Laser soldering temperature control works best when the process is closed-loop: the joint temperature is measured in real time by a non-contact sensor (pyrometer or thermopile) and the laser power is modulated by a PID l…
Laser soldering temperature control works best when the process is closed-loop: the joint temperature is measured in real time by a non-contact sensor (pyrometer or thermopile) and the laser power is modulated by a PID loop so the measured temperature — not a fixed power/time recipe — follows a defined profile. In practice that means setting four things per solder alloy and per joint geometry: preheat temperature, ramp rate, peak joint temperature and time above liquidus, with the loop holding peak temperature within roughly ±3–5 °C part-to-part. Everything else — power headroom, spot size, emissivity calibration, sensor field of view — exists to make that measurement trustworthy.
What does closed-loop temperature control actually hold constant in laser soldering?
A laser soldering head delivers energy in milliseconds; without feedback, joint temperature is a function of laser power, absorption at the surface (which changes as flux activates and solder melts), and the local thermal mass of the pad, lead and substrate. Open-loop recipes therefore drift as soon as the supplier changes a lot of paste, the copper pad thickness changes, or the component lead mass varies from reel to reel.
The control loop shifts the problem from “how many watts” to “what temperature, measured where”:
- Sensor: a two-colour (ratio) IR pyrometer or a fast thermopile, focused on the joint, not the laser spot. Two-colour sensors tolerate emissivity changes better; single-colour sensors need per-part emissivity compensation.
- Sampling: 100–1000 Hz is typical for laser soldering because ramp rates reach 80–150 °C/s. A 10 Hz sensor simply cannot capture the ramp.
- Actuator: PWM or analogue modulation of the laser diode, with a control cycle of 1–5 ms.
- Power headroom: size the laser so steady-state soldering needs 50–70% of maximum power — enough margin for cold boards or heavy ground planes, without overshoot.
- Spot size: matched to the joint, typically ±0.2 mm of the pad-plus-lead footprint; an oversized spot heats adjacent parts and biases the sensor reading.
The acceptance criterion for the loop itself is repeatability: peak joint temperature should stay within ±3–5 °C across a 30-piece capability run (Cpk above 1.33) before the recipe is released.
How do you set the temperature profile for each solder alloy?
Profiles are alloy-driven, then trimmed for component mass. Laser heating is fast, so total time from ambient to peak is usually 1–4 s, and time above liquidus is deliberately short to limit intermetallic growth and flux burn-off.
| Alloy | Melting range | Preheat / soak | Ramp rate | Peak joint temp | Time above liquidus |
|---|---|---|---|---|---|
| SAC305 (Sn96.5Ag3Cu0.5) | 217–220 °C | 130–150 °C, 0.3–1 s | 80–150 °C/s | 235–248 °C | 0.5–1.5 s |
| Sn63Pb37 | 183 °C eutectic | 110–130 °C, 0.2–0.8 s | 80–150 °C/s | 205–220 °C | 0.5–1.2 s |
| Sn42Bi58 | 138 °C eutectic | 80–95 °C | 60–120 °C/s | 165–180 °C | 0.3–1.0 s |
| AuSn20 (die attach / optical packages) | 280 °C eutectic | 180–200 °C | 100–200 °C/s | 310–330 °C | 0.5–1.5 s |
Practical rules of thumb from production lines:
- Set peak 15–30 °C above liquidus for SAC alloys. Going beyond 250 °C on FR-4 with thin traces buys nothing and accelerates flux charring and pad delamination.
- Keep ramp below roughly 150 °C/s on ceramic capacitors and chip resistors to limit thermal shock cracking; slow to 50–80 °C/s for 0402 and smaller.
- Preheat matters more than it looks: a 20–30 °C board preheat (or a short in-head preheat step) narrows part-to-part variation and lets you lower peak power.
- Cooling should be natural or low-pressure nitrogen — forced air on a 0.4 mm pitch connector can shift the part before the joint solidifies.
How should the temperature sensor be calibrated and kept honest?
Sensor drift is the most common root cause of “the recipe worked last month”. Establish three checks:
- Thermocouple correlation. Attach a 0.1 mm K-type or N-type thermocouple to a representative joint (or a coupon with the same pad and lead geometry) and run the laser profile. Compare TC peak with pyrometer peak; adjust emissivity/offset until they agree within ±5 °C. Repeat for each joint family, not once per machine.
- Field-of-view verification. Confirm the sensor spot sits inside the joint area, typically 0.5–1.5 mm diameter. A sensor aimed partly at bare laminate reads low and drives the loop to over-power — the classic “cold joint with a perfect profile” symptom.
- Shift-level checks. Clean the protective window and check lens condition at every shift change; flux spatter on the optics attenuates the signal and biases the reading. Log a weekly verification run with a TC coupon and trend the deviation — a drift of more than 5 °C over a week means recalibrate before releasing the line.
Also verify nitrogen coverage if you use it: oxygen above roughly 1000 ppm at the joint raises the temperature needed for good wetting and encourages solder-ball spatter, which then contaminates the optics and restarts the drift cycle.
What do temperature-control failures look like, and how do you troubleshoot them?
| Symptom | Likely cause | Check and corrective action |
|---|---|---|
| Grainy, dull fillet; incomplete wetting | Measured temperature reads higher than the true joint temperature (sensor off-joint, contaminated window) | Run a thermocouple correlation; realign sensor FOV; clean optics; re-check emissivity setting |
| Peak overshoot above setpoint by more than 15 °C | Excess power headroom, integral windup, aggressive derivative term | Reduce max power limit to 1.5× steady-state need; lengthen integral time; verify PWM cycle matches loop rate |
| Joint-to-joint variation on the same panel | Different thermal mass or surface finish (ENIG vs OSP vs HASL); no preheat | Split recipes by pad/lead family; add preheat step; verify fixture clamping and repeatability of part seating |
| Solder balls and flux charring around the joint | Ramp too fast, TAL too long, spot oversized | Reduce peak 5–10 °C, shorten TAL to under 1 s, shrink spot to joint footprint, check paste deposit volume |
| Head-in-pillow on BGA or connector balls | Uneven heating; ball oxidized before reflow; insufficient time above liquidus | Confirm the sensor sees the ball, not the lead; extend TAL to 1.0–1.5 s; check nitrogen and paste shelf life |
Acceptance after any profile change should include: peak temperature within ±5 °C of target on a TC coupon, intermetallic layer of roughly 1–3 µm on a cross-section, voiding under 25% on thermal pads by X-ray, wetting angle below 30° on a visual/AOI check, and a pull or shear test that meets the drawing minimum.
FAQ
Why is closed-loop control better than a fixed power-and-time recipe?
Because absorbed energy varies with surface finish, paste lot and part mass. A closed loop measures the result and corrects within milliseconds, so one recipe covers normal material variation; a fixed recipe needs re-tuning every time a variable shifts.
Should the pyrometer measure the solder joint or the laser spot?
The joint — specifically the solder fillet area where the alloy melts. Measuring the laser spot or the component body gives a reading that does not represent the joint and will bias the loop.
How often should the profile be re-verified?
At every product or paste change, after any optical cleaning or sensor replacement, and on a fixed weekly schedule with a thermocouple coupon. Trend the deviation rather than only pass/fail, so drift is caught before it becomes scrap.
Can laser soldering reach 240–250 °C without damaging the board?
Yes, because the heated volume is small and the dwell is short — typically 1–3 s total. Damage risk comes from oversized spots, long time above liquidus and excessive peak, which is exactly what a properly tuned closed loop constrains.
How is this different from a reflow oven profile?
Reflow heats the whole assembly over minutes with a gentle ramp; laser soldering heats one joint in seconds at a much higher ramp rate. The profile logic (preheat, ramp, peak, TAL, cooling) is similar, but the control bandwidth and sensor placement requirements are far stricter.
If you share your alloy, pad geometry and component mix, we can suggest starting profile windows and sensor setup for your specific laser soldering application.