English Views: 0 Author: Elsa Long Publish Time: 2026-10-16 Origin: Site
A stencil printer degrades quietly. The slide from a printer that deposits 100% of target paste volume to one that produces intermittent insufficient joints is not a breakdown — it is an accumulation of small, unmeasured losses: a slightly worn squeegee lip, a film of dried flux on the stencil underside, a conveyor rail fraction of a millimeter out of parallel, a camera lens with one smudge too many. None of these announce themselves on the machine display, and all of them show up hours later as reflow defects that three different departments will argue about. An I.C.T solder paste printer product hub will show you what a printer can do when it is healthy; this checklist is about keeping it that way.
What follows is a complete SMT Solder Paste printer preventive maintenance checklist organized the way maintenance actually gets done: by subsystem — stencil underside cleaning, squeegee and print head, conveyor rails, vision cameras, board support and fixtures, pneumatics, and calibration — and by cycle, from the checks an operator performs every shift to the annual calibration work that keeps the machine's measurements honest. Every section closes with a cycle table you can lift directly into your CMMS or paper system, and the final section explains which record fields matter and how SPI data turns calendar-based maintenance into condition-based maintenance.
Table of Contents
Printing is the highest-variance step on the SMT line, and it is also the step where the machine condition leaks most directly into product quality. A pick-and-place machine that loses its calibration misplaces components loudly and immediately; a stencil printer that loses its condition degrades slowly and quietly, distributing its losses across deposit volume, positional accuracy and gasketing performance in proportions that shift week to week. The diagnostic consequence is brutal: without a maintenance record, a bridging rate that doubled over two weeks has no cause history, so every investigation starts from zero. With a checklist and records, the same trend points straight at the squeegee that passed 400,000 strokes or the underwipe solvent that nobody replaced.
This is why a structured SMT Solder Paste printer preventive maintenance checklist is not an administrative document but a yield instrument. It converts the machine's slow decay into scheduled, visible, attributable work — and it converts maintenance from a reaction to complaints into a process with its own metrics. The sections of this guide treat each subsystem separately because each degrades differently, but the logic connecting them is identical: know which component wears, know how fast it wears under your conditions, inspect it on a cycle shorter than its failure interval, and write down what you found so the next engineer inherits data instead of folklore. Lines that run this discipline consistently report fewer print-defect excursions and, just as valuable, faster and calmer ones when they do occur.
The degradation paths are worth tracing once in detail, because every item on the later checklists exists to interrupt one of them. The squeegee blade wears first: its lip loses straightness and sharpness, aperture filling becomes less complete at the same pressure, and deposit volume drifts down — fastest at the ends of the stroke, where blade deflection is greatest. The stencil underside accumulates a crust of dried flux and paste residue that the underwipe no longer fully removes, which first causes subtle bridging between fine-pitch pads and later smearing that contaminates board carriers downstream. Conveyor rails collect adhesive dust and solder balls in their guideways, and their width repeatability drifts, so boards enter the print nest a few tenths of a millimeter off their previous position — visible in the data as a slowly widening offset distribution rather than a shift.
Vision components fail in their own characteristic way: lens contamination and LED aging reduce contrast on fiducials, the alignment algorithm starts reporting low-confidence matches, and the machine compensates with poorer placement of the print. Support pins lose their seating positions and magnetic bases weaken; boards that used to print flat now bow a few tenths under print pressure, producing per-pad volume variance that looks superficially like a paste problem. And the pneumatic system — the least visible subsystem of all — leaks and sticks in ways that make print pressure inconsistent between strokes, an instability no parameter change can rescue. Each of these paths has a maintenance action and a cycle in the sections that follow; the point of tracing them is that none of them can be diagnosed from the front panel while they are happening.
Maintenance organizations occupy three postures, and the differences show up directly in yield data. Reactive maintenance waits for the failure or the defect spike and then responds — it is the most expensive posture not because the repair is costly but because the failures are discovered by the product, at full production speed, with the cause history already cold. Preventive maintenance inspects and replaces on a fixed calendar or usage cycle, interrupting the degradation paths before they reach the product; its weakness is that fixed intervals are a compromise, sometimes early and sometimes late. Condition-based maintenance uses measurement — from machine counters, from sensor trends and from inspection data — to schedule work when the evidence says the component needs it, which is the direction every mature line eventually moves.
The three postures are summarized in the table below. The practical reading is not that reactive is wrong and condition-based is right; it is that each posture has a legitimate territory. Truly random failures (a connector works loose, a sensor dies) will always be reactive, and the checklist exists to make sure they are rare. Predictable wear (blades, wipes, filters) belongs firmly in the preventive tier. And the middle ground — components whose condition is measurable but whose wear rate varies with product mix — is exactly where SPI data and machine counters convert a preventive program into a condition-based one without new hardware.
Posture | Trigger | Typical cost profile | Best-fit work | Chief risk |
|---|---|---|---|---|
Reactive | Failure or defect spike | Highest: scrap, downtime, root cause cold | Truly random events | Product discovers the failure |
Preventive | Calendar or usage interval | Predictable; some work done early | Predictable wear: blades, wipes, filters | Intervals wrong for the mix |
Condition-based | Measured wear or trend data | Efficient; needs data infrastructure | Variable-rate wear tied to print quality | Bad data drives bad scheduling |
Experienced line technicians genuinely do know their machines, and a skeptic will ask why their judgment needs a document. The answer is organizational, not technical. A checklist makes the work independent of who is on shift: the Friday technician and the Monday technician perform the same inspection, record the same fields, and escalate on the same criteria. It makes the work auditable — when a customer asks how print quality is assured, the completed checklists are the answer. It makes the work trainable, because a new technician learns the machine through a structure that already encodes ten years of someone else's failures. And it makes the work improvable: when a defect escapes, the checklist shows which item missed it, and either the item's cycle tightens, its method changes, or its acceptance criteria sharpen.
This is also the right moment to note that maintenance begins at purchase. Machines differ enormously in how maintainable they are — how quickly the print head opens for blade service, whether the underwipe system reports its own solvent level, how visible the rail guideways are, whether support tooling has home positions. When comparing platforms on a solder paste printing machine for an SMT line, the honest evaluation includes a maintenance chapter: ask for the recommended PM schedule, the consumables list with prices, and the time required for each tier of service, because a printer that saves ten minutes per shift of maintenance pays that back every day of its life. Buying the machine is a one-week decision; maintaining it is a ten-year relationship.
Before the subsystem chapters, the framework that holds them together deserves its own section, because the most common failure of PM programs is not a wrong technical item — it is a structure that does not survive contact with production pressure. A workable program assigns every task three properties: a tier (who may perform it), a cycle (when it must be performed), and a record (what evidence it leaves). Tiers respect the reality that operators, line technicians and field engineers have different training, tools and authority. Cycles respect the reality that some tasks make sense every shift and others only annually. Records respect the reality that unlogged work is, for diagnostic purposes, work that never happened.
The five-tier framework below is deliberately conventional — it matches what most printer OEMs assume in their service manuals and what ISO 9001-style audits expect to find. Your machine's manual will supply the OEM-recommended starting intervals; your own defect and downtime data will adjust them. What the manual cannot supply is the discipline, and that is what the tier structure exists to enforce: each task appears exactly once in exactly one tier, no task is optional, and no tier can be skipped on a busy day without the skip itself being recorded.
Tier | Cycle | Performed by | Typical scope | Record form |
|---|---|---|---|---|
T1 Operator check | Every shift start | Line operator (trained) | Visual inspections, wipes and consumable levels, pressure readings, first-board verification | Shift card, tick boxes + exceptions |
T2 Weekly service | Weekly or per print-count | Line technician | Blade lip inspection, underwipe deep service, rail cleaning, sensor checks | PM checklist with measured values |
T3 Monthly PM | Monthly | Line technician + engineer | Lubrication, filter changes, alignment repeatability test, pneumatic leak survey | PM record with before/after values |
T4 Quarterly deep PM | Quarterly | Engineer (OEM procedure) | Print head rebuild items, rail parallelism, camera calibration, full pneumatic service | Formal PM report, deviation log |
T5 Annual calibration | Yearly | OEM or metrology service | Sensor calibration, geometry verification, safety systems, metrology traceability | Calibration certificate |
OEM intervals are a starting point, not a verdict, and the honest method for tuning them uses three data sources in a fixed order. Start with the manual's recommendation, because it encodes the manufacturer's fleet experience with the mechanism. Then adjust by usage intensity: a printer running three shifts on high-mix small boards wears its underwipe and blade far faster per calendar week than one running single shifts on stable panels, so intervals should key on print strokes and paste cycles rather than on dates wherever the machine provides counters. Finally, adjust by evidence: if SPI volume trends stay flat for six months, the blade interval is conservative and can stretch; if defects cluster in the days before scheduled blade changes, the interval is too long, and tightening it will pay for itself within a quarter.
Two failure modes of interval-setting deserve explicit warning. The first is interval inflation under production pressure — PM gets postponed, then postponed again, and nothing breaks immediately, which teaches the organization that PM is negotiable. The evidence usually arrives as a defect excursion weeks later, by which time the connection to the skipped service is invisible. The defense is structural: postponement requires a supervisor's sign-off and carries a maximum deferral, so the cost of skipping is at least visible at the moment of skipping. The second is interval deflation by fear — extending every cycle after one bad experience with premature replacement wastes consumables and, worse, teaches technicians that PM results are fiction. Both failure modes are cured by the same thing: measured values in the records, which turn every interval debate into a data conversation.
A checklist entry that records only "done" is nearly worthless; the value is in the measured values and exceptions. Each entry should carry, at minimum: date, shift and machine identity; the tier and item ID from the checklist itself; the measured value or visual finding (not just a tick); any deviation from the acceptance criterion and what was done about it; relevant counters (stencil print count, blade stroke count, paste batch age) at the time of service; and the signature or login of the person who performed it. This is not bureaucracy for its own sake — every field answers a question that will be asked during the next defect investigation, and the cost of recording it at service time is seconds, while reconstructing it later is impossible.
The acceptance criteria deserve as much care as the items. "Inspect blade lip" is not a criterion; "blade lip under 10x magnification shows no nicks and straightness deviation under 0.02 mm across the working length; replace if exceeded" is. Criteria do two jobs: they make the inspection trainable (a new technician can perform it without years of intuition) and they make the record binary (either the measured value was within criterion or an exception was logged). Writing criteria is real engineering work, and it is done once per item — the acceptance column in the tables that follow gives reasonable starting values for a mid-range automatic printer, to be tuned to your platform and product mix.
Modern printers expose usage counters, and a mature PM program maps every interval to the most relevant counter available. The stencil's print count drives underwipe intervals and deep-cleaning frequency. The squeegee stroke count drives blade inspection and replacement. The underwipe roll or cloth consumption drives its own service cycle. Paste batch age drives stencil and squeegee cleaning at every changeover. Machine power-on hours drive lubrication and fan or filter service. Where the counter exists, the interval should be expressed in counter units — "every 5,000 prints" survives product-mix changes that "every Monday" does not, because a Monday means a very different number of prints depending on the week's schedule.
Where the counter does not exist, the PM record itself becomes the counter: log the counter value at every service, and the difference between consecutive records is the true usage per interval. Over a quarter, those differences tell you whether the calendar-based interval you inherited corresponds to sane usage-based intervals, and the conversion is usually illuminating — high-mix lines routinely find that their weekly tasks are actually every 40,000 prints in busy weeks and every 9,000 in slow ones, which is exactly the information needed to decide which items can move to counter-based triggers. The counters also close the loop with the condition-based tier: Section 10 shows how SPI data joins these counters to produce maintenance triggers that neither source provides alone.
If the printer has a single maintenance item that decides print quality most directly, it is the cleanliness of the stencil underside. The foil's bottom face is the gasket that seals paste inside the apertures during the print stroke; when that gasket surface carries dried flux, partially cured paste and lint, the seal fails in small ways thousands of times per panel — paste escapes under the stencil where it should not, fine-pitch apertures bridge, and the deposits that do print carry the volume inconsistency of a compromised seal. No downstream correction can repair a bad gasket, which is why underside cleaning appears in every tier of the framework: wiped automatically every few prints, serviced deeply every week, and inspected thoroughly at every stencil change.
The cleaning system itself is a machine within the machine — vacuum slots to lift residue out of apertures, a solvent-wetted wipe to dissolve and lift the foil surface film, and a dry section to leave the foil print-ready — and each of those three functions degrades on its own schedule and needs its own checklist line. The subsections below separate the automated cycle from the manual deep clean, because they catch different failure modes: the automated cycle manages the continuous film of residue, while the periodic deep clean catches the accumulation the wipes can never remove and inspects for foil damage that cleaning merely reveals. Both end in records, because a stencil's cleaning history is half of its life story — the other half being its print count, which travels with it on every line that runs multiple products.
The automatic underwipe runs its vacuum–solvent–dry cycle on a print-count interval, and the interval itself is a tuning variable, not a fixed setting. Fine-pitch products with dense apertures accumulate residue faster than large-pad panels; Type 4 and Type 5 pastes with higher metal content leave more film than coarser powders; high-humidity environments shorten the interval further. A sensible starting point is an underwipe every 5–10 prints for fine-pitch work and every 10–25 for standard panels, tuned by watching SPI volume variance and bridging escapes over the first weeks of a new product. The three sections of the wipe system each need their own weekly inspection: vacuum slots cleared of paste plugs and their suction verified, solvent reservoir level and dosing checked so apertures are wetted but boards are not flooded, and the dry section or air knife confirmed to be actually drying rather than smearing solvent across the foil.
Solvent selection and freshness are the two silent killers of underwipe performance. The solvent must match the paste's flux chemistry — a solvent that works for a no-clean paste may barely soften a water-soluble one — and stale solvent loaded with dissolved flux stops dissolving and starts depositing, turning the wipe into a residue applicator. The checklist items are concrete: verify solvent type against the paste in use at every changeover; replace or top up solvent on the weekly cycle regardless of level, because concentration, not volume, is the failure mode; and inspect the wipe material itself weekly for saturation and debris embedding, replacing it at the manufacturer's interval or sooner when inspection shows it loaded. A wipe that has stopped cleaning does not announce itself — the first symptom is usually an SPI trend, weeks later, and the record of solvent dates is what connects the two.
The automated wipe manages the film; the manual deep clean catches what the film leaves behind. On a weekly or per-changeover cycle, the stencil is removed to a bench, its underside wiped with fresh lint-free cloth and compatible solvent, its apertures cleared individually where inspection shows plugs, and its surface inspected under magnification. This is also the natural moment for the stencil health inspection that keeps consumable life honest: check foil tension against the supplier's specification, inspect for scratches and dents that print as permanent defects, examine aperture walls for burring from aggressive cleaning, and verify that tension is uniform across the foil. A stencil that fails any of these checks should be quarantined and reviewed against its print count — foils have finite lives, and the deep-clean record is where end-of-life decisions get their evidence.
Method discipline matters as much as frequency. Never use metal tools against the foil; never scrape apertures with anything harder than the approved plastic or fiber implements; never clean a warm stencil with cold solvent, because thermal shock is a documented cause of tension loss in framed foils; and never let cleaning solvent dry on the foil, because it leaves the dissolved residue behind as a harder film. Paper towels and shop rags carry lint and often carry cutting oils from other factory tasks — the checklist specifies lint-free wipes and dedicated solvent dispensers for the same reason the medical field specifies gloves. These rules sound fussy until the first scratched foil costs a week of print defects and a stencil replacement, and then they sound cheap.
The cleaning record earns its keep at two moments: when print quality drifts, and when the stencil budget is defended. At each deep clean, the entry should log stencil ID and product family, print count since last deep clean, inspection findings (plugged apertures, foil condition, tension if measured), actions taken, and consumables used. Over weeks, these records answer the drift question directly: if bridging escapes rose while the underwipe solvent replacement was logged late twice running, the connection is documentary rather than speculative. And over quarters, they answer the budget question: stencils whose deep-clean records show consistently heavy residue loads on specific products justify either an underwipe interval change or an aperture design conversation with the foil supplier — a conversation that is far more persuasive with records attached.
Item | Cycle | Action | Acceptance criterion / record |
|---|---|---|---|
Underwipe cycle verification | Every shift (T1) | Confirm wipe ran on schedule; visually check first-board print after changeover | No visible residue on first board; log exceptions |
Vacuum slots | Weekly (T2) | Clear paste plugs, verify suction | Full suction at all slots; findings logged |
Solvent level & freshness | Weekly (T2) | Top up or replace; verify type matches paste flux | Solvent type + date recorded |
Wipe material | Weekly (T2) | Inspect for saturation/debris; replace per OEM interval | Condition noted; replacement date |
Manual deep clean | Weekly / per changeover | Remove stencil, wipe underside, clear plugged apertures | Print count since last clean; findings |
Stencil health inspection | Monthly (T3) | Check tension, scratches, aperture burring under magnification | Tension within supplier spec; defects photographed |
Stencil life review | Quarterly (T4) | Review print count vs. findings; retire or re-tension foils at end of life | Disposition recorded per stencil ID |
The table's last row closes the loop with consumable economics. Stencil life is consumed by print strokes and degraded by cleaning aggressiveness in roughly equal measure, and the discipline that stretches it is choosing the gentlest effective cleaning method at every tier — which is why the stencil and squeegee consumable choices documented in the stencil and squeegee selection guide matter to maintenance as much as to print quality: a foil chosen for cleanability and a wipe system matched to its foil type will consume less life per print than a mismatched pair, at any cleaning frequency.
The squeegee is the printer's most intentional consumable — the only component whose regular replacement is planned into the cost of every panel — and it is also the component whose degradation is most easily mistaken for something else. A worn blade lip does not produce a dramatic failure; it produces a slow volume decline, an end-of-stroke gradient, and aperture filling that needs more pressure to reach the same target, and because pressure is adjustable, the machine's response to a worn blade is usually to raise pressure silently until something else fails first. The checklist's job in this subsystem is to break that pattern: inspect the blade on a fixed cycle, verify pressure against a known reference rather than trusting the setting, and record blade identity and stroke count so that wear rate becomes data instead of folklore.
The print head around the blade deserves equal billing. Its holder clamps the blade straight and transmits pressure evenly; its lift mechanism seats and retracts the blade between strokes; its rails or guides carry the stroke itself. Contamination in any of those paths produces symptoms that look exactly like blade wear — uneven pressure along the lip, chatter in the stroke, inconsistent separation — and the diagnostic difference between a blade problem and a holder problem is twenty minutes of guided inspection, provided the checklist puts the holder on the same service visit as the blade. The subsections separate the three concerns: the blade as a wear item, the pressure system that drives it, and the head mechanism that everything else depends on.
Blade inspection is a magnified look at the lip, and the criteria are specific enough to train on: the working edge should be straight along its full length within a few hundredths of a millimeter, free of nicks, glazing and embedded debris, and the lip profile should match the blade type's specification — metal blades show wear as a rounded or hooked lip, polymer blades as flattening and glazing. The inspection cycle should key on stroke count where the machine provides it: a common baseline is a detailed lip inspection every 50,000 strokes with replacement typically between 200,000 and 500,000 depending on paste abrasiveness and product mix, but the honest interval comes from your own records. High-volume LED lines wear blades dramatically faster than prototype shops — on platforms like the I.C.T 5151 fully automatic screen printer for LED, running large panels at high duty cycles, the blade budget is a genuine line item and wear-rate tracking pays for itself in a single quarter.
Blade change is a procedure, not a swap. The checklist should sequence it: clean the old blade before inspection (a dirty lip hides wear), record the outgoing stroke count, install the new blade with the correct orientation and clamp torque, verify lip parallelism to the stencil plane with the machine's gauge or a feeler reference, run the machine's blade-break-in or test print cycle, and record the incoming blade ID and count. Miss one step and the new blade inherits a mystery — most commonly, a blade installed without parallelism verification prints a volume gradient that takes a shift to diagnose and blames the paste. The two-minute verification is the difference between a blade change and a blade event.
Print pressure is set on the display but delivered by pneumatics, springs and mechanical linkages, and the delivered value drifts even when the setting does not. The weekly check compares displayed pressure against measured reality — most platforms offer a verification mode, and where they do not, a calibrated gauge in the supply line serves — with the acceptance criterion being agreement within the machine's specification, typically a few percent. Record the comparison at two setpoints, one light and one heavy, because a delivery system can agree at mid-range and diverge at the ends where fine-pitch and large-panel work actually operate.
The holder and its contact surfaces get cleaned at the same visit: paste that dries on the blade seat tilts the lip fractionally, and a tilt that costs 2% of volume on standard panels costs 20% on 0.3 mm pitch apertures. Check the holder's clamping for looseness, its guide surfaces for scoring, and the lift mechanism for consistent seating — each item takes seconds and each prevents a failure mode that would otherwise announce itself as a print defect. A lift mechanism that seats inconsistently also produces stroke-to-stroke separation differences that look like a machine setting; cleaning and checking seating first is the cheapest diagnostic on the entire checklist.
On the monthly and quarterly tiers, the print head receives its deeper service: drive rails or belts cleaned and checked for wear, bearings and wheels inspected per the OEM schedule, stroke speed and acceleration verified against specification, and — on dual-lane or dual-squeegee heads — both sides serviced symmetrically so the head does not learn a lean. The quarterly print-head session is also the right moment to re-verify the squeegee's parallelism and stroke synchronization with a print test on a dedicated verification panel, because cumulative small wear across rails, holders and blades surfaces as a combined signature no single component inspection reveals. Lines that skip the quarterly head service typically discover it when a new paste or new stencil arrives and the machine cannot print the combination to spec — the new variable gets blamed while the head's accumulated drift was the real author.
Item | Cycle | Action | Acceptance criterion / record |
|---|---|---|---|
Blade visual check | Every shift (T1) | Look for paste buildup, visible damage, seating | Clean seat, no visible damage |
Blade lip inspection | Per 50k strokes (T2) | Clean, inspect lip under magnification | Straight within 0.02 mm, no nicks; stroke count logged |
Blade replacement | Per wear data (T2/T3) | Full change procedure incl. parallelism check and test print | New blade ID + count; test print within spec |
Pressure verification | Weekly (T2) | Compare displayed vs. measured pressure | Within machine spec (typically ±3%) |
Holder & seat cleaning | Weekly (T2) | Remove dried paste from blade seat and clamp surfaces | Seats clean; clamp torque correct |
Head rails / drive | Monthly (T3) | Clean, inspect wear, lubricate per OEM | No scoring; lubrication logged |
Head verification print | Quarterly (T4) | Test panel print; verify parallelism and stroke sync | Volume uniformity within process limits |
Every positional capability the printer has — fiducial finding, alignment correction, print accuracy — assumes the board arrives in the same place every time, and the conveyor rails are the mechanical promise behind that assumption. Rails that are dirty, worn, or out of parallel deliver boards inconsistently into the clamping nest, and the alignment system then corrects for a different error on every board; the visible symptom in SPI data is an offset distribution that widens gradually over weeks, which is easily misread as a vision or support problem. Rails are also the subsystem most exposed to the line's other residents: adhesive dust from upstream processes, solder balls and clipped leads from rework, and the general grime of a working factory floor, all of which collect in exactly the guideway grooves where width precision lives.
The maintenance logic for rails separates three concerns that share one section: cleanliness, which is weekly and procedural; width mechanics, which is monthly and measured; and clamping with its sensors, which is a mix of both. Long-board and LED applications stress the subsystem hardest — a platform like the I.C.T 1500 long-board stencil printer carries boards well beyond a meter, where rail parallelism errors accumulate over length and a rail cleaning schedule sized for standard panels is simply wrong. The subsections give the cycle structure for all three concerns, and the checklist table at the end consolidates them by tier.
Rail cleaning is undramatic and non-negotiable: weekly, the guideways, their rails and the transport surfaces are wiped clean, with attention to the corners where flux residue pools and the rail teeth where dust packs. The width adjustment mechanism gets its own monthly attention — the lead screws or belts that set rail width are cleaned, checked for backlash, and lubricated per the OEM schedule, because a width mechanism with backlash sets width differently depending on approach direction, an error invisible on stable products and ruinous on changeover-heavy schedules. The verification that closes the loop is a width repeatability check: set a width, cycle the mechanism fully open and closed, return to the setpoint, and measure — acceptance is repeatability within a few hundredths of a millimeter, and the measured value goes in the record, not a tick.
Parallelism is the rail system's slow failure and belongs on the quarterly tier. Rails that are out of parallel — one side's width mechanism worn, a rail nudged by a cart, a frame shifted — grip boards unequally, and the symptoms migrate: edge boards print offset in one direction, wide and narrow boards misbehave differently, and clamping stress shows up as micro-bowing during the print. Quarterly verification uses the machine's own gauge or a calibrated reference board at multiple positions along the transport, comparing rail gap at entry, center and exit. The measured values, recorded across quarters, turn parallelism from a suspicion into a trend — and they protect the maintenance budget, because "the rails need realignment" with data attached is an hour of scheduled work, while the same sentence without data is an argument.
The clamping mechanism holds the board against the print force, and its condition decides whether that force goes into paste transfer or into board movement. Weekly checks cover the clamp pads for wear and paste contamination, clamping force verification against specification on the monthly cycle, and the edge or vacuum sensors that confirm board presence — cleaned and function-tested weekly, because a sticky sensor's most common failure is the intermittent kind that delays or double-triggers transport. Board stop and support surfaces get the same pass: paste residue on a board stop transfers to the next panel's edge, and edge contamination is one of those defects that travels two machines downstream before anyone connects it to the printer.
Long-board applications add their own entries to the checklist, because every error the rails make accumulates over the transport length. The entries are proportional to the platform: for extended-length machines, parallelism verification moves up to monthly, rail support along the extended span gets checked for level, and the synchronization between the printer's transport and the upstream or downstream equipment it hands boards to — normally a SMEMA handshake — is verified for timing consistency, because long boards spend enough time in transit that a delayed handshake shows up as a transport fault rather than a print one. LED and backplane lines running these formats should also widen their record to note product family per service: rail wear on a line cycling between 300 mm and 1,200 mm boards follows the wide panels, and the record should say so.
Item | Cycle | Action | Acceptance criterion / record |
|---|---|---|---|
Rail guideway cleaning | Weekly (T2) | Wipe guideways, rails, transport surfaces; clear packed debris | No residue film; corners clean |
Presence & edge sensors | Weekly (T2) | Clean, function-test board detection | Consistent trigger; no double-fires |
Clamp pads | Weekly (T2) | Inspect wear and contamination | Pads intact, clean; wear noted |
Width mechanism service | Monthly (T3) | Clean lead screws/belts, check backlash, lubricate | Backlash within OEM spec |
Width repeatability test | Monthly (T3) | Set–open–close–return cycle, measure setpoint | Repeatability ≤0.05 mm; value logged |
Clamping force check | Monthly (T3) | Verify force against specification | Within spec; deviations logged |
Rail parallelism verification | Quarterly (T4, monthly for long-board) | Measure gap at entry/center/exit with reference | Parallel within OEM tolerance; values recorded |
The vision system is the printer's claim to precision, and it is also the subsystem whose failure is most easily misattributed. When alignment quality degrades — offsets creeping upward, occasional fiducial failures, alignment confidence scores dropping — the instinct is to suspect the algorithm or the mechanical axes, but in working factories the most common cause is optical housekeeping: a lens with a fingerprint, a ring light dimmed by age, a calibration target smudged, a camera bracket carrying a season of flux condensate. The vision system makes thousands of critical measurements per panel, and every one of them travels through glass and illumination that a weekly check keeps honest and a neglected month lets degrade.
The maintenance structure mirrors the system's own layers: optics and illumination cleaned and inspected on the weekly tier; the camera's mechanical mounting and its motion path on the monthly tier; and the measurement itself — verified with a known reference — as a standing acceptance test that catches everything the cleaning checks miss. Semi-automatic platforms deserve a specific note here: a machine like the I.C.T P6 semi-automatic SMD solder paste printer leans on its vision support differently than a fully automatic line, with the operator in the loop, and its checks therefore include the operator-facing elements — display brightness, reticle or overlay condition, and the ergonomics of the alignment workflow — because a system used by humans a hundred times a shift wears in human-facing ways.
Optical cleaning is delicate work disguised as janitorial: camera lenses and their protective windows are cleaned only with proper optical wipes and lens fluid, never with shop solvent or dry cloths that drag grit across coatings, and the cleaning cycle is weekly or immediately after any paste-related event near the camera. Illumination is checked in the same visit — LED ring and bar lights dim gradually, and the human eye adapts so smoothly that nobody notices until fiducial detection starts failing intermittently; the checklist therefore records light verification against the machine's own test mode, and lamp replacement follows the OEM hour rating rather than visible failure. The calibration target or reference plate gets its own line: stored in its case, cleaned per its specification, and inspected monthly for the smudges and scratches that would otherwise corrupt every calibration performed with it.
The mechanical environment around the optics completes the weekly picture. Camera brackets and focus lock screws are checked for tightness — a bracket that has vibrated a fraction of a degree changes the optical axis in ways that show up as subtle offset bias. Cable dressing to the camera heads is inspected at the same time, because a cable tugging its connector is a classic source of intermittent vision faults that mimic software problems. And the air path matters more than most teams expect: in flux-heavy environments, condensate film builds on protective glass between service visits, so environments running high paste throughput may need the weekly optical check promoted to mid-week — a decision the record data will justify within a quarter.
Cleaning keeps the optics honest; verification proves the whole measurement chain. The standard test is simple and belongs on the monthly tier: present the same reference board ten or more times, let the system find fiducials and compute alignment each time, and record the spread of the computed positions. The acceptance criterion is machine-class dependent but a common expectation is repeatability within a few hundredths of a millimeter — what matters is the trend against your own baseline, recorded each month, because a repeatability that has doubled over six months is telling you something real even though it still passes every print. When the test fails or drifts, the checklist routes the diagnosis in order: clean and re-verify (optics), check mounting and cables (mechanical), re-run calibration with the reference target (measurement), and only then contact service (software or hardware internals).
Item | Cycle | Action | Acceptance criterion / record |
|---|---|---|---|
Lens & window cleaning | Weekly (T2) | Clean optics with proper wipes and fluid | No film under magnification |
Illumination check | Weekly (T2) | Verify light output via machine test mode | Output within baseline; hours logged |
Mounting & cables | Monthly (T3) | Check bracket tightness, focus locks, cable dressing | All fasteners torqued; no cable strain |
Calibration target care | Monthly (T3) | Inspect and clean reference plate | Target undamaged; condition noted |
Alignment repeatability test | Monthly (T3) | ≥10 repeated finds on reference board, record spread | Spread ≤ machine spec; value recorded |
Full camera calibration | Quarterly (T4) / on drift | Recalibrate against reference target | Calibration accepted; before/after values |
During the print stroke, the board must refuse to move and refuse to bow, and the support system — pins, magnetic bases, tooling plates and their adjustment mechanisms — is what makes that refusal physical. Support problems produce some of the most confusing defects in printing because they are product-dependent: a support configuration that prints one panel flat can let a different panel flex, and per-pad volume variance that looks like a paste or blade problem is often a board flexing a few tenths of a millimeter under the squeegee. The maintenance consequence is that support components need both physical maintenance (clean, seated, magnetized, adjusted) and configuration discipline (the support layout for each product recorded and reproduced), and the checklist covers both.
This subsystem also connects to print quality more visibly than any other, which is why its deep treatment — how support layouts are designed, how pins interact with fiducial strategy, and how flatness is diagnosed — gets its own guide, the board support and fiducial alignment reference. The present section confines itself to the maintenance question: keeping the support hardware in the condition its design assumes. That question divides into the hardware itself — pins and their seats, magnets, tooling plates — and the verification that proves the assembled support is doing its job, which is a flatness test with a sacrificial or reference board.
Support pins are small, numerous and constantly handled, which makes them the subsystem's wear items: they lose their tips' coating, their seats loosen, their magnetic bases weaken, and — most insidiously — they wander. The weekly check is a seating audit: every pin in place, seated in its nest or magnet, no pins missing from the configured layout, no stray pins left where a product change left them. The monthly tier adds condition: pin tips inspected for wear and paste buildup (wiped, not solvent-soaked, to protect coatings), magnetic bases tested for holding force against their specification, and tooling plate surfaces checked for dents, solder balls and adhesive residue that would print their shape into the next panel's underside. Every removed-and-replaced pin is logged, because a pin that needed reseating twice in a month is a pin whose seat needs attention, not a pin that needs luck.
Configuration discipline is the other half of the hardware work. The support layout for each product — pin positions, tooling plate selection, edge support settings — belongs in the product's recipe record, and the changeover checklist includes verifying the assembled support against that record before the first print. Lines without this discipline accumulate a quiet chaos of half-remembered layouts that works until a new shift, a new product or a new engineer asks why this panel prints differently, and nobody can answer. The record answers; the discipline is checking against it every changeover, which costs two minutes and prevents the support-related defect class almost entirely.
The verification that closes the subsystem is a flatness test, performed monthly and after any support hardware change: a reference or sacrificial board is printed with the product's support configuration, and the SPI volume variance across the board — or a simpler micrometer check of post-print bow — tells you whether the support is holding the panel plane. The acceptance criterion is process-specific, but the record's trend is the real product: flatness verified monthly and logged per product family converts "sometimes this panel prints badly" into "this panel's support configuration drifted after the March plate service," which is a solvable statement. Quarterly, the full support inventory gets audited — every pin, base and plate accounted for, condition-graded, and the count reconciled — because support hardware migrates between machines and products, and an unaccounted pin is a future defect sitting in a drawer.
Item | Cycle | Action | Acceptance criterion / record |
|---|---|---|---|
Pin seating audit | Every shift (T1) | Verify all configured pins present and seated | Layout matches product recipe |
Pin & magnet condition | Monthly (T3) | Inspect tips, clean buildup, test magnet holding | No wear beyond spec; force verified |
Tooling plate inspection | Monthly (T3) | Check surfaces for dents, balls, residue | Surface clean and undamaged |
Support config vs. recipe | Every changeover | Verify assembled support against product record | Match confirmed; deviations logged |
Flatness verification print | Monthly (T3) | Reference print; check bow / SPI variance | Within process limits; trend recorded |
Support inventory audit | Quarterly (T4) | Reconcile all pins, bases, plates; grade condition | Full count; replacements ordered |
The pneumatic system is the printer's muscle — it drives squeegee pressure, stencil separation, clamping and a dozen smaller actuators — and it is the subsystem most affected by conditions outside the machine. Compressed air arrives carrying water, oil and particles from the factory supply, and every one of those contaminants lands in the printer's valves, cylinders and tubing. The system's failure style is drift rather than drama: a filter loading up, a valve sticking intermittently, a fitting weeping, and print pressure that is subtly different from what the display claims. Because the squeegee pressure regulation lives here, pneumatic drift is print-quality drift, and the maintenance logic is to stop contaminants at the door and to verify delivered pressure as close to the point of use as the machine allows.
The checklist divides the work by where it happens. At the machine's air inlet, the filter-regulator-lubricator (FRL) unit is the front line and gets the most frequent attention. Inside the machine, valves, actuators and tubing age on slower cycles and are mostly quarterly or annual work. And in the records, every pressure reading taken during other subsystem checks doubles as pneumatic monitoring, which is why the checklist asks technicians to log supply pressure alongside their other measurements — the pattern across weeks is the earliest symptom of a system headed for trouble.
The FRL unit's filter bowl is where the factory air's water and oil condense, and it is checked every shift: drain the bowl (or verify the auto-drain functioning), inspect the filter element condition, and confirm the regulator holds its set pressure with the machine idle and again during a print cycle — a regulator that sags under flow is a classic silent fault that no static check reveals. The weekly tier adds a leak survey: with the machine pressurized and idle, listen and feel at fittings, or apply approved leak fluid to suspect joints; bubbles mean a fitting that will cost pressure consistency for months before it costs downtime. The daily and weekly pressure readings all go into the shift record, because pneumatic decline is a trend, and trends are only visible in records.
Air quality beyond the machine's inlet is a facility responsibility that the maintenance program should still track. If the factory supply's dew point rises in humid season and the machine's local treatment cannot compensate, the symptom arrives as valve sticking and shortened filter life — and the honest response is to raise the issue with facilities with the filter replacement record as evidence, not to keep replacing filters as if the problem were the filters. Coordinated lines and multi-machine plants should track filter element consumption per machine: a printer consuming elements twice as fast as its neighbors is telling you something about its supply line, its internal leaks, or both, and the consumption record is what makes the comparison possible.
Internal pneumatic components age on the quarterly and annual tiers. Solenoid valves are checked for response consistency during the quarterly service — a valve that responds a few milliseconds differently from its twin in a dual-squeegee application is a pressure asymmetry in disguise. Cylinders are inspected for seal wear indicated by drift or creep under load. Tubing is inspected along its whole length for the UV embrittlement, abrasion and nicking that precede failure by months, and any section showing damage is replaced on the spot — tubing is cheap and its failures are not. The annual tier, usually bundled with the calibration visit, includes a full pneumatic functional test of every actuated function with pressures and response times recorded, which becomes the baseline the next year's test is compared against.
Item | Cycle | Action | Acceptance criterion / record |
|---|---|---|---|
FRL filter bowl | Every shift (T1) | Drain / verify auto-drain; check element | Bowl dry; element condition noted |
Regulator stability | Every shift (T1) | Check set pressure at idle and during print | No sag under flow; value logged |
Leak survey | Weekly (T2) | Inspect fittings and joints under pressure | No leaks found; leaks tagged and fixed |
Valve response check | Quarterly (T4) | Verify response consistency of solenoid valves | Response within baseline |
Cylinder & seal inspection | Quarterly (T4) | Check for drift/creep under load | No drift; seals per OEM life |
Tubing inspection | Quarterly (T4) | Full-length inspection; replace damaged sections | No abrasion/embrittlement found |
Full pneumatic function test | Annual (T5) | Test all actuated functions; record pressures and times | Baseline established and archived |
Calibration is the discipline that keeps the machine's numbers honest — the pressures it displays, the positions it reports, the widths it sets — and it is the tier of maintenance where the cost of neglect is quietest and largest. A printer with drifted calibration still runs, still prints, and still passes casual inspection; it just prints with a pressure that is not what the recipe says, an alignment that is not quite the offset it reports, and a width that is a few hundredths away from its setting, and every one of those gaps costs deposit accuracy in ways that get blamed on paste, stencils and operators. The calibration program's structure is three concentric rings: frequent verification (does the machine agree with a reference?), periodic calibration (reset the machine to the reference), and annual metrology (traceable certification that the references themselves are honest).
The verification ring runs constantly inside the other checklists — every pressure comparison, repeatability test and width measurement in the sections above is a verification event. This section covers the ring that remains: the structured monthly verification suite that catches cross-subsystem drift, and the annual calibration that anchors everything to metrological reality. Equipment class shapes the work: a fully automatic platform like the I.C.T 4034 fully automatic stencil printer carries internal verification routines and automated reference checks that the checklist should exercise on schedule, while simpler platforms lean more on external references and the technician's procedure — the principle, verify against something you trust and record the comparison, is identical.
The monthly verification suite is a fixed sequence, performed the same way every time and recorded in the same fields, so that any drift appears as a comparable number against previous months. A representative suite for a mid-range automatic printer: squeegee pressure at three setpoints against the machine's reference mode or gauge; print stroke speed and acceleration against specification; alignment repeatability with the reference board (Section 6's test, logged here too); rail width at two setpoints against a calibrated reference; stencil separation distance against specification; and a standard verification print whose SPI-measured volume, offset and shape become the month's print-health snapshot. Run the suite in the same order every month and by the same technician where staffing allows — part of the suite's value is that its numbers are comparable, and comparability dies when methods wander.
The suite takes under two hours, and its value compounds: each month's numbers against the previous months' numbers is the machine's true health curve, and every escalation in later sections starts from it. When the suite shows a value outside criterion, the disposition is recorded the same day — the adjustment made, the re-verification result, or the escalation to the T4 tier — because an open verification failure left to next month has contaminated a whole month of production by the time anyone returns to it.
The annual calibration is performed by the OEM service organization or a qualified metrology provider, and its scope is everything the technician tiers deliberately do not touch: internal pressure sensor calibration against traceable references, axis geometry and scale verification, vision system metrology calibration, safety system function, and the electrical checks that keep the machine compliant with its certification. The deliverable is a calibration certificate listing each item, its as-found and as-left values, and the reference standard used — and the as-found values deserve a specific habit: they are reviewed against the monthly verification records, because an as-found value far outside what the verifications showed means the verification method itself needs recalibration, and an as-found value at the limit means the annual interval is too long for this machine in this environment.
Traceability is the concept that makes the whole ring trustworthy: each reference used — gauges, weights, the reference board itself — carries its own calibration status, traceable through the provider to national standards, and the PM program tracks those reference calibrations on their own schedule. A reference gauge that is itself two years out of date has been silently validating drift for two years, and the only defense is the reference register: every reference instrument, its calibration due date, and a checklist line in the monthly suite confirming that the references used that month are in date. It is the same honesty principle that governs the SPI machine's gauge in the next section — measurements are only as good as their worst-calibrated component, and in a print cell that component is usually a forgotten reference tool, not the expensive machine.
Item | Cycle | Action | Acceptance criterion / record |
|---|---|---|---|
Pressure verification | Monthly (T3) | 3 setpoints vs. reference mode/gauge | Within machine spec; all values logged |
Speed & separation check | Monthly (T3) | Verify stroke speed, acceleration, separation distance | Within specification |
Alignment repeatability | Monthly (T3) | Reference board repeated finds | Spread within spec (ties to s6) |
Width verification | Monthly (T3) | Two setpoints vs. calibrated reference | Within tolerance; values logged |
Verification print | Monthly (T3) | Standard panel through SPI; record volume/offset/shape | Print health snapshot archived |
Annual calibration | Yearly (T5) | OEM/metrology full calibration | Certificate filed; as-found reviewed |
Reference register review | Monthly (T3) | Confirm all references in calibration date | Register current; no expired tools used |
Everything the previous sections prescribe produces value only when it lands in records, and this section is about making those records earn their storage. A SMT Solder Paste printer preventive maintenance checklist generates its real payoff not at service time but months later, when a defect trend needs a cause, a consumable budget needs evidence, or an auditor needs proof — and the difference between records that answer those moments and records that merely exist is structure: consistent fields, consistent units, consistent dispositions. The first subsection defines the field structure; the second shows how the records join with inspection data to produce the maintenance program's most valuable upgrade.
That upgrade is condition-based triggering, and the print cell already owns the sensor it needs. An SMT stencil printer and SPI integration — the measurement loop described in the dedicated SPI process control guide — produces, as a byproduct of everyday quality control, exactly the trend data that maintenance scheduling has always wanted: volume variance that rises as underwipe solvent loads up, offset spread that widens as rails drift, per-pad nonuniformity that follows support degradation. Correlated with the PM records, those trends convert maintenance from "every N prints because the manual says so" to "when the data says the component is wearing" — the condition-based tier that Section 1 identified as the destination, reached without new hardware.
The mechanism is simpler than its reputation. Start with the PM records producing their normal baseline: months of interval-based service with measured values. Then overlay the SPI process's aggregate statistics — not per-board rejects, which belong to quality control, but the slow control-chart quantities: volume mean and variance, offset mean and spread, per-pad uniformity. Now the correlation work happens once, not continuously: for each subsystem, plot the SPI quantity most sensitive to it against the measured value recorded at each PM visit. The plots that show a relationship — underwipe service date against volume variance is the classic first hit — become candidates for a condition trigger: "if volume variance exceeds X for three consecutive runs, perform the deep-clean procedure early." Each trigger is bounded, logged and reviewed exactly like the correction rules a closed-loop print cell already uses, because it is the same discipline applied to maintenance instead of parameters.
Two cautions keep the mechanism honest. First, condition triggers supplement the calendar tier rather than replacing it: lubrication does not care what SPI says, and optics cleaning cannot wait for a volume trend, so the fixed-tier checklist remains the backbone while triggers pull specific items forward. Second, triggers need their own hygiene — a trigger that fires repeatedly without a confirming finding at service is mistuned, and the review that retunes it is the same false-call review discipline the SPI loop already runs. Lines that respect those two cautions typically find that condition triggers first earn trust on cleaning items, extend to blade timing, and stabilize there; the categories that never develop reliable triggers simply stay on the calendar, which is the system working as designed rather than failing.
Whether the checklist lives in a CMMS, a spreadsheet or on paper, the field structure is what matters, and the table below is the minimum viable structure. Digital systems add genuine value where it counts: counter integration (print and stroke counts arriving automatically), due-date escalation that does not depend on someone remembering, and trend dashboards that make the monthly verification curves visible without manual plotting. But the discipline is portable to paper — factories run excellent programs on clipboards — and the failure mode to design against is the same in both: records that are complete for routine passes and thin on exceptions, which is exactly backwards, because the exceptions are where the diagnosis will need to look. The fix is procedural: an exception field that is mandatory to fill (even if the entry is "none") and a shift-handover line that carries any open deviation forward until it is closed.
Field | Content | Why it matters |
|---|---|---|
Date / shift / machine | When, who was on, which printer | Attribution — connects the record to product and personnel |
Tier & item ID | Framework tier and checklist line | Auditable structure; enables interval analysis per item |
Measured value | The number or finding, not a tick | Trends, criteria enforcement, drift detection |
Deviation & disposition | What failed, what was done | The field defect investigations actually read |
Counter snapshot | Stencil prints, blade strokes, paste batch | Usage-based intervals; condition triggers |
Consumables used | Blade, wipes, solvent, filters | Budget evidence; wear-rate tracking |
Sign-off | Performer and, for T3+, reviewer | Accountability; audit readiness |
The handover line deserves its last word here. The single most common way PM programs decay is information falling between shifts — a technician notices something borderline on Thursday evening, mentions it verbally, and by Monday it is folklore. A handover field that carries open deviations forward, with closure required before the entry is archived, closes that gap for the cost of one sentence per shift. Combined with the field structure above and the SPI-driven triggers, it completes the system this guide has built: every subsystem inspected on a cycle shorter than its failure, every inspection leaving a measured record, and the records and the process data continuously tuning each other — which is the whole difference between a printer that is maintained and a printer that is merely cleaned.
Build your maintenance program on a machine that supports it
A checklist is only as effective as the machine it serves: service access, counter availability, self-reporting consumables and stable mechanical platforms all determine whether the program in this guide runs smoothly or fights the equipment. If you are evaluating a new printer, planning a PM program for an existing line, or simply want a second opinion on the intervals in your current checklist, the I.C.T process team works with exactly these tables every week. Tell us your platform, your shift structure and your product mix — we will return a tier-by-tier starting checklist calibrated to your conditions, including the acceptance criteria and record fields each item needs.
Where your current machine cannot support the program — counters missing, service access awkward, consumables unreportable — we will show you the solder paste printer range that can, matched honestly to your volumes and mix rather than to a catalog's highlights. Preventive maintenance is a ten-year relationship with a machine, and the relationship is worth choosing deliberately: send us your current checklist and your worst print-defect trend, and we will tell you what the records say the machine is trying to tell you.
The automated underwipe runs on a print-count interval — a common starting point is every 5–10 prints for fine-pitch products and every 10–25 for standard panels, tuned against SPI volume variance and bridging escapes. The manual deep clean runs weekly or at every changeover, whichever comes first, and always at stencil removal. The interval that matters most is the one your data supports: if bridging trends rise before scheduled cleaning, shorten the cycle; if variance stays flat for months, you have room to stretch. Solvent freshness is checked weekly regardless of the wipe interval.
The T1 shift card takes about ten minutes: visual check of squeegee and seating, FRL filter bowl drained and regulator pressure verified at idle and during a print, support pin seating audit against the product recipe, consumable levels (solvent, wipe), and a first-board verification print after any changeover. Every item records a value or an exception — a tick with no number is the field the next investigation will regret. Operators escalate anything outside criteria to the technician tier rather than adjusting it themselves.
Start with the OEM manual's recommended schedule — it encodes fleet experience with the mechanism. Convert the intervals to usage counters (prints, strokes, hours) wherever the machine provides them, then tune by evidence after the first quarter: SPI trends and PM measured values show which intervals are conservative and which are too long. Expect the first honest revision at three months and a stable program at around six. The commonest error is interval inflation under production pressure, which is why deferral should require sign-off and carry a maximum.
Yes, for the subsystems whose wear prints: underwipe condition shows in volume variance, rail drift in offset spread, support problems in per-pad nonuniformity. The mechanism is correlation, not magic — overlay aggregate SPI statistics against PM records, adopt triggers only where a relationship is demonstrated, and keep triggers bounded, logged and reviewed like any correction rule. Calendar maintenance remains the backbone for items SPI cannot see (lubrication, filters, optics), and a trigger that fires repeatedly without a confirming finding gets retuned or retired.
Seven fields as a minimum: date/shift/machine, tier and item ID, the measured value (never just a tick), deviation and disposition, a counter snapshot (stencil prints, blade strokes, paste batch), consumables used, and sign-off. The two fields most often skipped and most needed later are the measured value and the counter snapshot — they are what turn a service log into trend data. An exception field that is mandatory even when the entry is "none" keeps the record honest, and a handover line carries open deviations across shifts until closed.
Cleaning addresses one subsystem; printing quality is the product of six. If underside cleaning is current and drift persists, work the checklist in diagnostic order: blade lip condition and pressure verification next, then rail width repeatability, then alignment repeatability on the vision tier, then support flatness with a reference print, and finally pneumatic stability under flow. Each tier's monthly verification produces the number that either clears or convicts its subsystem — this is exactly why the measured-value discipline exists, because the drift question is answerable only with records.
Depth, authority and tools. Operator T1 checks are visual, take minutes, require no disassembly, and adjust nothing — they verify condition and escalate. Technician T2/T3 work opens service areas, takes measurements against acceptance criteria, replaces consumables and lubricates, and records before/after values. The boundary protects both: operators are not asked to make mechanical adjustments they lack tools for, and technician time is not consumed by checks an operator can do better because they see the machine every shift. The T4/T5 deep and calibration tiers belong to engineers and OEM service.
The recurring load is smaller than it sounds: T1 is about ten minutes per shift, the weekly T2 service is one to two hours, the monthly T3 suite including verification prints is two to three hours, and the quarterly T4 deep service is typically a half-day scheduled into a planned window. The annual calibration is usually a full day performed by the OEM or a metrology provider. Most lines anchor T3 to a monthly changeover window and T4 to a quarterly holiday or weekend slot; the scheduling rule is that PM time is planned production time, because the downtime it prevents is never planned.
