Read the wheel with the PLC's high-speed counter input, not with scan logic. The rule for scan reads is a worst-case cycle time at or below 0.8 times the shorter of the ON and OFF times, so a 14 ms pulse on an 11 ms scan has 0.25 ms of margin. Tool a 180 degree tripper, convert cut length to counts from the measured circumference, and confirm one increment per revolution before the wheels are cut.
Counting a measuring wheel with scan logic works until the pulse gets short. A 14 ms pulse read on an 11 ms scan clears the documented timing rule by 0.25 ms, which is no margin: one scan that runs long steps over the pulse, the revolution is never counted, and the finished cut grows by the travel of one wheel turn. Put the sensor on the PLC's high-speed counter input, tool the tripper as a 180 degree arc, and confirm one revolution makes one increment before the wheels are cut. The timing rule, the arithmetic and the checks follow.
What the scan can read
Scan logic samples the input once per scan, at a point in the pulse it does not choose. The sizing rule for reading a pulse that way appears in PLC programming documentation: the worst-case cycle time stays at or below 0.8 times the pulse width or the pause width, whichever is shorter. Turned around for design, the ON time and the OFF time each need at least 1.25 times the worst scan. An 11 ms scan sets that floor at 11 ms ÷ 0.8 = 13.75 ms.
The proposed numbers fail it. A 14 ms pulse clears 13.75 ms by 0.25 ms, under 2 percent of the pulse. Scan time is an average that moves with program execution and interrupt load, so any single scan longer than the ON time steps over the pulse and drops that revolution from the count. The miss is silent and it accumulates: on a wheel traveling 6.00 in per revolution, every miss hides 6 in of travel and the cut lands that much long.
Input filtering stacks on top of the scan. A digital input with a 50 ms response cannot follow a signal changing state every 50 ms or faster, so set the counter input's filter to the smallest documented value and keep filter time out of the pulse budget.
Two escapes exist short of hardware counting. Some micro PLCs document a pulse-catch input that latches edges shorter than the scan, holds the bit for one scan, and takes a minimum pulse width set in microseconds; if your processor lists one, it answers a slow, clean signal. For new cut-to-length tooling, wire the counter input and close the question. The same scan interleaving that drops a wheel pulse also disturbs counter direction logic; PLC up down counter scan timing and direction logic covers that case.
Move the count to the high-speed input
A high-speed counter reads its terminal in hardware, between scans, so pulse width leaves the list of things that can go wrong. On the processor class this line uses, the manual documents the counter as clocked by an external pulse train at a named terminal, X0 on that unit, counting up from 0 to 99,999,999 and comparing the live count against as many as 24 preset values. Your manual governs: read its terminal assignment, count range and preset count off the high-speed I/O pages before wiring.
Three rules for the wiring:
- One counter per wheel. Two lines get two counters, each on the terminal the manual assigns to that function. A general-purpose input carries no counter.
- One edge per event. The count advances on the signal's leading edge. Edge-pair counting halves the length per count, so if you configure it, halve the arithmetic to match.
- One display path. The line's touch panels read the count for the operator; no panel update is part of the capture path.
Wheel pulses are slow for this hardware. A revolution every 50 ms is 20 pulses per second, so this line's counting rate sits far inside any input the manual calls a high-speed counter input. For the general case of counting a fast pulse train, see count a pulse signal with a PLC high speed counter input.
Size the tripper arc
The sensor sees the target for the same fraction of the revolution the target covers. ON time = (arc ÷ 360°) × revolution time, and revolution time = wheel circumference ÷ line speed.
Make the target a 180 degree arc on a new wheel. It doubles the ON time of a 90 degree target, it machines and inspects easily, and with a single target it makes ON and OFF equal, so the shorter of the two is half the revolution.
Run the numbers from this job. The 180 degree arc was called out at a 25 ms pulse, so the wheel turns a revolution in 50 ms, 20 revolutions per second. Against an 11 ms scan the limit is 0.8 × 25 ms = 20 ms, and 11 ms sits at just over half of it. The original 14 ms pulse works back to about a 100 degree arc (14 ÷ 25 × 180°), with the 0.25 ms margin already shown.
If the wheel carries an encoder instead of one target, the length arithmetic is the same, using the encoder's pulses per revolution; measuring wheel encoder operation in cut-to-length applications goes deeper on those builds.
Example: one wheel at 90 feet per minute
One line: a wheel with a measured 6.00 in circumference at the wire, one target, so one pulse per revolution, running at a top speed of 90 ft/min.
- Line speed: 90 ft/min × 12 in/ft ÷ 60 s/min = 18 in/s.
- Revolution time: 6.00 in ÷ 18 in/s = 0.333 s = 333 ms.
- 180 degree target: ON = 333 ms × 180 ÷ 360 = 167 ms, OFF = 167 ms.
- Scan limit: 0.8 × 167 ms = 134 ms. The 11 ms scan is 12 times inside it.
- Length per count: 6.00 in ÷ 1 pulse per revolution = 6.00 in.
- Cut length: 8 ft = 96 in. Target count = 96 in ÷ 6.00 in = 16 counts.
Presets from that: ramp the drive down at 14 counts, which is 84 in traveled with 12 in left, and stop at 16 counts. The stop preset carries the length; the ramp preset only shapes the stop.
Change one thing and the design fails: a 12 degree target gives ON = 333 ms × 12 ÷ 360 = 11.1 ms, under the 13.75 ms floor. Same wheel, same speed, and the count is at the mercy of the scan.
| Target arc | ON time at 18 in/s | Against the 13.75 ms floor |
|---|---|---|
| 12 degrees | 11.1 ms | fails |
| 45 degrees | 41.7 ms | passes, 3 times over |
| 180 degrees | 167 ms | passes, 12 times over |
Steps before the wheels are cut
Lock the line out before touching wiring. The rest runs on the bench.
- Measure each wheel's circumference at the contact surface. Roll the wheel along a tape for 10 revolutions: 60.0 in of travel is 6.00 in per revolution. Record per wheel; two wheels off one drawing can still differ.
- Fix pulses per revolution from the tooling: one target, leading edge, is one pulse per revolution.
- Compute length per count = circumference ÷ pulses per revolution, then target count = cut length ÷ length per count. A target count that is not a whole number means cuts quantize in steps of the length per count; pick the circumference so the common lengths divide evenly where the process allows.
- Compute ON and OFF at top speed from the arc formula and check both against 1.25 times the worst-case scan. Come up short on either and you widen the arc, slow the line, or use the counter input.
- Wire each prox to the terminal the manual names for its counter, confirm the sensor's output type matches the input, and set the input filter to its smallest documented value.
- Enter the presets in count units: ramp-down preset first, stop preset second.
- Turn each wheel by hand through 10 whole revolutions and read the live count.
How to check it worked
- Clip a meter with a millisecond scale, or a scope, across the input at top line speed. The measured ON and OFF should land on the computed 167 ms within sensing tolerance.
- Ten hand revolutions must read as a count of exactly 10. Eleven means a double edge; 9 means a miss.
- Run both lines together. Turning wheel 1 leaves wheel 2's count untouched.
- Confirm each run starts from the defined start count. A stale count shifts the next cut by exactly the leftover; save the previous count before a PLC counter reset lays out the pattern.
- If cuts run long, measure the pulse before touching the length scale. Scaling over dropped pulses hides the fault, and the error returns the moment the line speeds up.
Operators who need the number across the plant floor read it from the line's panels; for a count visible at a distance, large LED batch count displays for PLC presses serve that job.
What to decide
Recommendation: wire each wheel sensor to its own high-speed counter input and tool a 180 degree tripper on any wheel whose pulse would come in under 25 ms at top speed. Keep scan-based counting only where ON and OFF each run at least twice the worst-case scan, confirmed with a meter. On a line already missing counts with short pulses, move the input first and leave the length scale alone until one revolution makes exactly one increment.
Questions
How wide does a pulse need to be for scan logic to read it?
Use the rule stated in PLC programming documentation for main-program reads: the worst-case cycle time stays at or below 0.8 times the shorter of the pulse width and the pause width. With an 11 ms scan, both halves need at least 13.75 ms. Treat that as a floor, since the factor already covers sampling phase and scan variation.
Can a wider tripper fix dropped counts without a high-speed input?
Yes, when the arc puts both the ON and OFF times at least 1.25 times the worst scan, with room to spare. The 25 ms pulse from a 180 degree arc leaves the scan limit at 20 ms against an 11 ms scan. Remember the pulse narrows as line speed rises, which is why the counter input is the safer default on new tooling.
Why do the cuts come out long?
Dropped pulses. Each missed revolution hides its travel from the counter, so the line feeds extra wire before the count reaches the target: on the example wheel, one miss adds 6.00 in. Measure the ON and OFF times at the input and confirm one increment per revolution before adjusting anything in the length scale.
How do I set the counter presets for a cut length?
Convert first: target count = cut length ÷ (wheel circumference ÷ pulses per revolution), 96 in ÷ 6.00 in = 16 counts on the example line. Put the ramp-down preset earlier in counts, 14 against 16, so the drive decelerates with travel to spare, and let the stop preset carry the length. Check that the counter returns to its defined start after every run.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.