Understanding Multi-Reflection Suppression in SPI

Every 3D SPI system relies on projecting light patterns onto the board surface and measuring the reflected light to reconstruct a 3D profile. This works well on matte surfaces that scatter light uniformly. But solder paste, bare copper pads, and board finishes like ENIG or immersion tin are highly reflective. When light bounces off one surface and reflects onto another before reaching the camera, it creates measurement artifacts that can look like solder paste where there is none, or mask paste deposits that are actually present. Multi-reflection suppression is the collection of techniques used to solve this fundamental optical challenge.

The Reflection Problem in 3D SPI

To understand why MRS matters, you need to understand how 3D measurement works in SPI and where reflections cause problems.

How 3D SPI Measurement Works

Most 3D SPI systems use a technique called phase-shift profilometry. The basic process is:

1. Pattern projection - A projector casts a series of sinusoidal fringe patterns (alternating light and dark stripes) onto the board surface
2. Phase shifting - The pattern is shifted multiple times (typically 4-8 shifts) while capturing images at each position
3. Phase calculation - Mathematical algorithms compute the phase value at each pixel from the captured images
4. Height conversion - The phase values are converted to height measurements using calibration data that maps phase to physical height

This process assumes that the light reaching the camera at each pixel came from the corresponding point on the surface directly below. When this assumption holds, the system produces accurate 3D measurements.

Where the Assumption Breaks Down

The assumption fails when specular (mirror-like) reflections come into play. There are several scenarios where this happens:

Scenario 1: Direct Specular Reflection

When the projected light hits a smooth, reflective surface (like a bare ENIG pad or a very smooth paste deposit), it reflects at the angle of incidence rather than scattering. If the reflected beam happens to enter the camera, it creates an extremely bright spot that saturates the sensor. Saturated pixels produce incorrect phase calculations and therefore incorrect height measurements.

Scenario 2: Inter-Reflection (Multi-Bounce)

This is the most insidious problem and the primary target of MRS technology. Light from the projector hits a reflective surface (like a pad), bounces onto a nearby surface (like a solder paste deposit or solder mask), and then reflects into the camera. The camera now sees light at that pixel from two sources: the direct reflection and the inter-reflected light. The resulting phase measurement is a weighted average of the two signals, producing a height value that is incorrect.

Scenario 3: Subsurface Scattering

Solder mask and some board substrates are translucent. Light can penetrate the surface, scatter within the material, and exit at a different location. This creates a blur effect in the phase map that reduces measurement accuracy, particularly for small pads near the solder mask edge.

How Multi-Reflection Suppression Works

MRS is not a single technique but a collection of approaches that work together to minimize the impact of reflections. Different SPI vendors use different combinations of these techniques, but the underlying physics is common.

Technique 1: Multi-Angle Illumination

Instead of projecting light from a single direction, the system projects patterns from multiple angles. At each angle, the specular reflection goes in a different direction. By analyzing which angles produce valid data at each pixel and which produce saturated or corrupted data, the system can select or combine the best data from each angle.

Technique 2: High Dynamic Range (HDR) Capture

The system captures multiple images at different exposure levels. Short exposures prevent saturation on highly reflective surfaces. Long exposures capture adequate signal on dark or diffuse surfaces. The best exposure data is selected or combined for each pixel.

Technique 3: Binary / Non-Sinusoidal Pattern Analysis

In addition to the sinusoidal patterns used for phase measurement, some systems project binary (on/off) stripe patterns. The response to binary patterns can be analyzed to detect areas affected by inter-reflection. Where the binary response does not match the expected behavior, the system flags those pixels as potentially corrupted.

Technique 4: Signal Quality Metrics

Advanced SPI systems compute quality metrics for each phase measurement, including:

• Modulation depth - How strong the fringe pattern signal is at each pixel. Low modulation indicates potential reflection problems
• Phase consistency - Whether phase values from different projection angles agree. Disagreement indicates reflection artifacts
• Residual error - The difference between the measured data and the expected sinusoidal pattern. Large residuals indicate corrupted data

Pixels with poor quality metrics are either excluded from the measurement or corrected using data from unaffected angles or exposures.

Technique 5: Algorithmic Compensation

Some systems use mathematical models of the inter-reflection process to estimate and subtract the reflected signal. If the system knows the geometry of the surfaces involved and their reflective properties, it can model the expected inter-reflection and correct for it computationally.

When MRS Matters Most

MRS is important for all 3D SPI applications, but certain conditions make it critical:

High-Reflectivity Board Finishes

• ENIG (Electroless Nickel Immersion Gold) - Highly reflective and increasingly common. Creates strong specular reflections
• Immersion tin - Very reflective when fresh
• OSP (Organic Solderability Preservative) on copper - Reflective bare copper visible through thin OSP layer
• Immersion silver - Extremely reflective surface finish

Fine-Pitch Components

As pad sizes decrease and spacing gets tighter, the impact of inter-reflection grows:

• Small pads have less area for valid measurement, so even a few corrupted pixels matter
• Close spacing means adjacent pad reflections are stronger
• Fine-pitch BGA pads (0.3-0.4mm pitch) are particularly susceptible
• 0201 and 01005 component pads where the entire deposit may be only a few pixels

Lead-Free Solder Paste

Lead-free solder pastes tend to be more reflective than leaded pastes due to their alloy composition. The paste surface can exhibit mirror-like reflections, especially with newer formulations designed for fine-pitch printing.

Mixed-Surface Boards

Boards with a mix of highly reflective and diffuse areas present the greatest challenge. The system must handle extreme dynamic range within a single field of view, with some areas requiring very short exposures and adjacent areas requiring long exposures.

Evaluating MRS Capability

When evaluating SPI systems, use these approaches to assess MRS performance:

Test 1: Known-Height Standard on Reflective Substrate

Place a calibrated step-height standard on a highly reflective surface (like a mirror or an ENIG coupon). Measure the standard height. Compare the measured value to the certified height. A system with good MRS will match the certified value within specification; a system with poor MRS will show significant errors.

Test 2: Your Actual Boards

This is the most important test. Run your most challenging boards, particularly those with ENIG finish, fine-pitch components, and high component density. Compare measurements at the same locations using multiple consecutive measurements (gauge R&R). A system with good MRS will show consistent results; poor MRS will show high variation, especially on reflective pads.

Test 3: Empty Pad Measurement

Measure an unpopulated board (no paste). All pad measurements should read zero height (within noise). If the system reports significant non-zero heights on bare pads, particularly near board edges or adjacent to solder mask steps, this indicates reflection artifacts.

Test 4: Compare Results Across Surface Finishes

If you use multiple board finishes (ENIG, HASL, OSP), measure the same stencil pattern on different finishes. With good MRS, the paste volume measurements should be consistent regardless of the board finish. Significant variation between finishes indicates that reflections are influencing the measurement.

The Speed-Accuracy Tradeoff

MRS techniques generally require additional data capture (more projection angles, more exposure levels, more images), which takes time. This creates a fundamental tradeoff between measurement accuracy and inspection speed:

The best SPI systems offer configurable MRS levels, allowing you to select the appropriate balance of speed and accuracy for each product. Simple boards with HASL finish may need only basic MRS, while complex boards with ENIG finish and fine-pitch BGAs may require enhanced MRS.

What to Look For in SPI Systems

When evaluating SPI systems for MRS capability, prioritize these features:

1. Multi-angle projection - Systems with 4+ projection angles have more data to work with for MRS
2. HDR capability - Multiple exposure levels are essential for handling mixed-reflectivity boards
3. Per-pixel quality metrics - The system should be able to identify and flag measurements affected by reflections
4. Configurable MRS levels - Ability to adjust MRS intensity per program or per board region
5. Consistent accuracy across finishes - Demonstrate comparable accuracy on ENIG, HASL, and OSP boards
6. Speed with MRS enabled - Understand the throughput impact of MRS at the level your boards require

Conclusion

Multi-reflection suppression is a fundamental capability that separates high-quality 3D SPI systems from those that provide unreliable measurements on real-world boards. As board finishes become more reflective and component pitches continue to shrink, MRS capability becomes increasingly critical.

Do not take vendor claims about MRS at face value. Test with your actual boards, on your actual finishes, and verify measurement accuracy independently. The difference between good and poor MRS implementation can mean the difference between actionable process control data and misleading numbers that lead to incorrect decisions.

When evaluating SPI systems, make MRS performance a top-tier evaluation criterion alongside measurement accuracy, throughput, and ease of use. A system that delivers accurate measurements on a demo board but fails on your reflective ENIG boards is not a system you can trust for production process control.