Locked Is Not Retained: Lens Setting Evidence for a Forming Press Cell

A locked lens tells you the ring cannot turn. It does not tell you that the setting survived the stroke, and it does not tell you where the blur began. Before you redesign the mount, ask the maker for one thing: focus retention measured on the complete assembly you actually run, camera, lens, adapter, bracket and cable included, at the acceleration you have measured at the camera, with focus position and MTF recorded before and after a stated number of press strokes against an explicit pass or fail limit. Then ask what holds the glass inside the barrel, not only what holds the outer ring. A maker who answers both questions with numbers is showing you a lens suited to the cell. A maker who answers with the word locked is telling you the evidence does not exist yet.

For factory-automation builds, the Lensmind FA lens series lists the large-format models with sub-0.1% distortion grades.

Blur During the Stroke and Blur After the Stroke Are Different Faults

A forming press hands the camera a repeated impulse, not a gentle shake. The frame that goes soft on every stroke can be produced by at least four different mechanisms, and they look almost identical in a saved image: the focus or iris position can move inside the lens, the lens can move relative to the camera at the mount, the whole camera can move on its bracket, or the workpiece and camera can both be still while the exposure is simply too long for the moment of impact. A lock screw on the focus ring addresses the first of these, and only the outer part of it.

You can separate them in about forty minutes with equipment you already own. Put a stationary target at the working distance, keep the lens setting, lighting, exposure and gain exactly as they run in production, and trigger the camera in step with the press. If the frames taken while the ram is at rest are sharp and the frames taken during the stroke are blurred, the setting held and the blur is happening while the shutter is open, which points at exposure time, at vibration during exposure, or at bracket and camera deflection. If the frames stay blurred after the stroke and only a fresh focus brings the image back, the setting moved, and the next question is which layer moved.

Write the tolerance before you run the test. A usable limit is expressed the way your measurement reports it, as image-plane shift or as a change in the number your gauging tool returns at the edge of the field, not as “looks sharp”. That limit is yours to set, because only you know how much image shift your inspection can absorb before the pass and fail decision changes. Hand the same number to the maker and the discussion becomes a comparison of measurements instead of a comparison of adjectives.

What the Evidence Package Has to Contain

Start with the method and the levels. Sinusoidal vibration and repeated shock are different loads, and the IEC 60068 series keeps them as separate test methods: part 2-6 covers sinusoidal vibration and part 2-27 covers shock, as listed in the IEC 60068 series index. A press stroke is a repeated impulse, so a slow frequency sweep on its own does not represent your load. Ask which method was used, at what acceleration, over which frequency range, in which axes, for how many cycles, and how those levels compare with the waveform you measured at the camera rather than with the nominal rating printed in the press manual.

Second, ask what was on the shaker. A bare lens in a shop fixture says nothing about slip at the adapter ring, the bracket or the cable bundle, and those are often the interfaces that move. The test should be run on the installed stack, or the report should state the fixture and explain, honestly, how it represents the assembly.

Third, ask for optical measurements taken before and after, at your working distance: focus position or image-plane position, an MTF or edge-sharpness figure at named field positions, image scale or field of view, iris setting, and the lock torque together with the angular position of the lock ring. A report that says the lens passed tells you the supplier has a threshold; it does not tell you what the threshold is. Ask for the numbers and for the points in the cycle run where they were taken, because a setting that creeps under repeated strokes can pass a single shock test and still walk after twenty thousand of them. If the report includes MTF curves, check that they were recorded at your working distance and image height before you compare them, which is the same discipline as reading a machine vision lens MTF chart.

Ask What Holds the Glass, Not Only What Locks the Ring

Retention has two layers, and most data sheets describe only the outer one. The outer layer is the mechanism you can see: a clamp ring, a set screw, a knurled nut, thread preload, or a friction element. The inner layer is the fixation that keeps the glass elements and their spacers where the optical design put them, and it is usually cemented or elastically held rather than clamped. A ring can hold its position perfectly while the group behind it relaxes, and the picture degrades in exactly the same way.

Ask the maker which internal fixation this lens uses, and whether that fixation was qualified under the same shock load as the ring, on the same units, with the same measurement points. Ask for the resistance to rotation and to axial motion measured before and after the test cycles, and for anything found during teardown of tested samples, such as play, backlash, loose spacers or a bond that let go. If the answer is that only the ring was qualified, then the retention claim covers the ring, and you should treat the internal group as unproven rather than assume it is fine.

The Mount Is a Chain of Threads, and You Own Every Link

Before the label on the mount means anything, the stack has to be named. C-mount sits at 17.526 mm flange distance and CS-mount at 12.5 mm, and a 5 mm adapter ring moves one to the other. That ring is a second thread and a second face contact in the optical path, and each joint is a place where the axis can tilt or the load can relax under an impulse. A retention question asked about “the lens” quietly assumes all the other joints are rigid.

Thread fit is a controlled dimension, not an assumption. Gauging a mount thread with a go and a no-go gauge, as in the inspection above, is how a barrel is proven to engage as designed, and it belongs in the same conversation as retention, because engagement length and fit decide how much of the clamp load reaches the joint. When you specify the stack, specify the engagement length, the tightening torque with its permitted range, and the measured resistance to rotation before and after the cycle run. If your stack includes an adapter ring, ask for the retention data on the assembled stack, not on a lens alone.

Mount choice is part of the same decision. For board-level and embedded cameras that live on a vibrating frame, the guidance on this site is an M12 S-mount lens with a thread-locked focus, which is a different retention design with its own evidence trail, and it is worth comparing against a C-mount build before the drawing freezes, as set out in M12 FT/HT focus lock under vibration.

There is also an honest alternative if the setting keeps walking and damping the cell does not cure it: remove the mechanical setting. Electrowetting liquid lenses focus in about 20 ms with no moving parts, so there is no focus thread to walk. In the ZWA series this covers M0834AFC, M1238AFC, M1643AFC, H1670AFC-A and H1670AFC-B. It is not automatically the cheaper answer, because the stability questions move to the driver, to repeatability across cycles and to temperature behavior, but it does remove the part of the problem that shock can reach.

Make Retention a Routine Batch Document, Not a One-Off Test

The strongest retention evidence is not one heroic test run, it is a habit. On the FA line at the Hangzhou factory, every production batch ships with test documentation covering the agreed family parameters, typically MTF, distortion, BFL and DOF, plus batch number, test conditions, inspection coverage and an explicit pass or fail judgment, and those reports are the default rather than a special request. BFL and focus position are the two numbers that let you compare the lens as delivered with the same lens after three months on the press, and units that demand it get closed-loop active alignment at the 1 μm class to hold the setting in the first place.

The in-house metrology is what makes the numbers checkable: Trioptics centration and OptiSurf, a laser Fizeau interferometer, reflective centering instruments, a precision air-bearing stage, a UV spectrophotometer and collimation systems. A factory that can measure focus position and centration can answer a retention question with data, instead of promising that the design is tight.

On your side, build the as-installed record the first day the camera is commissioned and keep it: working distance, focus position or the value your tool reports at the field edge, iris, exposure, gain, and one reference image of the production part. Re-measure after a fixed number of strokes with the lens untouched, so a later complaint is settled with a delta instead of a memory.

Then write the acceptance test into the next order, so it is a deliverable and not a favour: stationary target at the working distance, production exposure, triggers synchronized to the press, a stated number of strokes, focus position and edge sharpness measured at defined intervals, pass or fail limits set by you, and the raw data attached. Documentation and 1 μm class measurement are already part of the production route at this factory, where prototype to mass production runs on a 60 working day written commitment and repeat orders on 30 days, so the test can be quoted as part of the build rather than added afterwards.

Decide with numbers, in this order. If the setting holds through the cycle run and the image still blurs only while the ram is moving, the mount and the exposure time are your problem, not the lens. If the setting moves, find out which layer moved before any metal is cut, because a stiffer bracket will not hold a cement line or a walking thread. And if the maker cannot produce the measurements at all, that absence is the answer: work with a supplier who documents the setting, not one who describes the lock.