The Lens Drifted and Nobody Knows What Changed: Build the As-Installed Record

A drifted lens is not the problem; the missing baseline is. Nobody can say what changed because nobody recorded what the unit was set to when it was installed, and a comparison against a memory is not a comparison. The fix has two halves. From the maker, a per-unit as-shipped record: the serial number and the batch test data for that unit, covering MTF, distortion, BFL and depth of focus under stated conditions. From your own team, a setup sheet completed at commissioning for the parts you control: back focus, aperture, working distance, light, target and the accepted image. Together they turn drift into a measurement.

Two records, and two owners

The maker’s record describes the unit as it left the factory. On every production batch Lensmind ships test documentation that covers the agreed family parameters, typically MTF, distortion, BFL and DOF, with the batch number, the test conditions, the inspection coverage and an explicit pass or fail judgment, and those reports are the default rather than a special request. For the FA line this sits behind the numbers that matter on a machine vision station: distortion held below 0.05 percent with 10K plus batch consistency, and closed loop active alignment down to the one micron class on the units that demand it.

Your record describes the unit as it was installed, and only you can create it. The maker cannot know your ring light, your sensor position, your working distance or the image your process accepts. That sheet is what makes the maker’s numbers useful, because a batch report says what the part can do and your sheet says what it was doing on the day it worked.

What belongs on the setup sheet

Nine fields are enough, and every one of them is checkable in ten minutes at the machine. The lens model and its serial number. The mount and the flange distance check that goes with it, which for a C-mount is 17.526 mm and for CS-mount is 12.5 mm, with the 5 mm adapter ring noted if one is used. The back focal distance or adjustment value actually set on the lock. The focus lock type, since an M12 with an FT or HT focus lock behaves differently from a plain thread under vibration. The aperture, if it is adjustable. The working distance and the measured field of view at that distance. The illumination, named by model and by angle. The target and the accepted image, stored with the serial. And the date plus the name of whoever did the work.

None of that requires a laboratory. It requires a target, a light, a tape measure and the discipline to write down what was set rather than what was intended. The sheet should live with the machine’s maintenance file rather than in an engineer’s notebook, because the person who needs it will be whoever is on shift when the image softens.

Why back focus is the field to record first

When an image softens on a station that has not moved, the first suspect is the distance between the lens and the sensor, because it is the adjustment most likely to have been nudged and the easiest to measure. Recording the value at commissioning, and re-measuring it before anything else is touched, separates a mechanical shift from an optical or an environmental cause in a single step. On a liquid lens unit such as the ZWA series, where focus is set electrically in about 20 ms with no moving parts, the equivalent record is the focus value and the drive setting rather than a mechanical distance, and it belongs on the same sheet.

The lock type matters here as much as the value. A thread locked focus that has never moved is evidence; a focus that has been adjusted twice in a year without a note is a station with a handling problem. The record makes that visible without anyone having to remember.

Build the reference image, and keep it current

The setup sheet describes the machine; the reference image describes the result. Capture one at commissioning with the production target, the production light and the production exposure, and store it with the serial number and the settings. From then on the first diagnostic step is to re-capture the same image under the same conditions and compare, not to adjust anything. A difference in sharpness, contrast or distortion then has a direction and a magnitude.

Re-capture the reference after any service, any lens change and any change of illumination, and note the reason. A reference image that is three years old and was captured under a lamp that has since been replaced is a reference that will send somebody chasing a change in the light as if it were a change in the lens.

What the maker can and cannot supply

What a lens maker can supply is factory data: batch level documentation on the parameters that were agreed, per serial metrology where the unit justifies it, and the traceability that comes with in-house coating, cementing, cleanroom assembly and final metrology. Lensmind’s Hangzhou line runs machining, coating, cementing, cleanroom assembly and final metrology in house, with centration and surface measurements on named instruments and interferometry for the surfaces that need it, and units that require it are aligned actively to the one micron class before the final test.

What no maker can supply is the number your station needs. A batch report cannot know your sensor, your light or your acceptance criterion. That is why the two halves of the record are not alternatives: the factory data tells you what the part is, and your setup sheet tells you what the station was doing when the process was good. Where a maker can help is with the format, and asking for the batch documentation to arrive as a file per shipment, named by batch and serial, is a reasonable request to put in the order.

What the batch documentation actually looks like

Ask for the report in a format that survives a filing system rather than a photograph of a screen. The useful document names the batch and the model, states the parameters that were measured with their conditions, gives the result against the pass or fail judgment, and identifies the coverage, meaning what proportion of the batch was measured and how the sample was taken. For a family like the FA C-mount range, where distortion is held below 0.05 percent across production, the report is what turns that claim into something a station can rely on; on the line scan side, the parameters that matter shift towards field flatness and image circle coverage for the 4K, 8K and 16K sensors, and the same principle applies: name the parameter, state the condition, state the judgment.

Where the application justifies per unit data, ask for it and be explicit about which parameters. On a liquid lens assembly the useful per unit record includes the focus response and the drive settings rather than a mechanical back focus, because the adjustment is electrical and a replacement unit has to be set up from the record rather than by eye. Getting the field list right at the order stage is what makes the record useful a year later.

Using the record when something drifts

The drill is short. Compare the current image with the reference. Measure back focus and compare it with the recorded value. Check the mount and the lock, then the light, then the target. Where three units on the same line drifted together, the common factor is usually environmental rather than optical, and the record will show it because all three sheets share a light model, a temperature and a shift pattern. Where a single unit drifted, the suspicion narrows to that unit’s mount or adjustment, which is exactly the case where a serialised as-shipped record earns its keep.

Escalate to the maker when the numbers point at the part rather than the station, and send the serial number with the question. A maker who has the unit’s record can answer in a way that a maker who is guessing cannot, which is the same argument that makes the batch documentation worth asking for at the order stage rather than at the failure.

What to write into the order

Five lines: the unit identification that will appear on the maker’s record and on your sheet, agreed as a format rather than assumed; the batch documentation that ships with each order, listing the parameters, the conditions and the pass or fail judgment; per serial metrology where the application justifies the cost; the focus or lock configuration of the ordered variant, since FT, HT and plain thread options lead to different setup sheets; and the acceptance criterion for the station, written by the buyer and kept with the reference image.

Together they make a station that can answer its own question. A drifting lens on a station with a record is a comparison; the same lens on a station without one is a fortnight of emails that ends with somebody replacing a part that may not have been the cause.