Ultra wide lens thumbscrew: what the lock screw clamps

An ultra wide CS-mount or M12 lens arrives focused, unmarked and held by one small thumbscrew, and the packing sheet explains none of it. You are about to fix it to a robot, a kiosk or a production cell, and you want to know what that screw holds before vibration finds out for you. This page says what the screw clamps, why loosening it appears to change nothing, how to prove which ring it locks in five minutes on a printed target, and how to lock focus so the image is still sharp after months of heat and shake.

What does the thumbscrew on an ultra wide lens actually do?

The thumbscrew is a lock, not a control: it presses a small setscrew against the moving focus barrel and clamps it to the stationary mount, so that vibration, handling and temperature changes cannot drift the focus position that was set when the lens was assembled or adjusted.

A setscrew has one job in any machine: stop relative motion between two parts it can press together. On a lens the two parts are a moving barrel and the fixed front of the mount, and the barrel it usually holds is focus. Ultra wide board lenses are focused once, close to infinity, because a 3 mm class focal length carries metres of depth of field; the designer spends that depth at assembly and gives you no near or far marks to move it again. The thumbscrew is the promise that the setting survives shipping, handling and the first week on the machine. The anatomy of a machine vision lens maps every part of the barrel, and the page on locking screws, focus, aperture and enclosure clearance covers the larger marked lenses where the same idea is labelled.

Why does nothing move when you loosen it?

Because the screw holds a clamp, not a spring: loosening it releases the friction that fixes the barrel, and the barrel only moves when you grip it and turn it against the lens thread.

Expect movement in microns, not millimetres, and expect resistance at first. Factory barrels are either staked with a small dot of compound or simply stiff in their thread, and the ring usually carries no knurling, because nobody was meant to turn it. Back the screw out one to two full turns so the tip clears its hole or dimple, then turn the innermost ring with even finger pressure. If it refuses, stop there: thin aluminium barrels dent and bind, and a dented barrel will not hold focus or pass the screw again. The safe checks before forcing a bound mount are in the thread binding checklist.

How do you prove which ring the screw locks?

A five minute bench test settles it: focus a sharp target, loosen the screw, turn one barrel at a time while you watch the live image, and the ring that changes sharpness is the one the screw was holding.

Run it once before the camera ships, with the target in place.

1. Set the printed target at the real working distance, light it evenly and open a live view at full resolution.
2. Zoom to pixel level on a high-contrast dot edge and note how sharp it is.
3. Draw a paint-pen index mark across the barrel and the mount, so any turn is visible and reversible.
4. Back the thumbscrew out one full turn.
5. Grip the innermost ring and rotate it a few degrees while watching the edge. Sharpness changing marks it as focus. Repeat with any other ring, one at a time.
6. Return to the index mark, tighten the screw to a firm snug and confirm the edge is as sharp as it was in step 2.

A ring that changes brightness instead of sharpness is an iris, and a lens where only the whole body turns has a mount lock rather than a focus lock. The M12 focusing procedure continues from here with corner checks and the final lock, and if the image was already soft, the ultra wide focus-check sequence traces the blur through focus, spacing and sensor tilt.

A printed dot grid makes the pass or fail call obvious at pixel zoom. The LS-LXR-0029 is a 76 x 76 mm dot-grid calibration target with 0.5 mm dots.

How far does focus move when the barrel slips?

Very little: on a 3.3 mm lens, moving focus from infinity to a 0.5 m working distance takes about 22 µm of barrel travel, so a screw left finger-tight can lose focus from a fraction of that movement.

For a thin lens, Newton's form of the imaging equation gives the barrel travel x' needed to bring an object at distance x into focus, and the blur that slip produces at the sensor:

Here x is the working distance measured from the front focal point, which for a wide lens is close to the working distance itself, and N is the f-number. This is a thin-lens estimate; real wide lenses land near it, which is enough for a decision about a screw.

Work the numbers. Squared, f² = 3.3 × 3.3 = 10.89 mm². Bring the target to 0.5 m: x = 500 mm, so x' = 10.89 / 500 = 0.0218 mm, about 22 µm of travel from the infinity position. At 1 m the same sum gives 10.89 / 1000 ≈ 11 µm. A human hair is 60 to 80 µm across, so the whole close-focus travel of the lens is a third of a hair, and half of that travel is spent crossing from 1 m to 0.5 m.

The LS-MBL-0159 is a 3.3 mm low-distortion M12 board lens for 1/2.3" sensors at 16 MP, exactly this class of ultra wide.

Slip converts to blur micron for micron at the sensor, divided by the f-number. Take an f/2.8 example aperture: 11 µm of slip smears 11 / 2.8 = 3.9 µm, and 22 µm smears 7.9 µm. A 16 MP 4:3 imager of 1/2.3" format runs about 4,600 pixels across 6.2 mm, so pixels sit near 1.3 µm and those two blurs are three and six pixels wide, visible at exactly the zoom your line will use. The aperture arithmetic behind this is on the f-number and depth of field page.

The same arithmetic explains why an infinity-set wide lens tolerates small errors. The hyperfocal distance is f² / (N × c) = 10.89 / (2.8 × 0.0013) ≈ 3,000 mm with those example numbers, so a subject at 3 m is already within one pixel of perfect and the error only shrinks beyond it. The dangerous direction is inward, toward close focus, where the entire budget is the 11 to 22 µm above. Drop your own focal length, pixel size and f-number into the depth of field calculator for your station.

When is the thumbscrew not a focus lock?

On some lenses the same small screw pins the iris ring, the field-of-view ring or the whole lens body to the mount, and the image test tells you which within a minute.

Three clamps wear this screw. The focus lock changes sharpness when its ring turns. An iris lock, rare on board lenses where the aperture is usually fixed, turns a ring that changes brightness only. A mount lock presses on the flat of the mounting thread so the whole lens cannot rotate or back out; no individual ring turns, but the body loosens in its ring. Vehicle and robot builders use that third kind on purpose, because a lens that unscrews itself a quarter turn per month changes focus and horizon together. The C mount to CS mount adapter page shows how much thread engagement the mount offers a screw like this.

Varifocal lenses add a fourth ring: field of view. Set the angle first, lock it, then focus, then lock focus, because every field-of-view change re-focuses the lens. The LS-MBL-0176 is a CS-mount varifocal pinhole lens covering 4-20 mm at 2 MP with a 12.53 mm back focal length, the kind of lens where that order matters.

One trap lives here: turning the whole body to level an image also screws the lens in or out along the sensor axis. CS-mount holds the sensor about 12.5 mm behind the flange, so a fraction of a turn is a real slice of the spacing budget, as the S-mount thread travel page measures for the same physics on M12.

How do you lock focus so it stays put through vibration and heat?

Snug the screw to roughly 0.1 N·m, two fingers on the short driver that fits it, then prove the result with a thermal cycle instead of trusting the feel of the driver.

Snug means the barrel will not turn under a light finger twist. Record the position: a paint-pen line across barrel and mount, photographed into the station log, turns a mystery into a five-second check a year later. On machines that move, add a pinhead of removable, medium-strength threadlocker to the screw tip so the screw itself cannot walk out.

Heat moves focus on its own. Aluminium expands about 23 µm per metre per °C, so a 25 mm flange stack grows 25 × 23 × 10⁻⁶ × 10 = 5.8 µm over a 10 °C rise, half of the 11 µm that carries a 3.3 mm lens from infinity to 1 m. A housing at 45 °C has spent half the focus budget before anything vibrates. Set final focus at operating temperature, not on a cold bench, and re-check after the first real thermal cycle; the temperature focus-drift diagnosis walks that test.

Vibration finishes the job on vehicles, which is why the fixed versus autofocus comparison for mobile robots so often lands on locked fixed lenses. Corners blur first when a barrel or sensor tilts, so end every check at the corner dots, not the centre.

What do you need to set and prove the lock?

You need a printed target at the working distance, a live view zoomable to pixel level, a driver that fits the screw, and a way to cycle temperature if the machine lives outdoors or in a washdown cell.

The bench kit is small, and each item catches one failure mode:

• Dot target at the working distance, 0.5 mm class dots: catches focus error.
• Pixel-level zoom on live view: makes a few microns of blur visible.
• The correct driver, short lever: prevents a crushed barrel.
• Paint pen: catches rotation of barrel or body.
• Removable medium-strength threadlocker: stops the screw walking out on moving machines.
• A warm re-check at operating temperature: catches thermal drift.

Choosing rather than fixing? Browse the M12 board lenses for the ultra wide S-mount range, the machine vision lenses for C-mount glass with printed scales, and the lens accessories category for targets and extension hardware. The fisheye lens selection guide for M12 board cameras works through image circle and back focal length before you order. An FA lens is the marked alternative when a bare board lens is too crude: the LS-FFL-1582 is an 8 mm C-mount FA lens for 2/3" sensors with a printed 57.6° field of view. Barrels of that class carry labelled focus and iris rings, so the anonymous-screw problem disappears.

Set the target, run the five-minute proof, lock to a firm snug with the index mark photographed, and re-check warm. The lens finder and the lens configurator narrow focal length, sensor and working distance to parts that fit your station, and more lens guides sit on the lenses topic page. If a screw still puzzles you after the bench test, send a photo through the contact page and we will look at it with you.