Laser delay tuner

Laser Scan-Delay Tuner

Operates under the rules in ../../INSTRUCTIONS.md (reference-first, reference-only, interactive mode).

When to use this skill

When marks or cuts show defects at line starts, line ends, or corners — and the suspected cause is a mistuned scan delay. This skill reads the symptom, identifies the offending delay, and proposes a specific adjustment anchored to the laser's pulse-to-pulse interval.

References to load at start

• references/scanner_optimization.md §8–§10 — primary

The fundamental constraint

Per scanner_optimization.md §10:

Adjustments smaller than one pulse interval may produce no observable change — between pulses the beam is off anyway, so a sub-interval shift is invisible. Coarse-tune in 1–2 pulse intervals; fine-tune within that.

This drives every step-size recommendation in this skill.

Inputs to gather first

Ask the operator:

1. Symptom description — what does the defect look like, and where on the line/shape?
   • Burn-in (large dot) at line start?
   • Line start missing — mark begins late?
   • Line end short?
   • Burn-in at line end?
   • Curve / oscillation at start of mark?
   • Small curve at end of mark?
2. Test pattern used — square / rectangle / triangle? Are direction arrows shown?
3. Material (anodized aluminum is recommended per scanner_optimization.md §9).
4. Current settings — scan speed, PRF, current LaserOn / LaserOff / Jump / Mark / Polygon delays.

⚠️ If the test pattern has no direction arrows, ask the operator to re-run with arrows. Per scanner_optimization.md §9, without arrows you cannot tell whether a corner defect is at the start (LaserOn) or end (LaserOff) of the line. Stop here until arrows are included.

→ Pause: confirm inputs and test-pattern direction is known.

---

Workflow

Step 1 — Match symptom to root cause

Per the symptom table in scanner_optimization.md §9:

Symptom	Root cause	Fix direction
Burn-in (large dot) at line start	LaserOn too short — laser fires before mirror at start	Increase LaserOn
Line start missing — mark begins late	LaserOn too long — mirror has moved past start	Decrease LaserOn
Line ends short of endpoint	LaserOff too short — laser cuts off before mirror arrives	Increase LaserOff
Burn-in (dot) at line end	LaserOff too long — laser stays on past mirror endpoint	Decrease LaserOff
Curve / oscillation at start of mark	Jump too short — mirrors hadn't settled before mark started	Increase Jump
Total scan time longer than necessary	Jump or Mark too long	Decrease the offender
Small curve at end of mark	Mark too short — mirrors hadn't reached endpoint before next jump	Increase Mark

→ Pause: confirm symptom match. Show the candidate delay + fix direction. Wait for operator to confirm before proceeding to step size.

Step 2 — Compute step size

Per scanner_optimization.md §10:

Pulse interval = 1 / PRF

Common values:

PRF	Pulse interval
20 kHz	50 µs
50 kHz	20 µs
100 kHz	10 µs
200 kHz	5 µs
500 kHz	2 µs

Recommended adjustments:

• Coarse step: 2 × pulse interval. Use this to find the right neighborhood.
• Fine step: 0.5 × pulse interval. Use this only after coarse-tuning has the symptom roughly correct.

Compute the recommended initial change for this operator's specific PRF and present:

"Increase LaserOn from <current> by 2 × pulse interval = <X µs>, giving <new value>."

→ Pause: wait for operator to confirm the proposed step.

Step 3 — Validate

Per scanner_optimization.md §9:

Recommend re-running the same test pattern at the same scan speed and material.

Re-run result	Action
Symptom went away	Done with this delay. Go to Step 4 to optimize Jump/Mark.
Symptom changed (e.g., fixed one defect, exposed a different one)	Loop back to Step 1 with the new symptom.
Symptom is the same magnitude	Step size was too small — increase by another 1–2 pulse intervals.
Symptom is inverted (now over-corrected)	Reduce step. Try 0.5 × pulse interval in the opposite direction.

→ Pause: ask the operator to report the new pattern result. Loop.

Step 4 — Optimize the slow delays

Per scanner_optimization.md §8 — Throughput note:

After LaserOn / LaserOff are correct, reduce Jump and Mark down to the smallest values that still give clean corners. Jump and Mark each extend cycle time; LaserOn / LaserOff do not.

The order matters:

1. During LaserOn / LaserOff tuning (Steps 1–3), Jump and Mark were deliberately long so they didn't mask other defects.
2. Now reduce them in 1–2 pulse-interval increments until corner quality starts to degrade.
3. Back off one step to the last clean setting.

→ Pause: present each Jump / Mark adjustment for confirmation. Loop until both are at minimum-clean.

---

Boundary call-outs

If the operator reports defects that do not match the §9 table, this is not a scan-delay problem. Examples:

• Edge of mark is wavy → likely galvo dynamic-response issue or feedforward misconfiguration (scanner_optimization.md §7).
• Depth varies along the line → laser power instability or rise-time effect (scanner_optimization.md §6).
• Pattern is shifted from commanded location → field calibration or scaling (scanner_optimization.md §4–§5).

In any of these cases, hand off to laser-troubleshoot (general triage) or laser-defect-diagnose (defect classification).

---

Output

A delay-tuning record (markdown):

• Symptom history (each iteration)
• For each iteration: which delay, before / after value, observed result
• Final settings (LaserOn / LaserOff / Jump / Mark / Polygon)
• Speed / power / PRF point at which they were validated
• Notes for any speed-power point variations the operator anticipates

Save to outputs/laser-delay-tuner/<YYYY-MM-DD>-<machine>.md.

Reminder: Settings tuned at one scan speed will likely not be correct at another. Per scanner_optimization.md §9, delays scale with speed.