| Episodes | 141 |
| Mentions | 215 |
| Cited here | 66 |
| First — last | #1 — #724 |
| Top guests | Michael, Jeri Ellsworth, Chris Church |
| Related | component sourcing · digi key · schematic · gerber · open source hardware |
A bill of materials (BOM) is the itemised list of parts, quantities and reference designators required to build a product, and in electronics manufacture it functions as the central document linking design to procurement and assembly: it is exported from the design data, priced against distributors, and delivered to the contract manufacturer together with the Gerber and placement files.[17][411] Every line item is simultaneously a cost driver and an independent supply risk, so the composition, formatting and maintenance of the bill of materials are treated as engineering and logistics disciplines in their own right.[282][130]
Contents and part identification
A schematic that carries only reference designators forces a cross-reference into the bill of materials to identify each device, so the device name belongs on the schematic itself, while internal company part numbers are held in the bill of materials for purchasing.[80] An off-the-shelf item can be pinned in a bill of materials by manufacturer, manufacturer part number and distributor SKU together; a custom-made item such as a bespoke cable must first be given its own SKU and drawings before it can be specified the same way.[351]
Manufacturer part numbers are not unique identifiers, and the electronics industry has no universally accepted part-identity scheme comparable to the GS1 barcode system used in consumer goods; some manufacturers issue suffixes such as a trailing R or T that denote packaging rather than a different die, and distributors sometimes price those variants differently.[542] Engineers working without a formal internal numbering scheme use the manufacturer part number as the part number, a gap that becomes visible when an assembly service asks for a customer part number that does not exist.[577] A formal internal part number is immutable and unambiguous, whereas identifying a line item by name invites naming conflicts and misinterpretation; bills of materials have been observed in which a line was named only by a description of who knew how to fit the part.[577]
Generic and meta parts
Commodity passives are commonly specified in a bill of materials by value, package and tolerance alone, with no manufacturer or part number, because a contract manufacturer stocks standard values; an assembly service receiving such loosely specified lines matches them to its own house part and automatically merges identical lines into a single bill-of-materials line.[216][243] For components with many valid substitutes, the recommended practice is an in-house numbering scheme of meta parts — for example a 0603 1 µF X7R 16 V capacitor — each of which contains the set of acceptable physical components; Jan Rychter, whose PartsBox component-management tool implements the scheme, carries the meta part through the design so that the choice of physical component is deferred until ordering and a shortage substitution does not require reopening the CAD tool.[542] Binding a schematic symbol to a unique manufacturer part number rather than to a generic device is the back-end capability on which automated bill-of-materials generation and one-click ordering depend; the Upverter founders identified that symbol-to-part association as the difficult core of their browser-based design tool.[163]
Control documents and generation
Two control-document models exist for a design’s part data: the spreadsheet, held as a CSV or Excel file, or the schematic, from which the bill of materials is generated on demand.[230] Where the spreadsheet bill of materials and the schematic are maintained separately, edits made in the spreadsheet are not pushed back to the schematic in time and the two drift apart.[230] Spreadsheet bills of materials are obsolete the moment they are finished, cannot be updated without manually revisiting every part, and have no mechanism for multiple currencies, minimum order quantities or order multiples.[542] The cost of treating the schematic as the control document is that producing a bill of materials requires opening the CAD tool and running the export, which in turn requires a licensed seat for anyone who needs the data.[230]
Under the schematic-as-control-document method, part numbers, supplier data and a populate or do-not-populate attribute all live in the schematic and are pushed forward to generate the spreadsheet that purchasing and the contract manufacturer actually work from, since neither will open a schematic to determine whether a part is fitted.[174] A zero-ohm resistor that is populated or omitted serves as a build-variant mechanism enabling different parts of a circuit, and the fit or no-fit state has to be carried as design data rather than left implicit.[174] The method yields no benefit if adopted partially; it works only where the whole flow is committed to it.[174]
A scripted output job in the CAD tool packages Gerber generation together with the bill of materials, so that a company-standard job produces an identical set of manufacturing files on every project; a misconfigured job fails silently and exports the wrong data.[434] Tagging a design revision as production-ready in a version-control repository can trigger a hook that regenerates the whole manufacturing set automatically — Gerbers, schematic and bill of materials — rather than leaving those exports to be run by hand.[505] The interactive HTML bill of materials is a single self-contained JavaScript file that opens in a browser, can be served behind a web-server password so a client can view a project’s parts, and in later versions also exposes traces for troubleshooting.[561]
In mechanical design, an assembly modelled down to every nut and bolt exports a complete bill of materials directly from the CAD system, limited only by the completeness of the model.[611] In one quick-turn electronics workflow, Mike Harrison keeps component type and value in the PCB rather than the schematic and exports the bill of materials from the board layout as the shopping list for parts, boards and stencil, with the in-house library orientations matched to the pick-and-place feeders.[224]
Product lifecycle management
In product lifecycle management (PLM), the bill of materials and CAD files are held together with revision control, part number generation, and controlled distribution of content to team members and suppliers.[577] Released manufacturing data is distributed as a link into the controlled repository rather than as an email attachment, because an attachment persists in an inbox and can be built from after it has been superseded, whereas a link can be revoked or expired when a newer revision exists.[577] When many copies of a bill of materials circulate to different vendors, one document — the schematic or the lifecycle-management system — has to be nominated as the single source of truth against which any build is checked.[577]
Assembly data and contract manufacture
A board sent for contract assembly requires three data sets — Gerbers, the bill of materials and the X-Y placement file — and the designer has to learn to export and sanity-check all three, because an exporter can be misconfigured.[411] The merged bill-of-materials and placement file, sometimes called the XYRS file, is what allows assembly to be automated; the parts list alone does not say where the parts go.[243] Reference designators present in the bill of materials but absent from the placement file, and placement coordinates supplied in inches, millimetres or thousandths without indication, are routine defects a contract manufacturer has to resolve before a board can be programmed into the machines.[411]
Under a turnkey arrangement the contract manufacturer is given the bill of materials and sources every part, assembles the boards and delivers finished assemblies, in contrast with a consigned build where the customer supplies the parts; on a small build the value of a turnkey quote lies in the sourcing rather than the assembly, because the hours spent buying a large parts list exceed the hours spent placing them.[445] Where inventory is consigned to the contract manufacturer and pulled per build, attrition and mis-ordering push the recorded stock out of agreement with the physical stock unless a full audit is run regularly; scanning reels as they are consumed turns each build into an inventory checkpoint.[722] When the contract manufacturer is to buy the parts, the exported bill of materials should carry the general specification rather than a specific manufacturer part number, so the assembler can use whatever qualified component it can obtain, supported by an approved-substitutes document tabulating every generic line against all its acceptable manufacturer part numbers.[542]
Assembly houses that stock their own parts publish export plugins for CAD tools that emit the bill of materials pre-formatted to the required layout and check it against the house parts database before submission; using those stocked parts is what makes the low-cost service work, but they are not guaranteed to be available on a later build, so a repeat order of the same board can require reworking the bill of materials each time.[700] The choice between an assembler’s house part and a specified market part inverts with volume: the house part can be cheaper in very small quantities while the market part saves a few cents per board once quantities reach a thousand.[243] A combined purchase list is built by multiplying every bill of materials by its planned build quantity and merging the results, so that parts common to several projects are bought in one larger order, with the specific components chosen only when that list is turned into orders.[542]
Line-item overhead
Every added line on a bill of materials carries handling overhead at the assembly house — inventory, handling and correct binning — independent of the component’s price, and the feeder slot count of the assembler’s pick-and-place machine is a hard limit: adding one more part to a board that already fills every slot changes what the assembly costs.[243][428] David Kronstein’s high-speed camera board carried about a thousand placements across roughly 130 unique line items and was quoted at about one hundred dollars of assembly per board at a few hundred units, a quantity at which no significant component price break applies.[325] Reviewing his long-lived MightyOhm kit product, Jeff Keyzer identified reducing the number of bill-of-materials line items as the change he most wished he had made, because every extra line is repeated labour in kitting each unit; his kitting procedure allocates one bowl per component and a second bowl for any component whose bill-of-materials quantity is two, so that assembling a kit means taking one item from each bowl without counting, in batches of a hundred parts per bowl.[229] Integrating the crystal and flash memory into the RP2350 microcontroller’s package removed two part-selection decisions and one unique bill-of-materials line, as Raspberry Pi’s James Adams described the packaging choice.[687]
Cost engineering
At a volume of one million units, one cent removed from the bill of materials is ten thousand dollars, which sets the amount of engineering time a cost reduction of that size justifies.[130] Jon Oxer worked the canonical example: swapping a six-cent tantalum capacitor for a one-cent ceramic at a million units justifies up to fifty thousand dollars of engineering effort, most of it spent testing the change rather than making it.[349] Where a part appears only once or twice on a board, the difference between a three-cent and a thirty-cent component is immaterial, and the engineering effort to exploit it is repaid only at volume.[428]
A commonly used pricing floor for a small hardware operation is a retail price of at least 2.4 times, and more safely 3 times, the bill-of-materials cost; the multiplier of roughly 2.5 to 2.6 is properly applied to the full cost of producing the product — assembly, testing, inbound and outbound shipping — rather than to the bill-of-materials cost alone, because distributors take their margin out of the same price.[215][201] In the consumer hardware segment the quoted figure is a fourfold multiple of bill-of-materials cost in the selling price for the business arithmetic to work.[495] Plotting unit price against order quantity for each component puts the useful break region for a small-scale design in the hundreds to low thousands of units.[81] Achievable bill-of-materials cost is set by the minimum order quantity a company can fund, and cost estimates built from the bill of materials alone omit the tooling, firmware development and legal costs of bringing a product to market.[327]
Cost estimating for a manufactured product counts every screw, piece of tape and item of packaging, then applies a waste adder and a factory markup adder; Julia Truchsess, costing submitted inventions for a product-development firm, found them routinely estimated at half their true figure without that discipline.[424] The recommended conservative method prices the whole bill of materials at worst-case catalogue distributor prices and sets the target selling price from that, so that cheaper sourcing found later is margin rather than a dependency.[126] Contracted volume pricing differs sharply from the catalogue price an engineer sees at a distributor, and pulling the true negotiated cost forward into the design cycle changes what features a board can carry; conversely, bill-of-materials costs reconstructed from a teardown understate what a large manufacturer pays, because they do not capture the purchasing power of a buyer committing to millions of units up front.[577][70] An electronics manufacturing service taking on a design will redesign parts of the circuit for cheaper topologies — substituting a linear regulator for a buck converter where the efficiency loss is acceptable — to remove cents from the bill of materials.[291] Dave Jones, costing a completed design’s full bill of materials in a spreadsheet at the end of development, found the parts cost alone ruled out the intended market price, grounds for not taking the product to market despite working hardware.[682]
Dominant line items
Roughly half of a typical bill of materials is jelly-bean passives, and a designer who uses zero-ohm resistors liberally as configurable links can find they account for something like a fifth of the board’s parts.[552] On Bunnie Huang’s eight-inch framed consumer device, the display glass was the single most expensive bill-of-materials item, and the response was to redesign the product around an external screen rather than to optimise the rest of the parts list.[84] Zach Fredin found connectors accounting for about thirty percent of one board’s bill of materials, and that moving from gold flash to full gold plating would have tripled their cost while adding only wear resistance after first insertion.[330] In Aaed Musa’s robot arm, the actuation hardware — motors and motor controllers — was around eighty percent of bill-of-materials cost, so cheaper motors and controllers accounted for essentially the whole of a two-thousand-dollar cost reduction between revisions.[712] Simone Giertz’s calendar device repeats a component 365 times, making that component’s unit cost dominate the product: mechanical switches at that count would have produced a fifteen-hundred-dollar product, so the input was implemented as capacitive touch pads on the board itself.[592] Packaging can approach the electronics cost: on the castAR headset, Jeri Ellsworth’s box went from a two-dollar corrugated box to fifteen dollars with magnets, four colours and foil embossing, comparable to the electronics bill of materials.[394] Bill-of-materials scales differ by orders of magnitude between market segments — a computing module offered at ninety dollars in single quantity was pitched to lighting manufacturers whose entire product parts cost was around ninety-five cents — and a product with a five-dollar bill of materials cannot absorb a one-dollar power-monitoring chip that is justified in a variant sold on capability, as Chris Osterwood specified for his power-distribution line.[526][425] Using an external dedicated display driver alongside a small microcontroller can be cheaper than moving to a larger part in the same family, at the cost of an extra bill-of-materials line with its own obsolescence and supply exposure.[393]
Availability and supply risk
Every line item on a bill of materials is an independent supply risk: a hundred-line board whose lines carry designed-in substitutes can absorb a shortage, whereas a design that depends on a module built from two hundred other parts concentrates the risk into one bottleneck.[282] One part missing stops the whole surface-mount line; during the shortage period a prototype-sized distributor order covered more than ninety percent of a client’s bill of materials while a single unavailable instrumentation amplifier still blocked the build.[628] A component change made partway through a production run produces units with the same schematic, the same board and the same revision number but a different bill of materials, and the difference has to be tracked to know which units in the field carry which parts.[178]
A bill of materials released into production should be re-run every few months to confirm the parts can still be bought; distributors forward obsolescence notices to customers who previously purchased the affected part.[382] Component availability turns bill-of-materials maintenance into an iterative search in which a converter available two months earlier is gone, one previously unavailable has returned, and each substitution restarts the process of finding alternatives, lowest cost and availability.[128] Supply-chain information has historically been opaque to design engineers because it is held by a separate team and applied weeks or months after the design decisions that depend on it.[577]
Risk assessment and second sourcing
A bill-of-materials health risk assessment grades every manufacturer part number on global availability, the number of distinct sources selling it, whether it is end-of-life or not recommended for new designs, whether footprint-compatible crosses exist to populate an approved vendor list, and compliance status such as RoHS and REACH; the point to run it is at roughly eighty percent design completion, once the prototype functions, because component decisions taken then cast long shadows over the product’s life.[451] An approved vendor list names several drop-in alternatives per line, so a factory unable to find the specified part has qualified options rather than a stalled build against a sole source.[451] Second sourcing is implemented by qualifying several vendors and several part numbers for the same function and listing all of them on the bill of materials; military work has required three qualified vendors and three qualified parts per position.[574] Alternatives that differ only in package or packaging suffix are read by a buyer as the same part, but a change of base part number is not and has to be explicitly called out as an approved replacement.[574] Long-obsolete part numbers stay in production because they remain specified in existing bills of materials, and qualified customers such as military contractors will not change a part number.[574] On the castAR programme, Jeri Ellsworth’s team arranged second sources and contingency plans for most bill-of-materials lines before the crowdfunding campaign launched, on the principle that no line should be single-sourced, and derived the funding target and unit price by working the bill of materials with contract manufacturers over a summer and extrapolating from the required unit volume.[173] Publishing the bill of materials for a system-in-package or module lets the integrator identify a component that is holding up delivery and request the board be built with that part depopulated, a practice Lukas Henkel adopts for his compact high-speed modules.[681]
Procurement practice
A bill of materials generated straight out of a CAD package is incomplete for procurement: it lacks manufacturer part numbers, approved alternatives, and stock levels and prices, so a separately maintained purchasing BOM is required.[17] The traditional quoting process sends the bill of materials as a spreadsheet to several distributors, compares the returned line-item prices, and splits the order so each block of parts is bought from whichever distributor priced it best; the split is worth the administrative cost only when the saving is material, since for a difference on the order of a hundred dollars it is simpler to buy the whole list from one source.[411] Bill-of-materials data can be scripted out of the CAD tool into a procurement pipeline in which a build quantity is entered, current prices are pulled from every vendor, and purchase orders are emitted in each vendor’s required template; where BOM pricing and lead time are fed back into the design tool, a substitution produces an immediate readout of its effect on manufacturing cost, making a cost-target check a form of design rule check alongside the geometric one.[342]
Uploading a bill of materials to a single distributor’s site turns a repeat build into a single ordering action at the correct quantities, with live stock position reported against each line before the order is placed, at the cost of buying everything from one source.[104] It is frequently cheaper in total to order the entire bill of materials fresh from a distributor, bagged and labelled, than to search existing stock and net it out of the order, though ordering the full list for every project accumulates duplicate inventory — in one case around thirty separate lots of 0402 zero-ohm resistors — which tracking stock against planned build quantities avoids by consuming from reels already held.[493][542] Filling a bill of materials with the cheapest locally obtainable parts can consume on the order of ten hours per project, work that has usually already been done independently by many other builders of the same design; Ian Lesnet’s open-hardware projects accordingly publish sourcing information alongside the design.[125] Publishing a distributor-hosted bill of materials as a public link, alongside the board files, removes the per-builder sourcing effort for an open design.[453] Keeping a bill of materials and its on-hand stock figures accurate while batching purchases across several builds is error-prone enough that a miscalculation has left a production run short of parts; at scale it is a full-time function.[197] One budgetary costing convention records a standard cost field per component populated with the thousand-piece price, so a bill of materials produces a defensible estimate for a nominal build of a thousand units, exclusive of shipping and other loaded costs.[542]
Specification errors surface at ordering time: where the CAD library does not bind a specific component and the choice is made when the bill of materials is assembled, an inadequate capacitor voltage rating can pass into the parts order, small ceramic capacitors being commonly rated only for six or ten volts; and a bill of materials can name the right part number while specifying the wrong package, discovered only when the boards arrive.[561][341]
Currency, tariffs and working capital
A bill of materials denominated in a foreign currency carries exchange-rate exposure directly into product cost; costing in United States dollars assumes a variation of about plus or minus five percent, which a twelve percent move breaks.[178] Purchasing in bulk rather than at catalogue quantities brings customs into the transaction, a cost and complexity that small-quantity distributor orders avoid.[178] Tariffs do not fall uniformly across a bill of materials, so exposure is established by having the whole list re-quoted rather than applying a blanket percentage; importing parts into China for assembly there attracts substantial tariffs, so an assembler sources locally where possible and imports only what it cannot find.[400][125] Shortages and tariffs land on the bill of materials as a working-capital problem: the entire stock has to be paid for before any unit can ship.[715]
Disclosure and open hardware
Distributors routinely solicit customers’ bills of materials in order to quote against them competitively when opening a trade account.[136] Disclosing a bill of materials to a distributor signals that a part is designed in with no alternative, which their purchasing side can price against; the counter is to require a second source before disclosing.[136] A common open-hardware practice publishes a reduced bill of materials carrying only descriptions and part numbers while retaining a fuller internal version holding vendors, vendor prices and alternatives, on the basis that the sourcing data represents most of the work.[136] The open-source hardware convention requires releasing the PCB and schematic files together with the bill of materials, not the board files alone, and including supplier information is what makes a published design turnkey-reproducible by a third party.[55][534] Greg Charvat published each worked radar design in his textbook with a full bill of materials alongside its schematics, block diagrams and software, so that a reader could rebuild the instrument rather than only follow the theory.[179]