System designChecklist

Camera IO board pin alignment: verify the pinout before you mount

Verify an IO or breakout board against a board-level camera before you mount it: count pins, measure header pitch, check the pin map and trial-fit safely.

LumaSpec Vision11 min read Share on LinkedIn

On this page 8 sections
  1. 01Why aligned screw holes still allow a wrong board
  2. 02Before you order: match the accessory to the camera variant
  3. 03On arrival: count and measure before anything is screwed together
  4. 04Worked example: one-row offset on a 6-pin GPIO header
  5. 05Dry fit, continuity and first power-up
  6. 06Mounting the verified stack to a stand
  7. 07When the usual answer is wrong
  8. 08Questions

An IO board for a board-level camera looks like a drop-on accessory: same outline, standoff holes at the corners, a header that should meet the camera's GPIO sockets. The failure is quieter than that. The screws go in cleanly, the stack sits flush, and the header is still one row out of position, so some pins never meet a socket at all. That is how trigger lines end up dead, inputs end up misrouted and, in the worst case, the camera ends up damaged.

This checklist is for anyone about to order, mount or power an IO board, breakout or adapter under a compact board-level camera. It gives you three checks that catch a mismatch before it costs anything: counted pins, caliper-measured pitch and offset, and a documented pin map that names your exact camera variant.

Why aligned screw holes still allow a wrong board

IO and breakout boards are drawn against one camera variant's mechanical drawing, not against the family. Outline and standoff pattern tend to survive a sensor change; the GPIO header does not have to. Between two variants of the same compact camera, the header can sit one row along its axis from where your variant puts it, while everything visible from above stays the same.

The case behind this checklist ran exactly that way. A six-pin header met six GPIO sockets, every standoff screw seated cleanly, and the connector was still one row out of position: only four of the six pins found a socket. Two lines were dead, and since nobody had proved which four pins were contacting, the connected wiring could not be trusted either.

Screw-hole agreement is weak evidence. It shows the boards share an outline, not that the connector was placed for your variant. The electrical half of the check, reading the pin map before anything is wired, is covered in the camera shield shared IO pins guide. This checklist handles the mechanical half that comes first.

Before you order: match the accessory to the camera variant

Run this stage in the catalogue, not at the bench: the mismatch is cheapest to catch before anything ships.

  1. Record the camera's full model number and hardware revision. Accessories are released against revisions, and a revised board can move or reassign header pins. The hardware revision compatibility guide shows where to read the revision.
  2. Get compatibility stated at model level. A description naming the camera family is a lead, not an answer; ask which exact variants the accessory's papers list. It is the same discipline as matching a USB3 bridge board to the camera before you stream.
  3. Compare connector data, not photographs. Four numbers decide the fit: pin count, number of rows, row pitch in millimetres, and the distance from header to standoff holes on each board. Photos hide a one-row offset; these numbers expose it.
  4. Check the accessory is released for your sensor variant. Accessory lines lag camera lines; some variants have no IO board yet, and "not yet released" is a real answer.
  5. Fix the camera variant itself. One sensor family ships as bare board, M12, CS-mount and C-mount, and the accessory must follow the variant you bought; the CS-mount versus S-mount ordering decision is the same choice made earlier. To see how wide the spread runs inside one category: the first GigE model below has an M12 lens mount and a global shutter, the second a CS-mount and a rolling shutter:
Board-level camerasHikrobot MV-CB016-10GM-M-W 1.6 MP GigE board-level camera, IMX296, Mono, M12 lens mount, WTB GigE portMV-CB016-10GM-M-W
Sensor format
1/2.9"
Lens mount
M12
$559.00View part
Board-level cameras6.3 MP board-level camera, IMX178, 1/1.8" rolling shutter, 18 fps (LS-BCM-0020)LS-BCM-0020
Sensor format
1/1.8"
Lens mount
CS-mount
$290.00View part

The board-level cameras category filters by mount, sensor format, shutter and interface precisely because every one of those choices narrows which accessories fit.

On arrival: count and measure before anything is screwed together

The board arrived. Four checks, one pair of calipers, ten minutes, and none of them involves power.

  1. Count both sides. Pins on the board's header, sockets on the camera's GPIO field. Equal counts are necessary and nowhere near sufficient.
  2. Measure the row pitch. Span several pin centres along one row and divide by the number of gaps, per the formula below. The two pitches you will meet on compact camera accessories are 2.54 mm (0.100 inch) and 1.27 mm (0.05 inch); they differ by exactly a factor of two, and a 1.27 mm header physically cannot drop onto a 2.54 mm pattern, so never judge pitch by eye.
  3. Measure the header-to-hole distance on both boards, centre of the nearest pin row to centre of the adjacent standoff hole. On a correct pair the two distances agree.
  4. Read the silkscreen against the pin map. The board's labels should match the camera manual's pin names one for one, in order; anything unlabeled stays unwired.

Both numbers fold into one rule:

pitch = span / (pins in span - 1) reject the board if displacement > pitch / 2

Pitch and position are the only two mechanical numbers that decide whether pins meet sockets, and the worked example below turns them into a single reject limit.

Worked example: one-row offset on a 6-pin GPIO header

Inputs, and where each comes from: a compact board camera with six GPIO sockets and its vendor's own six-pin IO board (the reported failure); the two reference pitches, 2.54 mm and 1.27 mm (standard header values); the symptom to explain, which is standoffs aligned, connector one row off, four of six pins in contact.

The rule in words: the header is placed for your camera when pitch matches and position matches. Position is checked as displacement, meaning how far the board's first pin row sits from the camera's first socket row once the standoff holes are coaxial. Reject the pair when displacement exceeds half a row pitch, because past that point at least one pin cannot be on its intended socket.

The arithmetic, all of it from calipers:

  • Span five pin centres along one row of the IO board header: 5.08 mm. Gaps = 5 - 1 = 4, so pitch = 5.08 mm / 4 = 1.27 mm. The camera's sockets span the same 5.08 mm over the same four gaps: 5.08 / 4 = 1.27 mm. Pitch matches, and pitch matching alone has proved nothing.
  • Lay the IO board on the camera with the standoff holes aligned and measure the displacement between the board's first pin-row centre and the camera's first socket-row centre: it reads one full row, 1.27 mm at this pitch.
  • Reject limit: pitch / 2 = 1.27 mm / 2 = 0.635 mm. The measured displacement of 1.27 mm is twice the limit.

Decision: the offset is real and equals one full row, so the connector was not placed for this camera variant. Do not shim it, do not jumper the missed pins, do not apply power. Return it and order the board whose papers name your model and revision. The same arithmetic at the other common pitch: a five-pin span of 10.16 mm gives 10.16 / 4 = 2.54 mm, and the reject limit becomes 2.54 / 2 = 1.27 mm.

One more inference worth writing down: with one row of offset, the pins that do find sockets may not be the pins the map assigns there. That is why the reported case, four of six in contact, was unsafe even before power: the contacting four were unverified.

Dry fit, continuity and first power-up

The measurements passed; now prove the stack mechanically before it sees power.

  1. Dry-fit with no screws. The header should drop onto the sockets under its own weight. Any sideways force to seat it means an offset you have not found yet: stop and re-measure.
  2. Check continuity with the stack unpowered. Work from each terminal to the header pin the board's papers assign, then confirm the camera-side signal against the camera's pin table. The GPIO wiring guide takes over from there, with levels, grounds and what kills an input.
  3. Confirm supply and signal positions before connecting anything to the terminals. On compact-camera GPIO headers, power and IO typically share one field, so a one-position error can put a supply onto an input.
  4. First power-up on a current-limited supply, with the limit set just above the camera's rated current from its manual and the trigger lines still open. If the camera enumerates and streams, the mechanical alignment was right.

Mounting the verified stack to a stand

Once the pair is proven, build it as a stack: standoffs through the aligned holes, screws into the camera's inserts, then the assembly onto a stand or bracket. Three details keep the stack healthy over time.

  • Standoff material and height. Near bare boards, nylon standoffs remove any doubt about a metal barrel touching a trace, and spacer heights should follow the camera's mechanical drawing so the header joint carries no stress.
  • Torque. Fine-pitch screws strip with wrist force; finger-tight plus a small turn is the target, with numbers in the FAQ below.
  • Cable relief. The IO board adds a lever arm; support its terminal wiring so the header joints never carry cable weight.

For bench alignment work, mount the stack on a slide-table stand so height and focus travel are one knob each. The professional slide-table model in the category is specified at a height of 700 mm over a 400 x 450 mm base:

Camera standsMachine vision camera stand, professional slide-table type, height 700 mm, 400 x 450 mm base (LS-MNT-0040)LS-MNT-0040
Type
Camera stands
$629.00View partBuild a system with it

The camera stands category also covers brackets, X/Y stages and rotary stages when the inspection needs an angle, and the accessories finder filters the whole line.

When the usual answer is wrong

The usual answer, "if the holes line up, the board fits", is the one that produced this failure in the first place. Two versions of it fail:

  • "Same camera family, so it fits." Accessory boards are drawn per variant, and variants built on new sensors can ship months before their IO board exists. In the reported case, the manufacturer's own answer to the misalignment was that boards for those sensor variants were still in development under new part numbers. The correct board did not exist yet, and no amount of shimming makes an old board right.
  • "I will jumper the pins that missed." Individual wires restore the contact count, not the design: they defeat the board's retention, strain fine-pitch pads, and work loose with vibration. Bench experiments only; a mounted machine gets the documented board.

If no board is released for your variant, treat that as the answer, and wire the camera's own GPIO connector directly until the board ships. When speccing the next machine, choose the camera so its accessories exist on day one; the board level camera selection guide covers that ordering.

Buying the camera side now? The camera finder fixes mount and interface first, the systems topic page collects the rest of the integration guides, and if a compatibility question survives this checklist, the contact page reaches the applications team, with free expert support with every order. This is exactly the question to send: camera model, revision, and the accessory you plan to pair with it.

Questions

Where do you find the documented pin map for a board camera?

In the camera's hardware or interface manual from the camera maker or its distributor, as a table listing each GPIO header pin with its signal name and electrical level. Match the manual to your camera's hardware revision, because revisions can move or reassign pins. If the manual covers several variants, use the section for your exact model suffix. Where the accessory's papers and the camera's manual disagree, stop and ask before wiring.

Can you adapt an IO board designed for a different camera variant?

Only as a bench experiment, never as the mounted solution. Soldering jumper wires across the missed pins defeats the board's retention, strains fine-pitch pads and works loose with vibration. It also hides the original error, because the next person assumes the stack is a validated pair. If no board is documented for your variant, ask for the planned part number and timeline instead.

Why do accessory listings say they fit a camera family?

Accessory descriptions often name the camera family rather than each variant. Families share outlines, lens mounts and screw patterns, which change slowly between generations, while the GPIO header position can change with a sensor revision. A family-level claim is a starting point, not a confirmation: before ordering, ask for the compatibility list at model-number level and compare it against the full model and hardware revision on the camera's label.

What torque should standoffs on a board camera stack get?

Camera makers rarely print a torque for M2 or M2.5 standoff screws. A workable ceiling is 0.2 to 0.3 N·m for M2 steel screws and about 0.4 N·m for M2.5, scaled down from standard screw-torque tables; nylon standoffs take less. Board laminates around 1.6 mm thick crack near the hole long before the head feels loose, so finger-tight plus about an eighth of a turn is the practical target, or a torque driver set to 0.25 N·m.

Can you verify pin alignment electrically instead of mechanically?

At the bench, unpowered: seat the stack dry, then run a continuity check from each terminal to the header pin the map says it reaches, with both pin tables in front of you. Continuity proves the path you planned exists. It does not make a misplaced pin harmless, because a pin on the wrong socket can still show continuity to somewhere unexpected, which is itself the finding. Mechanical agreement first, continuity second, power last.