How to Select a Hanwha Chip Mounter Based on BOM and Component Range

Pillar selection guide · Published by I.C.T · For SMT process, production and project engineers

A chip mounter is not chosen from a brochure; it is chosen from the bill of materials. Every line in that document — the smallest chip, the finest-pitch IC, the largest BGA, the longest connector, the odd-form parts, the split between tape reels and trays — maps directly to a machine capability: which placement heads, which cameras, how many feeders, which board handling. Read the BOM methodically and the Hanwha range answers by itself; skim it and even an excellent machine becomes a standing yield investigation. This guide builds that method, working from component class to configuration against the models stocked in the Hanwha chip mounter catalog.

The logic runs in the order the Hanwha chip mounter BOM component feeder selection actually follows: micro chips first, because 01005-class packages set the accuracy floor nothing downstream can fix; then ICs, QFPs and BGAs, whose lead and ball pitches decide what the vision system must see; then connectors and odd-form parts, which push height, reach and tooling; then trays and feeders, whose counts shape the machine's physical configuration; and finally the translation of that matrix into a specific machine choice and a defensible decision. Along the way the guide corrects the most common reading error in Hanwha datasheets — what "0402 (01005)" really means — and shows how the same BOM line can silently disqualify a machine that looks identical on paper.

Table of Contents

1. The BOM is the specification: from component mix to machine class

2. Micro chips and 01005: the accuracy floor that chooses the platform

3. ICs, QFP and BGA: what the cameras must see

4. Connectors and odd-form parts: height, reach and special tooling

5. Trays and feeders: the configuration that follows the count

6. From BOM matrix to machine choice

7. From the matrix to a defensible decision

8. Frequently asked questions

The BOM is the specification: from component mix to machine class

Most mounter selection disputes dissolve the moment both parties look at the same BOM summary, because the document contains everything the machine decision needs. A chip mounter does not place "components in general"; it places a specific distribution of packages — thousands of passive chips, dozens of ICs, a handful of BGAs, some connectors, an odd-form part or two — and that distribution is what the head count, camera set and feeder configuration must serve. Two factories with identical budgets and identical board counts can need completely different machines, purely because one BOM bottoming out at 0201 chips and the other at 01005 sit on different sides of an accuracy boundary. Reading the BOM as the specification, rather than as an attachment to the RFQ, is the discipline this guide builds on.

That reading has a fixed structure, and the Hanwha chip mounter BOM component feeder selection works through it class by class. Micro chips define the accuracy floor and therefore the platform generation. IC classes — QFPs, BGAs, CSPs — define what the vision system must recognize and at what pitch. Connectors and odd-form parts define height limits, camera reach and special tooling. Trays versus tape define feeder real estate and pick paths. Each of the next sections takes one class and traces it to the machine attribute it decides; this section sets up the method — how to extract the extreme cases, why the extremes matter more than the averages, and what a BOM matrix looks like when it is ready to drive a purchase.

Extracting the extreme cases that actually decide

The BOM read is not a census; it is a search for the handful of lines that constrain the machine. For every product family the line will run, record five extremes: the smallest chip package, the finest IC lead pitch, the largest body (BGA or connector), the tallest component, and the longest connector or odd-form part. Add two volume facts — placements per board by class, and the split between tape-fed and tray-fed line items — and the entire selection-relevant content of the BOM fits on one page. Everything else on that document, however many thousands of lines it runs, is routine volume that any machine in the considered class will absorb.

The reason extremes dominate is failure asymmetry: a machine that handles 99% of a BOM well and fails on the smallest chip or the longest connector produces boards that cannot ship, and no surplus capability elsewhere compensates. Hanwha's own model structure acknowledges this — the range divides precisely at the boundaries these extremes define, from the SM471Plus's chip-shooter focus through the SM481Plus's extended IC reach to the SM482Plus's wide-body tolerance — and the datasheet numbers only become meaningful once the extremes name which rows to read. A buyer who has not extracted them is comparing machines against no reference at all, which is how specifications end up purchased rather than used.

Why the averages mislead and the tails decide

BOM averages flatter the wrong machine. The mean component on most boards is a chip — and chips are the easiest thing any mounter in this class places, so a machine chosen to be excellent at the average is excellent at nothing difficult. The tails are different: the 0.5% of placements at the smallest packages and largest bodies are where yield leaks, where rework costs concentrate, and where the machine classes genuinely separate. A 01005 chip and a 25 mm connector each appear a handful of times per board, yet each can disqualify half the catalog, because the physics of placing them — nozzle scale, vision field, Z-travel, holding force — is not something the other 99.5% of placements ever exercise.

The disciplined posture is therefore to score every candidate machine against the tails first and the average second, in that order, and to treat any capability that serves only the average as a commodity. This inverts the way datasheets are usually read — speed first, extremes later — and it is the inversion that separates selections that hold for five years from selections that trigger a capacity review within two. When the tails are mapped, the machine conversation shifts from "how fast" to "how well at exactly these lines", and that is a question the Hanwha range answers distinctly rather than uniformly.

The BOM matrix: one page that drives the whole purchase

The output of the BOM read is a matrix: component class in rows, the machine attributes each class stresses in columns, and your project's extreme case filled in against each row. Micro chips stress accuracy and nozzle inventory. Fine-pitch QFPs and BGAs stress vision resolution and flight-path control. Large ICs and connectors stress camera field and Z-height. Odd-form parts stress tooling and force handling. Trays stress station configuration; tape reels stress feeder count and width mix. Filling the matrix takes an hour with a script and a BOM export, and it converts every subsequent supplier conversation from "what do you need?" — which produces vague answers — to "does this configuration meet this matrix?", which produces commitments.

The matrix also protects the purchase across time. BOMs drift: next year's product adds a smaller package or a taller module, and a matrix recorded at purchase time makes that drift visible the day the new BOM arrives, rather than the day the line starts rejecting boards. Factories that maintain the matrix as a living document — one row added per new component class — turn it into the standing contract between product engineering and equipment, and the standing agenda for every capacity review. The sections that follow populate it column by column, class by class, against the Hanwha range's actual numbers.

What the Hanwha range encodes in its model structure

Read correctly, the Hanwha catalog is itself a BOM-decision tree. The SM471Plus, with its dual-gantry speed and chip-oriented window, encodes the answer for chip-dense high-volume boards. The SM481Plus widens the IC window and adds the stage-vision option, encoding the answer for mixed boards with larger devices. The SM482Plus trades speed for the widest body and connector tolerance in the series, encoding the answer for industrial and lighting BOMs. The SM485 encodes insertion work. On the DECAN side, the S1/S2/F2/L2 ladder re-answers the same questions a generation later: micro chips to 03015, dual-gantry speed, and multi-function reach at the L2. The model structure is the vendor's own statement of where BOM boundaries fall — and it means the matrix you build will usually point at one or two models, not the whole catalog.

That pointing is the practical payoff of the whole method. A buyer with a completed matrix does not audition machines; the matrix eliminates machines, and the surviving one or two candidates get demonstrations designed around the exact rows that made them survive. Table 1 below summarizes the mapping from component class to machine attribute that the matrix uses — the skeleton the next five sections put flesh on, one class at a time.

Micro chips and 01005: the accuracy floor that chooses the platform

The smallest chip in the BOM is the single most decisive line in the document, and it is also the line most often misread — because of a notation trap built into the industry's naming. Package designators come in two systems: the inch code (0402, 0201, 01005) and the metric code (1005, 0603, 0402), and a datasheet that says "0402" without units usually means the inch code — a 1.0 by 0.5 mm chip — while Hanwha's component windows, written as 0402 (01005 inch), actually mean the metric 0402 body: 0.4 by 0.2 mm, the chip the inch system calls 01005. One digit of ambiguity, an entire package class of difference. Buyers who resolve the notation first read the Hanwha range correctly; buyers who do not routinely believe the SM series tops out at a package class it in fact handles.

Three sub-sections work the micro-chip dimension: what 01005 actually demands of a machine, where the SM and DECAN generations draw their floors, and what the micro-chip decision does to nozzles, feeders and the rest of the configuration. The section's practical claim is simple: the smallest chip in the BOM is not one constraint among many — it is the constraint that selects the platform generation, before speed, feeders or price get a vote.

What 01005 actually demands from a machine

A 01005 chip — 0.4 by 0.2 mm, weighing a few milligrams — stresses every subsystem the placement path touches, which is why it functions as a platform marker rather than a line item. The nozzle must be scaled to pick a body smaller than most dust particles without vacuum instability; the feeder must present the part with pitch-level precision, because a mis-presented 01005 is already a defect at the pick; the vision system must resolve and center a body whose pad geometry is measured in tens of microns; and the placement pulse must set it down without the bounce or airflow disturbance that would send a part this light wandering. None of these demands is exotic on a machine designed for the class; all of them are failure modes on a machine designed without it.

The consequence for selection is that 01005 capability is binary in practice. A machine either holds the accuracy class and the pickup reliability the package requires — and places it in production, not just on a demo board — or it does not, and there is no partial credit to be had from being "close." The BOM matrix therefore records the smallest package with its system (inch or metric) spelled out, and any candidate machine's answer is verified against that exact package, on that exact feeder, before any other dimension of the comparison is discussed. Factories that skip this verification usually rediscover it at first-pass-yield review, at a price several orders of magnitude above the demonstration that would have caught it.

Where the SM and DECAN floors actually sit

With the notation resolved, the two platform generations draw their floors clearly. The SM series' flying-vision windows open at 0402 (01005 inch) — the 0.4 by 0.2 mm chip — across the SM471Plus, SM481Plus and SM482Plus, with the SM482Plus's flying window starting a class larger on its standard configuration. So a BOM bottoming out at 01005 is inside the SM platform's envelope: the series that buyers often assume is "too old" for micro work in fact handles the package its windows name, with chip accuracy in the ±40 micron class and, on the SM481Plus and SM482Plus, stage-vision options tightening IC-class precision to ±30 microns.

The DECAN generation extends the floor one class further: its speed models place 03015 — 0.3 by 0.15 mm, a full step below 01005 — while holding chip accuracy around ±28 microns and IC accuracy near ±25 microns, and every model in the family inherits the new platform's vision pipeline. The decision logic for a BOM follows directly: if the smallest specified package across the product roadmap is 01005 or larger, both generations qualify and the selection moves to the next rows of the matrix; if the roadmap names 03015, the DECAN platform is the only answer in the range and the comparison ends there. The floor is that clean, which is precisely why it is read first.

What the micro-chip decision does to the configuration

Choosing a platform for 01005 work cascades through the configuration in three concrete ways. Nozzle inventory leads: the machine must carry and automatically change the fine-pitch nozzle set the package demands, and the head's spindle layout must accommodate that set alongside the nozzles the rest of the BOM needs — on the ten-spindle heads of both generations, nozzle mix is planned per product, not per machine. Feeder precision follows: 01005 arrives on narrow tape, and the electric feeder option, with its controlled feed indexing, is the presentation standard the package expects; pneumatic feeders remain viable for the coarser majority of the BOM, but the fine-pitch lines belong on electric drive.

Verification completes the cascade. A micro-chip BOM justifies a demonstration protocol the buyer scripts themselves: the exact 01005 part from an actual production reel, on the feeder type the line will run, placed across a full panel — not a supplier-chosen demo package under demo conditions. The same protocol naturally extends upward through the BOM, because the machine that proves the hardest row usually proves the easy ones incidentally. This is the pattern the rest of the guide repeats: the extreme case scripts the demonstration, and the demonstration converts the matrix from analysis into evidence. A SM471 dual-gantry chip shooter, for instance, earns its place on a chip-dense BOM by exactly this kind of run — thousands of small chips per panel, in production rhythm, with yield counted rather than asserted.

ICs, QFP and BGA: what the cameras must see

Where chips stress accuracy, IC classes stress sight. A QFP with 0.3 mm lead pitch, a BGA with 0.4 mm ball pitch, a CSP re-balled in-house — each demands that the machine's vision system find the part, characterize its leads or balls, and correct the placement in the milliseconds the production rhythm allows. The Hanwha range answers this demand with a two-camera architecture whose division of labor is the key to reading IC capability honestly: flying vision, riding the head, handles the high-volume majority in motion, while stage vision — a precision camera with dedicated lighting on the machine frame — handles the packages whose geometry exceeds what a moving camera can resolve. The speed and pitch figures describing that architecture are standardized under the IPC-9850 measurement discipline, which is what makes them comparable across models — and it is that comparability the Hanwha chip mounter BOM component feeder selection leans on when it maps IC rows to camera sets. The figure below shows the machines that carry this architecture in their production setting.

Four sub-sections unpack the IC dimension: what flying vision resolves and where its ceiling sits; what the stage camera adds and which BOM rows route there; how the pitch figures — 0.3, 0.4, 0.5, 0.65 — translate across the range; and how to read the camera decision against your own IC census rather than against datasheet superlatives.

Flying vision: the workhorse and its ceiling

Flying vision is the reason modern placement is fast: the camera recognizes the component while the head travels, so recognition adds no dwell to the cycle. On the SM471Plus it verifies chips up to roughly 14 mm IC bodies at 0.4 mm lead pitch and BGA ball pitch down to 0.4 mm; the SM481Plus extends the flying window to about 16 mm; the DECAN speed models carry the same in-flight philosophy with a generation-tighter accuracy envelope. For the IC majority of most BOMs — SOPs, QFPs at conventional pitch, standard BGAs — flying vision is not merely adequate; it is the architecture the whole platform is optimized around, and any machine comparison that starts elsewhere starts wrong.

The ceiling is optical, not electronic. A camera in motion works with limited integration time and a compact field of view, and as lead pitch narrows below roughly 0.4 mm, or as body size climbs past the flying window's edge, the confidence of recognition degrades before accuracy does — the system still centers the part, but with less margin against lead variation, reflectivity and tombstone geometry. The BOM rows that approach this ceiling are easy to spot: fine-pitch QFPs below 0.4 mm, large cosmetic-critical connectors, tall reflective shields. Those rows route to the other camera, and the buyer's task is knowing which rows those are — a census, not a judgment call.

Stage vision: the precision station for the hard rows

Stage vision trades speed for optical authority. The component pauses at a precision camera with controlled lighting and a larger field, and the system resolves what motion blurred: lead pitch down to 0.3 mm on the SM481Plus and SM482Plus stage options, IC bodies to roughly 32 mm standard and 42 mm with the larger field of view, BGA balls at 0.4 to 0.5 mm pitch, and — on the SM482Plus — bodies to 55 mm in the extended-field mode along with connectors to roughly 100 mm. The DECAN generation folds some of this reach into its upgraded flying system, but the structural lesson of the SM architecture stands across the range: the stage camera exists because some BOM rows need controlled light and a big field, and no amount of motion optimization changes that.

The selection translation is a count, not a philosophy. From the BOM matrix, list every IC row whose lead pitch, body size or reflectivity approaches the flying ceiling; if that list is empty, stage vision is an option you do not need to price, and the chip-shooter class serves the BOM intact. If the list runs to a handful of rows on every board, the stage-equipped models — SM481Plus and SM482Plus with their options, or the DECAN multi-function tier — become the natural configuration, and the speed cost of routing the hard rows through the stage station is a known, small quantity on boards whose IC count is modest. Buyers who make this count before the demonstration buy the right camera the first time; buyers who skip it buy a specification they cannot connect to their own boards.

Reading the pitch figures across the range

The pitch numbers are where IC capability becomes concrete, and they deserve to be read in a table rather than in prose. Lead pitch — the spacing of QFP and connector leads — and ball pitch — the spacing of BGA and CSP spheres — define what the vision system must discriminate; the Hanwha range's figures step down through the model ladder in a pattern the BOM census maps onto directly. The 0.4 mm lead pitch of the SM471Plus's flying window covers the conventional-pitch IC majority; the 0.3 mm lead-pitch capability of the stage-vision options on the SM481Plus and SM482Plus covers fine-pitch QFPs; the ball-pitch figures run from 0.4 mm on the flying systems through 0.5 mm and 1.0 mm classes on the stage configurations. The table below assembles the picture for the SM series; the DECAN tier holds the same structure with a tighter accuracy envelope throughout.

Two reading rules keep the table honest. First, pitch figures describe the vision system's discrimination, not the machine's yield: a 0.3 mm-pitch QFP placed with insufficient correction fails at reflow, not at recognition, so the figure must be paired with the accuracy class of the platform and verified on the actual package. Second, the figures are configuration-dependent — the stage camera is an option on some models and standard on others — so the buyer confirms which camera set the quoted configuration carries, in writing, per model. Both rules point to the same practice: the IC census names the packages; the demonstration proves them; the table only organizes the conversation.

From IC census to camera configuration

The census-to-configuration translation is mechanical once the rows are known. A BOM whose ICs top out at conventional-pitch QFPs and standard BGAs is fully served by flying vision on any platform in the range, and camera money is better spent elsewhere. A BOM carrying fine-pitch QFPs below 0.4 mm routes those rows to a stage-equipped configuration, and the model choice narrows to the SM481Plus or SM482Plus with the stage option — or up to the DECAN tier if the micro-chip row already forced that platform. A BGA-heavy BOM — server boards, high-pin-count FPGAs — adds the ball-pitch rows to the same logic and, in production, often justifies the upward-vision verification options that inspect ball arrays after pickup, catching co-planarity problems before they become reflow escapes.

The configuration question has a cost side worth pricing explicitly, because cameras are where "option" pricing concentrates. A stage camera adds acquisition cost, adds per-component cycle time on the rows that use it, and adds a calibration and maintenance item to the schedule — all three proportionate to how many BOM rows actually need it. That is why the census matters: three hard rows per thousand placements price stage vision one way; thirty do, another. The buyer who knows the count negotiates the configuration knowing exactly what it is for, which is the difference between an optioned machine and an over-optioned one — and the bridge into the next section, where the BOM's largest and strangest rows make their own demands on height, reach and tooling.

Connectors and odd-form parts: height, reach and special tooling

At the large and irregular end of the BOM, the placement problem changes character: chips and ICs are about precision, while connectors and odd-form parts are about reach — how big a body the cameras can verify, how tall a component the head can clear, how an awkward shape is held and released without tipping or skewing. These are also the rows where Hanwha's model ladder spreads widest, because the platform family deliberately spans from a chip shooter that tops out near 14 mm to machines that verify 55 mm bodies and place connectors approaching the 100 mm mark. A BOM with a single battery holder or a shield can does not just add difficulty; it can redraw the machine shortlist entirely.

Three sub-sections cover the dimension: what body size and height figures actually govern, and how the range's windows step up; which special tooling — grippers, force control, upward vision — the odd-form rows pull into the configuration; and how to price the odd-form decision honestly, including when the right answer is a different machine rather than more options.

Body size, height and the stepping windows

Two figures govern the large-part rows, and both must be read per camera set. Body size — the largest square or rectangular part the system can verify — steps up through the range: roughly 14 mm on the flying windows of the SM471Plus, 16 mm on the SM481Plus flying, 32 to 42 mm on the stage options, 55 mm on the SM482Plus's extended-field mode, and comparable or wider reach on the DECAN multi-function tier. Height — how tall a component clears under the head — follows its own ladder, from the SM471Plus's 12 mm through the 15 mm class on the stage-equipped models up to the 25-plus-millimeter territory of the machines built for magnetics and modules. The BOM matrix records the extreme body and the extreme height separately, because different rows produce them: a large flat connector stresses field of view, while a tall electrolytic or transformer stresses Z-clearance, and a machine can pass one and fail the other.

The connector rows deserve their own note because they combine both stresses and add a third: length. Terminal blocks, header strips and battery holders are verified across their full length for lead alignment, and the Hanwha windows reflect that — connectors to roughly 75 mm on the SM481Plus's stage configuration and to about 100 mm on the SM482Plus's extended-field stage, with the wide-body tier covering similar territory on the DECAN side. A BOM whose longest connector sits just inside a window is a demonstration candidate, not a datasheet certainty: reflective housings and keyed bodies are exactly the parts where recognition confidence, not quoted reach, decides the outcome.

Special tooling the odd-form rows pull in

Odd-form content adds hardware beyond cameras. Gripper nozzles hold irregular bodies that vacuum alone cannot secure; force-controlled placement — the SM485's specialty, with its dedicated fourth-plus-one spindle — sets pin-in-paste and insertion parts at a controlled force instead of a fixed stroke, protecting both the part and the paste; upward vision inspects fine-pitch leads and BGA ball arrays from below, catching co-planarity faults before they reach reflow; and tray handling brings its own stations into the configuration, covered in the next section. Each of these is a configuration decision triggered by specific BOM rows, which is why the matrix asks for odd-form extremes explicitly rather than folding them into "other."

The honest engineering question for every odd-form row is whether automated placement is the right answer at all. Pin-in-paste content that fits paste-and-reflow physics automates beautifully and justifies the force-controlled head; through-hole hardware outside those physics may still belong on a selective or manual station downstream, and forcing it onto the mounter adds tooling cost and cycle time for marginal gain. The BOM census frames that decision with numbers — count and dimensions of the odd-form rows, per board and per year — and the machine conversation follows the numbers rather than the reflex to automate everything. Factories that run this analysis usually discover their odd-form content splits: a core that automates economically, and a tail that belongs downstream.

Pricing the odd-form decision honestly

When the odd-form rows do justify automation, the pricing is a configuration exercise. The SM482 multi-function SMT mounter represents the SM platform's answer — wide-body stage vision, the extended connector reach, and the 1,200 mm board options for long industrial panels — while the DECAN range answers the same BOM shape through the Hanwha DECAN L2 multi-function mounter, whose two six-spindle gantries place wide and long components at 56,000 CPH with LED-optimized recognition. Between those poles sit the incremental options — grippers, force control, upward vision — each tied to a BOM row, each carrying acquisition and maintenance cost.

The selection rule that keeps this pricing honest is per-row justification: every odd-form-driven option traces to a named BOM line with a volume, and the option's cost is weighed against the cost of the alternative (downstream station, manual placement, design change) for that line's volume alone. Rows with real volume justify real hardware; rare rows justify engineering review — sometimes the cheapest configuration change is on the product side, moving one component one step down the BOM. This is also where the odd-form rows intersect the platform decision made in section two: a BOM whose micro-chip floor forced the DECAN platform must satisfy its odd-form rows within that platform's tooling range, and if a 100-millimeter connector exists on the same board as a 03015 chip, the honest answer may be two machines — the split-fleet strategy the closing sections return to.

Trays and feeders: the configuration that follows the count

Everything so far has decided what the machine can place; trays and feeders decide how much of it the machine can hold, and that is a configuration question with physical consequences — station count, pick paths, machine width and changeover behavior. The Hanwha platforms standardize a 120-position eight-millimeter feeder capacity across the main models, with docking-cart systems that make full setups swappable, and beyond that standard the configuration branches: narrow tape for micro chips, wide tape for large bodies, tray stations for the parts that do not come on reels at all, stick feeders for the tail of the BOM that does. Supply mix is the quiet half of the Hanwha chip mounter BOM component feeder selection — invisible in datasheets, decisive on the floor — and this section gives it the same matrix discipline the component classes received. The figure below shows the configuration in place, feeders loaded, the way it looks when the count has been done right.

Four sub-sections cover the dimension: counting the tape requirement correctly, including width mix; configuring tray stations and understanding what they cost in pick-path time; the stick and specialty feeders that absorb the BOM's tail; and the setup-and-changeover arithmetic that turns feeder counts into productive hours.

Counting tape feeders: totals, widths and the family peak

The tape count starts deceptively simply — number of distinct tape-fed line items — and becomes real through three corrections. Width correction: an 8 mm feeder position is not fungible with a 24 mm one, so the count segments by tape width, and a body-hungry BOM can exhaust wide-tape positions long before its headline count suggests. Duplication correction: two products sharing a feeder load simultaneously, but two products alternating on one machine do not, so the governing number is the peak concurrent requirement across the product families that change over fastest, typically 30 to 50 percent above any single board's count. Growth correction: whatever the census says today, the roadmap adds packages, and feeder positions are the first thing a growing BOM exhausts.

The Hanwha platforms' 120-position standard absorbs most single-machine censuses comfortably, which is exactly why the count is still worth doing: the BOMs that exceed it do so quietly, on wide tape and duplicated families, and they surface as mid-run feeder swaps — the most expensive kind of interruption a placement line suffers. The count also drives the electric-versus-pneumatic feeder mix: electric drive for the fine-pitch and micro-chip rows where presentation precision matters, pneumatic for the coarse majority, with the split recorded per family rather than per machine. A feeder census page that ends with a peak-concurrent figure and a width-segmented table is the deliverable this sub-section promises, and it feeds the machine configuration conversation directly.

Tray stations: configuration and pick-path cost

Tray-fed components — large ICs, BGAs shipped in JEDEC trays, connectors, modules — enter the machine through dedicated tray stations, and their configuration follows the same census logic as tape. The count question is how many distinct tray parts per board and how many tray stations the machine offers; the Hanwha configurations place tray stations at the machine's rear or side, and each station holds a defined matrix of tray sizes. A BOM with two or three tray parts runs happily on a single station; a BGA-heavy server board or a module-rich industrial product can fill several, and the machine's tray capacity becomes a real selection row rather than a footnote.

The cost side of trays is time, not space. Every tray pickup sends the head away from the feeder bank and back — a longer flight than any tape pick — so boards heavy in tray parts run below their tape-only rate even on a perfectly configured machine. The planning response is arithmetic, not anxiety: weight the tray rows in the realistic-rate calculation, compare the resulting boards-per-shift figure against the schedule, and let the result decide whether the configuration needs another tray station, a different machine, or a supply-chain change that moves a tray part onto tape. Buyers who price tray time into the throughput model are never surprised by it in production; buyers who treat trays as free capacity rediscover the cost on the first capacity review.

Stick and specialty feeders: absorbing the BOM tail

Beyond tape and trays, the BOM's tail arrives in other packaging, and the platform's specialty feeders absorb it. Stick feeders handle tubed components — the packaging of choice for some connectors and odd-forms in lower volumes — and occupy feeder positions at their own widths. Specialty presentations cover the rest: odd-form tooling, the force-controlled head for insertion rows, and the vendor-specific fixtures that particular parts sometimes demand. The census discipline extends to these rows the same way: name the packaging, count the parts, and let the count decide whether a stick feeder position is configuration or clutter.

The tail rows also reward a supply-chain look before they reward a hardware look. A connector available on both tape and stick, at the same piece price, converts a specialty-feeder requirement into a standard feeder position — a configuration simplification that costs nothing at source. Factories that review the odd packaging rows with their distributors during selection, rather than after commissioning, routinely shrink the specialty hardware list, and the savings fund feeder positions where the BOM actually needs them. The principle generalizes: the feeder configuration is negotiable at the packaging level in ways it is not at the machine level, and the negotiation happens earliest when the census makes it visible.

From feeder counts to setup and changeover behavior

The feeder census becomes operational through the docking-cart system the platforms share: feeders load onto carts off-line, carts verify and stage the next product's setup, and the changeover becomes a cart swap rather than a feeding session at the rail. Configuration questions follow: how many carts the purchase includes, whether the verification scanner is included per cart or per machine, and how the software ties cart contents to product recipes — all standard items that nonetheless appear as options on quotes, and all cheaper to specify than to retrofit. For high-mix operations, the cart count is as consequential as the feeder count, because setups queue on carts, and a machine with more feeder positions than staged carts runs changeovers at half its potential tempo.

The last arithmetic is time. Each product family's setup — its feeder map, its tray stations, its nozzle assignments — consumes real minutes at changeover, and the annual total is the census multiplied by the schedule's switching frequency. The Hanwha platforms' recipe management compresses that time, but the census determines what there is to compress: a peak-concurrent requirement that fits the 120 positions with margin changes over in one cart swap, while one that forces mid-map feeder movement changes over in several. When the changeover arithmetic lands close to a boundary — one cart or two, one machine or a second — that is the moment to run the demonstration with the actual census, timing the setup end to end. The result feeds the machine-choice translation of the next section, where every row assembled so far finally points at a model.

From BOM matrix to machine choice

Every dimension so far has produced rows of the matrix; this section reads the matrix and names the machines. The translation is deliberately mechanical, because its value lies in being repeatable: micro-chip floor to platform generation, IC pitch rows to camera set, body and height extremes to model tier, feeder census to configuration, throughput requirement to machine count. Run in order, the sequence usually eliminates all but one or two candidates before any commercial factor enters — which is exactly how the Hanwha chip mounter BOM component feeder selection is meant to conclude: with a shortlist the BOM built, not a preference the brochure sold.

Three sub-sections do the translation. The first walks the elimination logic step by step against the range's actual boundaries. The second works the throughput overlay — the layer that decides not which machine can place the BOM but how many of it the schedule needs. The third handles the outcome the matrix occasionally forces: a BOM that no single machine in the range serves completely, and the split-configuration answer that follows.

The elimination sequence, step by step

Step one reads the smallest chip. A BOM specifying 03015 routes directly to the DECAN platform and no further discussion; a BOM bottoming at 01005 qualifies both generations and moves on. Step two reads the IC rows: fine-pitch QFPs below 0.4 mm lead pitch or fine-ball BGAs route the candidate toward stage-vision-equipped configurations — the SM481Plus or SM482Plus with their stage options on the SM side, the corresponding capability in the DECAN tier. Step three reads body and height extremes: anything past the flying windows routes to the wide-body tier, and connector length becomes the tiebreaker between the SM482Plus-class and L2-class answers. Step four applies the feeder census: peak concurrent requirement against the 120-position standard, width mix against the banks, tray count against stations.

By step four, the catalog has typically narrowed to two candidates — for example, a mixed industrial BOM with 01005 chips, 0.3 mm QFPs, a 90 mm connector and 85 feeder positions points at the SM482Plus-class answer on the SM side and the L2-class answer on the DECAN side, with every other model eliminated on a named row. Those two then face the throughput overlay, and the DECAN F2 dual-gantry chip mounter enters the conversation exactly when the schedule, not the BOM, starts binding — the machine that keeps wide capability while lifting the rate ceiling. Buyers who run the sequence in writing produce a shortlist with reasons attached, and reasons are what survive the inevitable procurement review.

The throughput overlay and machine count

Capability decided which machines survive; volume decides how many. The overlay converts each surviving candidate's rated optimum into a realistic rate — the efficiency factor from your board's geometry, applied to the rated figure — and divides the schedule's placements per shift into the result. Chip-dense boards run 0.5 to 0.7 of rated on machines in this class; mixed boards with stage-vision rows and tray content run 0.3 to 0.5. The output is a machine count, and it sometimes revises the shortlist: a BOM whose only qualifying machine needs a partner unit to hold takt may be better served by a slightly wider-capability single machine at a higher rate, which is precisely the trade the DECAN speed tier exists to offer.

The overlay also prices the stage-camera rows correctly at last. Stage-vision processing costs cycle time on the rows that use it, and the realistic-rate calculation is where that cost appears — a few percent on an IC-light board, materially more on a board that routes half its placements through the precision station. The platform-level discussion of the two Hanwha generations, laid out in our SM versus DECAN platform comparison, supplies the rated figures and architecture context for this overlay; the BOM supplies the factor. Together they answer the question buyers most often ask late and should ask early: not "which machine" but "which machine, at what count, at what margin over the schedule."

When the matrix splits: mixed configurations

Some BOMs refuse to fit one machine, and the matrix says so plainly: a 03015 chip row and a 100 mm connector row on the same product pull in different directions, since the platform that serves the former does not offer the extreme reach of the latter in a single configuration. The honest answer is a split configuration — the micro-chip stream on a DECAN machine, the large-body stream on an SM482Plus-class unit, either as two machines on one line or as parallel lines per product family — and the matrix makes that answer visible before purchase rather than after a failed demonstration. Splitting is not a failure of the method; it is the method correctly reporting that the BOM spans two capability regimes.

The split decision carries its own arithmetic, and the project-first method for sizing it line by line — which stream gets which machine, in what order, with what feeder strategy — is exactly the workflow of our BOM-driven Hanwha mounter selection framework, which treats the whole-line question this guide's matrix feeds into. The short version: total the two configurations' cost against the single-machine alternatives including any downstream manual station the single machine would force, and let the numbers confirm or reject the split. Factories that run this comparison on genuinely mixed BOMs usually find the split pays for itself in yield and changeover hours within the first year — but only when the split was designed from the census rather than improvised at commissioning.

From the matrix to a defensible decision

The matrix has named the machines; the last mile is turning that name into a signed, defensible purchase, and the discipline that closes the gap is evidence design. Every row of the matrix implies a demonstration, every demonstration implies a measured result, and every measured result belongs in the proposal file against the row that required it. A selection defended this way survives procurement scrutiny, budget reviews and the passage of engineers; a selection defended by datasheet comparison survives none of the three. That is the closing discipline of the Hanwha chip mounter BOM component feeder selection: evidence before signature. This section assembles the closing protocol: what to demonstrate, how to read the five-year numbers, and the checklist that compresses the entire method into one page.

Three sub-sections complete the guide. The demonstration protocol converts matrix rows into test scripts, including the two or three that matter most for any Hanwha selection: the smallest chip at production rhythm, the hardest IC at its true pitch, the largest connector on its true reel. The cost view frames the configured price against five years of running, so the option-laden configuration is judged on cost per good board rather than on line items. The checklist closes the method — the same one-page structure the earlier sections built, ready to carry into the supplier meeting.

The demonstration protocol your matrix writes

A Hanwha demonstration earns its day when the buyer scripts it, and the matrix writes the script. Bring production reels for the three hardest rows — the smallest chip, the finest-pitch IC, the largest or longest connector — and require each candidate machine to run them on the exact feeder and camera configuration quoted, across a full panel, at production cadence, with yield counted. Add one changeover: the next product's setup performed end to end with the census's real feeder map, timed from last good board to first. These four runs — chip, IC, connector, changeover — separate machines faster than any datasheet column, because they test the rows the BOM actually stresses, on the parts the BOM actually ships.

The protocol also disciplines the vendor conversation in a way price negotiation cannot. A supplier who engages the scripted runs enthusiastically is telling you how the machine behaves in year one; a supplier who steers toward the standard demo board is telling you where the margins live. And the protocol travels: the same scripts re-run at commissioning verify that the delivered configuration matches the demonstrated one, closing the loop between selection and installation. Factories that institutionalize the scripted demonstration — same parts, same metrics, every machine purchase — build a comparison record across vendors and years that no single datasheet exercise approaches.

Five-year cost of the configured machine

The configured price is the beginning of the money conversation, not the end. The five-year view adds the streams the configuration implies: consumables scaled to the throughput model, preventive maintenance per the vendor's schedule, the spares stock the optioned hardware adds, service response as priced risk, and the efficiency value of the realistic-rate calculation — the machine that produces more good boards per machine-hour is amortizing differently than its price suggests. Feeders and carts belong in this view at their real configuration counts, not their bundle counts, and the stage-camera options earn their place only if the IC census justifies them at volume.

Run honestly, the five-year model frequently reorders the shortlist the capability matrix produced — usually toward the configuration that matches capability to census without surplus, and away from the option-maximal quote. It also exposes the trade the mixed-BOM buyer faces explicitly: the split configuration's second machine against the single machine's downstream workarounds, priced over the same horizon. Neither answer is universal; the arithmetic decides, and it needs the census to run. That dependence is the final argument for the matrix method: every financial question in mounter selection reduces to a BOM question the census has already answered.

The one-page checklist before you sign

The method compresses to seven lines, and the purchase is ready to sign when each line has an answer written against it. The table below is that page. It reads top to bottom in decision order — the same order the guide followed — and its seventh line, the demonstration record, is what converts the first six from analysis into evidence. A proposal that arrives without answers to these rows is not a proposal; it is a brochure with a price.

Run your BOM matrix against the Hanwha range with I.C.T

The matrix method in this guide turns chip mounter selection from a datasheet debate into an evidence chain, and building that chain with a real BOM is a conversation our engineers run every week. Send your component-range extremes, feeder census and board outlines, and we will map them across the SM471Plus, SM481Plus, SM482Plus and the DECAN S1, F2, S2 and L2 — returning a configuration proposal with the camera set, feeder plan and machine count the matrix justifies, plus the scripted demonstration protocol for your hardest three rows. Whether the answer is one machine, a stage-optioned configuration or a deliberate split fleet, it arrives as evidence against your own production records, not as a brochure summary. Browse the range in the Hanwha chip mounter catalog, or share your BOM summary to schedule a demonstration on the exact packages that will decide your purchase — your smallest chip, your finest-pitch IC and your longest connector, run in production rhythm before any commitment is requested.

Frequently asked questions

What does "0402 (01005)" mean in Hanwha chip mounter specifications?

It resolves the industry's dual naming: the inch code 01005 and the metric code 0402 describe the same 0.4 by 0.2 mm chip. Hanwha datasheets write both together — 0402 (01005 inch) — meaning the machine's component window opens at the 01005 micro chip, not at the larger inch-code 0402 (1.0 by 0.5 mm). Confirm the naming system on any datasheet you read, because one digit of ambiguity shifts the smallest supported package by a full class and, with it, the platform decision.

Which Hanwha models place 01005 chips?

All three main SM Plus models — SM471Plus, SM481Plus and SM482Plus — open their flying-vision windows at 01005, and the entire DECAN family qualifies as well, extending to the smaller 03015 class on its speed models. The practical distinction is downstream: at 01005 both generations compete, while a BOM specifying 03015 routes exclusively to the DECAN platform. Verify the specific package on the specific feeder in a scripted demonstration either way, because 01005 capability is binary in production.

When do I need stage vision instead of flying vision?

Stage vision earns its place when the IC census contains rows flying vision cannot resolve confidently: lead pitches below roughly 0.4 mm, IC bodies beyond the flying window (about 14–16 mm), large reflective components, or BGA verification that benefits from controlled lighting and a larger field. Count those rows per board first — a handful justifies pricing the stage option on models like the SM481Plus or SM482Plus; none means flying vision alone serves the BOM and the option money is better spent elsewhere.

How do I calculate the feeder positions my BOM needs?

Start from distinct tape-fed line items, then apply three corrections: segment by tape width, because 8 mm and 24 mm positions are not fungible; convert to peak concurrent demand across the product families that change over fastest, typically 30 to 50 percent above any single board; and add roadmap growth. Compare the result to the platform's 120-position standard with margin, and remember tray parts consume stations, not feeder positions — count them separately in the same census.

Can one Hanwha machine handle both 03015 chips and long connectors?

Check the connector length against the DECAN platform's reach: within roughly 75 mm-class windows, the speed and multi-function models cover both extremes in one configuration. Beyond that — connectors approaching 100 mm alongside a 03015 chip — the matrix is pointing at a split configuration: the micro-chip stream on a DECAN machine, the large-body stream on an SM482Plus-class unit. The census makes the conflict visible at selection time, when the split can be designed rather than improvised.

Do tray-fed components slow the line down?

Modestly and predictably. Tray pickups send the head to a dedicated station and back — a longer flight than any tape pick — so boards heavy in tray parts run below their tape-only rate even on a correctly configured machine. The fix is arithmetic: weight the tray rows in the realistic-rate calculation, and let the result decide whether the configuration needs another tray station, a different machine, or a packaging change that moves the part onto tape at source.

How many BGAs can a Hanwha chip mounter verify reliably?

Reliability tracks ball pitch and the camera set, not BGA count. Flying vision verifies standard-pitch arrays — around 0.4 mm ball pitch on the SM471Plus and SM481Plus — at production speed, while finer or verification-critical arrays route to stage-vision configurations with their tighter optical control, and upward-vision options can inspect ball co-planarity after pickup. Census the ball pitches in your BOM, match them to the camera table, and prove the hardest array in the demonstration rather than trusting the class description.

What margin should I hold over my schedule when sizing the machine count?

Twenty to thirty percent above current demand, held in the configuration as purchased. The margin absorbs the mix drift every real product family drifts through, the efficiency decay between maintenance cycles, and the growth the roadmap promises — and the realistic-rate calculation is what makes the margin measurable rather than hopeful. A line bought to run at its ceiling converts every process hiccup into a missed shipment, while a line bought with calculated margin treats those hiccups as the rounding error they are.