Closed-Loop Temperature Control in Laser Solder Paste Processing

Closed-loop temperature control in laser solder paste processing means the machine measures joint temperature in real time — typically with a pyrometer aimed co-axially with the beam — and continuously modulates laser power so the paste follows a programmed reflow profile instead of a fixed power-and-time recipe. It matters more for paste than for wire or ball feeding, because printed paste volume, flux burn-off and pad thermal mass vary joint to joint: a fixed recipe that reaches 245 °C on one pad can leave a cold, non-wetted joint on the next and scorch flux on a third. The loop turns the laser from a constant heat gun into a controlled heat source with a measurable, recordable temperature trace per joint.

Why does an open-loop laser power/time recipe drift on solder paste joints?

With a fixed recipe, the only variables the machine controls are laser power, pulse duration and spot size. Everything that determines how hot the paste actually gets is left to chance:

Paste deposit variation. Stencil print volume typically varies ±10 % across a panel; a 30 % thicker deposit needs noticeably more energy to reach the same peak, and it also absorbs heat differently while the flux volatilises.
Flux endotherm. Flux activation and solvent evaporation consume energy in the 120–180 °C window. A fresh, solvent-rich deposit pulls the joint temperature down; a dried deposit does not. Open loop cannot tell the difference.
Absorptivity differences. A dark moulded package, a ceramic body and a bare copper pad absorb 980 nm laser energy at very different rates, so the same power produces very different ramp rates.
Thermal mass and heat sinking. A 0402 pad on 0.5 mm FR-4 and a connector pin on a 2 mm aluminium heat spreader need an order of magnitude different energy for the same profile.
Optical drift. Window contamination and diode ageing reduce delivered power by several percent over a production shift, silently shifting every joint colder.

Closed loop removes all of these as first-order effects: the controller holds the measured temperature to the setpoint and lets power float wherever it needs to be, within limits you set.

What hardware and control law does the loop actually need?

The practical architecture is a fast pyrometer input feeding a PID controller with feed-forward, driving the laser’s analogue or PWM modulation input. Typical specifications for a paste process:

Element	Typical specification	Why it matters for paste
Pyrometer wavelength	1.5–2.5 µm (InGaAs or extended InGaAs), filtered away from the 915/976 nm laser	Prevents the laser’s own light from biasing the reading
Response time	1–10 ms	Cycle is 1–3 s; a 100 ms sensor is too slow to catch the melt transition
Spot size	0.3–2 mm, ≤ joint footprint	Oversized spot averages in cold surrounding board and reads low
Sample / loop rate	1–5 kHz sampling, PID update ≥1 kHz	Needed to hold peak without 20–30 °C overshoot
Emissivity handling	Recipe table per joint, or two-colour (ratio) pyrometer	Paste is matte (~0.9) before melt and shiny (~0.1–0.4) after; a single fixed ε reads falsely cold at melt and the loop overshoots
Control law	PID + feed-forward from a thermal model, with ramp-rate limiting	Pure PID lags on a 1 s cycle; feed-forward gets you into the right power band immediately
Power ceiling / saturation alarm	Alarm at 95–100 % of commanded power for N consecutive joints	Early warning of window contamination or a bad joint fit
Verification channel	Per-joint temperature trace stored with board ID	Traceability for automotive and medical builds

One design point that is frequently missed: with a galvo-scanned head working multiple joints, the pyrometer only sees the joint it is aimed at. Make sure the sampling window is synchronised with the dwell on that joint, and do not let the loop average across a jump between two pads.

What profile window should the loop hold for SAC305 paste?

Laser paste reflow is much faster than oven reflow, so the numbers look different from a tunnel profile. A workable starting window for SAC305 (217 °C solidus) on a typical SMT pad:

Stage	Target	Loop behaviour
Preheat / flux activation	120–170 °C, 0.3–1.0 s	Ramp-limited, not full power; avoids flux spatter
Ramp to peak	100–300 °C/s	Feed-forward active, PID trims
Peak temperature	235–250 °C, held 0.2–1.0 s	Hard clamp; overshoot should be ≤10 °C
Time above liquidus	0.3–1.5 s total	Verified by thermocouple, not by recipe
Cooling	Natural or forced, ≥50 °C/s to below 150 °C	Laser off; loop records decay for AOI correlation

Treat these as a starting envelope, then measure. For paste with a different alloy or a no-clean flux that needs a longer soak, extend the activation stage first — peak temperature is the wrong knob to turn, because every extra 10 °C at peak increases spatter and intermetallic growth without improving wetting.

What do the failure symptoms look like, and how do you troubleshoot them?

Symptom	Likely loop cause	Check	Action
Cold joint, paste not fully fused, grainy surface	Loop holding a low reading — emissivity set too high, or spot oversized	Compare pyrometer trace against a K-type thermocouple on a dummy board	Recalibrate ε, shrink spot to joint size, raise peak setpoint 5 °C
Solder spatter / flux splatter around pad	Ramp too fast through 120–170 °C	Look at ramp slope in the trace, not just the peak	Enable ramp-rate limiting, extend activation stage
Overshoot alarm at melt, shiny joint with burn ring	Emissivity drop at melt causing false-cold reading	Trace shows a dip just before peak	Switch to ratio pyrometer or clamp power after melt detection
Power saturation alarm on a few joints only	Those pads have excess thermal mass or paste	Cross-check paste volume on those pads	Per-joint recipe, or add local preheat; investigate print defect
Gradual shift to cold joints across a shift	Window contamination reducing delivered power	Power meter check at the nozzle at shift start	Clean window, log delivered vs commanded power trend
Voids in X-ray after reflow	Too short a time above liquidus, or peak set too low	TC-verified TAL vs X-ray void percentage	Extend hold time at peak before raising temperature

How do you validate closed-loop control before release to production?

Thermocouple correlation. Attach a 0.1 mm K-type wire to a paste deposit on a real board, run the loop, and compare peak and TAL against the pyrometer trace. Agreement within ±5–10 °C is the pass criterion; larger gaps mean the emissivity or spot geometry is wrong.
Baseline recipe across the panel. Run 30 consecutive joints and check that peak spread is ≤10 °C. Wider spread points to per-joint thermal mass that the loop cannot overcome within its power ceiling.
Deliberate disturbance test. Add 30 % extra paste to one pad and remove paste from another; the loop should still land inside the peak window on both, or raise a saturation alarm rather than silently under-heat.
Destructive checks. Cross-section for wetting angle and intermetallic thickness, shear or pull test per J-STD-001 / IPC-A-610 criteria, and X-ray void measurement on the first article.
Trace review. Confirm every joint in the log has a complete trace with no truncation, and that saturation events are flagged rather than averaged away.

FAQ

Can one closed loop handle joints with very different thermal mass on the same board?
Only within its power ceiling. Small pads and large ground-plane pads need separate recipes with their own power limits and, often, different ramp rates. If a joint needs more than about 80 % of available power to follow the profile, add local preheat instead of pushing the loop harder.

Do I need a two-colour pyrometer, or is single-colour enough?
Single-colour works if you characterise emissivity before and after melt per joint type and clamp power at the melt transition. Ratio pyrometers cost more but remove the emissivity-shift problem entirely, which is worth it on high-mix lines where re-characterising every product is impractical.

What sampling rate is enough for a 1-second cycle?
1 kHz sampling with a ≥1 kHz PID update is the practical minimum; 5 kHz gives visibly tighter peak control and lets you detect the melt plateau reliably. Below 500 Hz you will see 20–30 °C overshoot that no tuning fully removes.

Does the loop replace profile development?
No. The loop holds whatever profile you programme; it does not decide the correct peak or TAL for your alloy, flux and component. Profile development on thermocoupled samples comes first, then the loop guarantees repeatability of that profile in production.

How often should the loop be recalibrated?
Verify delivered power at the nozzle and pyrometer accuracy at every shift start and after any window cleaning or optics change; do a full thermocouple correlation whenever paste supplier, stencil, or board design changes.

If you are evaluating closed-loop laser solder paste systems and want the temperature trace, pyrometer configuration and alarm logic reviewed against your specific boards, send us your joint list and we will map it to a suitable configuration.