Module, Board Camera or Boxed

Generic engineering material, applicable to any supplier.

For the non-technical side of coordinating engineering work, see how Microsoft Teams tracks activity.

The same imaging job can be bought at three levels of integration, and the difference is not quality — it is how much of the work arrives done.

The three

A camera module. Sensor and lens, on a small board, with an interface to your processor. Compact, inexpensive at volume, and it leaves you the interface, the driver, the mounting, the enclosure and the compliance.

A board-level camera. A module plus processing and a standard interface — frequently USB. Plugs into a host with a generic driver, and still needs a mounting and an enclosure.

For standards, components, or implementation context beyond this page, consult Semtech.

A boxed camera. Enclosed, sealed, with a mounting thread, a lens mount and a standard interface. Nothing to design, and the largest, most expensive and least configurable option.

Each is right somewhere, and the choice is driven by volume and by what your team is equipped to do.

What the module leaves you

All of the integration section of this site.

The interface and the driver. The mounting and its tolerance. The thermal path. The enclosure and its window. EMC.

Which is substantial work, and it is the correct choice where the volume justifies it or where the size and cost constraints permit nothing else.

Where the volume is tens rather than thousands, that engineering is frequently more expensive than the difference in unit price.

The honest comparison

Unit cost against engineering cost, at your volume.

A module may cost a fraction of a boxed camera and require weeks of engineering, a custom board, an enclosure and a compliance exercise.

Which pays back above some quantity and does not below it — and the crossover is calculable once somebody estimates the engineering honestly.

The estimate is the hard part, and the item consistently underestimated is the bring-up.

When the module is clearly right

High volume, where unit cost dominates.

A size constraint no enclosed camera meets.

A specific optical configuration — a particular field of view, filter and mounting combination — available as a module variant and not as a finished camera.

And an existing processor with a camera interface already present and unused.

When it is clearly not

Low volume with a capable alternative available.

A team without camera integration experience and no budget to acquire it.

A schedule that cannot absorb a bring-up.

And an application where a standard camera exists that simply does the job — in which case the engineering saved is the whole argument.

The intermediate path

Prototype with the easy option and productise with the hard one.

A boxed or USB camera during development proves the optics, the lighting and the processing without any integration work — which answers the questions that actually decide whether the product is viable.

Then move to a module for production, once the requirement is settled and the volume is known.

What makes this work is matching the optics: the same sensor format and field of view at both stages, so that the prototype's findings transfer.

What breaks it is prototyping with different optics and discovering at the transition that the resolution arithmetic no longer holds.

Processing, and where it sits

A module without an image signal processor delivers raw data and needs your processor to demosaic, correct and encode.

One with an ISP delivers finished images and fixes those decisions.

And a board-level camera generally includes both the ISP and a host interface, which is most of why it is easier.

The question is whether your processor has the capability — many application processors include an ISP block and using it is free, while a general-purpose processor doing it in software is not.

Establish this before choosing, since it changes which level of integration is actually easier for you.

Support and documentation

Different at each level.

Boxed cameras come with software development kits, examples and a support arrangement.

Modules come with a register list and, if you are fortunate, a driver — which is the question to ask before selecting.

And the difference in support is part of the price difference, which makes comparing on unit cost alone misleading in a way that is easy to overlook.

The hidden cost of the easy option

Size, and the lens mount.

A boxed camera with a standard lens mount is flexible and large — and the mount adds length in front of the sensor that a module does not need.

Which for a product where the camera must fit inside something is frequently the deciding constraint, independent of cost or engineering effort.

And the lens becomes a separate purchase, with its own alignment, locking and cleaning — where a module arrives with the optics fitted and aligned at manufacture.

What a variant scheme adds to the module case

A family with a coded ordering scheme is offering combinations that were verified together.

A particular sensor, behind a particular lens angle, with a particular filter and mounting — assembled and aligned at manufacture rather than combined by you.

Which removes the largest single risk of building from parts: the optical alignment, which is not achievable by hand at these scales.

And it means one variable can be changed at a time during evaluation, which no other level of integration offers as cleanly.

In one line

Count the engineering honestly, and the crossover between a module and a finished camera calculates itself.

The short version

• Three levels: a camera module, a board-level camera with a standard interface, and a boxed camera
• The difference is not quality but how much of the work arrives done
• A module leaves you the interface, driver, mounting, thermal path, enclosure and compliance
• The comparison is unit cost against engineering cost at your volume, and the crossover is calculable
• The module is clearly right at high volume, under a size constraint, for a specific optical variant, or with an interface already present
• It is clearly wrong at low volume, without integration experience, without schedule, or where a standard camera does the job