Publish Time: 2026-10-13 Origin: Site
Procurement guide for buyers and process engineers · 28 questions across working area, alignment, support, squeegee, stencil, program, documentation, training and service
Short answer: a competent SMT solder paste printer buying checklist is built around nine evaluation areas — working area, alignment, support, squeegee, stencil, program, documentation, training and service — and turns each area into a small number of measurable questions that can be answered on the shop floor before any vendor visit. The list is not a questionnaire for the salesperson; it is a procurement tool that ranks candidate machines against the actual production conditions of the buyer's plant.
If you need the full family of solder paste printers and how they break down by automation class, start at our solder paste printer product hub. This guide goes deeper into the questions buyers should ask before they sign a purchase order, the trade-offs behind each answer, and the practical way to assemble an evaluation matrix for the meeting room.
Every SMT printer vendor has 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 a line succeeds or fails. The reason is the combination of small, line-specific details that the brochure never mentions — how the bottom side of a thin board flexes during separation, how the fiducial recognition behaves on a textured copper surface, how the squeegee pressure drift interacts with the solder paste batch your line actually uses, and how the vendor's local service engineer is going to reach your plant on a public holiday.
An SMT solder paste printer buying checklist is what turns those hidden details into questions that have to be answered before a purchase order is signed. It also gives the buyer's engineering team a shared vocabulary: when someone says "we need to verify squeegee pressure control on the worst paste batch", the team knows which bench setup, which paste type and which pressure range the conversation refers to. Without a checklist, the same conversation takes three meetings and still ends with the vendor's best slide on the projector.
A buying checklist is a fixed list of questions grouped by evaluation area, where each question has a measurable answer and a pass/fail threshold. It is not a comparison spreadsheet of vendor specifications, because vendors rarely quote the same operating condition for the same number. A cycle time of "8 seconds per board" at the vendor's demo lab is not the same number as a cycle time of "8 seconds per board" on your worst product, with your paste, at your conveyor speed. The checklist forces both sides to agree on the operating condition first, and on the number second.
The checklist is also not a price negotiation tool. Price negotiations happen after the technical shortlist is complete, and only between machines that have passed the same set of questions. Mixing price into the technical shortlist almost always produces a winner that is the cheapest machine capable of meeting the easier half of the requirements — and that winner usually costs more in the second year of operation than the machine that was actually the right fit.
The most common failure mode is the machine that is technically capable of running the line but cannot be integrated without modifications. A 1 200 mm long-board printer is a perfectly good machine, but it is the wrong machine if the line is 800 mm wide and the conveyor cannot be extended. A semi-automatic printer is also a perfectly good machine, but it is the wrong machine if the line runs three shifts and the labour cost of one operator per printer exceeds the saving of buying the cheaper class. Both are the same kind of failure: the specification was read in isolation, not against the operating envelope of the buyer's plant.
The second most common failure mode is the machine that arrives without the documentation, training or spares that the line needs to stay productive. The first 30 days of a new printer are when the line loses the most money, because the operators are learning the machine, the recipes are being tuned, and any delay in service response translates directly into lost production hours. Skipping the documentation and training questions in the buying stage is the single most expensive way to save ten minutes during the procurement process.
Sections 2 through 10 walk through the nine evaluation areas one by one. Each section gives the questions the buyer should be able to answer, the threshold that separates a "pass" from a "fail", the operating condition under which the number should be measured, and the failure mode that appears when the question is skipped. Section 11 maps the I.C.T product range to the most common checklist profiles. Section 12 collapses all of the above into the one-page table that goes into the meeting room. If a buyer reads only section 12, they will still be ahead of most of the brochures in the industry; if they read all of it, they will be able to challenge any vendor's number on the spot.
The order of the sections is intentional. Working area comes first because it is a hard constraint that filters the candidate list faster than any other question. Alignment, support, squeegee and stencil follow because they are the four process areas that determine first-pass yield, and they are evaluated as a group on the same worst-case panel. Program, documentation, training and service come last because they determine the total cost of ownership rather than the technical capability, and they are the four areas where the vendor's answers are most easily verified by asking an existing customer.
The first evaluation area is the working area — the largest board or panel the printer can accept, the way the board enters and exits, and the practical size of the conveyor. Working area is the single specification that decides whether a machine is even physically capable of running the buyer's product, and it has to be answered before any of the other eight areas matter. A machine that fails this question cannot be rescued by any of the other criteria, no matter how good its alignment repeatability or squeegee control turns out to be, because a board that cannot be clamped cannot be printed at all.
The section breaks the question into three sub-questions that together define the working envelope: the largest bare board and the largest panelized array, the long-board and oversized formats that push the envelope beyond the standard 400 × 400 mm class, and the conveyor geometry that determines how the board enters and leaves the print area. Each of the three has its own failure mode, and each of the three has a number that the buyer's engineering team can measure before the vendor meeting rather than during it.
The question to ask is not "what is the largest board the machine can run" but "what is the largest panelized array the machine can run". Real production rarely loads a single bare PCB; it loads either a panelized array with break-away rails or a board mounted on a custom carrier or pallet, and both formats add 20 to 60 mm of border per side for tooling holes, fiducials and the clamp footprint. The number on the spec sheet is the bare board; the number the line needs is the panelized array, with margin.
A practical rule is to take the largest bare PCB, add 40 to 80 mm per side for panel rails and clamp footprint, and that is the working area the line should be looking for. Anything smaller forces a redesign of the panel, an awkward custom jig, or a workflow where operators manually reposition the board mid-print — all of which are expensive workarounds that defeat the purpose of buying a printer in the first place.
LED panel lighting, home appliance controllers, industrial power boards and some automotive electronics routinely use boards longer than 600 mm, often in the 1 200 mm range, and some LED lines go further. A compact 400 × 400 mm working area cannot clamp these panels, and forcing them onto a small machine with oversize jigs is one of the more expensive mistakes buyers make. The right answer for these applications is a long-board class printer with a working area that comfortably accepts the full panel plus a small margin for clamping and fiducial recognition.
For very long panels — 1 500 mm and above — buyers should also verify that the printer's stencil frame can be sized to the application. A 1 500 mm long panel typically needs a custom-cut stencil, and the stencil frame size (23 × 23 in, 29 × 29 in, or custom) becomes a real procurement constraint. A vendor that does not stock oversize stencil frames in its local market will cost the line several days of waiting every time the panel design changes.
The conveyor is the third working-area question, and it is one that buyers frequently under-specify. The conveyor has to be wide enough to accept the panel, plus the edge clearance the printer's fiducial camera needs to recognise the board edge, plus any mechanical guide rails. A 400 mm wide conveyor can run a 350 mm wide panel; a 350 mm wide conveyor on the same printer can run a 320 mm wide panel, and the difference is rarely a problem — until the line is redesigned and the panel grows by 30 mm.
Buyers should also check the panel direction convention: some printers accept panels only in the long-axis-forward direction, while others accept both directions. For lines that run a mix of long LED panels and compact boards, dual-direction capability removes a real source of operator error. For single-format lines, the convention is less important, but it should still be specified in the checklist so the line is not surprised by a custom carrier later.
Typical application | Bare PCB size (mm) | Recommended working area (mm) | Conveyor width (mm) |
|---|---|---|---|
Consumer electronics, IoT, instrumentation | ≤ 350 × 350 | 400 × 400 or larger | 350 – 400 |
Industrial control, telecom base-station modules | ≤ 500 × 400 | 500 × 500 or larger | 450 – 500 |
LED panel lighting, signage | 600 – 1 200 | 1 200 × 300 class long-board | 350 – 400 (panel rides long-axis) |
Home appliance controllers, large industrial | 1 200 – 1 500 | 1 500 × 350 class long-board | 350 – 400 (panel rides long-axis) |
Double-sided or panelized automotive | 300 × 400 (panel 400 × 600) | 600 × 600 or larger | 400 – 450 |
For a step-by-step walk through how PCB size interacts with the rest of the selection criteria, our solder paste printing machine selection guide covers the same six dimensions in a single decision framework. Buyers who already know the rough size of the printer they need can jump straight to section 11 and use the I.C.T range as a sanity check against the numbers they have collected.
Alignment is the evaluation area where a small number on the brochure can hide a very large difference in line behaviour. The headline number — typically expressed as "alignment accuracy ±X µm" or "repeatability ±Y µm at 6σ" — is almost always quoted on a perfect test board in a vendor demo lab. The number that matters in production is the repeatability on the buyer's actual product, on the buyer's actual fiducials, after the buyer's operators have loaded the machine 200 times.
The alignment section is therefore built around five questions rather than one: whether the vision system can recognise the fiducials the buyer's boards actually carry, whether the camera resolution and field of view support the buyer's smallest aperture, what the difference is between accuracy and repeatability and which of the two the buyer should weight, how the printer compensates for in-plane board skew, and how the alignment system feeds the downstream inspection loop. Vendors rarely volunteer all five; the checklist asks for all five.
The first alignment question is whether the printer's vision system can recognise the fiducials the buyer's PCBs actually carry. A fiducial that looks trivial in a CAD viewer can be difficult for a vision system if it sits on a textured copper pour, next to a silkscreen mark, or inside a panel break-away rail. Buyers should bring two or three worst-case panels to the vendor demo and ask the vendor to run a full print cycle on each one. A vendor that confidently says "yes, the camera will find it" without seeing the panel is selling the brochure, not the machine.
Two fiducials is the minimum for a standard panel; three or four fiducials are common for large panels and for panels that need skew compensation across the print area. The printer has to support the right number of fiducials, in the right positions, and the recipe editor has to let the buyer define asymmetric fiducial layouts without resorting to scripting. A printer that requires all four fiducials to be in a regular grid will fail the moment the buyer's panel design uses an L-shaped fiducial layout for cost reasons.
Camera resolution and field of view are the two numbers that determine whether the vision system can hit the alignment tolerance the buyer's product requires. A 5 MP camera with a 10 × 7 mm field of view gives a pixel pitch of roughly 2 µm, which is the practical floor for ±12.5 µm repeatability on small-pitch QFN. A 2 MP camera with the same field of view gives a pixel pitch of roughly 3.5 µm, which is the practical floor for ±25 µm — fine for 0.5 mm pitch QFP, marginal for 0.4 mm pitch QFN, and inadequate for 0.3 mm pitch wafer-level packages.
Buyers should ask the vendor for the camera's pixel pitch at the field of view used for production alignment, and they should verify that the field of view covers the fiducial plus a margin for the search area the vision algorithm uses. A camera that is technically high-resolution but has too narrow a field of view forces the algorithm to take multiple sub-images and stitch them, which slows the cycle and adds a small but real source of alignment drift.
Repeatability is the number buyers should weight most heavily; accuracy is the number vendors tend to quote. Accuracy is the distance between the nominal position and the average measured position, and a well-calibrated machine can hit very high accuracy on a single fiducial. Repeatability is the spread of measured positions when the same fiducial is recognised 100 times in a row, and it is the number that actually determines yield on small-pitch components. A machine that is accurate to ±5 µm but only repeatable to ±25 µm will print fine on 0.5 mm pitch QFP and badly on 0.4 mm pitch QFN.
For 0.4 mm pitch QFN and small BGA, the practical repeatability target is ±20 µm or better. For 0.3 mm pitch wafer-level or very fine-pitch BGA, the target tightens to roughly ±12.5 to 15 µm. Buyers should always confirm these numbers on a fine-pitch test board rather than on the headline six-axis stage accuracy, because the difference between the two is the difference between printing QFNs reliably and not printing them at all.
Real panels are not perfectly rectangular: they have bow, twist and in-plane skew that depend on the substrate, the panelization scheme and the reflow history of the bare PCB. The vision system has to compensate for these effects, and the compensation algorithm has to be both fast and visible. Buyers should ask the vendor to demonstrate PCB-side skew compensation on a panel that the buyer knows is at the edge of the supplier's flatness specification, and they should see the compensation on the HMI screen as a measurable value, not as a hidden internal flag.
Compensation that is invisible to the operator is compensation that cannot be audited when the line starts producing defects. The recipe editor should display the measured PCB-side compensation as a number, the audit log should record the compensation value at each print, and the print should be paused automatically if the compensation exceeds a configurable threshold. A printer that hides any of these three steps from the operator is a printer that will surprise the line one day.
Alignment and inspection are linked. A printer that can align a panel to ±12.5 µm and then send a "print successful" signal to a downstream SPI is a different class of equipment from a printer that prints the panel and assumes the SPI will catch whatever was missed. The first is part of a closed-loop process; the second is part of a serial process with a yield hole in the middle. The buying checklist should ask whether the printer supports inline SPI feedback, and at what protocol — SECS/GEM, OPC-UA, or a vendor-proprietary link — because the protocol determines how easily the printer can be integrated into the buyer's factory software stack.
For a deeper treatment of how printer alignment and SPI feedback combine to lift first-pass yield, our SMT printer and SPI matching guide walks through the closed-loop architecture, the typical yield uplift on a fine-pitch line, and the failure modes that appear when the two systems are not properly matched. Buyers who already know they will run inline SPI should read that guide before they sit down with a printer vendor, because the questions in this section will be the bridge between the two documents.
Support is the area of the buying checklist that is least visible in a brochure and most expensive to fix after the machine is installed. Support is what keeps the board flat, in the right place, and at the right height during the print stroke; a machine with excellent alignment and a poor support system will produce boards that pass the vision system and fail the print. The buying checklist should treat support as a first-class evaluation area, not a footnote.
Support is also the area where the buyer's existing tooling — the pallets, carriers and jigs the plant already owns — either fits the new machine or does not. A printer that is technically excellent but cannot accept the buyer's existing carriers will force a re-tooling programme that can cost 10 to 20% of the machine price and add months to the ramp-up. The four questions in this section cover the top clamp, the bottom support, the carrier interface and the repeatability test that catches all three.
The top clamp is what holds the board against the support system during the print stroke, and the geometry of the clamp determines how the board is held. A flat top clamp that bears on the centre of the board will bow a thin board at the corners; a clamp that bears on the edges of the board will keep the board flat but can interfere with tall components on the top side. Buyers should ask the vendor for the top clamp force range, the clamp footprint, and a video of the clamp operating on a board that is the same thickness and the same component density as the buyer's worst product.
Pneumatic top clamps are the most common in mid-range and high-end printers, and they should support a configurable force range — typically 5 to 50 N per clamp — so that the same machine can run a 0.8 mm FR4 board and a 1.6 mm FR4 board without manual adjustment. Servo-driven clamps are more precise but more expensive, and they tend to appear on high-end fully automatic printers where cycle time is the priority. Buyers should specify which class of clamp they need based on the product mix, not based on what the vendor's sales engineer recommends.
The bottom-side support is what keeps the board from flexing downwards when the squeegee passes over it, and the choice between magnetic support, pin support and vacuum support is one of the most consequential decisions in the buying checklist. Magnetic supports are fast to set up but cannot be used with ferrous or thick-copper boards. Pin supports are flexible and can be reconfigured for almost any panel, but they take longer to set up and require the operator to count and place every pin. Vacuum supports are the most repeatable but require a sealed support plate and add cycle time.
The buying checklist should ask which support systems the printer supports, how many pins the magnetic block can hold, and whether the support layout can be saved as part of the recipe. A printer that requires the operator to re-set the pin layout manually between products is a printer that will eat 5 to 10 minutes of changeover time per product, and that overhead is often invisible in the brochure cycle time number. The recipe editor should be able to recall the support layout, the top clamp force and the bottom support height from a single product name.
For double-sided boards, panelized arrays with break-away rails, and any application where the board needs to be presented to the printer in a custom carrier, the printer has to accept the carrier without modification. The carrier sits on the conveyor, the board sits on the carrier, and the printer's clamp and support system has to deal with the carrier height, the carrier weight and the carrier fiducials as if the carrier were part of the board. Buyers should bring their worst carrier to the demo and ask the vendor to print a panel through the carrier — not on a bare board.
The carrier's contribution to the alignment budget is often missed. A heavy carrier adds mass to the moving system and can slow the print cycle; a flexible carrier can flex under the squeegee and produce inconsistent paste deposits. The buying checklist should ask for a documented maximum carrier weight, a maximum carrier deflection under load, and a recommended carrier material. The answers will quickly sort serious vendors from sales offices.
The most common support-related failure mode is the board that passes the alignment vision system and fails the print, because the support is set too high in one corner and the squeegee pushes the board down before the paste transfers. The second most common is the board that prints well on the first 50 boards and then starts producing inconsistent deposits because the support pins are wearing and the operator has not noticed. The third is the carrier that introduces a 0.5 mm parallax error in the fiducial camera and pushes the alignment out of tolerance on the first shift.
All three failure modes are caught by a single, simple test: print the same panel 50 times in a row on the vendor demo, and measure the deposit height and the alignment offset on every board. A vendor that refuses to run a 50-board repeatability test, or that runs the test and refuses to share the data, is a vendor that does not want the buyer's engineering team to see the actual spread of the machine. The buying checklist should refuse the machine before it refuses the test.
The squeegee is the only part of the printer that physically touches the solder paste, and the four squeegee parameters — angle, pressure, speed and blade material — together determine whether the paste transfers cleanly to the board or smears, skips or skids across the apertures. The buying checklist should treat the squeegee as a sub-system, not as a single number, and the questions in this section are designed to make the vendor prove the sub-system works on the buyer's paste, not on a generic Type 4 reference paste.
The reason the squeegee deserves its own section is that it is the sub-system most sensitive to consumable variation. Solder paste batches vary in viscosity from lot to lot, blades wear at different rates depending on the abrasive content of the paste, and the enclosure temperature shifts the effective viscosity through the shift. A printer whose squeegee sub-system can absorb those variations without operator intervention will run the same recipe all week; a printer whose squeegee sub-system needs constant re-tuning will drift on the night shift and produce the defects that the morning shift then has to rework.
The squeegee angle sets the geometry of the paste roll in front of the blade, and the angle that works best depends on the paste viscosity, the aperture aspect ratio and the board finish. A 60° blade is the most common and is the right starting point for fine-pitch QFP and QFN. A 45° blade produces a stiffer paste roll and is favoured for high-viscosity pastes and for boards with a rough surface finish. A 75° blade produces a thinner paste roll and is used for very low-viscosity pastes and for very small apertures where the blade has to clear the aperture walls cleanly.
Buyers should ask the vendor which squeegee angles the printer supports, how the angle is changed (mechanical re-set, motorised adjustment, or a tool-less quick change), and whether the angle is recorded in the recipe. A printer that requires a wrench and a calibration fixture to change the squeegee angle is a printer that will run at 60° forever, because the operator will not change it. A printer that supports motorised angle adjustment and records the angle in the recipe is a printer that will run at 60° for product A and 45° for product B without the operator thinking about it.
Squeegee pressure is the parameter that drifts the most during a shift, because the paste on the blade changes shape as it transfers, the blade wears, the temperature in the printer enclosure changes, and the viscosity of the paste changes as it ages in the stencil aperture. A printer with closed-loop pressure control measures the actual force on the squeegee and adjusts the pneumatic or servo pressure to keep it constant; a printer without closed-loop control relies on the operator to re-set the pressure every 50 boards or every shift, and most operators do not.
Closed-loop pressure control adds cost, but the cost is small compared with the cost of a single paste-related defect batch. The buying checklist should ask whether the printer supports closed-loop pressure control, what the pressure range is (typically 1 to 30 N for a 200 mm blade, scaled to the actual blade length the buyer intends to use), and what the documented pressure drift is over a 12-hour shift. A pressure drift above ±10% over a 12-hour shift is a sign that the closed-loop system is not closed; a pressure drift below ±3% is a sign that the system is properly designed.
Print speed is the horizontal speed of the squeegee across the stencil; separation speed is the vertical speed at which the stencil peels off the board after the print stroke. The two are independent, and the right combination depends on the aperture size, the paste type and the board finish. A fast print speed (150 to 200 mm/s) is fine for large apertures on a smooth surface; a slow print speed (20 to 50 mm/s) is required for small apertures on a rough surface, because the paste needs time to wet the aperture walls.
Separation speed is the parameter that is most often set wrong, and it is the one that drives the small-pitch defects that show up three stations later. A separation speed that is too fast causes the paste to drag on the aperture walls and leave a "tail" of paste on the stencil, which then smears onto the next board. A separation speed that is too slow does not cause a defect, but it does add cycle time. The buying checklist should ask for the separation speed range (typically 0.5 to 3 mm/s for fine-pitch), the resolution (typically 0.1 mm/s), and the documented impact of separation speed on transfer efficiency for the buyer's aperture size.
Polyurethane blades are the most common and the cheapest. They work well for most applications and they are forgiving of small variations in board flatness and stencil tension. Metal blades (typically stainless steel) produce a cleaner aperture definition on small apertures and are required for some ultra-fine-pitch applications, but they are unforgiving of board flatness and they wear the stencil faster. Composite blades combine a metal edge with a polyurethane body and try to get the best of both, but they are more expensive and not all printers support them.
Buyers should ask the vendor which blade materials the printer supports, what the recommended blade change interval is for the buyer's application, and what the cost of consumable blades is. A printer that supports only one blade material is a printer that will lock the buyer into a single consumable supply chain; a printer that supports multiple materials is a printer that will let the buyer optimise the consumable cost against the product mix. The buying checklist should treat blade material compatibility as a procurement question, not a technical question.
Squeegee parameter | Typical range | What it controls | Checklist threshold |
|---|---|---|---|
Blade angle | 45° / 60° / 75° | Paste roll stiffness and aperture clearance | At least 2 angles, motorised or tool-less change |
Blade pressure | 1 – 30 N per 200 mm blade | Paste fill depth and stencil wear | Closed-loop, ≤ ±3% drift over 12 h |
Print speed | 20 – 200 mm/s | Paste wetting time on aperture walls | Recipe-controlled, 1 mm/s resolution |
Separation speed | 0.5 – 3 mm/s | Paste release and tail formation | 0.1 mm/s resolution, recipe-controlled |
Blade length | 200 – 400 mm | Coverage of the largest panel | ≥ largest panel width + 20 mm |
Blade material | Polyurethane / metal / composite | Aperture definition and stencil wear | ≥ 2 materials supported |
The stencil is the consumable that defines the volume of paste that lands on every pad, and the four stencil parameters — thickness, aperture geometry, frame size and foil tension — together determine the transfer efficiency and the consistency of the deposit. A buying checklist that treats the stencil as an afterthought is a checklist that will produce a machine that prints well on a 0.15 mm stencil and badly on a 0.10 mm stencil, and the line will have to buy a second machine when the product mix changes.
The stencil section is where the buyer's product roadmap matters most. A plant that intends to introduce 0.4 mm pitch QFN within two years needs a printer that can hold a 0.10 mm stencil under tension, and a machine that was specified for 0.15 mm stencils will not be able to follow the roadmap without a stencil-frame upgrade. Reading the roadmap into the stencil questions during procurement is cheaper than discovering the constraint during a new-product introduction three quarters later.
Stencil thickness is the single number that has the largest impact on the volume of paste deposited, and the aspect ratio (aperture width to stencil thickness) is the single number that has the largest impact on transfer efficiency. For a 0.4 mm pitch QFN with 0.25 mm apertures, a 0.10 mm stencil gives an aspect ratio of 2.5, which is at the edge of what most pastes can release cleanly; a 0.12 mm stencil gives 2.08, which is comfortably within range; a 0.15 mm stencil gives 1.67, which is fine for the QFN but produces too much paste on the surrounding 0.5 mm QFP pads.
Stepped stencils solve the conflict by using a thicker foil in some areas and a thinner foil in others, but they cost more and take longer to manufacture. Buyers should ask the vendor which stencil thicknesses the printer supports (most modern printers support 0.10 to 0.20 mm without modification), whether the stencil frame clamping system is compatible with the foil tension required for thin stencils (a thin foil that is not held under enough tension will buckle during the print stroke), and what the recommended stencil supplier is for the buyer's local market.
Aperture geometry is the second stencil parameter, and it is the one that separates a 90% transfer efficiency from a 95% transfer efficiency on small apertures. A trapezoidal aperture (wider at the bottom than at the top) releases paste more cleanly than a straight-walled aperture, because the paste has a tapered surface to release from. A nano-coating on the aperture walls reduces the adhesion between the paste and the foil, which improves transfer efficiency on small apertures and reduces the frequency of under-stencil cleaning.
Buyers should ask the vendor what the recommended aperture manufacturing process is (chemical etching, laser cutting, or electroforming), whether the printer supports nano-coated stencils without modification, and what the documented transfer efficiency is for the buyer's aperture size and paste type. A vendor that quotes a single transfer efficiency number without specifying the aperture size and the paste type is selling the brochure, not the machine.
The practical consequence for the buyer is that transfer efficiency is not a single number on the data sheet; it is a function of the aperture area ratio (the ratio of the aperture opening area to the aperture wall area). Below an area ratio of about 0.66, most pastes will not release cleanly no matter what the printer does, and the fix is a stencil design change rather than a printer change. The buying checklist should therefore ask the vendor to compute the area ratio for the buyer's smallest aperture on the buyer's intended stencil thickness, and to state whether that ratio is inside the paste supplier's recommended window.
Stencil frame size is the third stencil parameter, and it is the one that is most often under-specified in the buying stage. The industry standard frame sizes are 23 × 23 in (584 × 584 mm) for small printers, 29 × 29 in (737 × 737 mm) for mid-range printers, and custom sizes for long-board printers. The printer's frame holder has to be compatible with the frame size the buyer's stencil supplier ships, and the alignment system has to be able to clamp the frame without distorting the foil.
For long-board printers, the frame size is a real constraint: a 1 500 mm long panel typically needs a custom-cut stencil that is wider than the standard 29 in frame, and the printer's frame holder has to accept the oversize frame without manual modification. Buyers should ask the vendor what the maximum frame size is, whether the printer can be ordered with a larger frame holder, and what the lead time is for the oversize frame. The answers will determine whether the buyer's local stencil supplier can support the line, or whether the line will be locked into a single regional supplier.
Foil tension is the fourth stencil parameter, and it is the one that determines whether the stencil prints consistently over its working life. A stencil that is mounted under too little tension will sag during the print stroke and produce inconsistent deposits; a stencil that is mounted under too much tension will stretch and crack around the apertures after a few hundred print cycles. The recommended foil tension depends on the foil thickness, the mesh count of the carrier mesh, and the aperture density on the foil.
Buyers should ask the vendor for the recommended foil tension range, the maximum aperture density the printer's stencil frame can support, and the expected stencil life (in print cycles) for the buyer's typical aperture density. A stencil life of 30 000 cycles on a fine-pitch QFN panel is normal; a stencil life of 10 000 cycles is a sign that the foil tension is too low or the mesh count is wrong. The buying checklist should treat stencil life as a recurring-cost question, because the cost of stencils over the life of the machine is comparable to the cost of the machine itself.
The program is what turns a stencil printer into a specific machine on a specific line. Every board on the line has a recipe, and every recipe contains a dozen parameters — print coordinates, squeegee direction, stencil frame offset, fiducial layout, support pin layout, top clamp force, print speed, separation speed, stencil cleaning cycle. A buying checklist that does not ask how those parameters are stored, edited and recalled is a checklist that will produce a machine that is technically capable but operationally difficult.
Program questions matter more as the product mix grows. A line that runs five products can manage recipes on a spreadsheet; a line that runs two hundred products cannot, and the printer's recipe library, CAD import and backup strategy become operational infrastructure rather than nice-to-have features. The three subsections below address the three layers of that infrastructure: the size and structure of the library, the automation that creates a new recipe from CAD data, and the version control that protects the library from an accidental edit.
The first program question is how many recipes the printer can store. A small line with 20 active products can live with a library of 100 recipes; a large line with 200 active products needs a library of at least 1 000 recipes, with headroom for the 50 new products the line will introduce in the next year. A printer that ships with a 50-recipe library and an upgrade path to 200 is a printer that will run out of headroom in the first quarter; a printer that ships with a 1 000-recipe library is a printer that will not.
The second program question is the structure of the recipe. A well-structured recipe separates the line-level parameters (squeegee pressure range, print speed range, top clamp force range) from the product-level parameters (board dimensions, fiducial coordinates, support pin layout), and it lets the operator edit the product-level parameters without touching the line-level ones. A poorly-structured recipe puts everything in a single flat list, and every product change risks accidentally changing a line-level parameter. The buying checklist should ask the vendor to show a real recipe on a real HMI, and to demonstrate how a product engineer would add a new product without breaking the existing ones.
The second program question is whether the printer can import CAD data directly into the recipe, and how much of the recipe is generated automatically from the import. A modern printer imports Gerber or ODB++ data, extracts the board outline, the fiducial coordinates and the aperture positions, and generates a draft recipe that the operator can fine-tune. A printer that requires the operator to enter the board outline and the fiducial coordinates by hand is a printer that will eat 30 to 60 minutes of engineering time per new product, and that overhead is rarely visible in the brochure cycle time number.
Buyers should ask the vendor which CAD formats the printer supports, how much of the recipe is generated automatically from the import, and whether the import log is auditable. An auditable import log is essential for lines that have to demonstrate traceability to a customer or to a regulatory body, and the log should record the source file name, the import date, the operator and the changes that were made to the draft recipe before it was saved. A printer that overwrites the source data on import is a printer that breaks traceability.
The third program question is how the recipe library is backed up, how versions are tracked, and how the audit trail is stored. A printer that stores recipes only on the local hard drive is a printer that will lose all of its recipes the day the hard drive fails. A printer that supports automatic backup to a network share, to a USB drive, or to a factory database is a printer that will survive a hard drive failure. The buying checklist should ask the vendor what the recommended backup strategy is, how the backup is verified, and how long a restore takes.
Version control is the second part of this question. A line that has been running for three years will have hundreds of recipes, and a recipe that was working last month may have been changed accidentally this month. The printer should support recipe versioning, with a clear way to roll back to a previous version, and the version history should be auditable. A printer that does not support recipe versioning is a printer that will produce a yield drift that no one can diagnose, because no one can prove which version of the recipe was running when the drift started.
Documentation is the evaluation area that buyers most often skip because the vendor "always provides documentation", and the area that causes the most pain in the first year of operation because the documentation that was provided is incomplete, out of date, or in a language the line does not read. A buying checklist that asks for the documentation list up front turns a year of frustration into a single negotiation that happens before the purchase order is signed.
Documentation is unusual among the nine areas because it costs the vendor almost nothing to provide and costs the buyer a great deal to be without. A complete operator manual, a complete service manual and a complete parts list are three files that the vendor already has; the only question is whether the buyer asks for them before the purchase order or after the machine has been installed. The three subsections below cover the operator and service documents, the mechanical and electrical drawings, and the certifications and regulatory paperwork.
The first documentation question is whether the vendor provides a full operator manual, a full service manual and a full parts list with part numbers and current prices. The operator manual should cover every operator-level task — recipe creation, recipe recall, stencil change, squeegee change, support pin change, top clamp adjustment, paste loading, daily cleaning, weekly maintenance. The service manual should cover every service-level task — pneumatic system adjustment, vision system calibration, conveyor alignment, PCB-side compensation calibration, software update.
Buyers should ask the vendor for the manual index, the total page count, the language(s) the manual is available in, and the format (PDF, printed, or both). A vendor that provides the manuals only in English and only in PDF is a vendor that does not sell to lines where the operators do not read English; a vendor that provides printed manuals in the operator's language is a vendor that has thought about the day-to-day reality of the line. The parts list should be searchable by part number and by symptom, and it should be updated every time the vendor issues a parts revision.
The second documentation question is whether the vendor provides the mechanical and electrical drawings of the machine. Mechanical drawings are needed for floor planning, for utility connection planning, and for any future line integration. Electrical drawings are needed for the line electrician who has to connect the printer to the factory power and the factory network, and for any future upgrade that has to interface with the printer's I/O. Buyers should ask for the drawings in their native CAD format (DWG, DXF, EPLAN) and for the I/O list with the voltage and current rating of every signal.
A vendor that provides only a PDF of the mechanical outline and a one-line diagram of the electrical connections is a vendor that has not done the engineering work to support the line. A vendor that provides a full mechanical model, a full electrical schematic and a full I/O list is a vendor that the line's engineering team can integrate with confidence. The buying checklist should treat the documentation package as part of the machine, not as a freebie that comes with the machine.
The third documentation question is which certifications the machine carries, and which regulatory paperwork the vendor provides. For a line in the European Union, CE marking is mandatory; for a line in North America, UL or CSA certification is often required by the customer's quality team; for a line that ships product into the automotive supply chain, IATF 16949 compliance of the vendor's quality management system is often required. The buying checklist should ask for the certificate copies up front, with the certificate number and the issuing body, and should verify that the certificate covers the specific model being purchased.
Buyers should also ask the vendor for a Declaration of Conformity, a REACH compliance statement, a RoHS compliance statement and a conflict-minerals statement. These documents are not always required by the buyer's local regulator, but they are often required by the buyer's customer, and producing them on demand is a sign of a mature vendor. A vendor that does not have these documents ready is a vendor that will not have them ready when the buyer's customer asks for them either.
Training is the evaluation area that decides whether the first 90 days of the new printer are productive or painful. The first 90 days are when the operators are learning the machine, the recipes are being tuned, the line balance is being adjusted, and the first customer audit is likely to happen. A buying checklist that does not ask what training the vendor provides, who provides it, and how long it lasts is a checklist that will produce a machine that arrives with a 200-page manual and a 4-hour training session, and the line will spend the next six months learning what was not covered.
Training is also the area where the vendor's commitment is easiest to test before the purchase. Ask the vendor to name the engineer who will deliver the commissioning, the training centre where the advanced course is delivered, and the last three customers who completed the advanced course in the buyer's region. A vendor that cannot name all three is a vendor whose training promise is a line in the quotation rather than a service the vendor has actually built.
The first training question is what the vendor's installation engineer does during commissioning, and how long the commissioning takes. A serious vendor sends an installation engineer to the buyer's site for 3 to 5 days, and the engineer is responsible for unpacking, positioning, levelling, connecting utilities, calibrating the vision system, calibrating the print parameters, training the first operator, and signing a commissioning report. A vendor that ships the machine in a crate and tells the buyer's line team to follow the manual is a vendor that is not investing in the success of the line.
Buyers should ask the vendor for the commissioning checklist, the typical commissioning duration for a machine of the buyer's class, and what the buyer's team is expected to prepare before the engineer arrives (power, compressed air, network, exhaust). The buying checklist should also ask whether the commissioning report is signed by both the vendor's engineer and the buyer's process engineer, and whether the report includes the measured alignment accuracy and the measured cycle time on the buyer's first product.
The second training question is how the operators are trained, and how the training is sustained across shifts. A serious vendor provides a 2-day operator training course that covers every operator-level task, and the course is delivered on the buyer's site with the buyer's machine. The course should include hands-on practice on the actual machine, not on a simulator, and the operators who complete the course should be issued a certificate that names the operator, the date and the machine serial number.
Shift handover is the operational challenge that training has to address. A line that runs three shifts has three different operator teams, and the recipe knowledge that exists at 7am is not the same as the recipe knowledge that exists at 11pm. The buying checklist should ask the vendor how the training is structured to handle the shift handover — ideally with a written handover document, a shared recipe library, and a designated lead operator on each shift. A vendor that delivers the training once and walks away is a vendor that will leave the line to discover the shift handover problem on its own.
The third training question is what training the vendor provides for the buyer's process engineers, who are the people who will tune the recipes, integrate the printer with the rest of the line, and diagnose the first yield problem. A serious vendor provides a 3 to 5 day advanced training course that covers the vision system, the squeegee control, the stencil and paste interaction, the recipe editor, the CAD import, the SPI integration, and the diagnostic logs. The course is typically delivered at the vendor's training centre, not on the buyer's site, and it includes hands-on practice on a development machine.
Buyers should ask the vendor for the advanced training syllabus, the maximum class size, the prerequisite knowledge, and the language of instruction. A vendor that delivers the advanced training only in English and only at the vendor's headquarters in another country is a vendor that is pricing the buyer's process engineers out of the training budget. A vendor that delivers the advanced training in the operator's language and on a regional schedule is a vendor that is investing in the long-term capability of the buyer's team.
Service is the evaluation area that decides whether a printer that is technically capable of running the line will actually keep the line running for the next ten years. A stencil printer is expected to operate for ten years or more, and the difference between a one-day repair and a one-week line stop is almost entirely a function of the vendor's local service network. The buying checklist should treat service as a first-class procurement question, not as a footnote that gets handled after the machine is installed.
Service is the only one of the nine areas whose cost is not fixed at the purchase order. The machine price is negotiated once; the service cost accumulates over the life of the machine, and a vendor with a weak local service network will recover the discount offered at the purchase stage through parts pricing, response-time fees and emergency visits. The three subsections below cover the response-time commitment, the spare parts inventory, and the warranty and end-of-life policy that together define the total cost of ownership.
The first service question is what the vendor's documented response time is for a typical service call, and what the on-site coverage looks like in the buyer's region. A vendor that has a service engineer in the buyer's city can offer a 4-hour response time; a vendor that has to fly an engineer in from another country can offer a 48-hour response time at best. The buying checklist should ask the vendor for the response time in writing, with a service-level agreement that names the time-to-arrival, the time-to-repair and the time-to-replacement-parts targets.
Buyers should also ask the vendor whether the service engineers are employed by the vendor or by a third party. A vendor-employed engineer carries the vendor's training, the vendor's spare parts and the vendor's accountability; a third-party engineer carries none of these. A vendor that outsources service to a third party should be asked to provide the third party's credentials, the third party's spare parts inventory and the third party's escalation path back to the vendor's engineering team.
The second service question is what the vendor's local spare parts inventory looks like, and what the recommended spare parts kit is for a machine of the buyer's class. A well-stocked local inventory includes the high-wear items (squeegee blades, stencil frame clamps, conveyor belts, vision lights) and the high-impact items (vision cameras, servo drives, pneumatic cylinders, control board spares). A vendor that ships the machine without a recommended spare parts kit is a vendor that has not thought about the first failure.
Buyers should ask the vendor for the recommended spare parts kit, the part numbers, the unit prices, the local stocking policy and the lead time for items that are not stocked locally. The total cost of the spare parts kit is typically 5 to 10% of the machine price, and that cost is recovered many times over the first time a critical part fails. The buying checklist should treat the spare parts kit as part of the purchase order, not as an optional add-on.
The third service question is what the warranty covers, how long it lasts, and what the process is for software updates and end-of-life planning. A typical printer warranty is 12 months from commissioning or 18 months from shipment, whichever comes first, and it covers parts and labour for manufacturing defects. The warranty should be documented in writing, with a clear escalation path for disputed claims, and the buyer should ask the vendor for the warranty exclusions — typically consumables, misuse and unauthorised modifications.
Software updates are the second part of the service question. The printer's control software will receive updates over the life of the machine, and the buyer needs to know whether the updates are free, what the update policy is for major version changes, and how the updates are delivered (on-site by an engineer, remote download, or USB stick). End-of-life planning is the third part: a printer that the vendor plans to support for ten years is a different procurement from a printer that the vendor plans to discontinue in three years. The buying checklist should ask the vendor for the documented product life cycle and the end-of-life support policy.
The nine-area checklist in sections 2 to 10 is generic — it applies to any vendor's printer. This section maps the checklist to the I.C.T product range, so that a buyer who has shortlisted I.C.T can pick the configuration that matches their plant profile. The mapping is not a substitute for the checklist itself; it is a sanity check that the numbers on the I.C.T data sheet line up with the numbers the buyer's engineering team has collected.
The range is organised around five reference configurations, from the compact fully automatic machine through two long-board classes to two semi-automatic classes. Each configuration is described in terms of the checklist areas it addresses, not just in terms of the headline specification, so that a buyer reading this section can see immediately which of the 28 questions each machine answers and which questions will need a configuration option. The mapping also names the situations where a configuration is the wrong answer, because knowing which machine not to buy is as useful as knowing which one to buy.
The I.C.T-6534 is a fully automatic stencil printer designed for the 400 × 400 mm working area class, which covers most consumer electronics, IoT, instrumentation and industrial control boards. It supports up to four fiducials, closed-loop squeegee pressure control, motorised squeegee angle adjustment, and inline SPI integration via SECS/GEM or OPC-UA. The typical cycle time on a 200 × 200 mm board with a single print stroke is 8 to 10 seconds, and the typical alignment repeatability on a fine-pitch QFN test board is ±15 µm at 6σ. For a deeper look at how the 6534 fits the mid-range line profile, the I.C.T-6534 product page has the full data sheet, the supported options and the recommended applications.
For buyers who have already worked through the SMT stencil printer selection guide, the 6534 is the default answer for the 400 × 400 mm working area class. It is also the default answer for buyers who have already decided between manual, semi-automatic and fully automatic using the automatic versus semi-automatic printer comparison and have concluded that a fully automatic machine is the right answer for their line. The 6534 sits in the middle of the I.C.T range, between the long-board 1200/1500 series and the semi-automatic P-series, and it is the workhorse of the I.C.T installed base.
The I.C.T-1200 is a long-board class printer with a 1 200 × 300 mm working area, designed for LED panel lighting, home appliance controllers, and any application where the panel length exceeds 600 mm. It supports the same vision and squeegee systems as the 6534, but the conveyor, the stencil frame holder and the support system are sized for the long-panel application. The 1200 can also be ordered with a custom stencil frame size for panels that exceed the standard 29 in frame, and the lead time for the custom frame is typically 4 to 6 weeks. The product page is at I.C.T 1.2 m SMD stencil solder printer machine.
Buyers who are evaluating the 1200 against a compact-class machine should be careful about the working-area question in section 2: a 1 200 mm panel cannot be run on a 400 × 400 mm machine, and forcing it through a custom jig is a workaround that costs more than the saving of buying the smaller machine. The 1200 is also the right answer for lines that are planning to migrate from compact boards to LED panels within the next 18 months; buying the 1200 once and migrating the product mix later is cheaper than buying a compact machine and replacing it within two years.
The I.C.T-1500 extends the long-board concept to 1 500 mm panels, which covers the upper end of the LED panel lighting market and the larger home appliance and industrial control boards. The 1500 shares most of the 1200's sub-systems, but the conveyor, the frame holder and the support system are sized for the 1 500 mm application. Buyers should confirm the exact panel length and panel weight with the I.C.T engineering team before ordering, because the 1500's support system is optimised for panels in the 1 200 to 1 500 mm range, and panels outside that range may require a custom carrier. The product page is at I.C.T 1500 long-board fully automatic LED PCB stencil printer.
The 1500 is a specialist machine and the buying decision is more about the application than the checklist. Buyers who are evaluating the 1500 should treat the working-area question in section 2 as a hard constraint, and they should run a sample panel through the 1500 at the I.C.T demo lab before signing the purchase order. A 1 500 mm panel that prints well on the demo machine will print well on the buyer's line, because the long-panel application is more sensitive to support geometry and stencil tension than the compact-board application.
For lines that have already concluded that a fully automatic machine is the wrong answer — usually because the volume does not justify the capital, or because the line runs prototyping and NPI work that does not benefit from automation — the I.C.T P-series offers two semi-automatic options. The P6 is the entry-level machine in the range, with a 400 × 400 mm working area, manual board loading, and a single squeegee direction. The P12 is the precision-grade machine, with a higher-spec vision system, a more precise squeegee control, and a heavier frame that produces a more stable print stroke. The P6 product page is at I.C.T P6 semi-auto SMD solder paste printing machine and the P12 is at I.C.T P12 high-precision semi-automatic SMT screen stencil printer.
Buyers who are choosing between the P6 and the P12 should focus on two questions: what is the smallest aperture on the buyer's product, and what is the alignment repeatability the buyer needs on the worst product. A buyer running 0.5 mm pitch QFP only can use the P6 comfortably; a buyer running 0.4 mm pitch QFN should pay the extra for the P12's higher-spec vision system. The decision is rarely about the printer and almost always about the product mix, which is why the checklist in sections 2 to 10 has to be completed before the P6 vs P12 decision is made.
The nine areas in sections 2 to 10 produce 28 measurable questions. The table below collapses all of them into a single page that can be taken into the meeting with the vendor. A "pass" is a value that the buyer's engineering team has measured on the buyer's product, not a value that the vendor's sales engineer has quoted on a demo board. A "fail" is a value that the vendor cannot meet on the buyer's product, regardless of what the brochure says.
The table is deliberately organised in the same order as the sections, so that a buyer who wants to dig into a particular row can jump back to the corresponding section and read the full discussion. The threshold column carries the value that the buyer's engineering team has decided to require, and it should be filled in before the vendor meeting rather than during it, because a threshold that is negotiated in the meeting is a threshold that the vendor has already influenced. The final three paragraphs of this section describe how to use the table, how to write the evaluation report, and what the table cannot catch.
Area | Question to ask | Threshold | Pass / Fail |
|---|---|---|---|
Working area | Largest panelized array the machine can run | ≥ largest PCB + 80 mm per side | |
Working area | Conveyor width | ≥ panel width + 30 mm | |
Working area | Maximum stencil frame size | ≥ largest panel + 100 mm | |
Alignment | Vision repeatability on a fine-pitch QFN test board | ≤ ±20 µm at 6σ | |
Alignment | Maximum number of fiducials supported per recipe | ≥ 4 | |
Alignment | PCB-side skew compensation range | ≥ ±2° | |
Alignment | Closed-loop SPI integration (SECS/GEM or OPC-UA) | Available | |
Support | Top clamp force range (per clamp) | 5 – 50 N, configurable | |
Support | Magnetic support, pin support, or vacuum support | ≥ 2 of 3 | |
Support | Support layout saved in the recipe | Yes | |
Support | Maximum carrier weight | Documented value | |
Squeegee | Supported squeegee angles | 45°, 60°, 75° | |
Squeegee | Closed-loop pressure control | Yes, ±3% drift over 12 h | |
Squeegee | Print speed range | 20 – 200 mm/s | |
Squeegee | Separation speed range | 0.5 – 3 mm/s, 0.1 mm/s resolution | |
Stencil | Supported stencil thickness | 0.10 – 0.20 mm | |
Stencil | Recommended aperture manufacturing process | Laser cut or electroformed | |
Stencil | Maximum frame size | ≥ 29 in or custom with lead time | |
Stencil | Expected stencil life on fine-pitch QFN panel | ≥ 20 000 print cycles | |
Program | Recipe library size | ≥ 1 000 recipes | |
Program | CAD import format | Gerber or ODB++ | |
Program | Recipe versioning and audit trail | Yes, with rollback | |
Documentation | Operator manual language and format | Operator's language, printed or PDF | |
Documentation | Service manual and parts list with prices | Yes, with part numbers | |
Documentation | CE / UL / IATF certifications | Certificate copies provided | |
Training | On-site commissioning duration | 3 – 5 days | |
Service | Documented response time SLA | ≤ 24 h on-site, in writing | |
Service | Spare parts kit recommended for buyer's class | Provided with part numbers and prices |
Bring the table to the vendor meeting with the 28 thresholds already filled in from the buyer's own measurements. Walk through each question in order, and ask the vendor to answer with a number, a document, or a demo. A vendor that answers every question with a confident "yes" is a vendor that is selling the brochure; a vendor that pauses on a question and says "let me check with engineering" is a vendor that has engineering. Both responses are useful, but only the second is the basis for a serious procurement.
After the meeting, the completed SMT solder paste printer buying checklist is the document that the buyer's engineering team uses to write the final evaluation report. The report should include the table, the vendor's answers, the buyer's own measurements, and a recommendation that names the machine, the configuration, the price, the delivery time, the commissioning duration and the spare parts kit. A report that does not name all six is a report that the buyer's procurement team cannot act on, and the buying process will have to start again.
The checklist catches the things the buyer's engineering team can measure. It does not catch the three things that only emerge after the machine is installed and running. The first is the operator's subjective experience: a machine that the operators find awkward to use will produce more operator errors than a machine that the operators find comfortable, and the checklist cannot measure operator comfort. The second is the line balance: a machine that is technically excellent but does not match the takt time of the surrounding equipment will become a bottleneck, and the checklist cannot measure line balance without a line simulation. The third is the long-term relationship with the vendor: a vendor that is responsive in the first year and unresponsive in the third year is worse than a vendor that is uniformly responsive, and the checklist cannot predict the long-term relationship.
For all three, the mitigation is the same: visit an existing customer of the vendor, run a sample product on the vendor's demo machine, and ask the customer for the unfiltered answer to the question "what do you wish you had known before you bought this machine?" The answers are rarely what the vendor expects, and they are the most valuable input the buying team will collect.
Talk to the I.C.T engineering team
Bring the 28-question checklist to a 30-minute call with the I.C.T engineering team, and we will walk through every question on your list, with numbers measured on your product, not on a brochure. The call is free, and the output is the configuration that matches your plant profile.
The most important number is the largest panelized array the printer can run, not the largest bare board. Real production rarely loads a single bare PCB; it loads a panelized array with break-away rails or a board mounted on a carrier, and both formats add 20 to 60 mm of border per side. A machine that meets the bare-board size but fails the panelized size will produce a line that has to redesign its panels, and that is a costly workaround. The second most important number is the alignment repeatability on a fine-pitch test board, because that number is the one that determines whether the line can print the smallest aperture in the buyer's product mix.
All 28, in the sense that the engineering team should be able to produce a documented answer for every question. In practice, a buyer who can answer the first 12 (working area, alignment, support and squeegee) has a 90% complete picture; the remaining 16 (stencil, program, documentation, training and service) refine the picture and surface the long-term risks. A vendor meeting that covers only the first 12 will produce a working machine; a vendor meeting that covers all 28 will produce a working machine and a service relationship that survives the first three years.
For 0.4 mm pitch QFN and small BGA, the practical repeatability target is ±20 µm or better at 6σ. For 0.3 mm pitch wafer-level or very fine-pitch BGA, the target tightens to roughly ±12.5 to 15 µm. Always confirm these numbers on a fine-pitch test board rather than on the headline six-axis stage accuracy, because the difference between the two is the difference between printing QFNs reliably and not printing them at all. The vendor's demo lab should run a 50-board repeatability test on a fine-pitch QFN panel as part of the evaluation, and the test data should be shared with the buyer's engineering team.
Physically yes — most fully automatic printers can be slowed down, run with manual loading and used for NPI work — but the operating economics are poor. A printer that spends half its life being used as a semi-automatic machine is wasting its capital cost. Most serious lines run a smaller semi-automatic or manual printer for NPI and engineering work, and reserve the fully automatic machine for the high-volume production line. The two machines together usually cost less than the productivity loss of misusing the flagship machine, and they also reduce the risk of a high-volume line stop caused by an NPI experiment.
Ask for the documented response time SLA in writing, with the time-to-arrival, the time-to-repair and the time-to-replacement-parts targets named explicitly. Ask whether the service engineers are vendor-employed or third-party, and ask for the third party's credentials if applicable. Ask for the recommended spare parts kit, with part numbers, unit prices and local stocking policy. Visit an existing customer of the vendor in the buyer's region, and ask the customer for the unfiltered answer to the question "what do you wish you had known about the service network before you bought this machine?" The answers will quickly sort serious vendors from sales offices.
Whether the recipe library is automatically backed up, and how the backup is verified. A printer that stores recipes only on the local hard drive is a printer that will lose all of its recipes the day the hard drive fails, and a recipe loss on a high-mix line is a 3 to 5 day recovery effort. The second most overlooked question is whether the recipe editor supports versioning with rollback, because a recipe that was working last month may have been changed accidentally this month, and the buyer's engineering team needs to be able to roll back to the previous version without rebuilding the recipe from scratch.
For a fully automatic printer in the 400 × 400 mm class, the typical timeline is 8 to 12 weeks from first vendor contact to commissioning on the buyer's site. The breakdown is roughly 2 weeks for the technical evaluation and vendor shortlist, 2 weeks for the detailed commercial negotiation and configuration finalisation, 4 to 6 weeks for the machine to be built and shipped, and 1 to 2 weeks for installation, commissioning and operator training. For a long-board printer (1 200 or 1 500 mm class), the build time can extend to 6 to 10 weeks because the long-board frame and conveyor are often built to order. The buying checklist should be completed in the first 2 weeks; the rest of the timeline is largely outside the buyer's control.