Publish Time: 2026-11-08 Origin: Site
Buyer verification guide · Published by I.C.T · For SMT process engineers, production owners and equipment buyers
The new vs used JUKI pick and place machine question is usually framed as a price question, and that framing hides the part that actually decides the outcome. A used machine is not a cheaper version of a new one; it is a different asset with a different risk structure. What separates a good used purchase from an expensive one is not the discount, it is whether the buyer verified the four things that do not appear on a quotation: what the machine actually is, how it has been used, what the feeder package really contains, and what documentation travels with it. Get those four right and a used machine can be the better buy; skip them and the saving becomes a repair schedule.
This guide gives a verification framework rather than a preference. It separates the three things buyers call "used", computes an annual-cost comparison with stated assumptions, then walks the checkpoints that differ between a new and a used asset: model year and generation, usage history, the feeder bank, calibration records, software licences, transport and installation, and acceptance criteria that fit a machine which has already been in service. Every calculation is labeled as an assumption so it can be replaced with your own figures, and the platforms referenced are the ones listed on the placement platform range used throughout this framework.
The word used covers three assets that behave nothing alike, and most of the confusion in this purchase starts there. A machine that has been running in a plant until last month is a known quantity with an unknown future, and its condition is documented only to the extent that the previous owner bothered. A machine that has been rebuilt by a specialist has had specific assemblies replaced, and its value depends entirely on which ones. A machine assembled from parts in a dealer's warehouse is a third category again, sometimes better than described and sometimes a collection of compatible components that has never run together. Treating all three as "used" means applying one discount and one level of trust to situations that deserve different tests. Sorting the asset before valuing it is the first rule of any new vs used JUKI pick and place machine comparison, and it is the rule most often skipped when a discount is on the table.
The framework below sorts the asset first, then applies the checkpoints that matter to that class. It also puts a number on the decision, because the discount is the least informative figure in the negotiation. What matters is the annual cost of owning the machine through the period you will actually keep it, including the money you will spend to make it reliable and the production you will lose while that happens. Several buyers discover at that point that the used machine is not cheaper at all, and a few discover the opposite and buy with confidence. Both outcomes are useful, and both require the same three-or-four hours of verification.
A new machine carries a build risk and a delivery risk, but almost no condition risk: the configuration is what the order says, the wear is zero, and the documentation is produced by the process that made it. A rebuilt machine carries a scope risk: the assemblies that were replaced are new, the ones that were not are as old as the machine, and the boundary between them is the single most important piece of information in the offer. A used machine sold as-is carries a condition risk across everything, which is why the inspection budget should scale with how little is documented rather than with how much the machine costs.
The three classes also differ in what a seller can prove. For a new machine, the proof is the build sheet and the factory test. For a rebuilt machine, the proof is the rebuild scope with part numbers and the warranty that covers the replaced items. For an as-is machine, the proof is what you generate yourself: meter readings, photographs to a specification you set, and a run on your own board. Buyers who ask a dealer for the documentation appropriate to a new machine usually get silence, and then conclude the dealer is hiding something, when the real issue is that the question belonged to a different class.
Asset class | What the seller can prove | What the buyer must verify | Typical failure later |
|---|---|---|---|
New | Build sheet, factory test report, full warranty | That the configuration matches the order | An option assumed included but quoted separately |
Rebuilt | Rebuild scope with replaced part numbers | Which assemblies were not replaced | Wear in the assemblies outside the scope |
Used, as-is | Whatever the previous owner left behind | Condition, hours, feeder inventory, calibration | Two or three subsystems needing work in year one |
The comparison worth making is annual cost of ownership, not purchase price, because the two assets depreciate differently and consume different amounts of engineering time. Take a hypothetical new machine priced at 100 units, with a resale value of about 45 units after six years. The straight-line depreciation is 55 units over six years, or about 9.2 units a year. Now take the same platform used, at 55 units, with a resale value of about 15 units after six years: depreciation is 40 units over the same period, or about 6.7 units a year. On depreciation alone the used machine wins by roughly 2.5 units a year, which is exactly the gap buyers expect to see.
Then add the costs that belong to the used asset. A reconditioning allowance of 8 units in year one, spread over the six years, adds about 1.3 units a year. A difference in unplanned downtime of forty hours a year, valued at the contribution margin of the line rather than at the labour rate, can easily add two or three units a year. Put those together and the advantage shrinks to almost nothing, or reverses if the downtime estimate is worse. The arithmetic does not say used is bad; it says the discount has to survive the ownership costs, and that a 45 percent discount on a six-year-old machine is roughly the price of parity rather than the price of a bargain. All figures here are illustrative assumptions, not market data.
Line item (assumed, units of new price) | New machine | Used, six years old |
|---|---|---|
Purchase price | 100 | 55 |
Resale value after six years | 45 | 15 |
Depreciation over six years | 55 | 40 |
Depreciation per year | ≈9.2 | ≈6.7 |
Reconditioning in year one, spread over six years | 0 | ≈1.3 |
Extra unplanned downtime, 40 h/year at line margin | — | ≈2.0–3.0 |
Net annual cost of ownership | ≈9.2 | ≈10.0–11.0 |
There are conditions under which no discount justifies the purchase, and naming them early saves the weeks that would otherwise go into an inspection. The first is a product whose accuracy requirement sits at the edge of the platform's specification when new: a machine quoted at ±35 µm with a capability index of at least 1 has no margin to give away, and a used unit that has drifted will pass a demonstration and fail a capability study. The second is a certification or customer-audit requirement that the asset cannot satisfy, because traceability, calibration intervals or documentation are contractually specified and an undocumented machine creates an audit finding rather than a saving.
The third condition is time. A used machine needs inspection, possibly reconditioning, shipping, installation and a learning curve, and none of those compress well. If the product launch is eight weeks away, the used route is a schedule risk that no engineering effort removes, and the correct comparison is not price against price but date against date. The fourth is the absence of local support: a platform that nobody in your region services turns every fault into a shipping event, and that cost belongs in the model long before the discount is negotiated. A purchase that has to be defended to a customer, an auditor or a bank is usually better served by the asset with the shorter paper trail of questions, which is the new one.
Where the purchase is part of a wider programme rather than a single decision, the framework in the evidence-and-clause checklist written for a new machine purchase is worth running first, because it establishes what has to be confirmed and what has to be written into the order before the asset class is even discussed. The nine checkpoints there apply to both classes; what changes with a used machine is the evidence standard, not the list. Running the general checklist first prevents the common error of negotiating a discount on a machine whose configuration nobody has yet described in writing.
Sellers describe assets in the vocabulary of the class they are selling, and the vocabulary is informative. A new-machine seller talks about configuration, lead time and options, because those are the variables he controls. A rebuilt-machine seller talks about the scope of the rebuild, because that is what he is selling. An as-is seller talks about price, availability and "good working condition", because those are the only claims he can make without documentation. None of these is dishonest, and reading them correctly prevents the buyer from asking the wrong class of question and interpreting the resulting silence as concealment.
The phrase to treat as a request for evidence rather than a statement is "fully serviced". It can mean a documented service with replaced consumables, or it can mean somebody wiped the covers and ran a board. The follow-up is not argumentative: ask which consumables were replaced, on what date, by whom, and whether the work order exists. A seller with a real service history answers in a sentence and sends a photograph of the sheet. A seller without one answers in adjectives, and that answer is itself the finding, because it tells you the machine's maintenance record is going to be reconstructed by you rather than inherited.
Two more phrases deserve translation. "Low hours" is a claim about the meter, not about wear, and should be checked against the production history rather than accepted. "Ready to run" is a claim about the moment of demonstration, and the useful question is what it will need in the first month of your production rather than in the seller's warehouse. Neither phrase is a red flag on its own; both become one when they are the only evidence offered for a checkpoint that the framework below asks to be proven differently.
The model year is the first number buyers ask for and the least useful on its own, because a machine's capability depends on its build configuration more than on the calendar. Two units of the same model, built two years apart, can differ in head type, vision package, feeder interface and software generation, and those differences decide what the machine can place. The year matters mainly as a proxy for three things: which generation of the platform it belongs to, which spare parts are still available quickly, and how much of its service life has already been consumed.
Generation is the concept worth spending time on. Manufacturers revise platforms without renaming them, and a revision can change the head, the camera arrangement, the conveyor, the controller or the software architecture while the badge stays the same. A buyer who compares a used machine against a brochure from the current year is comparing two different products. The right reference is the specification that applied to the build year of the specific unit, which is why the build sheet and the serial matter more than the model name in the advert.
The serial is the only identifier that ties a specific machine to a specific build, and it should appear in three places: the nameplate, the build or configuration sheet, and the service history if one exists. Ask for all three, and check that they agree. A serial that appears on the nameplate but not on any document suggests the document belongs to a different unit, which happens more often than buyers expect when machines are bought in lots and paperwork is copied between them.
The build sheet is where the model year becomes useful, because it lists the assemblies fitted at build. From it you can establish the head type, the vision configuration, the feeder interface and the software level as built, and compare them with what is on the machine today. Any assembly that differs is either an upgrade, a replacement, or a swap from another unit, and each of those has a different implication for price and for support. Recording the differences takes twenty minutes and gives you the only reliable description of the asset you are actually buying.
Where documentation is missing, the year can be inferred from the software version and the nameplate format, but the inference should be written down as an inference. The practical consequence of an uncertain year is not the year itself but the spare-parts question: an older build may use assemblies that are no longer stocked, and the lead time for those parts becomes the machine's real availability risk. That question can be answered directly by the platform's service organisation, and it is worth answering before the discount is agreed rather than after the first failure.
Three kinds of change move a platform between generations in a way that affects the purchase. Head architecture is the first: a rotary head and an inline flexible head place different component ranges at different rates, and a machine's suitability for a BOM follows the head rather than the badge. Vision configuration is the second: the difference between recognizing chips reliably and recognizing formed or dark parts reliably is a camera and lighting package, and it is frequently the option that a used machine lacks. Conveyor and clamping is the third, and it decides the largest panel the machine will accept, which for many buyers is the constraint that removes a candidate entirely.
These are the questions to settle before comparing prices, and they are the same questions that decide a new-machine shortlist. The screening order that turns a BOM into a platform requirement is set out in the platform shortlist that precedes any inspection, and running it against a used candidate is a sanity check on the seller's recommendation. If the framework eliminates the platform, the discount is irrelevant, which is a cheaper discovery than an inspection trip.
A practical test for generation differences is to ask the seller to run a board that carries the smallest and the largest component on your BOM, rather than a demonstration board. The smallest component tests the vision package and the nozzle set; the largest tests the head, the clearance and the feeder path. A machine from an earlier generation can pass the first and fail the second, and the failure is usually visible in the first ten placements rather than in a specification table.
Model names are stable while their specifications are not, and this is where buyers get caught. A model that was published with a single-clamp board size may later be quoted with a double-clamp figure that is much larger, and both numbers can appear in circulation for the same badge. The RS-1R, for example, is described on this site with a maximum board size of 650 × 370 mm in single clamping and up to 1,200 × 370 mm with double clamping, while the RS-1XL extends the single-clamp width to 650 × 560 mm. A buyer who needs the larger panel has to confirm which clamping arrangement the specific used unit carries, because the badge alone does not answer it.
The same drift applies to feeder counts, component height limits and accuracy statements, and it is worst where an option changes the limit. A feeder maximum that assumes a specific feeder type, a height limit that depends on the head, an accuracy figure that assumes a particular vision package: each of these is a number that belongs to a configuration, not to a platform. When a listing quotes the largest number in circulation, the useful reply is a request for the line in the build sheet that establishes it for this unit.
Listings also vary in where they state the identity. Some put the platform in the title and the fitted model in a separate specification field, which is a listing where the fitted model is stated in its own field rather than in the headline. Read the field, not the headline, and carry that reading into the question you put to the seller. It sounds trivial and it prevents the most basic error in this market, which is inspecting one machine and receiving another.
The option list is the closest thing a used machine has to a birth certificate, because it records what the original buyer paid for and therefore what the machine was built to do. It is also the document most likely to be incomplete, since options added later by a distributor or a previous owner rarely appear on the original sheet. The task is to reconcile three sources: the original option list, the physical evidence on the machine, and the software's own view of what is licensed.
Reconciliation is straightforward when you know what to look for. Physically, an option leaves a trace: a mounting point, a connector, a camera, a station, a changed panel. In software, the licence screen names what is enabled, which can differ from what is installed. Where the two disagree, the software usually wins for capability and the hardware wins for what can be enabled later without a retrofit. Record both columns and treat the differences as negotiation items with a price attached, because that is what they are.
The last piece of the birth certificate is the original application. A machine built for a specific product family often carries options that are valuable for that family and irrelevant for yours, and the reverse. A unit configured for LED panels may carry the wider clamping and the larger-board conveyor, and may lack the fine-pitch vision package your board needs. The purchase is not the platform, it is the configuration, and the option list is the only place where the configuration is written down before you pay for an inspection.
Every machine records something, and no machine records the thing buyers most want to know. The counters describe volume: hours powered, boards produced, placements made. They do not describe duty, which is the combination of component mix, feeder changes, shift pattern and operator care that actually determines wear. A machine with 20,000 hours of stable single-product production and a machine with 12,000 hours of high-mix work with three changeovers a day can arrive with very different amounts of life left, and the counter will suggest the opposite conclusion.
The verification task is therefore two-part. First, extract the numbers the machine will give you and check them against the seller's story. Second, reconstruct the duty from the evidence that surrounds the numbers: the programs stored, the feeder widths present, the nozzle set, the alarm history, and the answers of the person who ran it. The second part is where the useful information is, and it is also the part that a buyer can do in an hour with a notebook, which makes it the best value in the whole inspection. For a new vs used JUKI pick and place machine decision, that hour usually answers more than another day of photographs.
Three counters are usually available: powered hours, production hours or run time, and a placement or board counter. Ask for all three on the same screen, photographed together, because the ratios between them are more informative than any single value. A machine showing many powered hours against few production hours has spent much of its life idle or in maintenance, which changes the wear profile. A machine showing production hours close to powered hours has been run hard in continuous shifts, which changes the consumable budget.
What the counters hide is the relationship between volume and rate. A machine that placed a large number of components may have done so comfortably within its capability, or may have been pushed at its limit with a mix that maximized head travel and feeder index cycles. The distinction matters because the published rate is an optimum figure measured under favourable conditions; what the machine delivered in service is a function of the boards it ran. The reasoning behind what the published rate becomes in a balanced line explains why a high-volume counter is not evidence of a hard life, and why a low counter is not evidence of an easy one.
One counter is worth requesting specifically because sellers rarely volunteer it: the number of programs stored. A machine with two hundred stored programs has run high-mix work and has been through the changeover cycle two hundred times. A machine with three has run a stable family. Neither is better in the abstract, but the number tells you where to look for wear, and it also tells you something about the operator skill the machine has been used with, since high-mix work either trains people well or wears equipment quickly depending on how the plant is run.
Campaign production means long runs of one product with few feeder changes. The wear concentrates in the head, the nozzles and the conveyor, and the feeder bank sees comparatively little handling. High-mix production means frequent changeovers, and the wear spreads: feeder connectors, cart mechanics, the nozzle changer if fitted, the clamping mechanism and the software's recipe library all see constant use. The two histories produce machines that look similar in a photograph and behave differently in the first month.
The practical consequence is that the inspection should follow the history rather than a generic route. For a campaign machine, spend the time on head and nozzle condition, on vision cleanliness, and on a capability run at the edges of the board. For a high-mix machine, spend it on the bank: seat a feeder in a dozen positions rather than one, check the connectors, look at the cart mechanics, and ask how many programs were edited by hand. The second inspection takes longer and finds different faults, and buyers who run the campaign route on a high-mix machine tend to discover the bank problems in production.
High-mix machines also carry a documentation advantage that is easy to overlook: they have a large recipe library, which is a transferable asset if the licence and the data come with the machine. A plant buying a used high-mix machine from a similar product family can inherit program structures that save weeks of programming. A plant buying a campaign machine inherits almost nothing beyond the platform, and should price its own programming time accordingly. The history therefore affects both the risk and the value, in opposite directions.
The component mix leaves physical traces that are readable without instrumentation. Count the nozzle stations and note which holders are empty: a machine configured for a broad mix will have a populated station, while one built for a narrow mix often has two or three holders in use and the rest empty. Look at the feeder bank for width diversity: a bank that is almost entirely 8 mm tape has run chip-heavy work, and a bank with a spread of 16, 24 and 32 mm positions has run boards with larger bodies. Both are legitimate; they simply predict different wear.
Then look at the board path. Wear on the rails, the clamp faces and the conveyor belts scales with the number of boards rather than with the number of components, so a machine that ran many small panels shows more mechanical wear than one that ran fewer large panels at the same placement count. If the rails are polished and the clamp faces are marked, the machine has handled a lot of boards, and the conveyor is the subsystem to inspect closely rather than the head.
Finally, look at what has been worked around. A tie-wrapped cable, a bypassed sensor, a missing cover screw, a nozzle holder with a replacement clip: each is a small repair that tells you the machine has had faults and somebody has kept it running. That is not disqualifying, and in a well-run plant it is a sign of care rather than neglect. What matters is whether the workaround touches something your production depends on, and the answer usually arrives by asking the operator rather than the seller.
If one part of the purchase explains most of the disappointment in used-machine deals, it is the feeder package. A new machine is quoted with a defined feeder inventory, and the configuration sheet states it. A used machine arrives with whatever feeders the previous owner owned, which is an accumulation shaped by their production history rather than a package designed for yours. The count quoted in the advert is usually the bank's maximum capacity, not the number of feeders in the crates, and the two numbers can differ by a factor of two without anyone lying.
The verification therefore has three parts: how many feeders exist, in which widths, and in what condition. The first is a counting exercise, the second is an arithmetic exercise against your BOM, and the third is a mechanical test that takes an afternoon. Skipping any one of them produces a machine that can place your board in principle and cannot load it in practice.
Feeder inventories grow by addition. A plant buys feeders for one product, adds a few widths for the next, retires the ones that fail and keeps the rest. Ten years later the inventory reflects ten years of product decisions, which is why a used package so often contains an excess of one width and a shortage of another. The seller's list usually reports the total, because a total is easy to state and sounds generous, and the buyer discovers the distribution after delivery.
There is a second reason the accumulation is misleading: some feeders in any large inventory are not serviceable. They may be intermittent, missing a component, or of a type the machine's interface does not support without an adapter. A used package of 90 feeders may contain 70 that are production-worthy, and the 20 that are not will be discovered one at a time over the first months, each discovery costing a pick error investigation. The count to negotiate against is the serviceable count, established by a bench test or by running them.
The practical check is to ask for an inventory table by width and type, then sample. Take five feeders at random from the package, load them, and run a short program that indexes each one a few hundred times. A package that survives random sampling is worth its stated count; one that fails two of five is worth a renegotiation, because the failure rate in the unsampled remainder is unlikely to be lower.
Usable positions are the number of feeder slots your BOM consumes after width conversion, compared with the number of slots the bank can serve with the feeders actually provided. The conversion uses the 8 mm equivalent, where a slot factor expresses how much bank width each tape width occupies. The factors below are an illustrative set for the calculation; the ones that apply to your machine and feeder type should come from the supplier, since they vary with feeder design and bank layout.
Take a hypothetical BOM with 96 lines: 58 on 8 mm tape, 22 on 12 or 16 mm, 10 on 24 mm, 4 on 32 mm and 2 on 44 mm. Applying the factors gives 58 + 44 + 30 + 16 + 12, which is 160 equivalents. Against a bank of 112 positions, that BOM needs either a second bank, a staging cart, or a split across two products. Against a delivered package of 90 feeders whose widths are concentrated in 8 mm, it needs a purchase of wide feeders before it can run at all. Both numbers belong in the offer, and neither appears in the advert.
Tape width | BOM lines (assumed) | Slot factor (assumed) | 8 mm equivalents |
|---|---|---|---|
8 mm | 58 | 1.0 | 58 |
12 mm | 14 | 2.0 | 28 |
16 mm | 8 | 2.0 | 16 |
24 mm | 10 | 3.0 | 30 |
32 mm | 4 | 4.0 | 16 |
44 mm | 2 | 6.0 | 12 |
Total | 96 | — | 160 |
The comparison that matters is between this total and the number of slots the machine can serve with the package in front of you, not the number it could serve if it were fully equipped. A used machine may be advertised with a 112-position bank while arriving with feeders that fill 60 of them, in widths that suit somebody else's boards. The gap is a purchase order, not a detail, and it is the single most common item missing from a used-machine budget. The method for turning a BOM into feeder and nozzle requirements is worked through in how BOM lines become feeder and nozzle requirements, and it produces the number to put next to the seller's count.
Condition is where a used package separates into three groups: feeders that will run for years, feeders that will run until the first awkward tape, and feeders that are already marginal. The swap test finds the third group quickly. Load a feeder, run it, remove it, load another in the same slot, and repeat across several slots. A bank whose slots grip firmly and release cleanly passes; one where a feeder rocks, needs force, or sits at an angle tells you the bank itself has wear that no feeder purchase fixes.
Then test the feeders against tape rather than against a bench. A feeder that indexes correctly with an empty tape can still fail with a full reel, because the added mass changes the pull. Run a reel of the width you actually use, through a program that indexes it several hundred times, and watch for missed indexes, tape lift and component loss. This is the test that separates a package priced on quantity from one priced on condition, and it takes about twenty minutes per width.
Spares belong in the same conversation. A used package with no spares puts the whole line behind one feeder failure; a package with two spare feeders in the dominant width is resilient in the way that matters. The tape feeders used in these banks, in the widths that correspond to the factors above, are described in the tape feeder widths and the bank they fit, which is the reference for pricing the gap rather than for judging the package.
Accuracy is the checkpoint where a used machine most often fails quietly. A machine can place a demonstration board perfectly and still be unable to hold a capability index on a fine-pitch product, because the demonstration is chosen to flatter and the real board is not. The difference between the two is measurement rather than impression, and the measurement is only meaningful if it is tied to a defined procedure, a defined board and a date.
The paper trail matters for the same reason. A calibration certificate is not a formality for an auditor; it is the record that tells you where the machine sits relative to its specification and how much of that margin has already been consumed. A machine with a current certificate and measured values is a known asset. A machine with a pass stamp is an unknown asset with a stamp, and the price should reflect which of the two you are buying.
Four fields make a certificate usable: the machine serial, the date, the procedure, and the measured results. The serial ties the document to the asset, which a model number cannot do. The date establishes whether the calibration is inside its interval, since any measurement decays with time and with use. The procedure identifies how the numbers were produced, because accuracy figures from different procedures are not comparable and a number without a procedure is a claim rather than a measurement.
The measured results are the part sellers most often omit, and they are the part that carries information. A certificate reporting offsets across a defined pattern tells you whether the machine is centred or skewed, whether it drifts at the field edges, and how much margin remains before it needs intervention. A certificate reporting only that the machine passed tells you that somebody was satisfied on the day, which is not the same thing and does not support a capability claim for your product.
Where the certificate includes the vision system, ask what was recognized and at what size. Recognition performance is part-dependent: a machine that reads chip bodies reliably may still struggle with dark moulded parts or reflective leads, and the certificate will not mention it unless the test covered those parts. Nozzle condition belongs in the same conversation, since a worn or wrong nozzle changes the effective accuracy of an otherwise well-calibrated head; the replacement part numbers and orifice sizes are listed by family in the nozzle part numbers by orifice size, which is the reference for checking that the set matches your component range.
There are three levels at which a used machine's accuracy can be established, and the buyer should choose the level that matches the product's requirement rather than the level that matches the seller's convenience. The first tier is documentary: a current certificate with measured values. It is cheap, it is often sufficient for a product with generous pitch, and it is worthless if the certificate is old or belongs to a different serial. The second tier is a witnessed sample: run a defined pattern on a board you bring, measure the offsets, and compare them with the specification.
The second tier is where the arithmetic becomes useful. Take a hypothetical sample of thirty boards with a measured mean offset of 8 µm and a standard deviation of 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 passes and it does not pass comfortably, which is exactly the kind of result a buyer needs to see rather than a green tick. Run the same calculation on a second pattern near the field edges and the number usually falls, which points at levelling or calibration rather than at the head.
The third tier is an independent measurement against a defined standard. The industry procedure for characterizing placement performance is IPC-9850, which specifies the test board, the pattern, the sample size and the calculation, and a measurement taken under it is comparable between machines and between suppliers. Where the product will be judged against solder joint criteria after reflow, IPC-A-610 supplies the visual acceptance language, and keeping the two standards separate in the report prevents a placement discussion from becoming an argument about joint appearance. This third tier costs more and removes the remaining ambiguity, and it is the right choice when the product's pitch leaves no margin.
A missing certificate is not a reason to walk away, and treating it as one eliminates most of the used market. It is a reason to change the purchase: the verification moves into the acceptance stage, the cost of a full recalibration is priced into the offer, and the payment is tied to a measured result rather than to delivery. A machine without documentation is buyable at the right price and with the right clauses, and the price should be reduced by at least the cost of the recalibration plus the risk that the recalibration reveals a fault.
The judgement call is whether the machine is likely to calibrate. A machine with clean mechanics, a serviceable head and a complete nozzle set usually does. One with visible impact damage, a history of head replacement, or a conveyor that has been repaired more than once may not, and paying for a calibration that fails is an expensive way to learn that. The middle course is a conditional agreement: a deposit, a witnessed capability run, and the balance payable only if the run meets the agreed criterion. Sellers accept this arrangement more often than buyers expect, because it costs them nothing if the machine is as described.
Where the requirement is contractual rather than internal, the acceptance criterion should be written in the language of the standard the customer will audit against, and the report should name the procedure, the board and the sample size. A certificate that satisfies an internal engineering review can still fail an automotive or medical audit because it does not say how the numbers were obtained. Deciding which audience the document has to satisfy is a ten-minute conversation that saves a repeat measurement later.
Software is the cheapest part of a used machine to verify and the most expensive part to fix after the fact, because a licence that does not transfer turns a working machine into a restricted one and there is no engineering workaround. The verification is mostly documentary: which version is installed, which options are licensed, in whose name the licence is registered, and whether the transfer is permitted. None of those questions requires a site visit, and all of them are worth settling before one is booked.
The data is the second half. A machine that arrives with its program archive arrives with years of somebody's work; a machine that arrives empty arrives with a programming project attached. The archive is usually free to request at the right moment and impossible to reconstruct later, which makes it the single highest-return item in this section. The third half, which is not data at all, is the knowledge in the heads of the people who ran the machine, and it can only be transferred by training.
Three licence conditions decide whether a used machine can do what its hardware allows. The first is ownership: a licence registered to the seller's company may not be transferable to yours, and a machine sold without the licence transfer is a machine whose options may stop working at the next software event. The second is scope: a licence may cover a site, a production line or a single machine, and moving the machine may invalidate it. The third is the dongle or activation mechanism, which is a physical item that has to be in the crate.
Ask for the licence document, not a statement that the machine is licensed. Check the name on it, the scope, the expiry if any, and whether a transfer procedure exists and who administers it. If the answer is that the seller will arrange it after the sale, the arrangement should be a condition of the payment rather than a promise, because the buyer has no leverage once the machine has shipped. The same applies to any subscription attached to remote support or software updates: it is either transferred or it is not, and it should be listed.
Used-unit listings make this easier to check than they used to, because the model field is usually stated explicitly. A listing for a used unit in the compact rotary class, for example, names the model in its own field, which lets you look up the correct software generation before you ask about the licence. Reading the field first prevents the common error of pricing a licence transfer for a platform the machine turns out not to be.
A used machine is usually several releases behind the current software, and that is not automatically a problem. It becomes one when the plant already owns a newer machine of the same family, because programs written on the newer release may not load on the older one, and the workaround is to maintain two program structures. The question to ask is not which version is installed but whether the installed version can exchange programs with the rest of your line.
Ask for the version screen and for one program file exported from the machine. Then have your engineer try to open it in your own environment, or ask the seller to load a file produced by your software. This test takes ten minutes when the machine is powered and it answers a question that specifications cannot. Where compatibility is impossible, the cost is engineering time to maintain a second program set, and that cost is predictable enough to be priced.
Updates are the second half of version drift. A machine behind on updates may lack the bug fixes that matter for your component mix, and catching up may require a service visit rather than a download. Ask whether updates are available to a second owner, whether they are chargeable, and whether the current hardware supports the latest release. A machine that cannot be updated is not disqualified, but its software is now a fixed asset and should be treated as one in the price and in the risk assessment.
Every stored program contains placement coordinates, feeder assignments, nozzle selections, vision parameters and mark settings, and together they represent the accumulated engineering of the previous owner. Where the products are similar, the archive is directly valuable: it provides starting points that reduce programming time from days to hours. Where the products are different, it is still valuable as a set of worked examples that shows how the machine's software is intended to be used.
Ask for three artifacts before the machine ships: a full backup in the manufacturer's format, a list of stored programs with the product each belongs to, and the mark and panel definitions, which live separately and are the part most often lost. Have the seller demonstrate a restore on the machine, not just an export, because an export that cannot be restored is a file rather than a backup. Ten minutes of demonstration is the difference between inheriting an archive and inheriting a folder.
Where no archive exists, the equivalent is training with a duration and a deliverable. An engineer who builds one program under supervision at handover converts a week of trial and error into a day, and the same session transfers the mark-recognition settings that are otherwise rediscovered by experiment. Write it as an obligation with a listed set of tasks rather than as a courtesy, and record who attended. The knowledge is the only part of the purchase that cannot be replaced by a purchase order later.
Used machines carry a training debt that new machines do not, because the people who knew the machine are not part of the transaction. On a new machine, the commissioning engineer transfers operating knowledge as part of the installation. On a used machine, that knowledge left with the previous operator, and the buyer inherits a machine that will run but whose idiosyncrasies are undocumented. The debt is real, it is measurable in weeks, and it is almost never in the budget.
The size of the debt depends on how unusual the machine's configuration is. A standard platform with a conventional feeder bank and a common software version can be learned from documentation and a few days of practice. A machine with a special head, a modified conveyor or an unusual option set requires somebody who has seen that configuration before, and if that person is not available the learning is done by experiment on production boards. Asking the seller for the name of the engineer who serviced the machine is worth more than it sounds, because that engineer may be available for a day of paid support.
Budget the debt explicitly: two to five days of engineering time for a standard configuration, more where the product is demanding or the machine unusual, plus the cost of a support visit if one is available. Compared with the discount, the number is small; compared with the schedule, it is often the difference between a machine that contributes in month one and one that contributes in month three. Buyers who price it in advance negotiate from knowledge rather than from hope.
The last stage is where a used machine most often disappoints, not because the machine is worse but because the assumptions around it are. A new machine arrives in a crate designed for it, with lifting points documented, with an installation scope agreed, and with an engineer who has commissioned the same model before. A used machine arrives in whatever packaging the seller could arrange, with a lifting arrangement that somebody has to improvise, and with an installation scope that is often just delivery. The machine may be excellent and the installation still consume a month. That gap is the part of a new vs used JUKI pick and place machine purchase that budgets forget, and it is also the easiest part to close before the machine ships.
Three things reduce that risk, and all three are contractual rather than technical. Specify the packing method and who is responsible for it. Specify the rigging route and who surveys it. Specify the acceptance test and the pass criteria, and link payment to the result. Each of those is a paragraph in the order, and each converts a stage that is normally discovered into a stage that is planned.
A new machine ships in a crate engineered for its mass, its centre of gravity and its lifting points, usually with shock and humidity indication. A used machine may ship in a similar crate, in a re-used crate, or on a skid with wrapping, and the difference matters because internal components move under shock. The request is not unreasonable: ask what the packing specification is, whether the machine will be bolted to a pallet or suspended, and whether the original lifting points are documented and accessible.
Rigging is where the improvisation shows. Measure the route before the machine ships: the goods door, the corridor at its narrowest, the turns, the lift capacity and the height of any duct crossing the path. Photograph each restriction with a tape visible and send the set to the seller or rigger. A route survey costs an hour and prevents the most expensive delay in the process, which is a machine that arrives and cannot enter the building while the schedule runs.
Protection belongs in the same scope. Moving a machine on rollers concentrates load on a small area, and a coated floor, a joint or a tile will mark or crack under it. Specify who supplies and lays the protection, and confirm the same for levelling: a machine that is not level will place accurately in one region of the board and poorly in another, and the resulting offset pattern looks like a calibration fault rather than a floor fault. For a used machine, the levelling step is more important than for a new one, because the frame has had years of load and may not sit flat without shimming.
Installation scope decides whether the purchase is a product or a project. At one end the seller delivers to the door; at the other he rigs, levels, connects, commissions and proves the machine. The price difference is real and legitimate, and a quotation that does not state which model applies leaves the buyer to discover it on the day. For a used machine the default is usually the first model, which means the utility work, the connections and the commissioning are the buyer's, and the electrical and pneumatic preparation has to be scheduled before the machine lands.
The utility gap is often larger than buyers expect because a used machine may have been configured for different site conditions. Voltage and frequency settings, the air specification, the conveyor height and the handshake direction all have to match your line, and each is a task with an owner. The conveyor interface between SMT stations conventionally follows SMEMA signalling, which defines the board-available, machine-ready and transfer signals; somebody has to configure that handshake, and on a used machine it is rarely included.
The practical approach is a one-page installation schedule written before the machine ships: who unloads, who rigs, who levels, who supplies power and air at the machine position, who connects the conveyor interface, who commissions, and in what order. Each line has a name and a date, and the schedule is attached to the order. Plants that write this page report installations measured in days; plants that leave it to the day report installations measured in weeks, with the machine standing wrapped while an electrician is scheduled.
Acceptance for a used machine cannot be the same test as for a new one, because some parameters have already drifted and some documentation does not exist. What it can be is a defined set of measurements with pass criteria agreed in advance, taken on your board, at your site, before the balance is paid. The four that carry the most information are placement offset on a defined pattern, capability index on that sample, feeder index and pick reliability across several widths, and a full production run long enough to include a reel change.
The run matters more than the measurements in one respect: it exposes the interactions. A machine can pass a capability sample and still fail a shift, because the failure comes from a feeder that lifts tape at a certain fill level or a conveyor that misplaces a board at a certain width. Running several hundred boards, with a reel change and a nozzle check inside the run, is the closest a buyer can get to the first month of production without spending the month. Where the product requires joint acceptance after reflow, the visual criteria belong to the process standard and should be reported separately from the placement result.
Payment structure does most of the work here. A deposit on order, a payment on delivery, and the balance on a signed acceptance report converts the acceptance test from a formality into an incentive. Sellers resist this less than buyers expect, particularly when the criteria are objective and the report format is agreed in advance. The clauses to avoid are the ones that define acceptance as "the machine runs" or make the balance payable on shipment, because both remove the only leverage the buyer will ever have over a used asset.
Acceptance item | New machine | Used machine | Evidence to keep |
|---|---|---|---|
Identity check | Order against build sheet | Serial, build sheet and plate reconciled | Photographs of plate and sheet |
Configuration | Option list as ordered | Option list as found, gaps priced | Signed configuration record |
Placement accuracy | Factory test plus site verification | Witnessed sample, 30 boards minimum | Offset data and capability calculation |
Feeder package | Inventory as ordered | Serviceable count by width, sampled | Inventory table and sample results |
Software | Licence in buyer's name | Licence transfer confirmed, backup restored | Licence document and restore log |
Run test | Commissioning run | Several hundred boards with a reel change | Production log and defect record |
Payment link | Balance on acceptance | Balance on signed acceptance report | Signed report |
The first week of a used machine is a settling period, and planning it as one prevents the conclusion that the purchase was a mistake. Three things typically happen: consumables that were marginal fail under continuous load, settings that were tuned for the previous owner's product need re-tuning for yours, and the operators discover behaviours that differ from the machine they used before. None of these is a defect, and all of them consume time that should be scheduled rather than discovered.
A first-week kit reduces the friction. It includes the nozzle sizes that cover your component range, at least one spare feeder in the dominant width, the filters and seals the manual lists, and a small stock of the consumables that fail first. The kit is inexpensive compared with the cost of a stopped line and it converts the first failure from an event into a routine. Buying it with the machine is easier than buying it urgently afterwards, and the parts list gives the negotiation a concrete item to attach to.
The second element is a named person. Somebody in the plant should own the machine's first month: track the faults, decide what is a defect and what is a setting, and hold the acceptance conversation with the seller while the evidence is fresh. Without a named owner, first-week issues dissolve into anecdotes and the acceptance report gets signed because nobody can reconstruct what happened. With one, the same issues become a list that either the seller fixes or the price reflects, which is the outcome the whole framework is designed to produce.
Put the two options on the same page before you choose
If you send us your board sizes, your BOM line count and tape widths, the accuracy your product actually needs, and the used candidates you are considering with their serial numbers and any documentation you have, we will build the comparison on the same basis: annual cost of ownership rather than purchase price, the feeder package converted to usable positions, the calibration position stated as measured values or as an open item, and the installation and training work that a used machine adds. 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 own assumptions, and we will say plainly when the used route wins and when it does not. A machine that arrives three weeks late, or one that cannot hold capability on your pitch, costs more than the discount it carried, and the only way to see that in advance is to write both options in the same units.
It can be, but only after the discount survives the ownership costs. Compare annual cost rather than price: depreciation over the period you will keep the machine, plus reconditioning, plus the contribution margin lost to extra downtime. A used machine at roughly half the new price, kept for six years, often lands within a few percent of the new machine on an annual basis. The usual deciding factors are schedule, documentation and local support rather than the discount itself.
Reconcile three artifacts: the nameplate photograph, the build or configuration sheet, and the software version screen. The serial must appear on the plate and on the sheet, and the fitted assemblies must match what the sheet describes. Then walk the machine and confirm each option physically. A serial that appears on the plate but on no document suggests the paperwork belongs to another unit. Where documentation is missing, treat the year as inferred and price the spare-parts risk explicitly.
Four things: the machine serial, the date, the procedure and the measured results. A pass stamp alone supports no capability claim for your product. The measured offsets tell you whether the machine is centred or skewed and how much margin remains. Ask whether the vision system was included and which parts were recognized, because recognition is part-dependent. If the certificate is older than the service interval, treat the machine as uncalibrated and plan a verification run at acceptance instead.
Enough serviceable feeders, in your widths, to cover your BOM. Convert each line to 8 mm equivalents using the slot factor for its width, add a working reserve of ten to fifteen percent, and compare the total with the positions the bank can serve using the delivered inventory. A package whose feeders are concentrated in 8 mm will not run a BOM built on 16, 24 and 32 mm tape, whatever the total count says. Ask for an inventory table by width and type, then sample five feeders by running them.
Sometimes, and the answer is documentary rather than technical. Ask for the licence document, check whose name is on it, what scope it covers, and whether a transfer procedure exists and who administers it. Confirm that any dongle or activation item is in the crate, and that remote support or update subscriptions move with the machine. Make the transfer a condition of payment rather than a promise after shipment, because once the machine has left the seller's site your leverage is gone.
Run your own board type, at a quantity that exposes repeatability, and measure rather than observe. Thirty boards on a defined pattern, with the offsets recorded and a capability index calculated, answers the accuracy question. Add a feeder sample across the widths you use, a program built by your engineer, and a production run long enough to include a reel change. Then collect the document pack: configuration record, calibration data, licence, backup and the signed report.
Usually the utility work, the connections and the commissioning, because the default scope for a used sale is delivery. Assume you own the power and air preparation at the machine position, the conveyor handshake configuration, the levelling and the first-week support. Write a one-page schedule naming who does each task and when it happens, before the machine ships. Plants that plan this page report installations in days; plants that leave it to the day report weeks of a wrapped machine waiting for an electrician.