How Component Selection Affects Product Cost

Price the part at the quantity you will buy for a named build, then add the process and the exposure it drags in. If those extras are unknown, leave the cell blank. An invented yield is not a model. Every figure below is hypothetical teaching maths: not a quotation, not a wage, and not a BrahmWorks price.

Read the price at a stated quantity, in a stated package, from a seller you can buy from, on a date you write down. A prototype price on cut tape and a production price on a full reel are different contracts. Quoting the reel price against a build that will buy cut tape makes the BOM look finished while the invoice is not.

Extended cost is unit price times quantity, plus anything the part forces you to buy and not fit: the rest of a reel, a stencil aperture change, a nozzle, a programming fixture, a licence for a tool the part requires. Keep that last group visible. Folding it into "overhead" is how a cheap part survives a review it should lose.

The same discipline applies to the development programme around the BOM. The hardware product development process is where you name the quantity assumption and the exclusions. A component choice made with no quantity is a preference. It is not a cost.

A connector that cannot survive the reflow profile of the rest of the board adds a hand step. A part that must be programmed before it is placed adds a station, or a fixture later. A tall part beside a tiny one changes the stencil and the inspection. None of these appears in the unit-price column. All of them appear in the product.

Do not fill the gap with a yield you do not have. Ask the assembler, on this package and this board, what they will charge and what they will not guarantee. If they will not say, leave the line open rather than borrow a percentage.

A tighter passive tolerance and a smaller body do not imply each other's cost. Get the piece price and the assembler's sentence separately.

A price break you cannot reach is not your price. A reel minimum larger than the build, for a part the design may still delete, is inventory with a chance of scrap. Write three numbers side by side: what this build needs, what you must buy, and what happens to the surplus if the design changes. The third is where "the reel was cheaper per piece" goes to die.

A second source is a part you can fit on this board and buy in the quantity this build needs. A different footprint with a hopeful note in the layout is a future spin. If you keep a single source, record the trigger that starts a redesign, and do not describe the part as low risk because the price was low. Risk and price are different columns.

Lifecycle belongs in the cost review. A part that is not recommended for new designs can still be cheap and available this month. The cost you are taking is the redesign when you cannot buy it for the next build. Name who watches that status, and at which gate they must look again.

A module can cost more per unit than the silicon it contains and still cost less as a product, when it removes a test, a layout risk, or a file you would otherwise build. It can also cost more in both senses, when you pay the module and still carry the test. The comparison is allowed only after both sides list the same inclusions: programming, inspection, the surplus buy, and any evidence the market requires. A module price next to a chip price, with the rest unspoken, is not a comparison.

Invented product: a small status indicator, external adapter, one button, a handful of LEDs, no radio. Two fictional connectors for the same cable, called P and Q.

Connector P is cheap and must be soldered by hand after reflow, because it cannot take the teaching profile used for the rest of the board. Connector Q costs more and reflows with everything else. The assembler has not given a yield. The model therefore has no yield. It has a time.

Teaching assumptions, invented for the sums:

• Connector P: INR 20, plus 40 seconds of hand work.
• Connector Q: INR 55, placed and reflowed with the board, no extra station.
• Labour rate, invented and not a wage: INR 6 per minute.
• This year's build: 1,000 units.
• A different part on the same board, part R, is offered at INR 25 each for 1,000, or INR 18 each on a reel of 3,000, and the design may still delete it.

Hand labour for P: 40 / 60 minutes × 6 = INR 4 per unit. P plus labour: 20 + 4 = INR 24. Q is INR 55. On this invented rate, P is cheaper by 55 − 24 = INR 31 per unit, and 31 × 1,000 = INR 31,000 on the batch, provided the 40 seconds is real and the hand step does not force a separate trip through the line. If the assembler charges a lot charge rather than seconds, replace the INR 4. Do not keep both.

Part R, if you buy only the 1,000: 1,000 × 25 = INR 25,000. If you buy the reel because the piece price looks better: 3,000 × 18 = INR 54,000. Extra cash this year: 54,000 − 25,000 = INR 29,000, for 2,000 pieces you may scrap if the design changes. The piece price fell from 25 to 18. The cash went up. Those are different results, and the review has to say which one it is optimising.

A sensitivity, not a measured rate: suppose the hand-soldered P has to be touched again on 1 unit in 10, and each touch costs a teaching INR 50. Added cost per unit shipped: 0.1 × 50 = INR 5. P plus labour plus that sensitivity: 24 + 5 = INR 29, still under Q at 55. The sensitivity does not flip the connector. It also is not data. Leave it out of the BOM until a pilot counts the touches. The arithmetic shows how large the rate would need to be before Q wins: the gap is INR 31, the touch is INR 50, so the rate would need to exceed 31 / 50 = 0.62, which is 62 in 100, before Q is cheaper on this invented pair of numbers. A rate like that would be a process failure, not a reason to hide inside a percentage.

• The price has a quantity, a package, a seller and a date.
• Prototype price and production price are not treated as the same contract.
• Extended cost includes surplus you must buy and cannot fit.
• Hand steps, programming and extra stations are written next to the part.
• Unknown yields are blank, not borrowed.
• The assembler has seen the package and either priced it or refused to.
• A second source fits this board, or the single-source trigger is named.
• Lifecycle status has an owner and a gate for the next look.
• A module is compared with the same inclusions as the discrete set.
• The review states whether it is optimising cash this build or piece price on a reel.
• A redesign triggered by this part is costed as a possibility, not assumed to be free.

To the price of the build you are about to place, labelled as such, with the production price held in a separate column so nobody presents it as today's invoice. Freezing a design on a reel price you will not reach makes later "cost down" look like a failure when it is only the first honest number.

When the module's price, plus what you still must do, is lower than the parts plus the test, layout and evidence you no longer must do. If you cannot name what the module removes, you cannot claim the saving. A higher unit price can still be the lower product cost. A lower unit price can still be the higher one.

As a risk with a trigger, not as a forbidden part and not as a non-issue. Write what you will do on a last-time buy or a stop-ship, who decides, and whether the footprint has a sibling you have actually checked. A blank "alternative" column is not a second source.

No. It reduces the piece price only if the price list says so, and it can increase assembly cost or remove an assembler. Ask both questions. A size chosen because another product used it is not a cost decision until those two answers are filled in.

Bring the quantity, the open lines, and the parts whose saving depends on a process you have not priced. A review can keep the cheap part, or it can show that the cheap part is only cheap in the column that was filled in.

Review the next stage with BrahmWorks