JUKI Pick and Place Machine Buying Checklist for Electronics Manufacturers

Buying checklist · Published by I.C.T · For SMT process engineers, production owners and equipment buyers

A JUKI pick and place machine buying checklist usually gets written after the quotation, which is exactly the wrong order. By then the model number is already in the subject line of six emails, the seller has a picture of the sale, and the questions that remain are the comfortable ones. The checklist that changes the outcome is the one written before the first quote is compared, and it is built around three columns rather than one: what has to be confirmed, what document or measurement proves it, and which sentence in the purchase order makes it enforceable. A model number on its own proves nothing, because two machines with the same badge can differ in head type, feeder count, vision package, software licence and warranty scope.

This guide runs nine checkpoints in the order a purchase actually moves: model and configuration identity, condition and wear, the feeder package, options and software, power and air, transport and installation, acceptance testing, and the service scope that decides what happens in year three. For each checkpoint it names the evidence worth asking for, the arithmetic worth doing before you commit, and the clause worth writing down. Teams that use it usually find that two or three items carry most of the risk, and that the two or three are rarely the items the seller wants to talk about. The platforms referenced throughout are the ones listed on the placement machine range this checklist was written against, and every calculated figure below is labeled as an assumption so it can be replaced with your own numbers.

Why a checklist that lists features fails

The failure mode is familiar to anyone who has sat in a post-installation review. The machine does what the brochure promised and the line still does not run as planned, because nobody wrote down that the quotation included four feeder carts rather than six, that the vision package covered chips but not the formed LED bodies on the BOM, or that the warranty covered parts but not the travel time of the engineer who carries them. None of those are machine faults. They are unconfirmed items, and they were unconfirmed because the checklist measured features rather than obligations. A feature list is easy to write and it flatters both sides: the seller lists what the platform can do, the buyer ticks what he recognizes, and the gap between capability and configuration stays invisible until the machine is on the floor.

The alternative is a checklist built as three columns. The first column is the item to confirm, stated as a question rather than a noun. The second is the evidence that answers it — a document, a photograph, a measurement taken in front of you. The third is the clause, the sentence that goes into the purchase order so the answer survives the salesperson who gave it. That last column is what separates a checklist from a shopping list, and it is the column most buyers never fill in. A checklist that changes the commercial outcome is short, evidence-driven and slightly uncomfortable to assemble, which is why it is worth building once and reusing for every machine you buy afterwards.

A feature list answers the wrong question

A feature list asks what the machine can do. A buying decision asks what this specific machine, in this specific configuration, will do on your boards in the third shift of the fourth month. Between those two questions sit the items that cost money: the head type fitted to the gantry, the number of feeder positions wired and enabled rather than merely available, the nozzle set that ships in the crate, the software options enabled by licence key rather than installed but dormant, and the calibration state at the moment of handover. Two RS-1R machines from the same production year can differ on all five without the badge changing, and a quotation that names only the model leaves all five open.

The practical correction is to require the seller to quote a configuration, not a platform. Ask for the machine described in the same terms the factory uses on its build sheet: head model, number of banks, feeder mix by width, nozzle list by part number, enabled software options, and the options fitted at build rather than added later. When a seller cannot produce that description, the honest reading is that the description does not exist yet, and a machine whose configuration is undefined at quotation stage will be defined by whoever assembles the crate. That is not a reason to walk away from a used or brokered machine, but it is a reason to write the configuration into the order and to price the missing items before signing rather than after arrival.

There is a second-order benefit that buyers underestimate. A configuration-level list makes two quotations comparable. Most quote comparisons fail not because prices are hidden but because the two documents describe different things: one includes a nozzle changer and six carts, the other includes neither and is cheaper by an amount that looks like a discount. Rebuilding both quotes into the same nine lines takes an afternoon and usually moves the apparent price gap by more than the negotiation did. It also removes the argument about what was agreed, because the agreed configuration is written in the same words on both sides.

One item on the list is not about the machine at all, and it is the one that most often exposes a gap between the quotation and the plant: what the published rate becomes once the machine is standing in your line. A head rated at a headline figure places fewer components per hour when the feeder mix indexes slowly, when the boards are long, or when the printer in front cannot keep up. The method for converting a published rate into boards per hour on your own line belongs next to the specification sheet during the comparison, because a machine bought on its optimum rate and installed behind a slower printer is a machine that will be blamed for a bottleneck it did not create.

Three columns: confirm, prove, write down

The three-column format is deliberately boring, and that is its value. Column one holds the question in plain language: which head is fitted, how many feeder positions are enabled, what the hour meter reads, which software version is installed, what the site needs in terms of power and air, who pays for the rigging, what test proves the machine is good, and what the service contract actually obliges anyone to do. Column two holds the artifact that answers the question. Where an artifact does not exist, that absence is itself the finding, and it should be recorded as an open item rather than filled with an assumption.

Column three is where the checklist becomes a contract. Each confirmed item needs a sentence in the purchase order that makes the confirmation binding: the machine is supplied with the head model stated, with the feeder inventory listed by width and quantity, with the nozzle set listed by part number, with the software options listed by name, with the acceptance test defined by procedure and pass criteria, and with the service obligations stated in days and scope. Without that sentence the answer is a memory of a conversation, and memories are what disputes are made of. Buyers who write column three consistently report fewer surprises at handover, not because sellers are dishonest but because a written obligation gets routed to the people who can meet it.

The columns also tell you where to spend your inspection time. Items whose evidence is a document can be checked at a desk before you ever travel. Items whose evidence is a photograph need to be specified in advance, because a photograph taken at the seller's convenience answers a different question than the one you asked. Items whose evidence requires witnessing need a visit, and the visit needs an agenda built from the list rather than from whatever the seller wants to demonstrate. Sorting the nine checkpoints by evidence type usually reduces a planned three-day inspection to one focused day plus a document review, which is both cheaper and more revealing.

The evidence ladder: paper, photo, witnessed

Not all evidence weighs the same, and a checklist that does not say so invites the weakest form. Sort every item into three rungs. The first rung is paper: a certificate, a build sheet, a service invoice, a licence document. Paper is cheap to obtain and easy to verify, and for many items it is sufficient, because a calibration certificate with a date and a serial number either exists or does not. The second rung is a photograph or a screen capture taken to a specification you set: the nameplate with the serial legible, the hour meter with the reading visible, the feeder bank with the carts in frame, the software version screen with the build number. The third rung is witnessed: something the buyer or his engineer observes operating, which is the only rung that proves function rather than intent.

The ladder matters because sellers default to the rung that is cheapest for them. A verbal assurance is rung zero. A brochure is rung zero with better typography. When a checkpoint sits on the third rung and the seller keeps answering on the first, that mismatch is the most useful signal in the whole process: it usually means the item has not been checked recently, or that checking it would reveal something. The response is not accusation but specification. Ask for the artifact by name, in the format you need, by a date, and record what arrives. A seller who can produce it moves up the ladder quickly; one who cannot has told you something about the machine's documentation that belongs in the price.

For a machine that is being bought new, most checkpoints can be satisfied at rung two plus the factory acceptance test, because the build is documented and the machine is not yet wearing. For a used machine, rung three becomes unavoidable on condition and accuracy, and the cheapest way to reach it is a witnessed run on your own board rather than a demonstration board. A demonstration board is chosen to flatter the machine: small chips, generous spacing, few feeders. Your board carries the parts that actually cause placements to fail, and running twenty of them tells you more about the head, the vision system and the feeder bank than an hour of catalogue video.

One practical note on keeping the ladder honest. Record the rung next to every answered item, and treat a rung-one answer to a rung-three question as unanswered. This single habit prevents the most common post-purchase argument, which is not about whether something was said but about whether it was verified. It also produces a document you can hand to a colleague who inherits the project, or to an insurer, or to the finance committee that wants to know why the second machine was specified differently from the first. The checklist is not paperwork for its own sake; it is the audit trail of a decision, and an audit trail with rungs in it is worth considerably more than a list of ticks.

Checkpoints 1 and 2: model identity and configuration

Identity is the checkpoint buyers think they have already completed, and it is the one that most often turns out to be half done. A model name identifies a platform. It does not identify the head, the build year, the fitted options, the software level or the serial, and every one of those changes what the machine can run. The reason identity matters beyond bookkeeping is that price, spare parts, nozzles, feeder compatibility and the service contract all key off the exact build. A nozzle that fits one head will not fit another on the same platform, and a service engineer who arrives expecting the standard head will spend the first hour looking for a part he did not bring.

Configuration is the second half of the same checkpoint, and it is where the commercial difference usually hides. Sellers quote what is in the crate, and what is in the crate depends on how the machine was ordered, how it was used and what was stripped before it was sold. A used machine's option list is a record of its previous owner's production plan, not of yours. The task is to compare that list against the options your BOM actually needs, price the gaps, and then decide whether the machine plus the gaps is still the better buy than the next candidate. Doing that arithmetic before signing is the entire purpose of this checkpoint.

Reading the nameplate and the configuration sheet

Ask for two photographs and one document before any site visit. The photographs are the nameplate and the hour meter, both taken close enough that the characters are legible without enhancement, and both showing enough surrounding machine that you can tell they were taken on this machine rather than reused. The document is the configuration sheet: the build list, the option list, or the invoice that accompanied the original sale, whichever the seller can produce. Cross-check the three against each other. The serial on the nameplate should appear on the configuration sheet, the build year should be consistent with the software version, and the options on the sheet should be visible on the machine in the photographs.

Look specifically for the items that get swapped between machines in a dealer's warehouse: heads, feeder carts, nozzle stations, monitors, cameras and the small option boards that enable a function. A head swap is legitimate and common, but it changes the machine's identity and therefore its price and its capability, so it must be documented rather than discovered. Ask directly whether any major assembly has been replaced and request the work order if so. A seller who answers precisely and produces the record is giving you a reason to trust the rest of the file; one who answers vaguely has told you that the configuration sheet is the only reliable document you will get.

Then read the sheet against your own requirement rather than against the platform's reputation. Take your BOM and list what it needs: component sizes at both extremes, the tallest body, the widest tape, the number of distinct part numbers, and the accuracy your product's pitch demands. Compare that list with the head type and the option list on the sheet. This is the stage where the platform choice is either confirmed or reopened, and it is cheaper to reopen it here than after a purchase order. A structured screening sequence for that comparison is set out in the screening order that comes before a model is named, and it is worth running even when the seller has already recommended a specific machine.

Where the five current platforms actually differ

The five platforms most often quoted to electronics manufacturers differ less in headline speed than in what they can physically hold, and the physical limits are the ones that decide whether a machine can run your board at all. The table below uses the published specification values from the product pages on this site rather than rounded marketing figures, and it deliberately leaves a cell blank where the page does not publish a value. A blank in a specification table is a question to put to the seller, not a number to estimate; estimating it is how buyers end up with a machine that cannot place the part they bought it for.

Read the table as a filter rather than a ranking. A board with 74 mm connectors removes the two rotary platforms immediately, whatever their published rate, because the largest component they accept is 25 mm and 5 mm square respectively. A board carrying 25 mm tall transformers removes the RX platforms for the same reason. A product family that mixes 03015 chips with a handful of large bodies is the classic case where a single platform cannot serve both ends, and where the honest answer is either a flexible platform with a slower rate or two platforms with a transfer between them. The RS series question of which build fits which board mix is worked through in the RS series comparison behind the plate-reading step, including the board-length and buffer trade-offs that decide between the three builds.

Two cells deserve a note because they are read carelessly. The KE-3010A feeder figure of 160 positions applies to 8 mm tape with double tape feeders, which is a specific configuration and not a universal maximum; quoting it as a general capability leads to disappointment when wider tape is loaded. The RX-7R component height of 3 to 10.5 mm is head dependent, so the head fitted to the specific machine decides the answer and the nameplate alone does not. Both are examples of why the configuration sheet outranks the model number, and why a specification table is a starting point for questions rather than a substitute for the answers. For small-board, high-count work where the rotary architecture pays, the compact platform is described in the compact rotary platform for small boards, including the 03015 and 76-position limits that come with the speed.

Options that change the machine's identity

Some options are accessories and some change what the machine is. The distinction is worth making explicit in the checklist because buyers group them together and sellers price them together, yet only the second group affects whether the machine can run the product at all. Options of the first kind improve comfort or speed: extra carts, additional monitors, a better light, a second nozzle station. Options of the second kind change capability: a different head, a taller component clearance, a tray or stick feeder for parts that cannot be taped, a vision package that recognizes formed leads or dark bodies, a dual-lane conveyor that changes the board handling envelope, or a bad-mark system the product needs.

For each second-kind option, write three things on the checklist: whether it is fitted, whether it is enabled, and whether it is included in the price. A machine can physically carry a tray feeder that was removed before sale, can have a vision option installed but disabled by licence, and can have an option listed in a quotation as available without being costed. All three conditions look identical in a specification sheet and behave completely differently on the floor. The check is a photograph of the option in place plus a line in the order naming it, and where a licence is involved, a copy of the licence rather than a promise to enable it later.

The commercial consequence is simple and worth stating plainly: an option that is not fitted at the factory is usually more expensive to add later than it would have been to order, because it involves a retrofit kit, travel and a software change rather than a line on a build sheet. That asymmetry belongs in the comparison. A machine quoted at one price without the vision package is not cheaper than a machine quoted higher with it, unless the gap is larger than the cost of the retrofit, and the retrofit cost is a number the seller has and the buyer usually does not. Requesting that number for each missing second-kind option converts the whole quotation into a like-for-like comparison in an afternoon.

Checkpoint 3: condition, hours and wear

Condition is the checkpoint that distinguishes a new purchase from a used one, and it is also the checkpoint where buyers most often accept a proxy. The proxy is the hour meter, the service sticker, or the seller's phrase that the machine is in good working order. None of the three is evidence of condition; each is evidence of something adjacent to it. Condition has to be assembled from several readings, some numeric, some visual, some witnessed, and the checklist works better when each reading is treated as an independent line with its own evidence rather than rolled into one judgement about whether the machine looks good.

The stakes are asymmetric, which is why the checkpoint deserves its own section. A used machine with low hours on a heavy feeder mix can be a better buy than one with high hours on a light mix, because head cycles and feeder indexes wear different parts than the hour meter suggests. Conversely, a machine that has been idle for two years can present low hours alongside dried grease, seized air cylinders and firmware that nobody has updated. The readings below are ordered so that the cheap ones come first: documents and photographs that cost nothing, then measurements that need a short visit, then the witnessed run that decides the purchase.

What the hour meter does and does not tell you

An hour meter counts the time the machine's power has been on, or in some builds the time a program has been running. It is a useful sanity check on the seller's story and a poor measure of wear, because machines accrue different wear per hour depending on what they place. A machine that has spent its life on a stable family with few feeder changes has different wear from one that has run high-mix work with a changeover every shift, even at identical hours. The readings that matter more are placement count where the software exposes it, the number of programs stored, and the date of the last recorded service, all three of which can be captured from the machine's own screens.

Ask for the hour meter reading and the placement counter in the same photograph, then compare with the service log. A machine showing 30,000 hours with a log that stops four years ago is a machine that has been maintained outside the record, which may be perfectly fine but cannot be assumed. A machine showing low hours with a placement count that implies dense production is a machine that ran short, intense campaigns; that is not a defect, but it changes which parts you inspect. The point of the cross-check is not to catch the seller in a lie. It is to decide where the wear most likely sits, so the inspection visit has a target rather than a route.

Two further readings are worth requesting because they are cheap and revealing: the error or alarm history, and the number of nozzle stations configured. The alarm history shows what the machine has been complaining about, which is often the first sign of a vacuum or vision problem that has been worked around rather than fixed. The nozzle station count shows whether the machine has been running a narrow part mix or a broad one, which tells you something about feeder and head wear before you look at either. Neither reading is proof of anything on its own, and together they narrow the inspection to the two or three subsystems that deserve the covers-off treatment.

Six wear points that carry the price

Six items account for most of the difference between a used machine that runs and one that becomes a project, and each can be inspected in minutes once the covers are off. The first is the head itself: nozzle holders, shaft play, and the condition of the vacuum path behind each holder. The second is the nozzle set, which is a consumable and is usually the largest single line of the reconditioning budget. The third is the feeder bank and its connectors, because a bank with damaged slots silently reduces the machine's usable capacity. The fourth is the conveyor and clamp mechanism, where worn belts and misaligned rails cause the intermittent transfer faults that are hardest to diagnose later.

The fifth is the vision system: camera lenses, lighting modules and the calibration state of the recognition routines. A cloudy lens or a dimmed ring light produces recognition failures that look like software problems and are not. The sixth is the vacuum generation and pneumatic circuit, including filters and regulators, which is cheap to restore and expensive to ignore because a marginal vacuum shows up as intermittent pick errors that consume weeks of engineering time. For each of the six, the checklist should carry an accept condition written in advance, so the inspection produces a decision rather than an impression.

Nozzles deserve separate handling because they are both a wear item and a specification item. The machine ships with a nozzle set sized to its intended work, and a used machine often arrives with a partial set, mismatched part numbers, or nozzles that fit a different head. Ask for the nozzle inventory by part number and compare it with the orifice sizes your component range needs; the series and orifice sizes for the heads in this range are listed on the nozzle series that fits this head, and the difference between a complete set and a partial one is a four-figure line that buyers routinely forget to price. A partial set is not a deal-breaker; an unquantified partial set is.

Wear inspection also has a measurement standard behind it, and knowing that standard helps frame the questions. Placement accuracy is characterized under defined conditions rather than asserted, and the industry procedure for that characterization is IPC-9850, which specifies the test board, the pattern, the sample size and the way the figures are calculated. When a seller quotes an accuracy figure, the useful follow-up is not whether it is true but how it was obtained: under the standard procedure, on a demonstration board, or inherited from a brochure. A machine that has been recalibrated to a defined procedure and can show the certificate is in a different category from one whose accuracy is a memory of how it used to run.

What a calibration record should contain

A calibration certificate is only useful if it identifies the machine, the date, the procedure and the result. Four fields, and a certificate missing any of them cannot be used to support a claim. The machine is identified by serial, not by model, because a certificate for the platform proves nothing about the unit in the crate. The date matters because calibration drifts and because a certificate older than the service interval is a statement about the past. The procedure matters because different procedures produce different numbers from the same machine, which is precisely why accuracy figures are not comparable across sources without knowing how they were derived.

The result should include the measured deviations rather than a pass stamp. A certificate that reports a maximum offset, or better a set of offsets across the field, tells you where the machine sits relative to its specification and how much margin remains before it needs attention. A pass stamp tells you only that someone was satisfied on the day. Where the manufacturer's stated accuracy is ±35 µm with a Cpk of at least 1, the certificate should show measurements that support that statement rather than restate it, and the same logic applies to the vision system's recognition results and to the feeder index accuracy if it was tested.

If no certificate exists, the checklist does not collapse; it changes shape. Add a witnessed verification run to the acceptance stage and treat the result as the baseline, with the cost of a full recalibration priced into the offer. A machine without a current certificate is not a bad machine, but it is a machine whose accuracy claim is unverified, and the buyer who prices the verification before signing pays for it once. The buyer who discovers it after installation pays for it twice, because the same recalibration now happens under production pressure, with the line stopped and the schedule already committed.

Checkpoint 4: the feeder package and how to count it

Feeders are where quotations quietly differ, and the reason is that a feeder count can be stated in at least four ways that all sound alike. The maximum number of positions on the bank, the number of positions enabled by the machine's configuration, the number of feeders physically included, and the number of 8 mm-equivalent slots those feeders consume are four different numbers, and a seller is free to quote the largest. Buyers who compare quotations on the word feeders are therefore comparing four different measurements without knowing it, which is a reliable way to buy the wrong package at the right price.

The checkpoint has a second half that matters more than the count: the mix. A bank filled with 8 mm feeders is a bank built for chip-heavy boards, and it will not run a product whose BOM is dominated by 16, 24 and 32 mm tape without somebody removing feeders to make room. The arithmetic that converts a BOM into slots is simple, and doing it before the purchase turns a vague worry about feeder capacity into a number you can compare against the machine's bank. That number, not the headline maximum, is what decides whether one machine can hold your product family or whether you need carts, a second bank, or a second platform.

Why two feeder counts never agree

The first count is the mechanical maximum: how many slots the bank physically provides. The second is the enabled count, which can be lower when the machine's configuration or software limits the number of active positions. The third is the delivered inventory: the feeders that arrive in the crate, which is usually well below the maximum and is often the only number the price actually reflects. The fourth is the equivalent count, which expresses the inventory in 8 mm slots and is the only figure that can be compared with the BOM. A quotation that says 112 feeders is meaningless until you know which of the four it means.

The fix is to ask for an inventory table rather than a number: width, quantity, electric or mechanical, and condition, with a total in 8 mm equivalents. Sellers who deal in machines regularly can produce this in an hour; sellers who cannot are usually assembling the package from whatever is on the shelf, which is useful to know before you commit. Add the physical count during the inspection, because removing a feeder takes seconds and an inventory that does not match the bank is a common and easily corrected discrepancy. Record both numbers on the checklist and treat the difference as an open item.

The enabled count deserves one specific question, because it is invisible in a photograph: is any position unavailable because of a hardware limit, a software limit, or a removed option? On some builds a lane, a bank section or a set of positions is tied to an option rather than to the frame, and a machine bought without that option has empty slots that look usable and are not. This is exactly the kind of item that a model number conceals and a configuration sheet reveals, and it is the reason the identity checkpoint comes first in the sequence rather than last.

Normalizing positions to 8 mm equivalents

Tape feeders occupy bank width in proportion to the tape they carry, and the convention that makes them comparable is the 8 mm equivalent: one 8 mm feeder occupies one slot, a 12 or 16 mm feeder occupies two, and wider feeders occupy proportionally more. The exact multiples depend on the feeder design and on the bank, so the numbers below are a worked example rather than a universal table. The method is what matters, and it takes one pass through the BOM to run.

Take a hypothetical product family with 120 BOM lines and assign tape widths from the component list. In the example below, 70 lines run on 8 mm tape, 28 on 12 or 16 mm, 14 on 24 mm, 6 on 32 mm and 2 on 44 mm. Multiply each quantity by the slot factor, sum, and compare with the enabled positions on the machine. The result is 204 equivalents against 112 positions, which means the family cannot be loaded on one bank at one time, and the decision moves from which machine to how the line is staged: two carts and a scheduled change, or a second platform. Neither answer is wrong, but the arithmetic has to happen before the purchase order, because the staging decision changes the price and the floor plan.

Two refinements make the number more honest. First, add a working reserve for the parts that change between orders: a bank loaded to exactly its capacity has no room for the engineering change that arrives the week after the program is proven, and the usual allowance is ten to fifteen percent of positions held free. Second, separate the parts that are loaded every run from the parts that appear occasionally, because the occasional parts can live on a cart and be swapped in during changeover rather than occupying a permanent slot. The combination of the equivalent total, the reserve and the split between resident and staged parts is a feeder plan, and a feeder plan is what the seller should be asked to quote against.

Where the equivalent total exceeds the bank, the choice between a second cart and a second platform is a real one and should be made with numbers. A cart costs a fraction of a platform and converts changeover time into a scheduled activity; a second platform costs more and converts changeover into parallel production. Which is cheaper depends on how often the family changes and how long the change takes, and that calculation belongs with the changeover time rather than with the machine price. For a product mix that needs help deciding, the accounting method that converts BOM lines into feeder and nozzle requirements is set out in the three-ledger method for counting BOM lines as feeder positions, and it produces the same equivalent total with a documented audit trail behind each line.

Electric and mechanical feeders on the same bank

Most machines in this range accept both electric and mechanical feeders, and the choice is a genuine one rather than a formality. Electric feeders index under their own control, hold their position after a power interruption, and are generally the better choice for small tape and for parts that must not be advanced by a mechanical stroke. Mechanical feeders are driven by the machine's cycle, are simpler and cheaper, and remain common in packages sold with older machines. A bank that mixes the two is normal, and the checklist should record which type each inventory line is rather than treating feeders as a single category.

The mixing matters because the two types behave differently in three places: changeover, maintenance and fault diagnosis. Electric feeders carry their own settings, which means a feeder moved between machines keeps its parameters and a faulty one can be swapped without reprogramming. Mechanical feeders inherit timing from the machine, which makes them sensitive to bank alignment and to wear in the drive mechanism. When a pick error appears on a mixed bank, the first diagnostic step is to know which type is involved, and an inventory without types turns a ten-minute check into an hour of swapping.

For a used package, ask how many of the delivered feeders have been serviced and whether any are known to be intermittent. Feeders are the most repairable part of the whole purchase and also the part most often sold in unknown condition, because a feeder that fails once a shift still passes a bench test. The practical protection is a clause that allows a defined number of feeders to be returned or replaced during the first weeks of production, which costs the seller little and removes the argument that a fault must have been caused after delivery. It is a small clause that changes the incentives on the seller's side of the table.

Carts, staging and the second set

A feeder cart is not a convenience; it is the mechanism that makes changeover a scheduled activity instead of a stopped line. The staging model is simple: one cart on the machine, one cart pre-loaded for the next order, and the changeover consists of swapping carts and loading the program rather than pulling and re-threading individual feeders. The time difference is large — the swap is minutes, the re-thread is tens of minutes — and it is recovered on every changeover for the life of the machine. Buyers who count carts as accessories usually discover their value in the first month of high-mix production.

The number of carts to buy follows from the feeder arithmetic rather than from a rule of thumb. If the equivalent total exceeds the enabled positions, at least one extra cart is structural rather than optional. If the product family changes weekly, two extra carts let one be loaded while another runs and a third be prepared. The cost of a cart is small against the cost of a stopped line, but it is not zero, and it belongs in the total project figure rather than in a later supplementary purchase where it will be scrutinised as an overrun. The tape feeders themselves, in the widths your BOM actually consumes, are the other half of that inventory and should be priced at the same time.

One more item belongs on this line of the checklist: the physical state of the bank itself. Slots wear, connectors fatigue, and a bank that has been loaded and unloaded for a decade can develop positions that hold a feeder loosely enough to cause intermittent index errors. Inspect the bank with the covers off and with a feeder in hand, and check that a feeder seats firmly in several positions rather than in the one the seller demonstrates. The tape feeder range that fits these banks, with the width options that correspond to the slot factors above, is described on the tape feeders in the width range you actually run. A bank with two or three loose positions is a manageable negotiation point; one with a worn section is a replacement cost that should appear before the price is agreed.

Checkpoint 5: options, software and the data you inherit

Options and software are the part of the purchase that nobody prices properly, because they do not look like hardware and they arrive without a crate. The pattern is consistent across machines: the option list is long, the seller quotes a subset, the buyer assumes the rest is included or unnecessary, and the gap surfaces in the first month when a product needs a feature the machine does not have. Software has the same shape but a different failure mode, because a feature can be installed and disabled by licence, and a machine can therefore be physically capable and contractually restricted at the same time.

The checkpoint's purpose is to sort the option list into three groups, price the gaps, and secure the software in a form that survives the seller's departure from the relationship. It also covers the least discussed asset in any used machine purchase: the program archive. A machine that arrives with its recipes is a machine that starts producing; one that arrives empty is a machine that starts programming, and the difference is measured in weeks of engineering time that the purchase price never mentions. Both belong on the checklist before the order, because after the order the seller's incentive to help with either has changed.

Which options are production-critical and which are comfort

Production-critical options change whether a specific product can be built. A tray or stick feeder is critical when the BOM contains parts that cannot be taped, such as certain connectors, modules and shields. A vision package that recognises formed leads or dark moulded bodies is critical when the product carries those parts, because a machine without it will reject them or place them badly. A tall-component capability is critical when the tallest body exceeds the standard clearance. A bad-mark system is critical when panels arrive with known defective cavities and the line must skip them without a manual decision.

Comfort options change how pleasant or how fast the work is rather than whether it can be done: additional nozzle stations, extra monitors, better lighting, a larger program storage, an automatic nozzle changer that saves a few minutes per shift. These are worth buying when the labour arithmetic supports them, and they are the correct place to trim a budget that has overrun. The mistake runs the other way far more often, with buyers trimming a vision option because it looks like an accessory and then discovering that a fifth of the BOM needs it.

For each critical option, write the clause in the present tense rather than the future: the machine is supplied with the option fitted and enabled, and the option appears in the configuration list annexed to the order. Future tense — the option can be added later, the seller will arrange it — is where the cost migrates to the buyer. It is also worth asking for the retrofit price of every critical option that is missing, because that number converts an incomplete offer into a comparable one, and it is often the single largest correction to an apparently cheap quotation.

The feeder count is the clearest example of an option that looks like a specification. A platform may be described by its maximum positions, while the number actually available depends on which feeder configuration was ordered; the six-nozzle single-gantry platform illustrated on the single-gantry six-nozzle platform with 160 feeder positions reaches that 160 only with double tape feeders on 8 mm tape, which is a configuration choice rather than a property of the frame. Buying the platform without the configuration buys the smaller number, and the quotation will not say so unless the checklist asks.

Software: version, licence and the CAD path

Software questions cluster into four groups and each has a specific artifact. The version group asks which software release is installed, whether it is current, and whether updates are available to a second owner; the artifact is a screen capture of the version page with the build number legible. The licence group asks what is licensed, in whose name, and whether the licence transfers; the artifact is the licence document or dongle, not a verbal assurance. The restriction group asks which functions are installed but disabled, which is answered by the option list rather than by the screen. The dependency group asks what the software needs to run at all: an operating system version, a particular PC, a specific network service.

The CAD path is the one buyers forget and the one that delays first production most reliably. Programs can be built from CAD data, from a teaching run, or from a library inherited with the machine, and each path has a prerequisite. If the plan is to import CAD, the machine needs the import option and a defined data format with a defined origin convention; if the plan is to teach, the machine needs a working vision system and a program editor someone knows how to use. Ask the seller to demonstrate the path your team will actually use, not the one that is fastest to show, and record how long it took to build one program from scratch on a board you bring.

Two commercial details decide whether the software remains usable in year two. The first is licence ownership: a licence registered to the seller's company may not transfer, and a dongle that stays with the seller leaves the machine unable to run. The second is the update path: a machine on an old release may be unable to read programs produced by newer software, which matters if the same product will be programmed on a second, newer machine later. Both are inexpensive to resolve at purchase and awkward to resolve afterwards, and both are visible in documents rather than in demonstrations.

The recipe archive and the backups you inherit

A working program archive is worth more than most options and is usually free if it is asked for at the right moment. Every stored recipe contains placement coordinates, feeder assignments, nozzle selections, vision parameters and mark recognition settings that took somebody hours to establish. When the machine changes hands, that knowledge can either travel with it or be wiped during a routine reset by an engineer who is preparing the machine for sale. The difference is not recoverable after the fact, because the knowledge left with the previous operator.

Ask for three things. First, a full backup of the machine's configuration and program data, exported in the manufacturer's own format and handed over on media rather than left on the machine. Second, a list of the stored programs with the product each belongs to, so the archive can be matched against your own products. Third, the mark and panel definitions, which are the part most often lost because they live in a different structure from the placement programs. If the seller cannot export data, that limitation is itself a finding about the machine's software level, and it should be priced as programming time.

Where no archive exists, plan the transfer of knowledge instead of data. A short training session at handover, with your engineer building one program under supervision, converts a week of trial and error into a day and it is a reasonable item to negotiate into the order. Record the sequence in the checklist as an obligation with a duration rather than as a courtesy, because courtesy evaporates once the invoice is paid. The same logic applies to the parameter set behind the vision system, which is where a new owner loses the most time: the machine will place parts, but only after somebody has taught it what each part looks like.

Checkpoints 6 and 7: power, air, transport and installation

The physical arrival of a machine is planned as a logistics event and experienced as an engineering one. Power, air, floor loading, access and rigging are all decided before the crate is opened, and any of them can stop an installation for days while the machine sits on the floor in its wrapping. The checklist items here are unusual in that they can all be resolved with arithmetic and a site visit, and yet they are the items most often left to the installer, who arrives expecting a prepared site and is not contracted to create one.

The right moment for this checkpoint is before the purchase order, not after, because two of the answers can change the choice of machine. A site with a single-phase supply and no intention of upgrading will exclude platforms that need three-phase input; a floor at first-floor level with a narrow goods lift will exclude machines whose crate cannot be turned in the corridor. Discovering either after signing converts a configuration decision into a construction project, and construction projects have their own schedule.

Power: from nameplate kVA to a real circuit

The nameplate gives a rating, and the rating is a starting point rather than a circuit design. Take a hypothetical machine rated at 3.5 kVA on a 200–240 V single-phase supply. At 230 V the full-load current is 3,500 divided by 230, which is about 15.2 A. A circuit is not sized at the load, it is sized above it, and a common allowance is 125 percent, which gives about 19 A and points to a 20 A circuit with the correct plug and breaker type. Add the printer, the oven and the conveyors on the same distribution board and the arithmetic becomes a panel schedule rather than a single line.

Three refinements keep the schedule honest. First, ask for the inrush or start-up current as well as the running current, because a machine with a large motor or a chiller can trip a breaker sized only on steady-state load. Second, ask whether the machine's internal supply is switchable for your voltage and frequency, since a machine built for one market may need a transformer in another, and a transformer is both a cost and a floor item. Third, confirm the earthing and neutral arrangement your site uses, because a machine expecting one configuration will misbehave on another in ways that look like intermittent faults rather than like an installation error.

The calculation also sets the schedule for the electrician, which is a resource most plants share. A single 20 A circuit can be installed in a day; a new three-phase drop with a transformer, a disconnect and a dedicated earth can take weeks and may need approval. If the machine you are buying needs the second and your site only has the first, that difference is a cost and a delay that belongs in the comparison before the order. Write the utility requirement into the order as a document the seller must supply, with the exact figures for your configuration rather than the platform's generic table.

Air, vacuum and floor loading

Compressed air is the utility that is most often already present and least often adequate. A machine needs a pressure at its inlet, a flow at that pressure, and a quality of air that will not contaminate the pneumatic circuit. Plants frequently have pressure at the compressor and pressure at the machine that differ by a bar or more, because the pipe run is long, the fittings are restrictive, or the air is shared with equipment that draws in bursts. The check is a measurement at the machine's intended position during production hours, not at the compressor gauge during a quiet moment. The consequence of a shortfall is not a stoppage but a gradual degradation of pick reliability that consumes engineering time for weeks.

Air quality belongs in the same sentence because it is invisible and cumulative. Moisture and oil carry through the circuit and degrade vacuum generation, valves and nozzle performance, and the damage shows up as intermittent placement errors rather than as a clear fault. Ask what filtration and drying the machine's manual requires, compare it with what the plant actually has, and price the difference. A dryer or a filter bank is a modest capital item and a large reduction in the diagnostic load on the process engineers, and it is one of the few items on this checklist that pays back without any change in production.

Floor loading is arithmetic rather than opinion. A machine of about 1,800 kg standing on a footprint of roughly 1.5 by 1.6 metres presents a load of about 750 kg per square metre, and whether that is acceptable depends on the slab and on where it sits. Ground-floor industrial slabs usually carry it without discussion; mezzanines, raised floors and upper storeys need a structural opinion, and the opinion takes time to obtain. Add the dynamic component from a moving gantry, which is small but not zero, and add the point loads at the feet, which are higher than the average. The safe course is to obtain the machine's mass and foot dimensions before the order and to have the floor checked while the quotation is still being negotiated.

Rigging: door width, floor protection, access

The rigging plan is a walk with a tape measure, and it is best done with the crate dimensions in hand rather than with the machine's dimensions, because the crate is larger. Measure the goods door, the corridor width at its narrowest point, the turn radius into the production hall, the lift capacity if there is one, and the height of any pipe or duct crossing the route. Photograph each restriction with the tape visible, and send the set to the seller or the rigger before the shipment is booked. A route survey that takes an hour prevents the most expensive kind of delay, which is a crate that arrives and cannot enter the building.

Floor protection is part of the same plan and is usually omitted from the installation quotation. Moving a machine on rollers or a pallet truck concentrates load and can mark a coated floor, damage a joint, or crack a tile, and the repair cost lands on whoever owns the building. Specify the protection method in the installation scope — steel plates, plywood runners, or a specific route — and confirm who supplies and lays it. The same question applies to the machine's own feet and levelling: a machine that is not levelled will place accurately in one area of the board and poorly in another, and the difference is often attributed to the head before anyone checks the floor.

The date of the rigging is worth controlling separately from the date of the shipment, because the two are often weeks apart and the machine is vulnerable in between. A machine that arrives before the site is ready sits in its wrapping, sometimes in a corridor, sometimes in a warehouse with uncontrolled humidity. If that is unavoidable, specify the storage conditions and who is responsible for them, and include a re-inspection at installation rather than assuming the condition recorded at the factory still holds. The checklist line is short: who takes custody, where the machine stands, and what is checked when the wrapping comes off.

Installation and commissioning: who signs what

Installation scope is the clause that determines whether the machine is a product or a project. At one end, the seller delivers to the door and the buyer does everything else; at the other, the seller rigs, levels, connects, commissions, trains and proves the machine against an agreed test. Both are legitimate, and the price difference between them is real, but a quotation that does not state which one it is leaves the buyer to discover the boundary on the day. The checklist item is a scope document with a signature, not an email confirming that support will be provided.

The technical part of commissioning has a defined edge worth knowing: the interface between the machine and the rest of the line. Conveyor handshaking in SMT lines conventionally follows the SMEMA interface standard, which defines the signals for board available, machine ready and board transfer. When a new machine is inserted into an existing line, that interface has to be wired and configured, and the responsibility for it sits with somebody. Asking who configures the SMEMA handshake, and whether the machine's conveyor height and direction match the line, turns a vague commissioning conversation into two specific tasks with owners.

Commissioning also produces documents, and they should be listed rather than hoped for: the as-built configuration list, the utility connection record, the calibration record, the backup of the software and the program archive, the training attendance record, and the signed acceptance report. Each of these is the artifact behind a checkpoint earlier in this article, and collecting them at commissioning is the last moment when the seller's team is on site and motivated. A file assembled then takes an hour; the same file assembled a year later takes a week of emails and produces a weaker version.

Checkpoints 8 and 9: acceptance, service scope and clauses

The last two checkpoints decide what happens after the money has moved, which is when the buyer's leverage is lowest and the seller's obligations are tested. Acceptance is the mechanism that ties payment to evidence, and service scope is the mechanism that decides how much of the next three years is predictable. Both are contractual rather than technical, and both are routinely left vague because the technical discussion was more interesting. The result is a machine that works on the day of installation and a relationship that becomes difficult the first time something needs to be resolved quickly.

There is a sequencing argument for putting them last in the checklist but first in the negotiation. A seller who knows that payment is tied to a defined acceptance test prepares for that test, brings the nozzles and the calibration kit, and arrives with a plan. A seller who knows that acceptance means "the machine runs" arrives with good intentions. The difference is not cynicism; it is the ordinary effect of specifying outcomes. The clauses below are short, and their value is that they convert a conversation into an obligation with a date attached.

FAT: what to run before the machine ships

The factory acceptance test is the last moment when problems can be corrected at the seller's site, with the seller's engineers, before shipping costs and customs complicate everything. For a new machine it is often a standard internal test, and the buyer's job is to add the product-specific items that the standard test omits. For a used machine the FAT is more important still, because it is the only structured opportunity to prove the machine before it is loaded. Either way, the test needs a procedure in advance and a pass criterion agreed in writing, otherwise the result is an opinion.

Four items belong in every FAT regardless of platform. Run a board that resembles your product rather than a demonstration board, and run at least the number of boards that exposes repeatability rather than luck. Measure placement offset on a defined pattern and record the values, since a mean and a spread tell you more than a pass stamp. Verify the feeder bank with your widths loaded, including one changeover, so that the mechanics of swapping are exercised rather than described. Verify the vision system on the parts that are difficult, which usually means the dark bodies, the reflective leads and the smallest chips on the BOM.

Record the FAT on video as well as on paper, with the serial number visible at the start. Video is the artifact that settles later questions about what was demonstrated, and it costs nothing. Ask for the measured data rather than a summary, and keep both with the order documents. If a FAT is not possible at the seller's site, the same procedure becomes the first half of the SAT, with the difference that remedies are now more expensive and the schedule is under pressure — which is precisely why the effort belongs earlier.

SAT: the first seventy-two hours

The site acceptance test is where the machine meets your building, your boards and your operators, and it should be defined as a sequence with pass criteria rather than as an impression. A workable sequence runs over three days: day one for installation, levelling, utilities and the first power-on with the seller's engineer present; day two for program building on your own product, feeder loading and a production run long enough to include a reel change and a nozzle check; day three for the acceptance measurements and the training handover. Compressing all three into one day is possible and removes the part of the test that finds intermittent faults.

The acceptance measurements deserve numbers. Take a sample of thirty boards and measure placement offset on a defined component pattern; if the mean offset is 8 µm and the standard deviation is 8 µm against a specification of ±35 µm, the capability index is (35 − 8) divided by three times 8, which is about 1.13. That is a passing result and a thin one, and knowing the number is more useful than knowing that the machine passed. The same calculation on a second pattern often reveals that the machine is comfortable in the centre of the field and marginal at the edges, which is a levelling or calibration question rather than a machine defect. These figures are an illustrative example with assumed values, not a prediction for any specific machine.

Solder joint acceptance, where the SAT includes a run through reflow, is conventionally judged against IPC-A-610, which defines the visual criteria for acceptable, acceptable with process indicator, defect and process indicator conditions. Referencing the standard in the acceptance procedure does two things: it gives both sides the same vocabulary, and it prevents a debate about aesthetics from being conducted as a debate about function. The machine's job is placement; the joint's acceptance belongs to the process, and separating the two in the report avoids blaming the placer for a printing or profile problem.

Four service questions that separate a contract from a promise

Service is the part of the purchase that is priced least precisely and valued most highly after the first unplanned stop. Four questions do most of the work of separating a real arrangement from a friendly one. Who responds, and through which channel — a named engineer, a regional office, or a general inbox? How fast, measured from what event — the call, the remote diagnosis, or the arrival on site? At whose cost — is travel included, are parts included, is there a charge for diagnosis? And for how long — does the commitment expire with the warranty, and what happens in year four?

Each answer should be a number or a name rather than an adjective. A response time of four hours during production hours and next business day otherwise is a commitment; prompt support is not. A spare parts stock held locally with a stated list is a commitment; parts available is not. The purpose is not to trap the seller but to make the obligation explicit enough that both sides can plan. A seller who cannot commit to a response time is telling you that the arrangement is discretionary, which may be acceptable if the price reflects it and unacceptable if it does not.

Remote support deserves its own line because it changes the economics of everything else. A machine connected to the seller's support service can often be diagnosed and corrected in an hour, while the same fault without remote access costs a visit and a delay. Ask whether remote support is available, whether it requires a network connection or a subscription, and what data it transmits. Then check that the site can actually provide the connection at the machine's location, which is a network port and a cable route rather than an intention.

Training belongs in the same clause. The useful measure is not hours of training but tasks performed unaided: building a program, loading a feeder, changing a nozzle, recovering from a pick error, and reading the machine's fault log. Write these as a list in the acceptance report and have somebody sign that each was performed. A training session measured in hours can be delivered entirely as demonstration; a training session measured in tasks produces operators who can start a shift without calling anyone.

Warranty exclusions and the spare parts list

Warranty language is usually generous in the first line and specific in the fourth, and the fourth is the line that matters. The exclusions to look for are consumables, wear parts, damage from utilities, damage from improper use, and the cost of access — travel, accommodation and freight. Nozzles, feeder tapes, filters and belts are consumables in most contracts and will not be covered, which is reasonable and should be budgeted rather than argued. The items worth negotiating are the ones in the middle: a head or a vision module that fails early is not a consumable, and the contract should say so.

Ask for the recommended spare parts list with prices and lead times, and compare the lead times with your production commitments. A list without lead times is a catalogue; the lead time is what determines whether a failure costs a shift or a week. For the parts that carry the line — a nozzle family, a feeder type, a camera module, a drive — the question is whether they are held locally or shipped from the factory, and whether the seller can state a delivery window rather than an intention to expedite.

The final item on this line is the initial spares purchase, which is easier to negotiate with the machine than afterwards. A modest kit bought with the machine is usually cheaper than the same kit bought in a hurry, and it converts the first failure from an event into a routine. The kit does not need to be large: the nozzle sizes that cover your component range, one spare feeder in the width you use most, filters, and the small consumables the manual lists. What matters is that the decision is made deliberately, with the parts list in front of you, rather than discovered on the night the line stops.

A scoring sheet you can take into the meeting

Nine checkpoints produce nine opinions, and opinions are hard to compare across three quotations. A scoring sheet fixes that by making the weights explicit before the offers are opened, which is also the only way to avoid weighting the criteria to fit the preferred machine. Keep it short: one line per checkpoint, a weight that sums to one hundred, a score from one to five, and a note recording the evidence behind the score. The note is the important column, because it is where a score of four with a document differs from a score of four because the seller was pleasant.

Suggested weights for a first machine, offered as an example rather than a rule: condition and identity together thirty, feeder package twenty, options and software fifteen, utilities and installation ten, acceptance ten, service fifteen. A second machine bought for capacity would shift weight toward feeders and service and away from installation. The weights should be agreed by the people who will live with the machine, not by the buyer alone, because the process engineer and the maintenance technician rarely weight the same items the same way and the disagreement is informative.

Use the sheet twice: once on the documents, before any visit, and once after the visits. The movement between the two scores is the most useful output, because it shows which seller's story failed under inspection. A quotation that scores well on paper and collapses after a visit is a common pattern and an expensive one to discover after delivery. Recording both scores, with the evidence notes, also gives the finance approval a defensible basis that is not simply the lowest price, which matters when the cheapest offer is not the one the plant wants.

New, used and rebuilt: which checkpoints get harder

The checklist does not change between a new machine and a used one; the evidence standard does. For a new machine, identity and configuration are established by the order and confirmed by the build sheet, condition is largely a matter of the acceptance test, and the software arrives with a licence in the buyer's name. The checkpoints that need attention are the ones where a standard quotation is silent: which nozzle set ships in the crate, which options are included rather than available, and what the acceptance procedure actually measures.

For a used machine, condition moves from a formality to the centre of the purchase, and three checkpoints become harder at once: wear inspection, feeder inventory and software transfer. The hour meter is unreliable as a wear measure, the feeder package is often a mixed accumulation rather than a designed set, and the software may be licensed in a way that does not travel. The remedy is not more caution but more artifacts: photographs to a specification, an inventory table, and a witnessed run on your own board before the machine is shipped.

A rebuilt machine sits between the two and needs one extra question: what was rebuilt, and by whom. A rebuild that replaced the head, the bank and the vision module is closer to new than to used; a rebuild that cleaned and painted is a used machine with better presentation. Ask for the rebuild scope as a list of replaced assemblies with part numbers, and treat any assembly not on the list as original. The distinction matters for price and for the warranty conversation, and it is the single question that most changes the value of a rebuilt offer. A current flexible build in production form, such as the prototype-scale build used as the reference machine, is a useful baseline for that comparison because the specification it carries is the one the rebuild should be restoring.

Turn the checklist into a quotation you can compare

If you send us the nine checkpoints with your own answers where you have them, we will return a configuration rather than a catalogue entry: the platform, the head, the feeder inventory in 8 mm equivalents, the nozzle list by part number, the options that your BOM actually requires, the utility requirement at your site conditions, and the service obligations in numbers. We will mark which figures come from the product pages on this site, which come from the manufacturer's published specification and which are our assumptions, so the comparison against any other offer is like for like instead of price against price. We will also tell you which checkpoint we think carries the most risk in your case, and if the honest answer is that a used machine with a documented feeder package beats a new one with an incomplete package, the review will say that. A purchase that starts with evidence usually ends with a machine that runs, and the checklist exists to make that the normal outcome rather than the lucky one.

Frequently asked questions

What should a JUKI pick and place machine buying checklist cover first?

Start with identity and configuration, because every later question depends on them. Confirm the platform, the head fitted, the build year and the serial from the nameplate, then compare the configuration sheet against the machine itself. A model number alone leaves the head, the options, the feeder inventory and the software level undefined, and those four items decide what the machine can actually run. Buyers who begin with price or speed usually revisit identity later, at a point when the answer is expensive to change and the seller has already built a quotation around an assumption.

How can I confirm the model and options of a used machine?

Ask for three artifacts and cross-check them against each other: a legible nameplate photograph, the configuration or build sheet, and photographs of the options in place on the machine. The serial on the plate should appear on the sheet, the build year should be consistent with the software version, and every option claimed should be visible somewhere in the photographs. Then ask whether any major assembly has been replaced and request the work order. A seller who produces the record promptly is giving you a reason to trust the rest of the file.

How many feeders should come with the machine?

Enough to hold your BOM in your tape widths, plus a working reserve, and that is a calculation rather than a number. Convert each BOM line into 8 mm equivalents using the slot factor for its width, add ten to fifteen percent for engineering changes, and compare the total with the enabled positions on the machine. A hypothetical family of 120 lines can easily consume more than 200 equivalents, which means one bank cannot hold it at once. Ask for the feeder inventory as a table by width and type, not as a single count.

What power and air should I prepare before delivery?

Prepare for the configuration you ordered, not for the platform's generic table, and ask the seller for the figures in writing before shipment. Take the rated kVA, divide by your voltage to get full-load current, and size the circuit above it, commonly at 125 percent. Confirm the inrush current, the voltage and frequency setting, and the earthing arrangement. For air, ask for pressure and consumption at the machine inlet, and measure the pressure at that position during production hours rather than at the compressor.

What should be tested before I sign the acceptance report?

Test on your own board type, at a quantity that exposes repeatability, and record measured values rather than a pass stamp. Run at least thirty boards, measure placement offset on a defined pattern, and calculate the capability index from the mean and the spread. Exercise one feeder change and one reel change, verify recognition on the darkest and most reflective parts, and confirm that your engineer can build a program unaided. Test the conveyor handshake with the rest of the line, then collect the document pack before the seller's team leaves the site.

What belongs in a service agreement beyond the warranty?

Four commitments, each expressed as a number or a name: who responds and through which channel, how fast measured from which event, at whose cost, and for how long. Add remote support, because a fault diagnosed in an hour is a different cost from a fault that waits for a visit. Ask for the spare parts list with lead times and whether critical items are held locally. Training should be measured in tasks performed unaided rather than in hours delivered, and the tasks should be signed off at handover.

Does the same checklist work for new, used and rebuilt machines?

The nine checkpoints stay the same; the evidence standard changes. A new machine establishes identity and configuration through the order and the build sheet, so attention moves to what a standard quotation leaves silent. A used machine moves condition, feeder inventory and software transfer to the centre, and needs photographs to a specification plus a witnessed run. A rebuilt machine adds one question: which assemblies were replaced, with part numbers, and by whom. Treat any assembly absent from that list as original when you compare the price.