Weight per meter drifting? Seven causes, checked in this order

First work out the shape of the fluctuation, then decide what to check

A large weight-per-meter fluctuation is the most common reason for a service call, but the fault is not often in the controller or the control algorithm. Most sources lie outside the system, are passed along layer by layer, and end up as a weight-per-meter curve swinging up and down.

Touch the control parameters first and the cause remains while the parameters are muddled: a curve that showed a clear pattern becomes the sum of two or three causes, and you can no longer judge whether the fluctuation is even periodic. So work from outside to inside and from cheap to expensive — start with what the eye can see, rule out what one record can rule out, and leave pulling the screw, recalibrating and parameter changes until last.

Sorting the fluctuation into three shapes saves half the work: a rhythmic fluctuation you can beat time with points to actions sharing that rhythm; one that goes wrong only when the speed is raised and settles when it drops points to mechanical wear on the conveying side; and only a strongly regular fluctuation of similar amplitude is the turn of the control parameters.

Item one: the raw material itself has changed

At a constant screw speed the volume extruded per unit time is broadly unchanged, so a change in density changes the mass with it; pellet size and shape alter packing and compaction in the feed zone; and a higher regrind ratio alters melt flow. All three push weight per meter away without the screw speed moving. To check, line up the moment the fluctuation began with the times of goods receipt, batch change, supplier change and regrind blend adjustment. Then take the material from before and after, measure one fixed volume of loose material from each and weigh them — this needs no instrument. Spread a handful on white paper and compare pellet size and powder content: a change in regrind ratio usually shows first as more powder. Raw-material fluctuation is usually a step change — the curve shifts to a new level after the material change and holds there instead of swinging continuously. Once confirmed, re-confirm the reference value for that construction: the system has no means of recognising a change in the density of incoming material.

Item two: unstable feeding and material level

Uneven supply shows directly as weight-per-meter fluctuation, and it is rhythmic: every suction, top-up and fall in material level is a disturbance. Check a large level change leaving the feed zone under-filled, bridging in the hopper making flow come and go, negative-pressure fluctuation at the loader, and uneven conveying of regrind. Four ways to judge: listen — note the loader's start-stop rhythm, then see whether the curve rises and falls at the same frequency; in step means essentially confirmed. Look — is the level gauge wandering over a very wide range? A stable level stays in a narrow band. Measure — watch the discharge at the hopper mouth for intermittent breaks. Compare — hold the level high and cut the top-up frequency, then observe; a narrowing fluctuation means you have found the right direction.

Item three: wear on the screw and barrel

Wear lowers conveying efficiency, and not linearly: at low speed the leakage is a small share and things may still look acceptable, but as speed rises the leakage along the screw flight clearance grows faster than the expected increase in screw speed, so actual output falls behind — hence wear's habit of going wrong the moment you raise the speed. To check, take two clearly different speed points on the same construction and raw material, run a stable period at each, weigh a sample and work out the actual extrusion output, then see whether the deviation is clearly larger at the higher point. Also look at the trend: wear is gradual, so instability worsens slowly rather than starting suddenly one day, and records from a few months back will show it. Distinguish "weight per meter is high overall" from "weight per meter has become unstable": early wear shows first as worse stability, with the average not yet drifting. This item is the most expensive, so put it after the first two, and before stopping the line and stripping the machine use the two sample weighings to narrow the range.

Item four: unstable temperature control

Feed-zone temperature affects conveying: too high and the material softens early and slips in the screw channel. Metering-zone temperature affects output: a fluctuation changes melt viscosity and pressure, and output follows. Three things have to be looked at. Heating bands: check the contact and any local failure — one failed band will not stop the line, but it leaves that zone low for a long time while other zones compensate, creating a deviation that is hard to notice. Thermocouple: check it has not worked loose and that the measuring point has not left the barrel wall — if it has, the instrument no longer shows the true barrel temperature and the control loop keeps adjusting the wrong way. Cooling fan: a fan that will not turn when it should leaves the temperature high, while one at full speed pulls it down again and again. To judge, set each zone's temperature record beside the weight-per-meter curve: steady temperature with steady weight per meter rules this out, while a temperature wandering in a small range with weight per meter following it points to that zone — check its heating, temperature measurement and cooling. Temperature-driven fluctuation is usually slow and small in amplitude.

Item five: haul-off and take-up

Haul-off slip, tension fluctuation and unstable take-up speed change the actual linear speed of the cable, and the material used per unit length changes with it — it looks like a weight-per-meter problem, but it is a speed problem. Weight per meter is weight divided by length, and the haul-off sets the length side. The on-site readings show this layer: screw speed 41.24 rpm, haul-off speed 60.013 m/min and extrusion output 508.70 kg/h move together, so if the screw speed hardly moves while only the haul-off speed drifts, the problem is on the haul-off side rather than the extrusion side. To check, mark a section of cable between the haul-off and the take-up, measure the actual length between the marks and compare it with the instrument's accumulated length; an unstable difference means slip. Then watch the cable on the haul-off wheel for jumping, wandering or rhythmic shaking — visible jumping is basically the answer. Also see whether the fluctuation grows as the take-up reel fills: reel diameter changes tension, and tension fluctuation feeds back into the linear speed.

Item six: the measurement stage itself

Once everything outside has been ruled out, come back to the measurement. Four common problems: zero drift — the sensor zero shifts with temperature and time, the system takes it for a real weight change and corrects for it, drifting further the more it corrects; out-of-date calibration — the proportional relationship between displayed and true value is wrong, which does not shake the curve but puts the whole curve in the wrong place; vibration interference in sampling — line vibration and nearby impacts are collected into the signal; accumulation or build-up in the hopper — false rises and falls appear in the weight signal, with one jump when the build-up falls away. Four steps to check it: empty the weighing section and see whether the zero returns to where it belongs — if not, that is zero drift; take a length of cable of known length, weigh it separately, work out weight per meter and compare with the display — a fixed difference is a calibration problem, a varying one means going on to vibration and interference; look at the shape — vibration interference is high-frequency, fine, irregular spikes, while build-up is a directional slow drift plus an occasional jump; finally check the records for how long it has been since the last calibration. That last step needs no work with the hands.

Two further on-site values indicate the direction of the deviation: reference wall thickness 1.650 mm and wall thickness compensation 0.020 mm. The compensation is a fixed correction entered by hand; if it is changed again and again, each time by a different amount, an un-eliminated factor underneath is moving.

Item seven: control parameters badly tuned

This is the only one of the seven that is caused by the system itself, and the easiest to overlook — the equipment has run this way since installation, so nobody suspects parameters that were only just commissioned. Three typical situations. The repeat-check interval is too short: extrusion has thermal inertia and material residence time, so an adjustment shows in the reading only after a delay, and the system keeps adding to an adjustment that has not taken effect yet, oscillating back and forth. Each adjustment is too large: the step overshoots, the next brings it back, and a regular swing appears. Only a proportional term, with no means of suppressing overshoot: a large deviation produces a large action, with no handling of accumulated deviation or trend, so it keeps crossing back and forth around the target value. This fluctuation is strongly regular, with a stable period and a similar amplitude, as if pushed along deliberately, unlike the random drifting caused by raw material or wear. The test: switch back to manual with a fixed output and it immediately flattens out. To locate it, look at the lag between the adjustment and the weight-per-meter response — a second adjustment issued before the response appears means the interval is too short; weight per meter overshooting the target and turning back after every adjustment means the size is too large. Stretch the repeat-check interval for the actual conditions, reduce the size of a single adjustment, and watch for a few cycles.

Once it is installed, start from these three steps when fluctuation appears

1. Work out the shape first: is there a period, how long is it, how large is the amplitude, and is it a shift or a swing?
2. Then check the cheapest items: raw material batch and regrind ratio, material level and suction rhythm, temperature records, haul-off slip — none of these needs a stop or a strip-down.
3. Only then touch measurement and control: empty and watch the zero, compare a known length, check the calibration records; once measurement is confirmed clean, deal with the repeat-check interval and the size of adjustment.

If the mechanical and process side are both stable and the fluctuation remains, attention should turn to measurement accuracy, the choice of control method and the interfacing with the existing line — the part of cable weight-per-meter control that genuinely has to be customised line by line. For how different line structures are matched, compare by product construction in Solutions; for the scenarios in which interface functions such as "construction confirmation", "material usage statistics" and "wall thickness compensation" are used, see the system overview. If the fluctuation appears as the average shifting as a whole rather than swinging up and down, it is most likely the conversion relationship — check the conversion inputs first, namely conductor diameter, density and target wall thickness; and to get the signal flow of the control loop straight first, start with how a weight-per-meter control system works.

Frequently asked questions

How do you tell whether a fluctuation comes from the control system or from the line?

The cheapest test is to switch to manual: take the system out of automatic, leave the output fixed, and see whether the weight-per-meter curve flattens out. If it does, the control action is producing the fluctuation, so check the repeat-check interval and the size of adjustment; if it still swings, the disturbance comes from outside the system, so work back through the first six items in order. This step touches no mechanical part.

Weight per meter fluctuates after a material change — does that count as a system fault?

Usually not. Once the density, pellet size or regrind ratio of the incoming material changes, the same screw speed corresponds to a different extruded mass, and the system has no means of recognising the density of incoming material, so a person has to re-confirm the reference value for that construction. This kind of fluctuation shifts the curve to a new level after the material change, rather than swinging back and forth continuously.

The fluctuation is very small but always there — does it need a stop for investigation?

Do not stop the line yet. Bring up the historical curves of weight per meter, screw speed, haul-off speed and each zone's temperature, compare them side by side and see which signal the small fluctuation shares a source with; that usually pins it to one direction before you decide whether to stop. Stripping the machine down without data to support it usually means the parts you open up are fine.