Yarn Clearing: Set Fault Limits Without Creating Excessive Splices

A yarn clearer can remove an unwanted thick place and leave a new problem at the joint. It can also make many cuts in yarn that would have produced acceptable fabric. The settings must remove the faults that matter. The resulting joins must also remain fit for use.

Good yarn clearer settings depend on the yarn and its next process. They must also meet the customer's quality needs. A curve copied from another count or blend may give a neat report while missing the relevant defect. The useful question is what each extra cut achieves in the delivered package.

This guide develops that decision through an original winding trial. All trial values are illustrative. They explain the calculations and trade-offs. They do not set machine limits or promise savings. None is a supplier's test result.

Define the fault you need to remove

Start with the actual concern in the cloth or the next process. Keep examples of the rejected appearance beside acceptable examples. Identify the yarn lot and the conditions under which the problem appeared. A vague request for cleaner yarn gives the winding room little basis for choosing a limit.

The same detected event may have different importance in two products. Its position, contrast and persistence can affect the finished appearance. A weak join may also cause trouble even when the original fault has been removed. These are reasons to include fabric and processing checks in a clearing trial.

Separate the quality requirement from the measurement used to enforce it. A customer may reject a visible streak. The clearer measures signals as yarn passes through its sensor. The link between those signals and the streak needs evidence from the product.

For example, a long thick place may be the suspected cause. Where the equipment permits it, collect the yarn removed for that event. Compare that material with the fault record and the affected fabric. This can test whether the selected channel addresses the complaint. It also helps reveal a cut caused by a different feature.

Do not assume that every defect named in a report is unacceptable. Decide which classes need removal and which need further study. Others may be part of the normal yarn appearance. For a fancy yarn, check the intended effects against the agreed design.

Read the units behind the clearing curve

A clearing curve relates fault size to fault length. The meaning of size depends on the sensor and the selected mode. A percentage based on mass is not the same measurement as a percentage based on diameter.

Carvalho and colleagues' research on yarn measurement uses capacitive sensing to measure mass. Optical sensors measure diameter and hairiness. The authors also discuss how mass can be inferred from diameter. That step needs assumptions about density and packing. This distinction matters when transferring a setting between systems.

Consider a simple geometry example. Assume a circular strand with constant material density and constant fibre packing. A diameter increase of 25% gives:

Cross-sectional area ratio = 1.25 × 1.25 = 1.5625.

Under those assumptions, mass per unit length rises by 56.25%. It does not rise by 25%. Working in the other direction, a 50% mass increase gives a diameter ratio of √1.5, or about 1.2247. That is a diameter increase of about 22.47%.

Real spun yarn is not a solid cylinder with fixed packing. Its surface fibres and structure also affect the measured appearance. These calculations explain why percentages cannot be exchanged by eye. They are not a conversion recipe for commercial clearer settings.

Check the selected measurement mode, reference value and units on the installed system. Confirm which channels and options are present. The Uster Quantum 4.0 technical data dated May 2025 set out the available sensors and options. Size faults and count changes have separate controls. Other channels address periodic faults, splices and contamination. A listed feature may not be active on every winding position. Check the installed setup.

If yarn count also changes, establish the correct reference before judging local deviations. ApparelDiary's yarn count guide explains the mass and length basis of tex and other count systems.

Consider length and size together

Imagine three events measured against the same mass reference. The values below are invented observations, not proposed clearing limits.

Ranking these events by percentage alone puts A first. That does not establish which one matters most in the finished product. Length gives a second part of the evidence. A limit aimed only at very large deviations could allow a lower-amplitude long fault to pass.

The reverse mistake is to tighten the whole curve for a problem in one long-fault class. That may add cuts among short events which have already proved acceptable. Adjust the relevant part of the curve if the evidence supports it. Check that the equipment permits this choice.

Loepfe's Matrix Clearing explanation describes classification using the fault's length and distribution. It also discusses repeated cuts associated with a long fault. One extended disturbance can differ from several separate faults. Keep that distinction in the review. The manufacturer's expected benefits still need checking on the actual yarn and machine.

Ask how the system defines the start and end of an event. Nearby excursions may be combined or treated separately according to its logic. This affects both the plotted population and the cut total. Thus two machines can give different counts even when the incoming yarn is unchanged.

Find the normal yarn population before moving the limit

Look at the distribution around the curve, not just the events already removed. The dense part of the plot describes common variations in that yarn. Moving a limit into this region may cause a large increase in cuts for a small change in the setting.

Uster's historical SE640 feature guide, dated September 2010, describes a yarn-body display. It explains the same idea. It treats suggested limits as starting points with a predicted cut effect. The document describes an older system. Its channels and operating values are not current instructions for every machine.

Use the current manual to interpret the displayed population. Check that the yarn used to establish the reference represents the intended lot. A mixed group of cops can produce an average that represents neither component well.

If most of the yarn falls outside an acceptable product range, repeated cutting may be a poor remedy. Investigate the incoming yarn and the upstream process. Clearing can remove unwanted exceptions. It may add repeated stops when the lot as a whole is unsuitable.

Compare trial settings on equal exposure

Prepare a trial that can distinguish a curve effect from a material or position effect. Use comparable yarn from the same defined lot. Spread the alternatives across suitable winding positions and operating periods. Keep the splicing setup and speed consistent. Check relevant machine conditions too. Change them only if they are intended trial factors.

Record the valid yarn length for each alternative. An event total without exposure can reverse the apparent ranking. For example, 40 cuts in 100 km is 40 cuts per 100 km. Sixty cuts in 300 km is only 20 cuts per 100 km, despite the higher raw count.

Cuts per 100 km = quality cuts ÷ valid kilometres × 100.

State what the counter includes. Some displays separate detected events, quality cuts, alarms and splice rejections. A detected event is not necessarily a cut. Nor is a cut necessarily proof that the intended defect was removed successfully.

An original three-setting trial

Suppose three candidate settings each receive 1,000 km of comparable yarn. This is unique incoming length, without counting remeasured sections twice. A, B and C progressively remove more events. The following simplified records cover only clearer-related recovery cycles. Cop changes, unrelated breaks and other stops are excluded.

Each initial cut in this example eventually produces one accepted join. An extra attempt replaces a rejected joint; it does not leave another joint in the final yarn. Thus C creates 600 retained joins from quality cuts, not 690. It nevertheless requires 690 recovery attempts.

Those distinctions matter when the splicer performs poorly. Counting only retained joins hides the repeated work. Each attempt is not another delivered join. Counting it that way overstates the joints in the package.

Take an average recovery cycle of eight seconds per attempt for this teaching case. The total stopped time attributable to these cycles is:

• A: 275 × 8 ÷ 60 = 36.67 position-minutes.

• B: 385 × 8 ÷ 60 = 51.33 position-minutes.

• C: 690 × 8 ÷ 60 = 92.00 position-minutes.

These are summed minutes at individual winding positions. They are not the elapsed delay of the whole machine. Some stops can overlap, and operator response or other losses may add time in a real trial. Measure the actual duration if retries differ from first attempts.

Convert removed length to material loss

Assume that collection during this trial gives an average waste length of 2.4 m per recovery attempt. This includes the removal and end preparation covered by the example. Use a measured average for each setting in practice, since long faults may change it.

Tex is grams per kilometre. For B, the mass calculation is:

924 m × 20 g/km ÷ 1,000 m/km = 18.48 g.

Moving from B to C adds 14.64 g of waste and 40.67 position-minutes in this case. The mass figure may look small, but it does not capture the effect of extra joins on the next process. Time, waste and delivered yarn quality need separate checks.

Do not use these assumed cycle and waste values to price another mill's production. Different yarn lengths, counts, fault patterns and recovery systems can change the result. The method is reusable; the input values must come from the operation being assessed.

Keep incoming and delivered length apart

The trial's cut rates use incoming length. A customer receives the yarn left after waste removal. A delivered-join rate therefore uses a different length. For a simple balance, assume that recovery waste is the only length loss. Also assume there is no stock change.

Setting C then delivers:

1,000 km − 1.656 km = 998.344 km.

Its 600 retained quality-cut joins give:

600 ÷ 998.344 × 100 = 60.10 joins per 100 delivered km, rounded.

The incoming-length figure was 60 cuts per 100 km. Both calculations are correct for their stated purpose. They should not be mixed in one trend or compared without the labels. Actual package records must include the other joins and losses. This teaching balance excludes them.

During recovery, yarn may pass back through the sensor. Check how the installed counter treats that length. Use its documented definition when extracting data. A total sensor passage distance may differ from the unique yarn length assigned to a trial. If this is unclear, retain the raw counter label. Resolve its meaning before claiming equal exposure.

Check what remains after clearing

Now inspect the output for the relevant remaining faults. Use an agreed method that can detect them. In the teaching trial, inspect an equal 100 km from each output. Suppose this gives eight objectionable events for A, three for B and two for C.

This suggests a useful next question. Does B remove enough of the problem to meet the agreed product requirement? Its splices and process performance must also be acceptable. If it does, C's extra work needs a clear benefit before adoption. If B still fails the requirement, its lower cut rate is not a reason to release it.

Two observed events are fewer than three. That small gap does not prove a stable difference between settings. The inspected length may be too short to judge a rare fault. Variation between packages or periods may be larger than the apparent improvement. Repeat or extend the trial as needed for the decision.

Choose the sampling plan before seeing the results. Cover the packages and positions needed for the question. Selecting only the cleanest-looking cone could bias the result. If start-up yarn is excluded, state why and apply the same rule to every setting. When later output will include that yarn, assess it separately. Keep rejected or interrupted runs visible in the trial record. The result should describe the production route being considered.

Use the same inspection rule for all alternatives. Keep the inspector's treatment of borderline defects consistent, and retain representative samples. Redefining an objectionable fault can make the process appear better even if it has not changed.

Some systems can compare candidate curves against stored events. Use that feature to narrow the trial where it is available. Each curve then uses the same recorded population, without a material change between comparisons. It can show which fault classes account for the predicted extra cuts. The replay still cannot measure a new splice's strength or the actual recovery delay. Those require output from a running trial. Keep predicted counts and observed trial counts in separate columns. Investigate a large difference before using the setting for the whole order.

Fabric trials should also preserve identity. Know which yarn package and setting supplied each sample. Compare after the intended preparation and finish where those steps affect the complaint. A clear greige sample does not settle a defect which becomes visible only after dyeing.

Assess the join left by each cut

Splice appearance and splice performance answer different questions. A smooth joint can still be weak. A strong joint can still be too bulky or visually unsuitable for the product.

NPTEL's splicing lesson discusses appearance assessment and strength retained at the splice. These measures provide a useful basis for a trial, but the required values must be agreed for the yarn and end use.

For an original calculation, suppose the mean breaking force of matched parent yarn is 3.50 N. The mean for the tested spliced specimens is 3.00 N under the same stated conditions.

Mean force retention = 3.00 ÷ 3.50 × 100 = 85.71%.

This ratio compares the two means. It does not mean that 85.71% of joints passed, or that every joint retains that share of the parent force. Keep the individual results and break locations. A few weak joints may matter even when the average appears satisfactory.

Define specimen preparation and the tensile method before comparing settings. Place the joint as required by the agreed procedure. Record invalid tests and keep gauge and loading conditions comparable. Also assess the joint's extension and appearance where the product requires them.

Uyanık's 2019 study of three splicing methods tested several yarn materials and counts. Results depended on those combinations. Visual appearance did not always agree with mechanical performance. The study supports testing the actual yarn. It does not establish one best method for every blend. Its trial settings are not a universal production recipe.

Remember the other sources of joins. Rieter's Autoconer splice report notes that joins also arise at cop changes. A package can therefore contain more splices than its quality-cut total. Keep those origins separate when investigating a rise in downstream breaks.

Treat contamination as a separate product question

A mass-clearing curve does not describe every form of contamination. Colour contrast, foreign fibres and unwanted material may need separate detection channels. Confirm the installed capability and assess it with relevant samples.

Loepfe's organic-matter discussion describes distinguishing foreign material from some non-disturbing organic matter. The word non-disturbing still needs a product basis. It is not permission to assume that all plant matter disappears in later processing.

Choose representative contaminants for assessment. Agree how to judge them after the actual bleaching, dyeing or finishing route. A sample which looks harmless in one colour or finish may not represent another order. A supplier demonstration shows results under its stated conditions. Check the local application before relying on them.

Keep the removed samples and remaining-fault assessment linked. A high cut count may reflect heavy incoming contamination. It may also mean the system is not separating unwanted material well. Both can occur together. Reducing that number without examining what passes can shift the problem into the fabric room.

Contamination found in the output should also prompt an upstream check. Compare the affected material and lot history before changing a broader set of clearing limits. Keep evidence from unrelated batches separate. This makes a problem from one source easier to trace.

Use automatic settings with a defined acceptance basis

Automatic proposals can shorten setup work. They still need a stable definition of acceptable output. Loepfe's Autoclearing explanation describes curve creation, fine-tuning and ongoing adjustment. Its stated initial observation length belongs to that product feature. It is not a universal trial sample size.

Review what an automatic proposal is based on. Check the material group, reference period and channels included. A forecast of cuts estimates the result under stated data and settings. It does not establish whether the remaining yarn meets the fabric requirement.

Save the proposed curve and the accepted curve separately when they differ. The system may change limits while running. Keep enough information to link the output to the limits actually used. Otherwise, a good trial cannot be reproduced from its starting screen alone.

Revisit the approval when the yarn type or intended product changes. The yarn population may shift, and an automatic adjustment may follow it. That could keep the cut total stable. Whether it also keeps acceptable quality is a separate question for the output checks.

Use the pattern of cuts to find the next action

The distribution of events can be more useful than a machine-wide average. Check where the increase occurs. It may follow a cop source, winding position, time interval or fault class. That helps decide what to inspect next.

One position may reject splices repeatedly across several yarn supplies. Review its joining and sensing conditions. A problem may follow one cop source across positions. Inspect that incoming yarn and its production history. These are investigation routes, not proof from the counter alone.

Several small events close together can also matter differently from isolated ones. The Quantum 4.0 brochure describes a swarm function for accumulations of faults. Such functions need their own settings and interpretation. They do not justify tightening all isolated-fault limits to address a cluster problem.

Plot comparable event rates over time after a change. Retain the material and exposure basis. A different production mix can look like an improvement if these details are missing. ApparelDiary's practical SPC guide explains how to distinguish process signals from ordinary variation.

The final choice should connect what the curve removes with what the recovery process adds. Check both against what reaches the customer. Keep the trial's accepted yarn and fabric examples with its settings. Use those results to explain the clearing limit. Confirm that it can be reproduced on the intended product.

References and technical reading

• Carvalho and colleagues: yarn mass, optical diameter and hairiness measurement research, 2011.

• Uster Quantum 4.0 technical data, May 2025.

• Loepfe: PRISMA Matrix Clearing.

• Uster SE640: historical description of yarn-clearing features, September 2010.

• NPTEL, IIT Delhi: splicing and splice-quality assessment.

• Uyanık, 2019: a comparative study of air, wet and twin splicing.

• Rieter: Autoconer X6 splice-quality report, 2022.

• Loepfe: PRISMA organic-matter filtering.

• Loepfe: PRISMA Autoclearing.

• Uster Quantum 4.0 brochure, May 2023.