Implementing Closed-Loop Feedback Between SPI and Stencil Printer

1. What Is Closed-Loop SPI-Printer Feedback?

In a conventional SMT line, the stencil printer and SPI system operate independently. When the SPI detects a print offset or volume issue, an operator must manually adjust the printer settings. This manual process is slow, reactive, and prone to over-correction.

Closed-loop feedback creates an automatic communication link between the SPI system and the stencil printer. The SPI measures paste deposit characteristics (position, volume, height) and sends correction data back to the printer, which automatically adjusts its parameters for the next print cycle. This creates a self-correcting system that continuously optimizes print quality without operator intervention.

Benefits of Closed-Loop Operation

• Reduced offset variation: Automatic X-Y-theta correction keeps print alignment within tighter tolerances than manual adjustment
• Faster response: Corrections are applied every print cycle, not after an operator notices a trend
• Consistent quality: Eliminates operator-to-operator variation in printer adjustment
• Reduced scrap: Catching and correcting drift before it produces defective boards
• Data traceability: Every correction is logged with timestamps, providing complete process history

2. Communication Protocols and Standards

The communication between SPI and printer can use several protocols depending on equipment manufacturer and generation:

Network Architecture

Closed-loop feedback typically uses a dedicated Ethernet connection between the SPI and printer, separate from the factory network. This ensures low-latency communication and prevents network congestion from delaying correction data.

• Direct connection: Point-to-point Ethernet between SPI and printer (simplest, most reliable)
• Dedicated switch: SPI, printer, and optional MES system on an isolated VLAN
• Factory network: Possible but not recommended due to latency and reliability concerns

3. Types of Automatic Corrections

Closed-loop systems can provide several types of corrections, ranging from simple alignment to complex process parameter adjustments:

Level 1: Print Alignment Correction (X-Y-Theta)

The most common and lowest-risk form of closed-loop control. The SPI measures the average paste deposit offset from pad center and sends X, Y, and rotational (theta) correction values to the printer.

• What it corrects: Board-to-stencil misalignment, fiducial recognition drift, thermal expansion effects
• Correction speed: Applied to the next print cycle
• Risk level: Low — alignment corrections are easily reversible
• Typical improvement: 40–60% reduction in offset variation (Cpk improvement from 1.0 to 1.5+)

Level 2: Squeegee Pressure Correction

When the SPI detects a systematic trend in paste height (e.g., front-to-back height gradient), it can signal the printer to adjust squeegee pressure or speed.

• What it corrects: Pressure drift, paste viscosity changes, squeegee blade wear
• Correction speed: Applied over multiple print cycles (averaged to prevent oscillation)
• Risk level: Medium — incorrect pressure adjustments can cause bridging or insufficient paste
• Typical improvement: 15–25% reduction in height variation

Level 3: Cleaning Cycle Triggering

The SPI can trigger automatic stencil under-wipe cycles when it detects patterns consistent with aperture clogging (e.g., progressively decreasing volume on specific pads).

• What it corrects: Aperture clogging, paste build-up, contamination
• Trigger criteria: Volume drop >15% on ≥3 adjacent pads for ≥2 consecutive boards
• Risk level: Low — cleaning is always beneficial; only cost is cycle time

Level 4: Separation Speed Adjustment

Advanced systems can adjust the stencil-to-board separation speed based on SPI volume data, optimizing paste release for different aperture sizes across the board.

• What it corrects: Poor paste release, especially on small apertures and high-aspect-ratio openings
• Risk level: Medium-high — separation speed affects all deposits simultaneously
• Recommendation: Implement only after Levels 1–3 are stable and validated

4. Step-by-Step Implementation Procedure

Prerequisites

• SPI system with closed-loop output capability (verify software version supports your printer)
• Stencil printer with closed-loop input capability (verify compatible firmware)
• Ethernet cable(s) and switch if needed
• Optimized SPI program (see our companion application note on Optimizing SPI Program Settings)
• Baseline process capability data (Cpk for offset and volume)

Phase 1: Physical Connection and Communication Setup

1. Connect the SPI and printer via dedicated Ethernet. Assign static IP addresses on the same subnet (e.g., SPI: 192.168.1.10, Printer: 192.168.1.20)
2. Configure the communication protocol on both machines (Hermes recommended)
3. Verify connectivity with a ping test from the SPI machine's diagnostic menu
4. Confirm board-tracking handshake — verify the SPI correctly identifies which board data corresponds to which print cycle
5. Run 5 boards and confirm the SPI is logging correction values (even before enabling automatic correction)

Phase 2: Offset Correction (Level 1) Implementation

1. Enable offset correction in “monitor only” mode — the SPI calculates corrections but does not send them to the printer
2. Run 25 boards and compare SPI-calculated corrections against the printer's own alignment data
3. Verify correlation: the correction direction and magnitude should match expected behavior
4. Set correction limits (see Section 5) — typically ±50µm for X/Y, ±0.02° for theta
5. Enable automatic correction with damping factor of 0.5 (the printer applies 50% of the calculated correction)
6. Run 25 more boards and verify offset Cpk improvement
7. If results are satisfactory, increase damping to 0.7–0.8 for faster convergence

Phase 3: Volume Feedback (Level 2–3) Implementation

1. After offset correction is stable for at least 1 week of production, proceed to volume feedback
2. Enable cleaning cycle triggering first (Level 3) as it is lowest risk
3. Monitor cleaning frequency for 1 week — excessive cleaning (>1 per 10 boards) indicates SPI threshold issues, not a printer problem
4. Enable squeegee pressure feedback in monitor mode
5. Validate suggested corrections against manual process engineering judgment for 50 boards
6. Enable automatic pressure correction with conservative limits (±5% of nominal pressure)

5. Setting Correction Limits and Damping

Correction limits and damping factors are the most critical tuning parameters for closed-loop stability. Incorrect settings can cause oscillation, where the system overcorrects in alternating directions.

Correction Limits

Correction limits define the maximum adjustment the SPI can request per cycle. Setting these appropriately prevents the system from making dangerously large corrections due to measurement errors or anomalous boards.

Damping Factor

The damping factor (also called gain or correction ratio) determines what fraction of the calculated correction is actually applied. A damping factor of 1.0 means full correction; 0.5 means half the correction is applied.

• 0.3–0.5: Conservative. Use during initial setup and validation. Slower convergence but very stable.
• 0.5–0.7: Standard. Recommended for production after validation. Good balance of speed and stability.
• 0.7–0.9: Aggressive. Use only with very low measurement noise and stable process. Risk of oscillation.
• 1.0: Not recommended. Full correction amplifies any measurement noise and frequently causes oscillation.

6. Validation Procedure

Proper validation is essential before committing closed-loop correction to full production. The following procedure provides documented evidence that the system is operating correctly.

Validation Test Protocol

1. Baseline run (25 boards, open-loop): Run with closed-loop disabled. Record offset and volume data for all boards. Calculate Cpk for X, Y, theta, and volume.
2. Closed-loop run (100 boards, corrections active): Enable offset corrections with validated damping factor. Record all correction values, measured offsets before and after correction, and volume data.
3. Intentional disturbance test: Introduce a known offset (e.g., shift a fiducial or stencil) and verify the closed-loop system detects and corrects it within 2–3 cycles.
4. Correction limit test: Introduce a disturbance exceeding the correction limit and verify the system alerts the operator rather than making an excessive correction.
5. Communication failure test: Disconnect the Ethernet cable during production and verify both machines handle the failure gracefully (printer should continue with last-known-good settings).

Acceptance Criteria

• Offset Cpk improvement of ≥30% compared to baseline
• No oscillation observed (corrections should not alternate sign for >3 consecutive boards)
• All corrections within defined limits for 100% of boards
• Proper alert generation when limits are exceeded
• Graceful failure handling with no defective boards produced during communication loss

7. Troubleshooting Common Issues

Oscillation (Hunting)

Symptom: Offset corrections alternate between positive and negative values, with increasing magnitude.

Root causes: Damping factor too high, measurement noise exceeding correction magnitude, or board-to-board variation (e.g., warped boards) being interpreted as systematic drift.

Solution: Reduce damping factor by 0.1 increments. Implement a moving-average filter (use 3–5 board average instead of single-board correction). Verify SPI measurement Gage R&R.

Corrections Not Taking Effect

Symptom: SPI sends corrections but printer offsets don't improve.

Root causes: Board tracking mismatch (SPI data for board N being applied to board N+1 or N+2), printer ignoring correction values due to internal limits, or correction direction inverted.

Solution: Verify board tracking synchronization. Check printer's correction log to confirm values are received. Test with an intentional large offset to verify correction direction.

Intermittent Communication Loss

Symptom: Corrections work sometimes but fail intermittently.

Root causes: Network congestion (if using shared factory network), cable quality issues, or timing conflicts between measurement completion and print cycle start.

Solution: Move to a dedicated Ethernet connection. Replace cables and test with cable certifier. Adjust SPI data transmission timing to ensure correction data is available before the printer starts its next cycle.

8. Advanced Considerations

Multi-Board Averaging

Instead of correcting based on a single board's data, use a rolling average of the last 3–5 boards. This filters out board-to-board variation (warpage, fiducial quality) and produces smoother, more stable corrections.

Zone-Based Corrections

Some advanced SPI systems can calculate different offsets for different zones of the board. This is useful for large boards where thermal expansion causes different offsets at the center versus edges, or where board flex creates local misalignment.

Integration with MES

Logging all correction data to a Manufacturing Execution System (MES) enables long-term trend analysis. Look for patterns such as progressive offset drift (indicating printer mechanical wear), time-of-day variation (thermal effects), or paste-lot-dependent volume changes.

Industry 4.0 and CFX

IPC-CFX (Connected Factory Exchange) is emerging as the next-generation standard for machine-to-machine communication in SMT. While Hermes handles board-level handshaking, CFX provides richer data exchange including process recipes, quality data, and machine state. Consider CFX readiness in your equipment selection for future-proofing.