Ceramic Circuit Camera Bands: Separate LED Flicker from Real Surface Stripes

Measurement purpose

Separate temporally generated image bands from physical stripes before a ceramic circuit inspection decision.

Specimens and conditions

Surface and location

Stationary as-found region, stable landmarks and representative conductor, resistor or glass feature.

Comparisons

Preserved surface with registered captures and a suitable uniform witness region where available.

Equipment and records required

• Camera timing: Known shutter arrangement, native image data, active height, exposure and row timing.
• Illumination control: Recorded light setting and verified temporal behavior; suitable synchronization and measurement where used.

Method sequence

1. Preserve

   Locate bands in both material and sensor-row coordinates.

   Record: Original view and capture state.

2. Challenge

   Compare controlled exposure timing and illumination while preserving feature identity.

   Record: Registered image sequence and timing model.

3. Qualify

   Confirm required physical details under the selected capture and escalate persistent indications.

   Record: Supported method condition and remaining material question.

Decision and uncertainty

Assign temporal banding only when controlled acquisition comparisons support it; do not accept material from corrected appearance alone.

Unknown light waveform, row timing, trigger phase, clipping and spatial reflections can prevent unique attribution.

Imaging-method owner verifies acquisition; drawing and quality owners decide physical conformance.

Traceable outputs

Separate specimen coordinates from sensor rows

Locate the suspect stripe relative to a pad, resistor edge or overglaze opening. Retain a wider view showing whether it continues through unrelated materials and onto the surrounding fixture. A band spanning ceramic, metal and background in the same sensor rows suggests an acquisition contribution. It does not establish that every material inside the band is acceptable, nor that all horizontal features are artifacts.

Keep the circuit stationary and preserve its as-found surface. Do not clean the stripe away or apply image smoothing before collecting comparisons. Save native image data with exposure, gain, light setting, camera shutter mode and trigger information. Disable automatic image adjustments where a controlled capture is supported. If the camera silently changes exposure between frames, a moving band can otherwise be confused with changing material contrast.

Check when each row actually collects light

A conventional rolling shutter starts equal-duration exposures sequentially across rows. Let row r begin at t0 plus r times delta-t, where delta-t is the interval between successive row starts. Each row then integrates the light arriving during its own exposure interval Te. A stationary uniform surface can therefore produce unequal row values if the illumination changes between those intervals.

Read the selected camera's actual mode and timing; frame rate alone does not define delta-t. A global shutter collects the rows over a common interval and avoids this particular row-time mechanism, but temporally unstable light can still change brightness between complete frames. Global reset with staggered exposure ends is another timing arrangement and must not be treated as an ordinary global shutter.

Calculate the light average over one exposure

For an independent screening model, suppose illumination is spatially uniform and varies as L(t) = L0[1 + m cos(2 pi fL t + phi)]. L0 is mean illumination, m is its fractional sinusoidal modulation between zero and one, fL is modulation frequency in hertz and phi is phase in radians. Assume a stationary, uniform-reflectance region and a linear, nonclipped sensor. Averaging this expression over each row exposure gives the equation below.

The result separates two effects. Row-start spacing changes the phase recorded in successive rows, while exposure duration attenuates the modulation by time averaging. It predicts an image-formation effect without assigning any stripe to the fired material. A PWM light has a nonsinusoidal waveform and may require direct integration of its measured output or multiple frequency components. Do not fit a single sine to a complex driver and call the residual a manufacturing defect.

Work a stationary-surface example before adjusting the process

Assume a 1,000 Hz illumination modulation with m = 0.20 and a row-start interval of 10 microseconds. The modulation advances by one complete cycle every 100 rows. At a 100 microsecond exposure, the averaging factor is sin(0.1 pi)/(0.1 pi), approximately 0.98363. The ideal row averages can therefore vary about the mean by approximately plus or minus 19.67 percent. This substantial banding requires neither a moving part nor a striped resistor.

At an exposure of exactly 1 millisecond, the assumed sine completes one whole period within every row interval and its average contribution becomes zero. That result is specific to the stated ideal waveform and exact timing. It is not a universal recommended exposure. Mixed ambient light, imperfect synchronization, other modulation frequencies, clipping and light-driver transients can prevent the cancellation. Measure the actual capture rather than treating a clean calculation as inspection qualification.

Change a temporal variable while preserving the physical evidence

First repeat the same stationary view at several recorded exposure durations with unchanged light geometry. Keep the useful regions nonclipped and distinguish brightness scaling from changes in band amplitude. Then compare with a suitably controlled light source whose temporal stability has been checked at the selected operating setting. Dimming a light can change its modulation, so a steady full-brightness condition does not automatically describe reduced brightness.

Where permitted, capture a uniform witness region in the field and examine whether the same row pattern appears there. A registered change in specimen orientation can provide another test: material features follow the part, while sensor-row timing remains tied to the camera. Rotation also changes reflection geometry, so this is supporting evidence rather than a single conclusive rule. Keep a persistent stripe unresolved if these comparisons do not separate the mechanisms.

Do not confuse an exposure signal with an all-row flash window

If a synchronized flash is considered, determine whether all relevant rows share an exposure interval. For the simple sequential model with N rows, the common interval has duration Te minus (N minus one) delta-t when that value is positive. With 1,000 rows and a 10 microsecond row offset, the last row begins 9.99 milliseconds after the first. A 1 millisecond exposure has no common all-row interval, despite each row receiving that exposure duration.

An assumed 12 millisecond exposure instead leaves a 2.01 millisecond common interval. This is a timing illustration, not a flash-pulse specification. Actual camera output delay, light rise and fall times, trigger uncertainty, ambient contribution and permitted pulse energy still matter. An Exposure Active signal can remain asserted from the first row's start until the last row's end; that wider interval does not mean all rows are simultaneously collecting light. Use the documented synchronization arrangement and verify its optical result.

Verify that the cure has not removed required information

A longer exposure can reduce a temporal artifact while increasing motion smear during routine handling. A new light can change glare from silver conductors or glossy green glass. A different shutter mode can change the usable field or acquisition sequence. Recheck the actual pad gap, printed resistor boundary and protective-layer opening after selecting the revised timing. Absence of bands is necessary only for this problem; it is not complete optical-method approval.

Do not remove stripes with a row-subtraction filter and then use the edited image as proof that the product was uniform. Such processing can remove genuine features aligned with the same rows. Preserve the original capture and compare an independently acquired image under a qualified illumination condition. If physical striping remains, investigate the applicable printing, surface or material mechanism with evidence tied to the original location.

Deliver a reproducible temporal-capture decision

Retain the native image sequence, row direction, active image height, shutter mode, exposure durations, trigger settings and light operating state. Include any measured illumination waveform with its acquisition bandwidth and timing relationship, rather than only a nominal driver frequency. Identify whether bands changed in row coordinates, remained on the specimen, or could not be separated within the available method.

For a ChipSimple ceramic circuit enquiry, send the relevant layer drawing and the actual feature being judged alongside those images. The review can then distinguish an invalid camera indication from a drawing-specific material question. Neither smooth brightness nor glossy glass color alone proves fired-layer coverage or electrical performance. The useful output is a supported capture condition and a clear disposition route for any indication that survives it.