Accuracy tells you how close a reading is to the truth. Repeatability tells you whether you'll get the same answer next time.
In semiconductor wet process — where the same chemistry, the same flow rate, and the same timing need to repeat thousands of times across a production run — only one of those actually keeps your process stable.
𝗧𝗵𝗲 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗹𝗼𝗼𝗽 𝗽𝗿𝗼𝗯𝗹𝗲𝗺
Wet process flow control follows a simple logic: set a target → meter reads flow → controller adjusts the valve → process window holds.
Every link in that chain depends on the meter giving consistent feedback. If the reading shifts from one moment to the next — not because the flow changed, but because the meter is inconsistent — the controller starts chasing a moving target. The valve hunts. The process window widens. Yield variation follows.
This happens even when the accuracy spec looks excellent on paper. Accuracy is measured once, under controlled calibration conditions. Repeatability is what the meter actually does on the line, cycle after cycle, with
temperature drift, media variation, and pipe vibration in play.
𝗪𝗵𝗮𝘁 𝗿𝗲𝗽𝗲𝗮𝘁𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗺𝗲𝗮𝗻𝘀 𝗶𝗻 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗲
Most flow meter datasheets lead with accuracy. Experienced process engineers look past that to repeatability first — because it's the one that determines whether a process is actually controllable.
The FM800H clamp-on ultrasonic flow meter is specified at:
✦ ±0.4% repeatability across the normal flow range (Qt to Qmax)
✦ ±0.8% at low flow (Qmin to Qt)
In practical terms: run the same flow through an FM800H (½" line, 0.5–20 L/min range) 100 times under identical conditions, and the readings stay within a ±0.4% band. The control loop receives consistent feedback. The valve moves predictably. The process window holds.
𝗪𝗵𝘆 𝗿𝗲𝗽𝗲𝗮𝘁𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗮𝗹𝗼𝗻𝗲 𝗶𝘀𝗻'𝘁 𝗲𝗻𝗼𝘂𝗴𝗵
Stable feedback is only useful if it arrives in time.
Semiconductor wet processes move fast. Bellows pumps and diaphragm pumps introduce rapid flow pulsation. CMP and cleaning steps have narrow timing windows. A meter with good repeatability but slow response still introduces lag into the control loop — the feedback is accurate and consistent, just late.
The FM800H addresses this with 7 user-selectable response time settings, from 0.25 seconds to 30 seconds, adjustable from the meter menu or remotely via Modbus RTU. For a fast-switching wet process, the 0.25s setting means flow deviations are captured almost immediately — the controller reacts before the process window has time to drift.
Repeatability and response time work as a pair. Stable feedback that arrives fast gives PID loops the cleanest possible signal. That's when process windows tighten and batch-to-batch consistency improves.
𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝘁𝗲𝘀𝘁: 𝗹𝗮𝗯 𝘃𝘀. 𝗹𝗶𝗻𝗲
Most flow meters look good on the spec sheet. The gap shows up in actual process conditions.
Calibration labs use steady, clean reference fluid at controlled temperature. The line has process chemicals, temperature swings, pump pulsation, and vibration from adjacent equipment. A meter specified only for accuracy makes an implicit promise about steady-state performance. A meter specified for repeatability is making a promise about what it delivers in the real process — measured in consistent output, not just in proximity to a reference value.
That distinction is why the FM800H publishes both numbers, rather than leading with accuracy alone.