Cameras & sensorsExplainer

Reported exposure time differs from set exposure: the firmware offset

Why the reported exposure time differs from the exposure you set after a firmware update, how the sensor offset shifts values, and what to adjust.

LumaSpec Vision10 min read Share on LinkedIn

On this page 10 sections
  1. 01The two exposure numbers a camera has
  2. 02What the newer firmware build reports
  3. 03Exposure offset and exposure efficiency
  4. 04Worked example: the same set-point on two firmware builds
  5. 05Why the minimum rises while the image does not
  6. 06Confirm the offset on your own camera in four checks
  7. 07Getting a firmware build the safe way
  8. 08When the usual answer is wrong
  9. 09The decision, case by case
  10. 10Questions

You set 20 µs on Monday. On Friday, after a firmware update, the same camera rejects that value and reports a minimum close to 38 µs. Nothing in the optics moved and the frames look identical, so the change sits in how the camera names its exposure, not in how it gathers light. Anyone running a fleet of area scan cameras meets this sooner or later, because two batches of one model can arrive on different firmware builds. What follows is the mechanism, the arithmetic that predicts the new number, and the decisions: what to re-tune (nothing), what to update (logs and acceptance limits), and how to obtain firmware builds the safe way.

The two exposure numbers a camera has

An industrial camera carries two exposure numbers. The first is the value you write: the commanded integration time in microseconds, held by the camera's exposure feature alongside a minimum, a maximum and an increment. The second is the value the camera reports back once you set it. Manuals for USB3 Vision camera families describe the loop plainly: the camera adjusts the exposure time inside a range whose limits the user sets, then reports what it accepted. While the firmware reports the commanded value unchanged, write and read match. They split apart when a build starts reporting a figure that includes overhead the commanded value never contained.

How a raw reported number maps to physical time is its own small subject; the guide on camera exposure and gain units works through it value by value.

What the newer firmware build reports

Take one camera model delivered in two batches. On the older build, the reported value was the commanded value: what you set is what you read, and the floor sat there too. The newer build reports the exposure as it actually runs, including the sensor's exposure efficiency — the time the shutter electronics take to open, hold and close. The camera maker's explanation fits in one line: the exposure time now reflects the actual exposure, so the minimum reads higher while the exposure itself is unchanged, because the old build listed the exposure without the sensor offset (the maker's note on this family puts that offset at 15–20 µs). Camera manuals state the same property in general terms: every camera has an exposure time offset, and its size follows the sensor and the camera design.

Exposure offset and exposure efficiency

The offset is dead time at the edges of the exposure: the sensor moves charge, opens its electronic shutter and settles before the first usable photon lands. Exposure efficiency is the share of the commanded integration time during which charge is actually collected; the EMVA 1288 standard measures it as a sensor parameter, and a real shutter always loses something to it. Vendors publish the same physics as an exposure time offset in microseconds, and the firmware decides whether that number stays hidden or lands in the report.

μ = t_collect / t_set exposure efficiency, μ < 1 t_report_new = t_report_old + t_offset t_offset ≈ 15–20 µs sensor offset, the camera family in this case

Folding the offset into the reported value lifts the smallest number the camera accepts by the same margin. The collection window behind that number does not move.

Worked example: the same set-point on two firmware builds

The case: a station tuned to a 20 µs exposure, now watching the camera refuse that value after the firmware moved to the newer build. Inputs: the set-point from the production recipe; the old build's reported minimum equal to it; the sensor offset from the maker's note on this family. Rule: the new build adds the offset to the value it reports. Arithmetic: 20 + 15 = 35 µs at the low end, 20 + 20 = 40 µs at the high end, roughly 38 µs in the middle.

Result: the same physical setting reads 35–40 µs on the new build, and the reported minimum climbs from the old commanded value to 35–40 µs, while the collection window stays near 20 µs. The test that proves it: aim the camera at a fixed grey target before and after the update and compare mean grey values. A move of no more than ±2 grey levels on an 8-bit image says the light-gathering time did not change.

Decision rule: frames within ±2 grey levels of before means the change is bookkeeping — keep the lighting, raise log bands and alarm limits by the offset, and enter the value the camera now reports. Frames measurably darker means a real exposure change — chase auto exposure, gain and pixel format first, not the lights.

What you observe Build before the update Build after the update
Reported minimum exposure 20 µs 35–40 µs
Reported value for the old set-point 20 µs 35–40 µs
Charge-collection window about 20 µs about 20 µs
Mean grey on a fixed target unchanged unchanged

Why the minimum rises while the image does not

The minimum exposure time is the smallest value the exposure feature accepts. Before the update the camera named the commanded number, so the floor sat at the set-point. After it, the camera names the true window including the offset, so the floor label sits at 35–40 µs and smaller entries are refused, while the photons per frame stay the same. Motion blur, which follows the collection window, is unchanged for the same reason.

Two neighbors of this floor behave the same way. Readout time hides inside the minimum on many cameras — does minimum exposure time include sensor readout time covers that case — and in trigger mode the floor can come from the trigger circuitry rather than the sensor, as minimum exposure time in trigger mode shows.

Confirm the offset on your own camera in four checks

  1. Read the exposure feature's minimum from both builds in a GenICam browser and note the unit; on SFNC-compliant cameras the value is in microseconds.
  2. Write the same commanded value on both builds and subtract the two reported numbers. The difference is that camera's offset.
  3. Compare mean grey values on a fixed target before and after the update. A move within ±2 grey levels on an 8-bit image says the collection window held.
  4. Read the camera manual for a stated exposure time offset. When the manual gives a number for your model, use that instead of this case's range.

The habit behind all four is reading the unit before you write a value. Exposure lives in microseconds here; a controller's scheduler cycle time reported in microseconds follows the same pattern elsewhere on the plant floor, and assuming the unit instead of reading it is the classic way to be wrong by a factor of 1,000. The camera-side version of the check is auto exposure limit units in microseconds: query before you set.

Getting a firmware build the safe way

Two batches, two builds: the reflex is to hunt the missing build number on a download page. The safe path is shorter. Firmware for a camera comes from its maker's support portal or from the distributor that sold the camera; send the model, the serial number and the current firmware version, and ask for the build that matches the rest of the fleet. If the camera came from us, contact us with the model and serial and we will point you at the right build.

Version-number quirks are not a camera specialty either: teams restoring an HMI project between panels that run different firmware versions meet the same batch-mixing problem one rack over.

If an update ever does leave a camera unrecognizable, camera bricked after firmware update: DFU recovery steps that work is the recovery path, and camera firmware version mismatch: fix detection failures safely covers the cousin problem where build differences break device detection.

When the usual answer is wrong

Three situations look like this problem and are not it.

If mean grey actually dropped after the update, the offset is not the cause, because a reporting change cannot move photons. Check whether auto exposure retargeted, a gain default reset or the pixel format changed, then re-shoot the grey target.

On a UVC camera the reported exposure may not be in microseconds at all: those controls step in powers of two, which is why a UVC camera's exposure in log2 units can read 0.25 s where you set 0.125 s. Different trap, same lesson.

And if the task genuinely needs a collection window below the old minimum, no firmware build will supply it. The floor is a hardware number and camera families differ: one publishes 26 µs as its minimum in every triggered mode. When the floor itself is the blocker, the answer is hardware chosen for short exposure — for example the MV-CH120-11UC, the high-speed 12 MP USB3 variant of its series.

Area scan camerasHikrobot MV-CH120-11UC 12 MP Area Scan Camera, USB3, IMX304, High-speed version, ColorMV-CH120-11UC
Sensor format
1.1"
Lens mount
C-mount
Interface
USB3
Resolution
12 MP
$2,339.00 · quotedView and quoteBuild a system with it

The decision, case by case

What you see after the update What it means What to do
Minimum reads higher by the offset, images unchanged The offset is now inside the report Keep the lighting; raise log bands and limits by the offset; enter values as the camera reports them
Minimum reads higher and frames are darker A real exposure change, not the offset Check auto exposure, gain, pixel format and AOI before touching lights
Two builds across one line Set-points read differently per batch Unify on one build through the maker's support channel, or store a per-camera offset
The old minimum was already too long for the task A hardware floor, not a report Shortlist cameras specified for short exposure and check the stated minimum

If the outcome is new hardware, the camera finder filters by sensor, resolution, shutter and interface, and the configurator walks the chain from sensor to cable. The cameras topic page gathers the rest of the camera-side guides, including the units and readout pieces linked above.

Questions

What unit is the exposure time value on an industrial camera?

On GigE Vision and USB3 Vision cameras that follow the GenICam SFNC naming, the exposure feature is defined in microseconds, and the values you read and write carry that unit. Confirm it anyway in the camera's own feature list, because other tool stacks differ: UVC exposure controls use logarithmic base-2 steps, so a one-step change doubles or halves the time. When you log values from mixed sources, store the unit next to the value.

Why does my camera now reject the 20 µs exposure it accepted before the update?

The newer firmware folds the sensor's exposure offset into the reported and accepted range. The value that used to read as the commanded set-point now reads roughly 35–40 µs, so entries below that floor are refused. The light-gathering window is unchanged; enter the value the camera reports for the old setting, or ask the maker's support channel for the build behavior you need.

Does the higher reported minimum increase motion blur?

No. Blur comes from the window during which the pixel actually collects charge, and that window is unchanged by the reporting change. The offset the newer build exposes is overhead the sensor always spent, which the old firmware simply did not show. If blur visibly worsens after an update, look elsewhere: an auto exposure target change, a gain default or a pixel format change will alter the image, a reporting change will not.

Can I run two firmware versions across the cameras in one line?

You can, but log it. The same commanded setting reads one offset higher on the newer build, so any recipe, alarm limit or throughput calculation that assumes one scale will misfire on the other batch. Either unify the fleet on one build through the maker's support channel, or store a per-camera offset and apply it where set-points and limits are written.

How do I get a firmware build that is missing from the download page?

Ask the camera maker's support portal, or the distributor that sold you the cameras, quoting the model, the serial number and the current firmware version. Support can supply builds that the public download area does not list and can tell you what changed. Do not pull camera firmware from forums or file mirrors; a wrong image can stop the camera completely, and recovery then runs through the maker's own tools.

Does every camera model have this exposure offset?

Every sensor spends some time opening and settling its shutter, and camera manuals say so in those terms: each camera has an exposure time offset whose size follows the sensor and the camera design. The 15–20 µs figure belongs to one family and one firmware change; your camera's number may be a few microseconds or tens of microseconds. Read it from the manual, or measure it as the difference between two builds.