SMT Solder Paste Printer FAT and SAT Checklist for Production Engineering Teams

Process engineering guide for SMT production managers and incoming-quality teams · FAT and SAT scope, accuracy and repeatability numbers, stencil and squeegee controls, program and recipe handover, trial-run criteria, and a one-page acceptance table you can actually print

Short answer: a credible SMT solder paste printer FAT and SAT checklist is two linked documents that together answer a single question — can this machine print your worst board, at your first-pass yield target, on the day it lands in your plant, and on every shift after that? Factory Acceptance Test (FAT) verifies the machine against the vendor's published numbers, on the vendor's floor, with the vendor's paste, the vendor's stencil and the vendor's engineer signing the report. Site Acceptance Test (SAT) re-runs the same numbers, on the customer's floor, with the customer's paste, the customer's stencil, the customer's boards, the customer's line integration and the customer's operators running the machine. The two documents are not interchangeable. Skipping FAT or compressing SAT is the single most expensive way to lose the first 30 days of a new line.

If you need the wider solder paste printer family first, start at our I.C.T solder paste printer product hub, then come back to this checklist. The rest of this guide goes deep on what each acceptance test should actually measure, the numbers that decide pass or fail, the failure modes that appear when an item is skipped, the way the two documents have to be linked so a SAT deviation can be traced back to a specific FAT row, and the one-page printable table a process engineer can take into the customer's plant on day one.

1. What FAT and SAT actually verify (and why the order matters)

Every SMT printer is sold on a brochure that lists the same eight headline numbers: maximum board size, cycle time, alignment accuracy, repeatability, squeegee speed, stencil frame size, program library size and CE certification. None of those numbers, by itself, is the reason an SMT line succeeds or fails. The reason is the combination of small, line-specific details the brochure never mentions — how the bottom side of a thin board flexes during stencil separation, how the fiducial camera behaves on a textured copper surface, how the squeegee pressure drift interacts with the solder paste batch the line actually uses, and how the vendor's local service engineer is going to reach the customer's plant on a public holiday. FAT and SAT are the two procedures that turn those hidden details into questions with measurable answers, and into documents the customer's process engineering team can sign without ambiguity.

The order matters for a simple reason: a finding that is fixed at FAT costs the vendor ten minutes on a bench, while the same finding fixed at SAT costs the customer a day of line downtime, an air-freight spare part and a renegotiated acceptance report. Shifting the burden left, from the customer's floor back to the vendor's floor, is the entire economic point of a structured FAT, and it is the reason the FAT document has to be sharper and longer than most customers expect when they first see it. SAT is then a confirmation, not a discovery exercise; the discovery should already be on the vendor's bench.

1.1 Why FAT comes first: shifting the burden left

The economic case for FAT is straightforward. If a vibration resonance only shows up on the customer's floor, it is the customer's line that loses the eight hours of diagnosis. If the same resonance is found on the vendor's granite bench with a vibration sensor and a known dummy load, it is the vendor's problem to redesign the foot, the frame or the drive before the crate is sealed. The same logic applies to every other acceptance item: vision tuning, squeegee parallelism, stencil release angle, conveyor geometry, recipe management, MES hookup, the lot. Every item discovered at FAT is a fix the vendor absorbs. Every item discovered at SAT is a fix the customer's line absorbs, with the loss of production hours the line was supposed to be earning from the new machine.

There is also a second-order reason to keep FAT first. The document that comes out of FAT is the reference baseline for SAT. When a SAT deviation is recorded, the line engineer should be able to point to a specific row in the FAT report and answer two questions: was this item measured at FAT, and if so, what was the number? Without the FAT document, every SAT deviation has to be investigated from scratch, the customer's engineer has to negotiate the acceptable range on the day the line is supposed to ramp, and the vendor has every incentive to redefine the specification on the spot. A connected FAT-then-SAT workflow is the only way to keep the conversation grounded in numbers rather than opinions. The trade-off between automatic and semi-automatic stencil printers for example changes several rows in the FAT/SAT matrix, because the semi-automatic class shifts two responsibility lines (board loading and stencil cleaning) from the machine to the operator, and those shifts have to be visible in the document.

1.2 What SAT can never make up for

There are a small number of items that no amount of SAT effort can rescue, and a mature acceptance programme knows what those items are before the machine ships. The most common is geometric: a machine that was not levelled correctly at FAT, or whose frame flexed during crating, will never hit its print-accuracy number on the customer's floor no matter how carefully the SAT engineer calibrates the fiducial. The geometric fix has to happen on the vendor's bench, with the machine still on its shipping skids, before the crate is welded shut. The same is true of motion-system resonances, of axis-orthogonality errors, of backlash in the squeegee drive, and of vision-system baseline noise on the actual board finish the customer is going to run. Each of these is a "design-level" defect that is expensive to repair on the customer's floor and cheap to repair on the vendor's floor.

A second class of unrescuable SAT items is contract and documentation. A machine can be technically perfect on the customer's floor and still fail SAT if the as-built documentation does not match the as-shipped hardware, if the CE / UL / CCC declarations on the nameplate do not match the declarations in the contract, if the operator manual is in a language the customer's shift does not read, or if the bill of materials in the maintenance manual is missing the spare parts the customer's maintenance team was promised. None of these items can be detected by running a board through the printer. They are detected by reading the acceptance document against the contract, and the only place that read-through is cheap is at FAT, where the vendor's documentation team is one desk away. The FAT document should explicitly record that this read-through has happened, with a signature from the customer's engineering lead, or SAT becomes a documentation discovery exercise the customer's procurement team was not expecting to pay for.

1.3 How to keep FAT and SAT from drifting apart

The single most common failure of a two-stage acceptance programme is drift: the FAT document measures the machine against one set of conditions, the SAT document measures the same machine against a subtly different set of conditions, and the two reports cannot be reconciled when a deviation appears. The fix is to use a single specification document, with the FAT and SAT columns populated from the same source rows, and to forbid either side from modifying the rows without a written change request that both parties countersign. The 28-row matrix in §4 is built on this principle: each row carries the same item, the same pass/fail threshold, the same measurement procedure, and the same record location, and the only thing that changes between the FAT column and the SAT column is the recorded value and the signature.

The second defence against drift is the equipment. The same stencil, the same squeegee blade lot, the same solder paste batch, the same board coupon and the same SPI program should be used at FAT and SAT wherever the customer is able to provide them. When the customer cannot provide them — for example because the production stencil is still being designed at FAT time — the vendor's reference stencil has to be characterised in enough detail that a substitution at SAT is a controlled experiment rather than a free parameter. The reference board coupon is the second-most-controlled item after the stencil: it should be a known-good production board, with a known fiducial layout, and it should be stored in a way that allows both the vendor's engineer at FAT and the customer's engineer at SAT to run the same measurement procedure on the same coupon. Without this discipline, the FAT and SAT numbers are not comparable, and the customer's process engineering team has no way to know whether a SAT deviation is a real drift in the machine or a drift in the test setup.

1.4 Who signs what: the responsibility split

The signature block on the FAT report is the vendor's quality manager, the vendor's application engineer, and the customer's incoming-quality or process engineering representative. The signature block on the SAT report is the customer's process engineering lead, the customer's maintenance lead, the customer's production manager, and the vendor's commissioning engineer. The two signature blocks overlap at exactly one name: the customer's representative who was present at FAT and who is also responsible for SAT on the customer's floor. That single overlap is what keeps the documents linked, and a programme that allows the FAT customer representative to be different from the SAT customer representative is a programme that has already accepted the cost of drift.

The other side of the responsibility split is the closure process. Every deviation recorded at FAT or SAT has to be classified as "fix before sign-off", "fix within 30 days", or "accepted with comment", and the three classes have to be visible in the cover page of both reports. A "fix before sign-off" item is a hard block on machine acceptance; a "fix within 30 days" item is a hard block on the final invoice, not on acceptance; an "accepted with comment" item is documented in the deviation log and reviewed at the 30-day check. Anything that is not classified is treated as "fix before sign-off" by default, and that default has to be stated explicitly in the cover page so the vendor cannot leave a finding unclassified and use the ambiguity to slip a defect past the customer.

2. FAT scope: the production-line reality check before the machine leaves the factory

The FAT document is longer than most customers expect, and that is by design. The scope has to cover every item that is expensive to discover on the customer's floor, every item that depends on conditions the customer cannot easily reproduce (vendor paste, vendor stencil, vendor line layout, vendor ambient), and every item that has to be linked back to a SAT row in the matrix. The scope breaks into six families: mechanical and motion, print accuracy and repeatability, vision and PCB handling, stencil and squeegee, program and recipe, and finally documentation and sign-off. Each family has its own pass/fail thresholds, its own measurement procedure, and its own deviation class. Skipping any of the six families turns FAT into a visual inspection, which is a polite way of saying that the customer has accepted the cost of finding the rest of the defects at SAT.

Two operational rules apply to the whole FAT scope. The first is that every measurement is taken on a board the customer has either supplied or formally approved, with a fiducial layout the customer's engineering team has signed, and with a solder paste batch the customer has accepted. The second is that the measurement procedure is written down before the measurement is taken, with the procedure version number recorded next to the result, so the same procedure can be re-run at SAT by name. A FAT report that records the result without the procedure version is a FAT report that the customer's engineer cannot use at SAT, and the only honest response to a result without a procedure is to re-measure it on the customer's floor.

2.1 Mechanical and motion acceptance at FAT

The mechanical and motion family is the first family to be measured at FAT, because every other measurement depends on the machine being geometrically sound. The items in this family are frame level, axis orthogonality, conveyor width and parallelism, squeegee carriage travel and parallelism, stencil frame clamp, vacuum table flatness, and the vibration signature of the machine under a known dummy load. Frame level is measured on the four corner pads with a precision level, to a tolerance of 0.05 mm/m or better, and the result is recorded on a sketch that shows the four pad locations. Axis orthogonality is measured by jogging the print head in X and Y against an optical square, to a tolerance that depends on the machine class but should be tighter than 0.01 mm over the full travel. Conveyor parallelism is measured at three points along the conveyor length with a dial indicator against a master rail.

The vibration signature is the item most often skipped at FAT, because it requires a vibration sensor and a known dummy load, and most customer FAT teams do not carry that equipment to the vendor's floor. The cost of skipping it is paid at SAT, when the customer's line has a structural resonance at a frequency the vendor's floor never excited, and the only way to find the resonance is to add stiffening to the floor or to the machine frame. The minimum viable FAT vibration check is a low-cost accelerometer taped to the print head during a 10-cycle print run, with the result captured as an FFT waterfall; the visual inspection of the waterfall is enough to flag a resonance, and a flagged resonance is enough to keep the machine on the vendor's floor until the frame is stiffened or the floor is re-isolated.

2.2 Print accuracy and repeatability targets at FAT

Print accuracy and repeatability is the second family, and it is the family the customer's process engineering team will look at first when the report lands on their desk. The two numbers that matter are 2D position accuracy (X, Y) and θ (theta, rotational) accuracy of the print deposit relative to the board fiducial, measured at FAT against the reference board coupon with the customer's fiducial layout. A typical FAT target is ±0.025 mm at 2 σ for a fully automatic machine in 2D, and ±0.05° at 2 σ for θ, with the Cpk on deposit volume above 1.33 for the reference board. The numbers are quoted at 2 σ because the customer's engineering team is going to compare them against their own Cpk baseline, and a single-sigma number is not directly comparable to a Cpk.

The repeatability target has to be measured on a real production-style coupon, not on a vendor demo coupon. The vendor's demo coupon is usually a 200 × 200 mm board with a single large aperture, and the Cpk on that coupon is not representative of the Cpk on the customer's 0.4 mm pitch QFN coupon. A practical rule is to ask the vendor to measure repeatability on the customer's three worst coupons — the smallest aperture, the largest board, and the highest-density BGA — and to record the Cpk for each. The three numbers, taken together, are what the customer's process engineering team will use to set the SPI alarm threshold. A FAT report that records only the demo-board Cpk is a report that forces the customer to redo the work on their own floor. For the same reason, the SMT solder paste printer buying checklist treats repeatability as a three-coupon measurement, not as a single number on a brochure.

2.3 Vision, fiducial, and PCB handling at FAT

Vision and PCB handling is the third family, and it is the family that the customer's process engineering team is most likely to underestimate. The items are fiducial recognition on the customer's actual board finish, fiducial recognition on the customer's actual solder mask colour, PCB entry and exit timing, top-side clamp force, bottom-side support, and the machine's behaviour when the board is presented slightly off the centreline. Each item is measured at FAT, recorded by name, and the result is compared against the customer's own production data. A fiducial camera that recognises the demo board at 0.2 seconds per fiducial and the customer's textured-copper fiducial at 1.1 seconds per fiducial is a camera that the customer's process engineering team will treat as a known cost, not as a surprise.

PCB handling covers a number of items the brochure rarely lists: the maximum board weight the machine will accept without vibration, the maximum board thickness the bottom-side support can hold, the minimum panel size the machine will index without a manual re-clamp, and the behaviour of the machine when the board is presented with a known defect (bent edge, missing fiducial, extra fiducial, foreign object on the conveyor). The last item is the most informative: a FAT engineer who deliberately runs a board with a missing fiducial through the machine, and records how the machine behaves, is producing a piece of evidence the customer's line team will refer to for the next five years. The same is true of a deliberately warped board: a machine that catches the warp and alarms is a machine the customer's line can trust to catch a real warp on a real production board.

2.4 Stencil, squeegee, and paste path at FAT

The stencil and squeegee family is where most of the late-stage SAT surprises are born, and a thorough FAT has to measure the items in this family on a real production stencil or a stencil that has been characterised in enough detail to be a controlled substitute. The items are stencil frame size and tension, stencil release distance, stencil release speed, stencil cleaning cycle, squeegee pressure, squeegee angle, squeegee speed, squeegee parallelism across the print width, and the paste path from cartridge to stencil aperture (including the under-stencil wipe frequency, the under-stencil vacuum if present, and the paste roll diameter at the start of the run). Each item is measured, recorded, and linked to a specific row in the SAT matrix.

Squeegee pressure deserves special attention because it is the single most common source of print defects. A FAT target for squeegee pressure is set by the customer's engineering team, with a typical range of 5 to 15 kg per squeegee depending on board size and paste type, and the FAT engineer has to demonstrate that the machine can hold the set-point pressure to within ±0.5 kg over the full squeegee travel, over a full 8-hour shift, and across at least three different paste batches. A squeegee pressure that drifts by 1.5 kg over a shift is a squeegee pressure the customer's process engineering team will treat as a known cost, and the cost is either a more frequent calibration cycle or a tighter paste specification. The same is true of squeegee angle (typically 60° down to 45° for fine-pitch work) and squeegee speed (typically 20 to 80 mm/s).

2.5 Program, recipe, and MES hookup at FAT

The program and recipe family is the family that is most often left to SAT, with the result that SAT becomes a software discovery exercise the customer's IT team was not expecting. A mature FAT includes a full program and recipe walkthrough: a reference recipe is built on the customer's representative board, the recipe is saved, the recipe is opened from disk, the recipe is re-opened by a second engineer, and the resulting print is compared to the first print. The walkthrough is repeated with the customer's barcode and MES connection, the customer's directory structure, the customer's naming convention, and the customer's recipe-versioning policy. The SMT solder paste printer selection guide for production lines treats recipe hand-off as a separate evaluation area, and the FAT walkthrough is the place where that evaluation is actually measured.

The MES hookup is the second item in this family, and it is the item most likely to fail at SAT if it is not exercised at FAT. The FAT engineer has to demonstrate the end-to-end flow: the customer scans a board barcode at the line entry, the printer receives the board ID over MES, the printer loads the corresponding recipe by name, the printer prints the board, the printer writes the print result (pass/fail, SPI id, deposit volume Cpk) back to MES, and the customer pulls the same record from MES by board ID. Each step has a timestamp, a log entry, and a sign-off, and any step that has to be re-implemented at SAT is a step the customer's IT team has to budget for. The minimum viable MES hookup test is a 10-board loop with all 10 records visible in MES; anything less is not an integration test.

2.6 FAT documentation and sign-off

The FAT documentation and sign-off family is the family that converts the other five families into a document the customer's engineering team can use. The cover page records the machine serial number, the software version, the configuration code, the FAT date, the customer's representative, the vendor's quality manager, and the vendor's application engineer. The deviation log records every item that did not pass, classified as "fix before sign-off", "fix within 30 days", or "accepted with comment". The as-built documentation includes the electrical schematic, the pneumatic schematic, the mechanical drawing, the bill of materials with the recommended spares list, the operator manual in the customer's language, the maintenance manual, and the CE / UL / CCC declarations on the nameplate. Each document is present, is the as-built revision, and is signed by the vendor's quality manager.

The sign-off block is the last item, and it has to be physically signed before the crate is sealed. The customer's representative, the vendor's quality manager, and the vendor's application engineer all sign the cover page, the deviation log, and the as-built documentation index. A FAT that is signed after the machine has already been crated is a FAT that has lost the economic point of the procedure: the burden has already shifted to the customer's floor, and the customer has no leverage left to require a fix before shipping. The rule is simple: no FAT signature, no crate.

3. SAT scope: the real-floor test where the machine meets your line

SAT is shorter than FAT, but it is the document the customer will live with for the next decade. The scope breaks into six families that mirror the FAT families: mechanical re-acceptance, utilities and line integration, first-article print, trial run and yield baseline, training, and SAT documentation and sign-off. The discipline is the same as FAT: every measurement is taken on the customer's coupon, with the customer's paste, with the customer's stencil, and with the customer's operators running the machine. The signature overlap with FAT is exactly one name, and that overlap is what keeps the two documents comparable.

Two operational rules apply to the whole SAT scope. The first is that SAT does not measure anything FAT was supposed to measure; SAT confirms the FAT numbers on the customer's floor. A SAT engineer who discovers a new defect has to flag it as a deviation, but the deviation is not the same as a FAT deviation: a SAT deviation is a finding the customer's line has to absorb, with the production hours already lost, while a FAT deviation was a finding the vendor's floor was supposed to catch. The second rule is that the customer's process engineering lead is on the floor for every SAT measurement, with sign-off authority. A SAT run that the customer does not sign on the day is a SAT run the customer will be asked to re-confirm in writing three months later, and the second confirmation is always more expensive than the first.

3.1 Mechanical re-acceptance after shipping

Mechanical re-acceptance is the first SAT family, and it is the family that catches the defects FAT was supposed to prevent. The items are frame level after uncrating, conveyor geometry, axis orthogonality after rigging, squeegee carriage parallelism, stencil frame clamp, vacuum table flatness, and the vibration signature under the customer's real dummy load. Each item is measured against the FAT baseline, with the same procedure version number, and the result is recorded on the same sketch the FAT engineer used. A frame that was level at FAT to 0.04 mm/m and is level at SAT to 0.06 mm/m is a frame the customer's engineering team will accept; a frame that is level at SAT to 0.12 mm/m is a frame that has to be re-shimmed before the print accuracy measurements start.

The vibration signature is the second item, and it is the item most likely to show a real difference between the vendor's floor and the customer's floor. The customer's floor may have a different structural resonance from the vendor's, especially if the line is mounted on a raised floor or on a steel mezzanine, and the difference shows up as a print-accuracy drift at cycle time. A SAT engineer who records the customer's floor vibration at three points (machine foot, conveyor entry, conveyor exit) is producing a piece of evidence the customer's facility team will use to plan any future line re-arrangement. The same is true of the local electrical environment: a SAT engineer who records the line voltage, the line frequency, the harmonic distortion, and the ground impedance is producing evidence the customer's facility team needs to decide whether the line needs a power conditioner or a dedicated transformer.

3.2 Local utilities, exhaust, and line integration

The utilities and line integration family is the second SAT family, and it is the family that the customer's facility team will read first. The items are compressed air pressure and quality, exhaust flow and routing, power quality, network connectivity, MES link, barcode reader, light stack, tower lamp, and the physical interface to the upstream and downstream machines in the line. Compressed air is the most common surprise: a machine that was specified for 6 bar at FAT is connected to a customer's 5.2 bar supply at SAT, and the difference shows up as slower cycle time and a squeegee pressure drift over the shift. The I.C.T P6 semi-automatic stencil printer for example has a different compressed-air requirement from a fully automatic inline machine, and the SAT document has to record which class the customer's line is connecting.

Exhaust is the second item, and it is the item the customer's EHS team will care about. The items are exhaust flow at the stencil-cleaning station, exhaust flow at the paste-cartridge loading station, exhaust routing away from operator breathing zone, and the make-up air the line needs to keep the workshop negative-pressure balanced. A SAT engineer who records the actual exhaust flow at the stencil-cleaning station and the actual operator-station breathing-zone measurement is producing evidence the customer's EHS team needs to sign off the line. The line-integration item is the third: the conveyor has to be physically connected to the upstream loader and the downstream SPI or AOI, with the correct hand-off timing, the correct board-stop position, and the correct reject-bin signal. A SAT engineer who runs 10 boards end-to-end through the line, with no manual intervention, has measured the integration; anything less is not an integration test.

3.3 First-article print and solder joint inspection

First-article print is the third SAT family, and it is the family the customer's process engineering team will remember. The items are the first printed board, the first SPI scan, the first AOI scan (if AOI is in the line), the first reflow profile, and the first solder-joint cross-section. Each step is recorded with the recipe name, the paste batch, the stencil serial number, the squeegee pressure, the squeegee speed, the squeegee angle, the print-accuracy result, the SPI result, and the AOI result. The first board is the one the customer's engineering team will keep as the production reference for the next twelve months, and a SAT engineer who does not capture a high-resolution image of that first board is leaving the customer without a baseline.

The solder-joint inspection is the second item, and it is the item that ties the print result back to the line's overall yield. A SAT engineer who pulls two boards from the first 100, cross-sections them at three locations (smallest aperture, largest BGA, highest-density QFN), and records the cross-section image is producing the evidence the customer's process engineering team needs to confirm that the print is within the SPI alarm threshold. A cross-section that shows bridging on a 0.4 mm pitch QFN is a cross-section that forces the recipe to be re-tuned before the trial run; a cross-section that shows clean pads and consistent fillets is a cross-section that confirms the FAT numbers have transferred to the customer's floor.

3.4 Trial run, OEE ramp, and yield baseline

Trial run is the fourth SAT family, and it is the family that converts the first-article print into a yield baseline. The items are a 4-hour continuous run on the customer's representative product, a yield measurement at the SPI station, a yield measurement at the AOI station (if present), an OEE measurement across the trial run, and a defect Pareto by defect type and by paste batch. A typical SAT trial run target is 95% first-pass yield at SPI and 98% first-pass yield at AOI for a representative consumer-electronics board, with the understanding that the numbers will be different for a high-reliability automotive or medical board. The I.C.T P12 high-precision semi-automatic printer for example is targeted at a different yield band from a fully automatic inline machine, and the SAT trial-run target has to be set against the actual product class.

The OEE ramp is the second item, and it is the item the customer's production manager will use to plan the first 30 days of production. The OEE is recorded at 1 hour, 4 hours, 8 hours, and 24 hours of continuous run, with a breakdown by availability, performance and quality. A typical SAT target is 60% OEE at 1 hour, 70% at 4 hours, 75% at 8 hours, and 80% at 24 hours, with the understanding that the line will hit steady-state OEE in the 80 to 85% range after the first month. A trial run that ends below the SAT target is a trial run that the customer's production manager will treat as a known ramp, and the ramp has to be visible in the deviation log so the line team knows what to expect on day one of production.

3.5 Training and knowledge transfer

Training is the fifth SAT family, and it is the family that converts a vendor's commissioning engineer into the customer's in-house capability. The items are operator training (every operator who will run the line on every shift), process engineer training (recipe build, SPI tuning, stencil change, squeegee change), and maintenance training (daily, weekly, monthly preventive maintenance, plus the first-line diagnosis of the top 10 alarms). Each training session has a duration, an agenda, a sign-off, and a competency check, and the sign-off has to include the trainee's name, the trainer's name, the date, and the pass/fail result. A training session that ends without a competency check is a training session the customer's shift lead will have to repeat in three months, when the operator has forgotten the procedure.

The knowledge-transfer item is the second, and it is the item that determines whether the customer's process engineering team can run the line without the vendor's help in year two. The knowledge-transfer package is the as-built documentation (already signed at FAT), the recipe library (built and tested at SAT), the SPI tuning guide (built and tested at SAT), the stencil and squeegee change procedure (built and tested at SAT), and the top-10-alarm diagnosis flowchart (built and tested at SAT). A SAT that does not produce the knowledge-transfer package is a SAT that the customer's engineering team will treat as a partial acceptance, and the partial acceptance has to be visible in the deviation log with a 30-day close-out target.

3.6 SAT documentation and sign-off

SAT documentation and sign-off is the sixth family, and it is the family that closes the loop. The cover page records the machine serial number, the software version, the configuration code, the SAT date, the customer's process engineering lead, the customer's maintenance lead, the customer's production manager, and the vendor's commissioning engineer. The deviation log carries forward every open item from the FAT deviation log, plus every new deviation discovered at SAT, classified as "fix before sign-off", "fix within 30 days", or "accepted with comment". The as-built SAT record includes the SAT measurements against the FAT baseline, the trial-run results, the training records, and the knowledge-transfer package index.

The sign-off block is the last item, and it has to be physically signed before the line is released to production. The customer's process engineering lead signs the SAT cover page after the trial run, the customer's maintenance lead signs the training records after the last training session, the customer's production manager signs the trial-run and yield baseline after the 24-hour run, and the vendor's commissioning engineer signs the deviation log with a commitment date for every open item. A SAT that is not signed on the day is a SAT that the customer's procurement team will treat as not-yet-accepted, and the invoice has to be held until the signature is in place. The 30-day check, where the customer's engineering team reviews the open deviations and confirms the vendor's commitments, is the final closure of the acceptance programme.

4. The side-by-side FAT vs SAT checklist

The single document a process engineering team should walk into a vendor meeting with is a 28-row FAT/SAT matrix. Each row is a measurable acceptance item, with the same item measured twice — once at FAT on the vendor's bench, once at SAT on the customer's floor — and the same pass/fail threshold applied to both columns. The two columns have to be populated from the same source row, with the same procedure version number, with the same record location, and with the same signature block. The matrix is not a comparison of two machines; it is a comparison of one machine against itself, on two floors, with two paste batches and two stencil lots.

The matrix also has to be written down before the FAT measurements start. A matrix that is written after the measurements are taken is a matrix that the customer's engineering team cannot use to argue a deviation at SAT, because the vendor will correctly point out that the acceptance criterion was defined after the measurement, and any number that does not pass is a number that the customer retroactively decided to fail. The order of operations is non-negotiable: define the matrix, agree the matrix, sign the matrix, then measure against the matrix. The 28 rows below are the standard matrix; a customer's matrix may add or remove rows, but the additions and removals have to be agreed in writing before FAT starts.

4.1 Why one combined table beats two separate ones

The alternative to a combined matrix is two separate documents: a FAT report and a SAT report, each with its own list of items, its own pass/fail thresholds, and its own format. The two-document approach is the default for most vendor templates, and it is the default for a reason: it is easier to write a single document than a combined one. The problem is that the two documents cannot be reconciled when a deviation appears. A FAT report that records "alignment accuracy ±0.025 mm at 2 σ" and a SAT report that records "alignment accuracy ±0.04 mm at 2 σ" tells the customer's engineering team that the numbers are different, but does not tell them why the numbers are different, whether the difference is in the measurement procedure, in the paste batch, in the board coupon, or in the machine itself. A combined matrix forces the answer into the same row, with the same item name, with the same procedure version, and with the same record location, and the answer becomes a deviation that can be classified and tracked rather than a number that can be argued about.

The combined matrix also makes the 30-day check easier to run. The 30-day check is the closure of the acceptance programme, and it has to compare the FAT numbers, the SAT numbers, the trial-run numbers, the first-month production numbers, and the open deviation list. A combined matrix is the only document that contains all five of these in the same row. A two-document approach forces the 30-day check to cross-reference between two documents, with the result that items are missed and the 30-day check becomes a three-month check. The combined matrix is the cheapest insurance against the late-stage surprises that the customer is paying FAT and SAT to prevent.

4.2 The 28-row FAT/SAT matrix

The 28 rows below are grouped into six families that mirror the FAT and SAT scope sections. The first column is the item, the second is the pass/fail threshold, the third is the measurement procedure, the fourth is the FAT record, and the fifth is the SAT record. The matrix is a template; a customer's matrix may add rows for items specific to the customer's product (for example, an LED panel with a known fiducial-recognition difficulty, a BGA line with a known stencil-cleaning cycle, or a contract manufacturer with a 50-product mix), but the additions have to be agreed in writing before FAT starts, and they have to use the same column structure as the rest of the matrix. A matrix with a different column structure for the customer-specific rows is a matrix that the vendor's commissioning engineer cannot sign on the day, and the sign-off block on the cover page will not match the row count on the matrix.

The row count itself is intentional. Twenty-eight rows cover the six families — mechanical and motion, print accuracy and repeatability, vision and PCB handling, stencil and squeegee, program and recipe, and documentation and sign-off — with the most expensive SAT deviations distributed across all six families. A shorter matrix (say 14 rows) leaves out the items that the customer's shift lead will be touching in production, and a longer matrix (say 50 rows) adds items that the customer's engineering team cannot measure without equipment they do not own. Twenty-eight is the number that fits both constraints, and a matrix that has fewer or more rows has to be justified in writing before FAT starts.

The threshold column is the second column that has to be set before the matrix is signed. A threshold that is set after the measurement is a threshold the vendor can argue, and the argument is the most common source of late-stage SAT deviations. The thresholds in the matrix below are starting points; a customer's matrix has to use the customer's product band (consumer electronics, high-reliability automotive, medical, or LED), the customer's paste type (lead-free SAC305, SnPb, or low-temperature), and the customer's aperture range (smallest aperture, largest BGA, highest-density QFN). The three numbers together determine the pass/fail threshold, and the three numbers have to be recorded in the matrix's appendix before FAT starts.

The matrix also has to be cross-referenced against the buyer's earlier evaluation work. A customer that has already gone through the how to choose an SMT stencil printer for your production needs evaluation will have already set the working area, the alignment class, the squeegee class, and the support class against the same product band, and the FAT/SAT matrix has to use the same numbers rather than re-deriving them. A matrix that contradicts the earlier evaluation is a matrix the customer's engineering team will be arguing with the vendor's application engineer on the day, and the argument is the most common reason the FAT runs over the three-day window.

Table 1 — The 28-row FAT/SAT matrix. Each row carries the same item, the same pass/fail threshold, the same measurement procedure, and the same record location. The only thing that changes between the FAT column and the SAT column is the recorded value and the signature.

4.3 Pass/fail thresholds and how to record deviations

The pass/fail thresholds in the matrix are intentionally written as numbers, not as adjectives. A threshold that says "alignment accuracy within specification" is a threshold the vendor's engineer can argue; a threshold that says "alignment accuracy ±0.025 mm at 2 σ on the reference coupon" is a threshold that the customer's engineer can measure. Every number in the matrix has to be measurable with the equipment the customer already owns, or with equipment the vendor brings to FAT, and the equipment list has to be recorded in the matrix's appendix. A threshold that requires a £50 000 optical comparator to verify is a threshold that the customer's engineering team will not be able to verify at SAT, and the threshold has to be re-stated in terms of equipment the customer does own.

The deviation log is the second half of the matrix, and it has to live in the same document, not in a separate file. The deviation log records the row number, the FAT value, the SAT value, the delta, the deviation class ("fix before sign-off", "fix within 30 days", "accepted with comment"), the vendor's commitment date, and the closure date. The deviation log is the document the customer's engineering team will use in the 30-day check, and the 30-day check is the closure of the acceptance programme. A deviation log that is missing a closure date is a deviation that the customer will still be chasing six months later, and the chase will cost more than the original fix would have cost.

5. Print accuracy and repeatability: the numbers that decide pass or fail

Print accuracy and repeatability is the family the customer's engineering team will look at first, and it is the family that most often separates a brochure-level pass from a production-grade pass. The four numbers that matter are 2D position accuracy (X, Y), θ rotational accuracy, deposit volume Cpk, and stencil release / aperture fill ratio. Each number has a measurement procedure, a pass/fail threshold, and a known failure mode, and each number has to be measured on the customer's representative coupon, not on the vendor's demo coupon. The I.C.T 5134 online fully automatic solder paste printer for example is designed for a 2D accuracy band tighter than a 400 mm class bench machine, and the matrix has to use the right threshold for the machine class.

The four numbers are also the four numbers that the customer will quote back to the vendor in the 30-day check. A vendor that agrees to ±0.025 mm at 2 σ at FAT and then quotes ±0.04 mm at 2 σ at SAT is a vendor that has changed the definition of the measurement between the two documents, and the customer's engineering team should treat the change as a deviation, not as a re-spec. The 30-day check is the place where the change is surfaced, classified, and either fixed or accepted in writing, and the matrix is the only place where the change is visible.

5.1 Position accuracy: 2D and θ

2D position accuracy is the X, Y offset of the print deposit relative to the board fiducial, measured across at least 30 printed boards on the reference coupon, with the mean offset subtracted to leave the random component. The pass/fail threshold is typically ±0.025 mm at 2 σ for a fully automatic inline machine and ±0.04 mm at 2 σ for a compact bench machine, with the understanding that the customer's product may demand a tighter band. A high-volume LED line running a 0.5 mm pitch CSP for example will demand a tighter band than a consumer-electronics line running a 0.65 mm pitch QFP, and the matrix has to use the customer's product band, not the vendor's default band.

θ rotational accuracy is the rotational offset of the print deposit relative to the board fiducial, measured across the same 30-board sample. The pass/fail threshold is typically ±0.05° at 2 σ for a fully automatic inline machine and ±0.1° at 2 σ for a compact bench machine, with the same product-band adjustment. A high-density BGA line will demand a tighter band than a through-hole-dominated line, and the matrix has to use the customer's product band. A FAT engineer who records the 2D and θ numbers on the vendor's demo coupon and not on the customer's reference coupon is producing numbers the customer's engineering team cannot use at SAT, and the SAT engineer has to re-measure on the customer's coupon, with the result that the FAT and SAT numbers are not comparable.

5.2 Volume repeatability: Cpk and IPC standards

Volume repeatability is the within-board and board-to-board variation of the solder paste deposit volume, measured by SPI on at least 30 boards and at least 5 apertures per board. The pass/fail threshold is Cpk ≥ 1.33 for the customer's worst aperture, with a secondary target of Cpk ≥ 1.67 for the customer's highest-density BGA. The IPC-7525 / IPC-7527 standards provide a reference band for stencil aperture design, and the IPC-9191 standard provides a reference band for the print process Cpk, but neither standard is a substitute for the customer's own Cpk baseline. A vendor that quotes a brochure-level Cpk of 1.67 is quoting a number on the demo board; the customer's Cpk on the production board is the number that the matrix has to record. The I.C.T 5151 solder paste PCB SMT screen printer is targeted at the LED panel class, where the aperture sizes are larger and the Cpk band is correspondingly wider, and the matrix has to use the right band for the machine class.

The Cpk calculation has to be recorded with the paste batch number, the stencil serial number, the squeegee pressure, the squeegee speed, the squeegee angle, and the ambient temperature and humidity. A Cpk recorded without the ambient conditions is a Cpk the customer's engineering team cannot reproduce at SAT, because the ambient on the customer's floor may be different from the ambient on the vendor's floor by 5 °C and 20% RH, and the difference is enough to shift the Cpk by 0.2 to 0.4. The full environmental record is the only way to make the FAT and SAT Cpk numbers comparable.

5.3 Stencil release and aperture fill ratio

Stencil release is the third number, and it is the number that most often goes un-measured. Stencil release is the height and the speed at which the stencil separates from the board after the squeegee has passed, and the two numbers together determine the aperture fill ratio. A typical FAT target is a release distance of 0 to 0.5 mm at a release speed of 0.5 to 3 mm/s, with the exact values depending on the aperture size and the paste type. A release distance that is too small or a release speed that is too high produces a poor aperture fill on fine-pitch apertures; a release distance that is too large or a release speed that is too low produces smearing and bridging on large apertures. The matrix has to record both numbers, with the recipe set-point and the measured value side by side.

Aperture fill ratio is the second number, and it is the number the SPI station will use to alarm on a real production board. A typical FAT target is 90% fill ratio on the customer's worst aperture and 95% on the customer's typical aperture, with the SPI alarm threshold set to 85% on the worst aperture. A fill ratio that drops below 85% on a real production board is a fill ratio that the SPI station will flag, and the flag is what the customer's process engineering team will use to drive a stencil-cleaning cycle or a recipe adjustment. A FAT engineer who records the fill ratio on the vendor's demo board and not on the customer's worst aperture is leaving the customer's process engineering team without a baseline, and the SAT engineer has to re-measure on the customer's coupon.

5.4 SPI feedback and the closed loop

SPI feedback is the fourth number, and it is the number that turns the standalone printer into a closed-loop process. The SPI station reads the print result on every board, the printer receives the SPI result over MES, and the printer adjusts the recipe (squeegee pressure, stencil-cleaning cycle, print speed) on the next board to compensate for the measured drift. A FAT target is a SPI round-trip time of less than 200 ms, with the closed-loop recipe adjustment visible in the MES log. A printer that is not in the closed loop is a printer that the customer's process engineering team will treat as a standalone asset, and the standalone asset requires manual recipe adjustment, which is the most common source of yield drift on a long shift. The matrix has to record the SPI round-trip time and the closed-loop recipe adjustment log, and the SAT engineer has to verify that the closed loop is still active on the customer's floor.

The closed loop is also the place where the most common FAT-pass / SAT-fail pattern appears. A FAT that records the closed loop as "functional" without recording the SPI round-trip time is a FAT that does not capture the number the customer's process engineering team needs at SAT, and the SAT engineer has to negotiate the SPI round-trip time on the day, with the vendor's commissioning engineer using the "I don't have that number" as a reason to slip the acceptance. A matrix that records the SPI round-trip time as a specific number, with a specific pass/fail threshold, is a matrix that prevents the slip.

6. Stencil, squeegee, and board support: where most SAT surprises live

Stencil, squeegee and board support is the family that produces the largest number of SAT deviations, and it is the family that the customer's process engineering team is most likely to underestimate at FAT. The four items are stencil selection and inspection, squeegee pressure / angle / speed / blade material, board support and clamp, and solder paste lot-to-lot variation. Each item has a measurement procedure, a pass/fail threshold, and a known failure mode, and each item has to be measured on the customer's representative stencil or a characterised substitute. A FAT that uses the vendor's reference stencil for the squeegee and board-support measurements is a FAT that forces the SAT engineer to re-measure everything on the customer's stencil, and the re-measurement is the place where the late-stage surprises appear.

The four items are also the four items that the customer's shift lead will be adjusting in production, and a SAT that does not produce a documented procedure for each item is a SAT that the shift lead will be inventing procedures for in the first month of production. The invented procedures are the most common source of yield drift, and the drift shows up in the 30-day check as a list of paste-related defects that the FAT matrix did not catch. The four items below are the four items the shift lead will be touching, and a mature acceptance programme measures all four on the customer's stencil, with the customer's paste, and with the customer's operators running the machine.

6.1 Stencil selection and inspection at SAT

Stencil selection at SAT is the first item, and it is the item most often left to the customer to figure out on the first day of production. The FAT engineer has to walk the customer through the stencil selection logic: frame size, foil thickness, aperture size, aperture shape, electropolish vs nano-coating, and the tensioning system. Each parameter has a number, and the number is the difference between a stencil that prints the customer's product and a stencil that does not. A 0.1 mm foil thickness stencil for example will not print a 0.4 mm pitch QFN; a 0.12 mm foil thickness with a nano-coating will print the QFN but will not print a 0.65 mm pitch QFP; a 0.15 mm foil thickness will print both but will over-print on the 0.4 mm pitch QFN. The matrix has to record the foil thickness, the aperture size range, and the surface treatment, and the SAT engineer has to verify that the customer's production stencil matches the matrix.

Stencil inspection at SAT is the second item, and it is the item the customer's incoming-quality team will own. The items are incoming stencil tension measurement, incoming aperture dimension measurement (across at least 5 apertures per stencil), incoming foil thickness measurement (across at least 5 points), and incoming visual inspection for dents, scratches and contamination. A tension measurement of less than 35 N/cm is a tension that the customer's incoming-quality team should reject; a tension measurement of 35 to 45 N/cm is a tension that the stencil is fit for use; a tension measurement above 45 N/cm is a tension that the stencil frame may be over-stressed. The matrix has to record the tension measurement at FAT (on the vendor's reference stencil) and at SAT (on the customer's production stencil), and the two numbers have to be within 10% of each other.

6.2 Squeegee pressure, angle, speed, and blade material

Squeegee pressure is the second item, and it is the single most common source of print defects. A typical squeegee pressure band is 5 to 15 kg per squeegee, with the exact value depending on the board size, the stencil thickness, the aperture size, and the paste type. A lead-free SAC305 paste with a 0.12 mm stencil and a 0.4 mm pitch QFN will demand a higher squeegee pressure than a lead-free SAC305 paste with a 0.15 mm stencil and a 0.65 mm pitch QFP. The matrix has to record the squeegee pressure set-point for each product, the measured pressure over an 8-hour shift, the pressure drift across three paste batches, and the cleaning cycle. A squeegee pressure that drifts by 1.5 kg over a shift is a squeegee pressure the customer's process engineering team will treat as a known cost, and the cost is either a more frequent calibration cycle or a tighter paste specification.

Table 2 — Squeegee parameter bands by paste type and stencil thickness. The bands are starting points; the customer's process engineering team has to tune them on the actual product.

Squeegee angle and speed are the third and fourth items, and they are typically set together with the pressure. The angle is usually 60° down to 45° for fine-pitch work, the speed is usually 20 to 80 mm/s, and the two numbers together determine the print quality. A squeegee speed that is too high produces a poor aperture fill on fine-pitch apertures; a squeegee speed that is too low produces smearing and bridging on large apertures. Squeegee blade material is the fifth item, and the typical band is polyurethane 70 to 90 Shore A, with the harder blade for the larger apertures and the softer blade for the smaller apertures. The matrix has to record the blade material lot, the blade age, and the blade change cycle, and the SAT engineer has to verify that the customer's blade matches the matrix.

6.3 Board support, clamp, and vacuum

Board support is the third item, and it is the item that most often goes un-measured. The board support has to hold the board flat across the print area, with a support force that does not bow the board, and with a vacuum that does not deform a thin board. A 1.6 mm FR-4 board can be held with a magnetic support; a 0.8 mm FR-4 board needs a vacuum support; a 0.4 mm polyimide board needs both a vacuum support and a top-side clamp. The matrix has to record the board support type, the support force, the vacuum level, the top-side clamp force, and the maximum board thickness the machine will accept. A support that is too soft for a 1.6 mm board is a support that will allow the board to flex during stencil separation, and the flex is what produces the position-accuracy drift that the SAT engineer will be chasing.

Clamp and vacuum are the fourth and fifth items, and they are typically set together with the board support. The top-side clamp force has to be enough to hold the board flat against the support, without over-clamping a thin board, and the vacuum level has to be enough to hold the board against the support during the stencil separation step, without deforming a thin board. The matrix has to record the clamp force set-point, the vacuum level set-point, and the maximum board thickness, and the SAT engineer has to verify that the customer's production board is within the matrix's envelope. A board that is outside the envelope is a board the customer's process engineering team will have to print on a different machine, and the routing of that board is a yield-cost the FAT matrix did not catch.

6.4 Solder paste lot-to-lot variation

Solder paste lot-to-lot variation is the fourth item, and it is the item that the customer's incoming-quality team will care about. The items are paste viscosity at the start of the shift, paste viscosity after 4 hours on the stencil, paste viscosity after 8 hours on the stencil, paste metal load, paste particle size distribution, and paste flux activity. A typical FAT band is viscosity within ±10% of the vendor's spec at the start of the shift and within ±15% after 8 hours on the stencil, with the metal load within ±1% of the spec and the particle size within ±2 μm of the spec. A paste lot that is outside the band is a paste lot the customer's incoming-quality team should reject, and a paste lot that is accepted but drifts during the shift is a paste lot the customer's process engineering team will be compensating for in the recipe.

The matrix has to record the paste lot number, the viscosity at the start of the FAT shift, the viscosity at the end of the FAT shift, the metal load, the particle size, and the flux activity. The SAT engineer has to record the same numbers on the customer's paste lot, and the two records have to be within 10% of each other. A FAT record that does not include the paste lot number is a FAT record that the customer's process engineering team cannot use at SAT, and the SAT engineer has to negotiate the paste lot acceptance on the day, with the vendor's commissioning engineer using the "I don't have that number" as a reason to slip the acceptance. A matrix that records the paste lot number as a specific item, with a specific pass/fail threshold, is a matrix that prevents the slip.

7. Programs, recipes, and line integration

Programs, recipes and line integration is the family that turns a standalone printer into a working part of the customer's SMT line, and it is the family that is most often left to SAT, with the result that SAT becomes a software discovery exercise the customer's IT team was not expecting. The three items are recipe structure and version control, MES and traceability integration, and vision program tuning. Each item has a measurement procedure, a pass/fail threshold, and a known failure mode, and each item has to be exercised at FAT with the customer's barcode, the customer's MES, the customer's directory structure, and the customer's naming convention. A FAT that uses the vendor's reference barcode and the vendor's demo MES is a FAT that the customer's IT team will have to redo at SAT, and the redo is the place where most of the late-stage IT surprises appear.

The three items are also the three items that the customer's process engineering team will be modifying in production, and a SAT that does not produce a documented procedure for each item is a SAT that the process engineering team will be inventing procedures for in the first month of production. The invented procedures are the second most common source of yield drift (after the squeegee pressure), and the drift shows up in the 30-day check as a list of recipe-related defects that the FAT matrix did not catch. The three items below are the three items the process engineering team will be touching, and a mature acceptance programme exercises all three on the customer's MES, with the customer's barcode, and with the customer's process engineering team running the machine.

7.1 Recipe structure and version control

Recipe structure is the first item, and it is the item most often left to the customer to figure out on the first day of production. The FAT engineer has to walk the customer through the recipe structure: product code, board revision, stencil serial number, squeegee pressure, squeegee angle, squeegee speed, stencil release distance, stencil release speed, stencil cleaning cycle, and the print-accuracy threshold. Each parameter has a number, and the number is the difference between a recipe that prints the customer's product and a recipe that does not. The matrix has to record the recipe structure template, the parameter ranges, the parameter default values, and the parameter validation rules. A recipe structure that allows the squeegee pressure to be set to 0 kg is a recipe structure the customer's process engineering team will treat as a known risk, and the risk has to be flagged in the deviation log.

Version control is the second item, and it is the item that determines whether the customer's process engineering team can roll back a recipe change. A mature recipe version control system records every change with a timestamp, an author, a change reason, and the previous version's value. The system has to support roll-back to any previous version, with a single click, and the roll-back has to be visible in the MES log. A version control system that records the change but does not support roll-back is a system the customer's process engineering team will treat as a one-way ratchet, and the one-way ratchet is the most common source of recipe drift. The matrix has to record the version control system name, the roll-back test result, and the audit log retention period.

7.2 MES, traceability, and barcode integration

MES and traceability integration is the second item, and it is the item most likely to fail at SAT if it is not exercised at FAT. The FAT engineer has to demonstrate the end-to-end flow: the customer scans a board barcode at the line entry, the printer receives the board ID over MES, the printer loads the corresponding recipe by name, the printer prints the board, the printer writes the print result (pass/fail, SPI id, deposit volume Cpk) back to MES, and the customer pulls the same record from MES by board ID. Each step has a timestamp, a log entry, and a sign-off, and any step that has to be re-implemented at SAT is a step the customer's IT team has to budget for. The minimum viable MES hookup test is a 10-board loop with all 10 records visible in MES; anything less is not an integration test.

Barcode integration is the third item, and it is the item that determines whether the customer's barcode system can talk to the printer. The items are barcode symbology (Code 128, DataMatrix, QR), barcode orientation, barcode position on the board, barcode reader type (laser, imager), barcode reader mounting, barcode reader trigger signal, and the barcode read rate. A barcode read rate of less than 99% is a read rate the customer's production manager will treat as a known cost, and the cost is either a barcode position redesign or a barcode reader replacement. The matrix has to record the barcode symbology, the read rate at FAT, the read rate at SAT, and the barcode position tolerance. A barcode position tolerance of ±2 mm is typical; a tolerance of ±0.5 mm is tight but achievable with a fixture.

7.3 Vision program tuning at SAT

Vision program tuning is the third item, and it is the item that determines whether the printer can recognise the customer's fiducial layout. The items are fiducial shape, fiducial size, fiducial contrast against the surrounding solder mask, fiducial position on the board, fiducial recognition algorithm, fiducial recognition time, and the false-reject rate. A fiducial recognition time of less than 1 second is typical; a time of more than 2 seconds is a time the customer's cycle-time budget cannot absorb. The matrix has to record the fiducial recognition time at FAT (on the vendor's demo board) and at SAT (on the customer's production board), and the two numbers have to be within 20% of each other. A recognition time that is twice as long at SAT as at FAT is a recognition time the customer's process engineering team will treat as a known cost, and the cost is either a fiducial redesign or a vision program re-tuning.

False-reject rate is the second item, and it is the item that determines whether the line can run unattended. A false-reject rate of less than 0.1% is typical; a rate of more than 1% is a rate the customer's production manager will treat as a known cost, and the cost is either a vision program re-tuning or an operator intervention. The matrix has to record the false-reject rate at FAT, the false-reject rate at SAT, and the false-reject root-cause analysis. A root cause that is "fiducial contamination" is a root cause the customer's process engineering team can address with a stencil-cleaning cycle; a root cause that is "fiducial design" is a root cause that requires a board redesign. The matrix has to record the root cause for every false reject, and the root cause has to be visible in the 30-day check.

8. Trial run, training, and continuous improvement

Trial run, training and continuous improvement is the family that converts the acceptance measurements into a production-ready line, and it is the family that determines whether the customer's engineering team can run the line without the vendor's help in year two. The three items are trial run acceptance criteria, training depth, and handover plus the first 30 days. Each item has a measurement procedure, a pass/fail threshold, and a known failure mode, and each item has to be measured on the customer's representative product, with the customer's operators running the machine. A SAT that uses the vendor's reference product for the trial run is a SAT that the customer's production manager will have to redo on the production product, and the redo is the place where the late-stage yield surprises appear.

The three items are also the three items that the customer's shift lead will be repeating in production, and a SAT that does not produce a documented procedure for each item is a SAT that the shift lead will be inventing procedures for in the first month of production. The invented procedures are the third most common source of yield drift (after the squeegee pressure and the recipe), and the drift shows up in the 30-day check as a list of process-related defects that the FAT matrix did not catch. The three items below are the three items the shift lead will be repeating, and a mature acceptance programme exercises all three on the customer's representative product, with the customer's operators running the machine, and with the customer's shift lead present for every measurement.

8.1 Trial run acceptance criteria

Trial run acceptance criteria are the first item, and they are the criteria that determine whether the line is ready for production. The items are the trial run duration (typically 4 to 8 hours), the trial run product mix (the customer's top 5 products, weighted by volume), the trial run yield target at SPI (typically 95% first-pass for consumer electronics, 99% for high-reliability), the trial run yield target at AOI (typically 98% first-pass for consumer electronics, 99.5% for high-reliability), and the trial run OEE target (typically 75% at 4 hours, 80% at 8 hours, 85% at 24 hours). A trial run that ends below the SAT target is a trial run the customer's production manager will treat as a known ramp, and the ramp has to be visible in the deviation log so the line team knows what to expect on day one of production.

The trial run defect Pareto is the second item, and it is the item that the customer's process engineering team will use to plan the first month of production. The Pareto has to be recorded by defect type (insufficient paste, bridging, misalignment, contamination), by paste batch, by stencil cleaning cycle, by shift, and by operator. A Pareto that shows a single defect type accounting for more than 30% of the total defects is a Pareto that the customer's process engineering team will treat as a known cost, and the cost is either a recipe adjustment or a stencil-cleaning cycle adjustment. The matrix has to record the Pareto at FAT, the Pareto at SAT, and the Pareto at the 30-day check, and the three Paretos have to be compared in the 30-day check.

8.2 Training depth: operator, process engineer, maintenance

Training depth is the second item, and it is the item that determines whether the customer's team can run the line without the vendor's help. The three training tracks are operator training, process engineer training, and maintenance training. Operator training covers the daily start-up, the daily shut-down, the board load, the stencil change, the squeegee change, the paste cartridge change, the basic alarm acknowledgement, and the basic recipe selection. Process engineer training covers the recipe build, the recipe tuning, the SPI tuning, the stencil-cleaning cycle tuning, the squeegee pressure tuning, the MES interface, and the closed-loop recipe adjustment. Maintenance training covers the daily, weekly and monthly preventive maintenance, the top-10-alarm diagnosis, the first-line replacement of the top-20 wear parts, and the firmware update procedure.

Each training session has a duration, an agenda, a sign-off, and a competency check. The sign-off has to include the trainee's name, the trainer's name, the date, and the pass/fail result. A training session that ends without a competency check is a training session the customer's shift lead will have to repeat in three months, when the operator has forgotten the procedure. The minimum viable training record is: every operator who will run the line on every shift, every process engineer who will tune the line, and every maintenance technician who will service the line, with each training record signed by both the trainee and the trainer. Anything less is a partial training, and the partial training has to be visible in the deviation log with a 30-day close-out target.

8.3 Handover, ramp-up, and the first 30 days

Handover is the third item, and it is the item that converts the vendor's commissioning engineer into the customer's in-house capability. The handover package is the as-built documentation (already signed at FAT), the recipe library (built and tested at SAT), the SPI tuning guide (built and tested at SAT), the stencil and squeegee change procedure (built and tested at SAT), the top-10-alarm diagnosis flowchart (built and tested at SAT), the preventive maintenance schedule, the recommended spares list, and the vendor's local service contact (with phone, email, and response-time commitment). A SAT that does not produce the handover package is a SAT the customer's engineering team will treat as a partial acceptance, and the partial acceptance has to be visible in the deviation log with a 30-day close-out target.

The ramp-up and the first 30 days are the second and third items, and they are the items that determine whether the line hits steady-state OEE in the first month. The ramp-up is tracked through a daily OEE measurement, a daily yield measurement at SPI and AOI, and a daily defect Pareto. The 30-day check is the formal review of the ramp-up, with the customer's engineering team comparing the FAT numbers, the SAT numbers, the trial-run numbers, the first-month production numbers, and the open deviation list. A 30-day check that finds a deviation in the open list is a 30-day check that triggers a vendor commitment to fix, with a commitment date. A 30-day check that finds a deviation that the vendor disputes is a 30-day check that escalates to the customer's procurement team, and the escalation is the place where the FAT/SAT documents earn their keep.

9. Common FAT-pass / SAT-fail failure modes

The most expensive way to learn a lesson in SMT acceptance is to learn it on the customer's floor, after the crate has been opened, after the line has been rigged, and after the production manager has scheduled the first 30 days of production. The three failure modes that appear most often are vibration / transport / re-leveling, paste batch and environmental drift, and operator skill gap. Each failure mode has a known signature at SAT, a known root cause, and a known prevention at FAT, and the customer's engineering team should walk into SAT with the three failure modes already on the deviation log template, so any signature that appears at SAT can be classified in seconds rather than in days.

The three failure modes are also the three failure modes that the 30-day check will surface even when the SAT numbers looked clean on the day. A SAT that records "all 28 rows pass" and a 30-day check that surfaces a vibration resonance on day 14 is a SAT that missed the vibration measurement at row 7, and the missed measurement is the place where the matrix earns its keep. The three failure modes below are the three failure modes the matrix has to guard against, and a matrix that does not guard against them is a matrix the customer's engineering team will be re-writing in the first month of production.

9.1 Vibration, transport, and re-leveling

The first failure mode is vibration, transport and re-leveling. A machine that is level to 0.04 mm/m at FAT can arrive at the customer's floor level to 0.12 mm/m, and the difference is enough to push the print-accuracy number from ±0.025 mm to ±0.05 mm at 2 σ. The root cause is one of three: the crate was dropped, the rigging was not done correctly, or the customer's floor is not as flat as the vendor's floor. The first two are vendor problems; the third is a customer problem, and the only way to know which is to measure the floor flatness before the crate is opened, and to record the result against the machine's rigging log. A floor flatness of more than 2 mm over 5 m is a flatness the customer's facility team will have to address, and the cost of the floor work has to be visible in the deviation log.

The second part of the failure mode is transport. A machine that is shipped by air has a different vibration profile from a machine that is shipped by sea, and the difference shows up in the squeegee parallelism measurement at SAT. A squeegee parallelism that was 0.03 mm at FAT and is 0.08 mm at SAT is a squeegee that has to be re-adjusted, and the re-adjustment is a vendor commitment with a 30-day close-out. The prevention at FAT is to record the squeegee parallelism before and after a simulated transport test (a 30-minute shake on a vibration table with the machine on its shipping skids), and the result is the baseline the customer's SAT engineer has to match. A simulated transport test that produces a 0.05 mm change in squeegee parallelism is a test that the vendor's engineering team has to address before the crate is sealed.

9.2 Paste batch and environmental drift

The second failure mode is paste batch and environmental drift. A machine that prints a Cpk of 1.67 on the vendor's paste at FAT can print a Cpk of 1.20 on the customer's paste at SAT, and the difference is the difference between a 4 σ process and a 3 σ process. The root cause is one of three: the customer's paste has a different metal load, the customer's paste has a different particle size distribution, or the customer's ambient is different from the vendor's ambient by more than 5 °C or 20% RH. The first two are paste specification issues; the third is a facility issue, and the only way to know which is to record the paste lot number, the metal load, the particle size, and the ambient conditions at both FAT and SAT. A paste that is 0.5% low on metal load is a paste the customer's incoming-quality team will have to accept or reject, and the acceptance has to be visible in the deviation log.

The environmental drift is the second part of the failure mode, and it is the part that the customer's facility team will own. The items are ambient temperature, ambient humidity, ambient particulate count, and ambient electrostatic field. A temperature drift of 5 °C over a shift is a temperature drift the squeegee pressure will follow, and the pressure drift shows up as a Cpk drift over the shift. The prevention at FAT is to record the ambient conditions at every measurement, and the prevention at SAT is to record the ambient conditions at every measurement, and the two records have to be within 5 °C and 20% RH of each other. An ambient that is outside the envelope is an ambient the customer's facility team will have to address, and the cost of the facility work has to be visible in the deviation log.

9.3 Operator skill and recipe hand-off

The third failure mode is operator skill and recipe hand-off. A machine that is operated by a trained engineer at FAT can be operated by a less-experienced operator at SAT, and the difference shows up as a yield drift over the first month of production. The root cause is one of three: the operator was not trained on the customer's specific product, the recipe hand-off was not documented, or the top-10-alarm diagnosis flowchart was not transferred to the customer's shift lead. The first two are training issues; the third is a documentation issue, and the only way to know which is to record the training record, the recipe hand-off record, and the alarm flowchart transfer record at both FAT and SAT. A training record that is missing an operator's name is a training record the customer's shift lead will treat as a known gap, and the gap has to be visible in the deviation log with a 30-day close-out target.

The recipe hand-off is the second part of the failure mode, and it is the part that the customer's process engineering team will own. The items are the recipe library (with every product the customer runs), the recipe change log, the recipe version control system, and the recipe roll-back procedure. A recipe library that is missing the customer's top 5 products is a recipe library the customer's process engineering team will have to build on the first day of production, and the build is a known cost the FAT matrix did not catch. The prevention at FAT is to build the recipe library with the customer's engineering team present, with every product the customer runs, and the prevention at SAT is to verify that the recipe library is intact, with every product the customer runs, and the verification has to be visible in the deviation log.

10. The one-page FAT/SAT checklist you can actually use

The 28-row matrix in §4 is the detailed version, but the one-page version a process engineer can actually take into a vendor meeting is a 14-row condensed version, with the same items, the same pass/fail thresholds, and the same column structure. The one-page version is the document the customer prints on A4, the document the vendor signs in the meeting, and the document the customer's engineering team refers to in the 30-day check. The one-page version is not a substitute for the 28-row matrix; it is a summary of the matrix, and any item the customer wants to add to the matrix has to be added to the one-page version as well.

The one-page version is also the document the customer's procurement team will use to compare two vendor proposals. A vendor that agrees to all 14 items in the one-page version is a vendor that has read the matrix; a vendor that agrees to only 10 of the 14 items is a vendor that has not read the matrix, and the customer's procurement team should treat the 4 missing items as a known risk. The one-page version is the cheapest insurance against the late-stage surprises that the customer is paying FAT and SAT to prevent.

10.1 How to print and use the table

The one-page table is the document the process engineer prints on A4 and takes into the vendor meeting. The print is one-sided, single-page, with the deviation report and the sign-off block on the back of the same sheet. A one-page table that the engineer has to print on two pages is a table that will not be used at the meeting, and a table that is not used at the meeting is a table that the vendor will quietly ignore. The font is the standard 10-pt monospace, with the row count visible in the header, and the column widths set to fit the longest item in each column. The one-page constraint is non-negotiable: a two-page table is a document the customer's engineering team will not refer to in the meeting, and the meeting is the only place the document has leverage.

The table is also the document the engineer uses to record the FAT and SAT values. The FAT engineer fills the FAT value column on the vendor's bench, with the date, the time, the procedure version, and the signature. The SAT engineer fills the SAT value column on the customer's floor, with the same fields, and the customer's engineering lead signs both columns at the end of each acceptance. The signature is the legal record of acceptance, and the signed table is the document the customer's procurement team will refer to in the invoice release. A table that is missing a signature is a table the customer's procurement team will treat as a partial acceptance, and the partial acceptance has to be visible in the deviation log with a 30-day close-out target.

Table 3 — The one-page FAT/SAT checklist. Print on A4, take into the vendor meeting, and fill the FAT value column on the vendor's bench, the SAT value column on the customer's floor.

The table is a fillable form: the FAT engineer fills the FAT value column on the vendor's bench, the SAT engineer fills the SAT value column on the customer's floor, and the customer's engineering lead signs both columns at the end of each acceptance. The signature is the legal record of acceptance, and the signed table is the document the customer's procurement team will refer to in the invoice release. A table that is missing a signature is a table that the customer's procurement team will treat as a partial acceptance, and the partial acceptance has to be visible in the deviation log with a 30-day close-out target.

10.2 Sample deviation report and sign-off block

The deviation report is the second half of the one-page document, and it has to live on the same page as the table. The deviation report records every row that did not pass, the FAT value, the SAT value, the delta, the deviation class ("fix before sign-off", "fix within 30 days", "accepted with comment"), the vendor's commitment date, and the closure date. The sign-off block records the customer's engineering lead, the customer's production manager, the customer's maintenance lead, the vendor's commissioning engineer, and the vendor's quality manager, with a signature line for each. The block has to be physically signed before the line is released to production, and the signatures are the legal record of acceptance.

A sample sign-off block is reproduced below. The block is a template; a customer's block may add rows for additional stakeholders (for example, the customer's EHS team for the exhaust flow row, or the customer's IT team for the MES round-trip row), but the additions have to be agreed in writing before FAT starts, and they have to use the same column structure. A sign-off block with a different column structure for the customer-specific rows is a sign-off block that the vendor's commissioning engineer cannot sign on the day, and the customer will end up with a partial acceptance that the vendor will treat as a full acceptance.

Table 4 — The sign-off block. Every row signed before the line is released to production. The signatures are the legal record of acceptance, and the signed block is the document the customer's procurement team will refer to in the invoice release.

11. Reference configurations from the I.C.T range

The I.C.T solder paste printer family covers four configurations that map onto four different FAT/SAT profiles, and the matrix has to be tuned for the configuration the customer is buying. A compact semi-automatic bench machine has a different signature at SAT from a long-board inline machine, and the differences show up in the working area, the vision system, the squeegee drive, the stencil-cleaning cycle, and the closed-loop SPI integration. The four configurations below are the four configurations the I.C.T range ships today, and the FAT/SAT profile for each is the profile a customer's engineering team should expect to see in the matrix.

Each configuration has its own strengths and its own failure modes, and the matrix has to record both. A compact bench machine is the right answer for a low-volume prototype line or a high-mix NPI line, and the FAT/SAT matrix is correspondingly short. A long-board inline machine is the right answer for a high-volume LED line, and the FAT/SAT matrix is correspondingly long. The three sub-sections below map the four configurations onto the matrix, with the rows that are unique to each configuration, the rows that are shared across all four, and the failure modes that are unique to each configuration. The mapping is not a sales document; it is an engineering document, and the customer's engineering team should use it to decide which configuration fits which FAT/SAT profile.

11.1 Compact semi-automatic (P6, P12)

The compact semi-automatic class covers the I.C.T P6 and the I.C.T P12, both of which are bench-top machines designed for prototype runs, NPI runs, and small-batch production. The P6 is the entry-level machine, with a working area of 400 × 300 mm, manual board loading, manual stencil cleaning, and a single squeegee drive. The P12 is the higher-end bench machine, with a working area of 500 × 400 mm, manual board loading, semi-automatic stencil cleaning, and a dual squeegee drive with independent pressure control. Both machines share the same vision system, the same fiducial recognition algorithm, and the same MES interface, and the FAT/SAT matrix is correspondingly consistent between the two models.

The FAT/SAT profile for the compact semi-automatic class is shorter than the inline class, and the rows that are unique to the class are: manual board loading force (the operator's hand force on the loading handle), manual stencil cleaning cycle (the operator's cleaning stroke), and the alignment accuracy under manual loading (the operator's loading repeatability). The rows that are shared with the inline class are: frame level, squeegee pressure, stencil release distance, fiducial recognition time, MES round-trip, and training. A matrix that records the manual loading force and the manual cleaning cycle is a matrix that captures the operator-skill failure mode at SAT; a matrix that does not is a matrix that misses the most common SAT deviation on the compact class. The I.C.T P6 semi-automatic stencil printer is the right fit for a customer's NPI line, and the FAT/SAT matrix for the P6 should be the 14-row condensed version with the three class-specific rows added.

11.2 Inline fully automatic (5134, 5151)

The inline fully automatic class covers the I.C.T 5134 and the I.C.T 5151, both of which are conveyorised inline machines designed for high-volume production. The 5134 is the LED panel class machine, with a working area of 1200 × 400 mm, automatic board loading, automatic stencil cleaning, dual squeegee drive, and the long-board vision system. The 5151 is the high-end LED class machine, with a working area of 1500 × 500 mm, automatic board loading, automatic stencil cleaning, dual squeegee drive, and the long-board vision system with closed-loop SPI integration. Both machines share the same conveyor interface, the same MES interface, the same vision system, and the same preventive maintenance schedule, and the FAT/SAT matrix is correspondingly consistent between the two models.

The FAT/SAT profile for the inline fully automatic class is longer than the compact class, and the rows that are unique to the class are: conveyor hand-off timing (the timing of the board hand-off from the loader to the printer and from the printer to the SPI), the closed-loop SPI round-trip time (the time from the SPI read to the recipe adjustment), the line-integration test (the 10-board end-to-end test), and the long-board stencil frame size (the 23 × 23 in or 29 × 29 in frame that the customer's stencil has to fit). The rows that are shared with the compact class are: frame level, squeegee pressure, stencil release distance, fiducial recognition time, MES round-trip, and training. A matrix that records the conveyor hand-off timing and the closed-loop SPI round-trip is a matrix that captures the line-integration failure mode at SAT; a matrix that does not is a matrix that misses the most common SAT deviation on the inline class. The I.C.T 5134 online fully automatic solder paste printer is the right fit for a high-volume LED line, and the FAT/SAT matrix for the 5134 should be the 28-row detailed version with the four class-specific rows added.

11.3 Which configuration fits which FAT/SAT profile

The choice between the compact semi-automatic class and the inline fully automatic class is a choice between three trade-offs: volume, mix, and integration. A customer with a high-volume, low-mix production profile (for example, a dedicated LED panel line running one product family) is the right fit for the inline class, and the FAT/SAT matrix is the 28-row detailed version. A customer with a low-volume, high-mix profile (for example, a contract manufacturer running 50 different products a week) is the right fit for the compact class, and the FAT/SAT matrix is the 14-row condensed version with the three class-specific rows added. A customer with a mixed profile (for example, a dedicated line for the top 5 products plus a bench machine for the long tail) is the right fit for a combination, and the FAT/SAT matrix is the union of the two class-specific matrices.

The integration trade-off is the second axis, and it is the axis that most often catches the customer's engineering team by surprise. A compact bench machine is a standalone asset, and the FAT/SAT matrix does not have to include the line-integration row; an inline machine is a working part of the customer's SMT line, and the FAT/SAT matrix has to include the line-integration row. A customer that buys an inline machine and applies a compact matrix is a customer that will discover the line-integration row at SAT, and the discovery will cost the customer a day of line downtime. A customer that buys a compact machine and applies an inline matrix is a customer that has over-specified the acceptance, and the over-specification will cost the customer a week of vendor negotiation. The right matrix for the configuration is the cheapest insurance against both mistakes.

Talk to an I.C.T process engineer about your FAT/SAT matrix

If you are planning a new SMT line, replacing an existing printer, or formalising a FAT/SAT procedure for an in-house engineering team, send us your product profile (board size, fiducial layout, aperture range, paste type, target Cpk, target OEE) and we will return a tailored FAT/SAT matrix with the rows your line needs, the pass/fail thresholds against your product band, and the I.C.T configuration that fits your volume and mix. The matrix is yours to keep, regardless of whether you buy from us.

This guide is maintained by the I.C.T process engineering team. Last updated: October 2026. For the latest revision, or to request a specific FAT/SAT row we have not covered, contact us through the product hub above.

12. Frequently asked questions

What is the difference between FAT and SAT for an SMT solder paste printer?

FAT (Factory Acceptance Test) is performed on the vendor's bench, with the vendor's paste, the vendor's stencil, and the vendor's engineer signing the report. SAT (Site Acceptance Test) is performed on the customer's floor, with the customer's paste, the customer's stencil, the customer's board, and the customer's operators running the machine. The two tests share the same matrix, the same pass/fail thresholds, and the same measurement procedure, but the conditions are different. FAT is the cheaper place to find a defect; SAT is the more representative place to confirm a number. A mature acceptance programme treats FAT as a defect-discovery exercise and SAT as a defect-confirmation exercise, and the two documents are linked row-by-row through the matrix.

How long does a typical FAT take for a fully automatic inline printer?

A typical FAT for a fully automatic inline printer takes two to three working days on the vendor's bench, with day one for mechanical and motion acceptance, day two for print accuracy and repeatability, and day three for program / recipe / MES / training. A FAT that is compressed into a single day is a FAT that the customer's engineering team will treat as a visual inspection, and the visual inspection is the most common source of late-stage SAT deviations. A FAT that extends beyond three days is a FAT that the vendor's engineering team is using to fix defects the customer's engineering team should have caught during the buying stage, and the extension is a sign that the matrix is being used as a repair tool rather than as an acceptance tool.

How long does a typical SAT take?

A typical SAT for a fully automatic inline printer takes three to five working days on the customer's floor, with day one for uncrating, mechanical re-acceptance, and utilities connection, day two for first-article print and recipe hand-off, day three for trial run and OEE ramp, day four for training, and day five for documentation and sign-off. A SAT that is compressed into two days is a SAT that the customer's engineering team will treat as a confirmation exercise, and the confirmation will miss the operator-skill and recipe-hand-off failure modes that show up only in the training and trial-run days. A SAT that extends beyond five days is a SAT that the vendor's commissioning engineer is using to fix defects the customer's engineering team should have caught at FAT, and the extension is a sign that the FAT matrix was not applied rigorously.

What pass/fail threshold should the customer use for print accuracy?

The pass/fail threshold for print accuracy depends on the product class. A consumer-electronics line running a 0.65 mm pitch QFP can use ±0.04 mm at 2 σ on the reference coupon; a high-reliability line running a 0.4 mm pitch BGA should use ±0.025 mm at 2 σ; an LED panel line running a 1206 resistor can use ±0.05 mm at 2 σ. The customer's process engineering team has to set the threshold against the actual product band, not against the vendor's default band, and the threshold has to be recorded in the matrix before FAT starts. A threshold that is set after the FAT measurement is a threshold the vendor can argue, and the argument is the most common source of late-stage SAT deviations.

What is the most common FAT deviation that shows up at SAT?

The three most common FAT deviations that show up at SAT are: squeegee pressure drift (the FAT pressure holds within ±0.5 kg, the SAT pressure drifts by 1.5 kg over an 8-hour shift), frame re-leveling (the FAT frame is level to 0.04 mm/m, the SAT frame is level to 0.12 mm/m after rigging), and the closed-loop SPI round-trip (the FAT SPI round-trip is 150 ms, the SAT SPI round-trip is 350 ms because the customer's MES is slower than the vendor's demo MES). Each of the three has a known prevention at FAT (record the pressure drift over 8 hours and 3 paste batches, record the frame level after a simulated transport test, record the SPI round-trip on the customer's MES), and a mature acceptance programme applies all three preventions at FAT.

Can a customer run FAT and SAT with a single engineer?

A single engineer can run FAT and SAT for a compact bench machine, but a fully automatic inline machine requires a minimum of two engineers: one from the customer's engineering team (the process engineering lead or the incoming-quality representative) and one from the vendor's commissioning team. The single-engineer model is the most common source of acceptance drift, because the single engineer is the single point of failure for both the FAT and the SAT, and a defect that the single engineer misses at FAT is a defect the single engineer will miss at SAT. A two-engineer model catches the defect at one of the two stages, and the two documents become a cross-check rather than a single point of failure.

What happens if a FAT deviation is discovered at SAT?

A FAT deviation discovered at SAT is a deviation the vendor's floor was supposed to catch, and the discovery at SAT is a sign that the FAT was not applied rigorously. The customer's engineering team has three options: re-run the FAT measurement on the customer's floor (with the vendor's engineer present, and with the cost of the re-measurement charged to the vendor), accept the deviation as a "fix within 30 days" item (with the vendor committing to a fix in writing, and the invoice held until the fix is verified), or escalate the deviation to the customer's procurement team (with the FAT report re-opened and the deviation re-classified). The third option is the most expensive for the vendor, and the most common outcome is the first option, with the vendor absorbing the re-measurement cost.

How does the FAT/SAT matrix interact with a vendor's standard warranty?

The FAT/SAT matrix and the vendor's standard warranty are two separate documents, and the two have to be aligned at the contract stage. The matrix records the acceptance criteria; the warranty records the post-acceptance coverage. A matrix that records a print-accuracy threshold tighter than the warranty's coverage is a matrix that the vendor will argue in year two, and the argument is the most common source of warranty disputes. A warranty that covers the print-accuracy threshold the matrix recorded is a warranty that the customer's engineering team can use to enforce a fix in year two. The two documents have to be cross-referenced at the contract stage, and the cross-reference has to be visible in the FAT cover page.