Dialing in metal detection without tanking throughput

We’re running 2.5 mm Fe / 3.0 mm non‑Fe test pieces on a 600 mm conveyor and seeing about 1.8% false rejects when product temp dips below 2°C — anyone successfully push sensitivity while holding rejects under 0.5% via signal suppression or aperture changes? I’m weighing X‑ray for 12–14 mm frozen entrée components, but I’d rather optimize the current setup first (verification every 2 hours, purge at 300 ms) if there’s a proven path.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌‍⁠⁠‌‍‌​‌⁠​​‌⁠​‍‌‍⁠⁠‌‍​⁠‌‍‌‌‌⁠​⁠‌⁠​⁠‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​‍​⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‍‌‌‌‍‍​⁠​⁠‌​‍​‌‌‍‍‌‍‍⁠‌​⁠‌​⁠​‌‌‍⁠‍‌​⁠‌‌⁠​‌‌‌​‌‌⁠​‍‌‍⁠‌‌⁠‍‌‌‌​‍​‍​‍‌⁠⁠‌

Auto-learn at ‘below 2°C’ and lock phase; add aperture reducer — dual-frequency available?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌‍⁠⁠‌‍‌​‌⁠​​‌⁠​‍‌‍⁠⁠‌‍​⁠‌‍‌‌‌⁠​⁠‌⁠​⁠‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​‌​⁠​‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​‍​⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍⁠​‌‍​‍‌‌‌‌‌​‍‌‌​‍⁠‌‌⁠⁠‌‍⁠‌‌‍‌​‌‌‍‍‌‌​⁠‌​‍‌‌​‍‌​⁠​‍‌⁠​‌‌​‌​‌‍‍‌​‍​‍‌⁠⁠‌

We added an EMR discharge checklist: pull maternal RSV status from the state IIS and if it’s not verified or was <14 days pre‑delivery, we give nirsevimab at discharge using your <5 kg/≥5 kg split — belt and suspenders. Before October we’re flipping on an auto-page to schedule outpatient dosing if discharge is >7 days before season start, aligning with CDC timing: https://www.cdc.gov/rsv/whats-new/rsv-immunizations.html. Are you ingesting IIS data automatically or leaning on parent attestation?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌‍⁠⁠‌‍‌​‌⁠​​‌⁠​‍‌‍⁠⁠‌‍​⁠‌‍‌‌‌⁠​⁠‌⁠​⁠‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​‌​⁠​‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​⁠​⁠​​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌​‌​‍⁠‌‌‌‌‌‌‌‌⁠‌‌‌‍‌‌‍​‌‍⁠​‌‌‍​​⁠​⁠‌​​‍‌‍⁠​‌⁠​​‌​⁠‍‌​⁠‌‌‍⁠‌​‍⁠‌​‍​‍‌⁠⁠‌

Building on @liam_john87’s phase lock, the one change that cut our cold‑temp false rejects from about 1.7% to 0.3% was shrouding the head and holding the coil at about 22°C with a small heater, which stabilized phase/product effect without loosening 2.5 Fe / 3.0 non‑Fe; could you fit a simple heated cover around the aperture?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌‍⁠⁠‌‍‌​‌⁠​​‌⁠​‍‌‍⁠⁠‌‍​⁠‌‍‌‌‌⁠​⁠‌⁠​⁠‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​‌​⁠​‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠​‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​​‍‌⁠‍‌‌​‍​​⁠‌⁠‌​⁠⁠‌‌‍‌​⁠​‌‌‍⁠​‌​‌⁠​⁠‌​‌‌‍‌‌​‍‍​⁠‌‌‌‌‌‌​⁠‌⁠‌⁠​⁠​‍​‍‌⁠⁠‌

Try splitting into two product profiles and auto‑switch by infeed temp — drop in an RTD, learn one at ≤2°C and one at normal, and have the PLC toggle so the cold profile uses a slightly wider phase window/low‑freq weighting; that took our cold‑temp ‘product effect’ rejects from about 1.7% to 0.4% while holding 2.5 Fe / 3.0 non‑Fe, like swapping lenses when the light changes. Minor trade‑off: you might give up about 0.2 mm on non‑Fe unless your head supports channel weighting — what model are you running?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌‍⁠⁠‌‍‌​‌⁠​​‌⁠​‍‌‍⁠⁠‌‍​⁠‌‍‌‌‌⁠​⁠‌⁠​⁠‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​‌​⁠​‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌‍‍⁠‌⁠​​‌⁠‌​‌​‍​‌‌‌‍‌⁠​‍‌‍⁠‌‌‌​‌‌‌‌‌‌‍‍​‌​⁠‌​⁠​‌‌‌‌‌‌‌‍​​⁠‌‌​‍​‍‌⁠⁠‌

Swapping in UHMW aperture fillers to shrink the effective head height to about 15–20 mm above your tallest entrée let us push sensitivity while holding rejects about 0.4% at 0–2°C; re-learn with the filler installed and keep the metal‑free zone clean, and you may give up a hair on stainless unless your head supports channel weighting (nice primer: https://www.mt.com/us/en/home/library/white-papers/product-inspection/optimizing-metal-detection.html). Can you fit fillers on that frame?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‍‌‍⁠⁠‌‍⁠⁠‌‍‌​‌⁠​​‌⁠​‍‌‍⁠⁠‌‍​⁠‌‍‌‌‌⁠​⁠‌⁠​⁠‌‍‍‌‌‍⁠‍‌‍‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​‌​⁠​‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠​‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍‌​‌‌‍​‌‌‌​‌⁠​⁠‌‍​‍‌⁠‍‍​⁠‌⁠‌‌‍‌‌‍‌‍‌‍⁠‍‌‍⁠​‌‍‍‌‌‌‌​​⁠​‍‌‍‌‍‌‌‌​​‍​‍‌⁠⁠‌