an MTF chart explains what the axes and the field positions are telling you, and it matters more here than in a visible light selection, because you are comparing two wavelengths rather than two lenses.

The fourth is transmission at 850 nm, ideally with the coating stack named, because a contrast drop after a wavelength change has two candidate causes and you need to know which one you own. A lens that transmits 850 nm poorly, or whose anti reflection coating was tuned for the visible band, will lose contrast even when it is perfectly focused. The coating comparison is set out in NIR versus VIS coatings on industrial lenses, and it is the second half of the same question.

Subtract the Station, Not Just the Lens

Whatever the maker measures on a lens alone will not match what your station shows, because the station adds material in front of the sensor and material in front of the lens. The sensor’s cover glass, the package window, any protective window on the housing, and the filter in the illumination path all sit in the optical train, and each contributes its own dispersion and path length. The result is that the focal shift you measure on the machine includes things the lens datasheet never described.

Two practical consequences follow. First, when you compare the maker’s number with your own, compare the direction of the shift first and the magnitude second, because the direction should survive the extra glass and the magnitude will not. Second, when you order a replacement lens or a second station, specify the window stack with it. Cover glass thickness and lens matching sets out why a sensor stack change can quietly move the best focus, and the same logic applies when you change wavelength instead of thickness.

There is a third contributor that catches teams out: the camera. A colour sensor’s quantum efficiency falls away in the near infrared, and a mono sensor may still have a filter in front of it. If your 850 nm image lacks contrast, part of the loss may be the camera rather than the optics. Compare the same lens on a mono camera before you decide the lens is the problem.

Split the Contrast Drop Before You Buy Anything

Run the diagnosis in an order that eliminates the cheap causes before the expensive ones. Confirm the illumination is actually uniform and stable at 850 nm, since an LED ring that was even in white light can be uneven in the near infrared if the emitters or the diffuser behave differently. Then confirm the focus shift by moving the focus deliberately in small steps and watching whether the image sharpens; if it does, the cause is defocus and the rest of the exercise is about how to hold the corrected position. If moving the focus does not recover the image, the cause is more likely transmission, coating or illumination, and no amount of focus travel will fix it.

Only then look at the lens data. Doing it in this order stops a common waste: buying a NIR corrected lens to solve an illumination problem, or reworking a mount to solve a coating problem. The bench discipline is the same one used when an image darkens between shifts and the lens is being blamed for something in the cell: separating the lens from the cell before spending money.

One Lens Model, Many Stations: The Per Unit Offset Record

This is the part the generic advice skips. A maker’s chromatic focal shift figure is a design or batch figure. Your line has several stations, and each has its own lens, its own camera, its own window and its own mounting distance. The offset that brings station four into focus is not automatically the offset for station one, and if you apply a single recipe across a line you will spend weeks chasing a difference that is really a per unit calibration.

Build the offset record from the bottom up. For each station, record the lens serial, the camera serial, the window and filter part numbers, the object distance and aperture in use, the focus position in the mount (a number from the stage or the scale, not a mark on the ring), the target and lighting condition, and the MTF or contrast figure achieved at that position. Then state the acceptance: how far from the baseline a station’s offset may sit before it is treated as a fault rather than a variant. When a station is serviced or a lens is replaced, the record is what tells the technician where to start instead of where to guess.

This is also where the maker’s own documentation becomes useful, and it is worth asking a prospective supplier whether the number can be tied to the unit. Lensmind’s production batches ship with test documentation covering MTF, distortion, BFL and DOF with batch number, test conditions and a pass or fail judgment, produced on in-house metrology that includes Trioptics centration and OptiSurf, a laser Fizeau interferometer and a UV spectrophotometer. That package is the right starting point for a per unit offset table, and the question to put to any maker is whether they can measure the same unit again later and report as-found values, which is the as-installed record that makes a baseline useful.

The Dual Illumination Duty Case: Switching Becomes a Controlled Step

Many stations do not choose between white and 850 nm; they run both, perhaps by part number or by shift. That turns the focus offset into a recipe, and changes where the risk lives. The optics are now understood, so the failure mode is no longer physics but process: the recipe is selected wrongly, or the focus is left in the previous position, or a new operator does not know there are two positions.

Control it the way you control any other changeover. Give the two positions names and numbers rather than descriptions, put the offset in the station program rather than in a notebook, and require the first article of each changeover to pass a defined check before production units are run. Lock the focus mechanism so that vibration cannot drift it between the two states, and record both positions in the station documentation so that a technician who has never seen the cell can restore the setup. If the station is going to run both wavelengths for years, check whether the lens model offers a version built for near infrared use, since a lens corrected for the band you actually use reduces the size of the offset you have to manage.

Three Routes From the Data and When Each Is Right

Once the numbers are in, there are three honest routes. Keep the existing lens and apply a fixed offset where the shift is small enough to sit inside the available focus travel, stable across units, and recoverable in MTF at 850 nm. That is the cheapest route and often the right one, provided the offset is documented and each station is checked. Change to a lens corrected for near infrared where the shift exceeds the travel, or where the 850 nm MTF even at the corrected focus is below what the inspection needs. Design a lens for the wavelength where the station is a product rather than an installation, where the volume justifies tooling, and where the 850 nm performance has to be guaranteed at named image heights rather than recovered by adjustment. A custom path has a schedule attached to it, and a maker that runs prototype to mass production on a written 60 working day commitment is telling you the shape of that route.

The decision rule is worth writing down in one line each, so that the next engineer does not relitigate it: shift inside travel and MTF acceptable, keep and offset; shift outside travel, change the lens; MTF unacceptable at any focus, change the design. Put the measured values beside each line.

The Requalification Record

Finish with a document that a future auditor, or a future version of your own team, can use. Name the station, the lens and camera serials, the window stack, the two illumination conditions, and the focus positions for each. State the reference target, the lighting condition and the measurement method used to accept the result, and give the pass figure for each named image height rather than a single number. Record who approved it and when.

Then add the part that keeps it true: the conditions that require the requalification to be repeated. A lens replacement, a camera change, a window or filter change, a change of working distance or aperture, a move of the station, or a maker’s design or coating change are all enough to invalidate the offsets. With those triggers written down, the record stops being a snapshot of one good afternoon and becomes the reason the station stays qualified when the illumination changes again.

When the data package is settled, the choice usually narrows to a lens family that is already corrected for the band in use. The FA C-mount and M12 S-mount families, and the wider machine vision lens portfolio, are where that comparison starts.