Machine Vision Lens FAQ

Ask for the numbers that appear in the error budget and the cycle time: focus range in millimetres or dioptres at the working distance used; response and settling time with the actual driver; focus-position repeatability, and its drift with temperature; the effect of the focus setting on magnification; the aperture range and its effect on resolution; supply voltage, interface and whether the driver adds latency; operating temperature range; expected lifetime or cycle count; and the mechanical envelope and mass. Ask for the same list for both technologies, so the comparison is like for like rather than a feature list. [Data required: POMEAS product data]

Three architectures cover most cases. A motorized zoom lens with a focus group, where both zoom and focus positions are stored per recipe; this gives the widest coverage but adds two degrees of freedom to qualify. A fixed-focal lens with a motorized focus axis, which is simpler and more stable but cannot change the field of view. Or a fixed zoom with a focus-tunable element, where the field changes mechanically between product families and the fast focus handles the height variation within a family. Choose according to which requirement changes more often: if products differ mainly in height, keep the field fixed and vary focus; if they differ in size, the field has to change.

It has to be part of the specification, because mechanical and electrowetting systems age differently. A motorized system wears at its bearing, gear or screw and typically shows a slow loss of positioning repeatability; a liquid lens has no moving parts but its optical response can drift with temperature and with time, and its driving electronics matter as much as the lens itself. For a production station the useful figure is the focus-position repeatability measured after a realistic number of cycles, not a datasheet value measured on a new unit. If the supplier cannot provide a lifetime curve, plan an in-house endurance test on one unit and use the result to set the verification interval. [Test required]

Store as many as your product range needs, but verify each one the way it will run. A stored position is only valid while the mechanical zero, the temperature and the optical assembly remain unchanged, so the recipe needs a reference feature it can measure at each position and compare with the value recorded at qualification. Verify after the first article of a production run, after any long stop, and after any mechanical intervention. It is also worth recording, per position, the focus metric value that indicated a good image at qualification, because a comparison of that number over time is an early indicator of contamination, drift or a failing actuator.

Yes. A liquid lens focuses by changing its optical power, so the effective focal length changes, and with it the magnification and therefore the scale factor in millimetres per pixel. For a measurement application this means the scale has to be calibrated as a function of the liquid lens setting — or at least at the settings used in production — and re-verified after any change to the driving signal. It also means that a small focus correction can move a measured dimension even when the image looks sharper. For pure detection this is irrelevant; for metrology it belongs in the error budget together with the temperature coefficient of the lens, which shifts the power setting needed for the same focus.

The chain, not the actuator. Exposure time has to be long enough to give the contrast the focus algorithm needs; the algorithm often has to search several positions to find the maximum, and low-contrast surfaces make that search longer and less certain; communication and command latency add a fixed delay; and on high-magnification stations, mechanical settling after a move takes time that must be respected before the image is used. A liquid lens removes the actuator from the chain but leaves the rest. The practical way to shorten the cycle is to avoid searching at all — store the focus position per product in the recipe and drive straight to it, then verify with a single image.

Five numbers, measured on the real station with real parts. Settling time from command to a stable image; overshoot and the number of oscillations, if any; focus-position repeatability in micrometres over at least 30 cycles in both directions of travel; the focus metric threshold that defines a usable image, and how much of the field must meet it; and the time from trigger to the first good image, which is what the cycle time actually depends on. Measure at the height extremes of the product range as well as in the middle, and at the temperature extremes of the cell. A vendor's response-time figure does not predict any of these on a specific station. [Test required]

More often than the market for autofocus suggests. If the height spread across all products fits inside the depth of field with margin, if a single product runs for long periods, and if the station can be built at one stable working distance, a fixed-focus lens gives the best MTF, the fewest variables and the lowest error budget. Autofocus adds a degree of freedom that has to be verified at every changeover, and each added motor is a component that can drift, wear or lose its reference. Fixed focus also removes an entire class of failure: the system measuring a part at the wrong focus because the search chose a local maximum on a low-contrast surface.

It depends on which cost dominates: cycle time or focus range. A liquid lens changes focus electronically in milliseconds with no moving parts, which suits high-throughput lines where each part stops briefly and the camera must be ready immediately. Its focus range, aperture and optical quality are more limited than a mechanical system, and its behaviour is temperature dependent. A motorized focus axis has a larger travel and can carry heavier optics, but it moves in tens to hundreds of milliseconds and has wearing parts. The usual split: liquid lens for small height variations at high rate; motorized focus where the range is large, the lens is heavy, or the focus position has to be held mechanically stable. [Data required: POMEAS liquid-lens and motorized-focus specifications — focus range, response time, repeatability, temperature coefficient]