Electronic Assembly Manufacturer: The Ten Proofs a Quote Never Carries
Every electronic assembly shop will quote your board, and most will quote it well. What separates them is not the price line. It is the paperwork behind it, which you only see if you ask for it before the order. The same test applies to a simple cable job, and it costs nothing to run.
The Short Answer
Ask ten questions, get ten documents. A shop that can produce all ten in a day is organised. A shop that produces three and promises the rest is selling hope.
The ten sit in three layers. Paper says who the shop is. Machine settings say how it runs your board. Records say what it actually did on the day.
Most buyers check the first layer and stop. That is why a supplier with good certificates still ships a bad board.
Collect the ten in one folder and score them against each other. A shop that answers nine and explains the tenth is a shop with a working system.
A shop that answers all ten in polished language but cannot show a single dated document is answering with marketing. The dates are the tell.
What an Assembly Shop Actually Sells
Three jobs sit under one name. Placing parts on a board, soldering it and testing it. Some shops also build the box around it.
The placing and the soldering are machine work, and machines are similar across the trade. The testing and the record keeping are where shops differ.
A small shop may run one line and buy its stencils outside. A large one runs ten lines and cuts stencils in house. Both can build a good board.
The risk changes with the size. A small shop gives your order attention and has fewer backups. A large shop has spare capacity and treats your order as a slot.
Some shops add work after the board. Coating, potting and building the box around it are separate skills, and each one brings its own record. Ask which of them the shop does itself and which it sends out.
So the question is not small or large. It is whether the shop can show you, on paper, which line will run your board and who will check it.
The parts market that feeds these shops sits a short drive away for most of them. That is why a missing part is normally a phone call and not a shortage.
Proof One: The Line List
Ask for a list of the lines, the machine model on each, and the feeder count. This is a one page document and it is the fastest filter there is.
The feeder count tells you how many part types a line can hold at once. A board with many small parts needs a line with room for them, or a second pass.
The machine model tells you the smallest part the line can place and the finest pitch it can hit. Both figures are published by the machine maker, so the claim is checkable.
Ask which line your board is planned on. A quote that does not name a line is a price without a plan.
One more line belongs on the page. The date of the last service on the placement machine. A machine that has not been serviced in years still places parts, and places them less accurately.
Proof Two: The Placement Report
The placement machine reports what it did. It counts parts placed, parts rejected and placement time per board. That report is generated by the machine, not typed by a person.
Ask for one from a recent job of similar size. Two things in it matter. The reject rate, and whether the machine re-checked a part after a failed pick.
A high reject count on small parts usually means the feeder is worn or the nozzle is wrong. Neither is a design problem, and both are visible in the report.
Two things the report does not show are worth asking about. How long the line was set up before the first board ran, and when the nozzles were last replaced.
Setup time decides how a small order is treated. A two hour setup on a fifty board run is a cost the shop carries. It comes back in the price or in the queue position.
The report also shows how long a board took. That figure is the honest basis for a price, far better than a price per board that appeared from nowhere.
Proof Three: Paste and Stencil Control
Paste has a shelf life and a storage rule. Ask where it is kept, how old the jar is, and how long it may sit on the line before it is replaced.
A shop with a written rule answers in one sentence. A shop without one answers with a shrug, and the shrug is the answer.
Ask whether the stencil is cut in house or bought. Both are normal. The useful part is the thickness, and whether it is chosen per board or fixed for everything.
One detail is worth asking because almost nobody does. What happens to paste left on the stencil at the end of a shift. Scraped back into the jar, it carries dried lumps into the next run.
Proof Four: The Oven Profile, Per Board
An oven profile is a chart of temperature over time. Ask for the profile for your board, not for the oven in general.
The document should carry the board name, the date, the oven number and the peak temperature. A profile without a board name is a picture of a different job.
Shops often run one profile for a family of boards. That is a reasonable shortcut, and it should be stated. A shortcut that is stated is a decision. One that is hidden is a risk.
Ask how the profile is checked. A thermocouple glued to a test board and run through the oven is the normal method. A profile reused from last year is a hope.
Boards that sat in storage before assembly need a bake, and the bake is also a record. Ask whether it happened, and where it is written down.
Proof Five: Inspection, and Who Reads It
Two machines do most of this work. A camera that compares joints against a stored image, and an X-ray machine for parts that hide their own pads.
Ask how many boards are inspected and what the machine is set to catch. A camera set to a loose standard passes a joint that a customer will fail.
The figure worth asking for is the false call rate. It tells you how often the machine flags a good joint, and how much a person has to recheck.
That number matters because of the second half of the question. Who looks at the flags, and how long do they spend on each one.
A shop that says the machine catches everything has answered badly. Every camera has blind spots under parts, and every X-ray image needs a person to read it.
A visit that checks the line, not just the goods is the outside option when the answers sound too clean.
Proof Six: The Test, and Who Owns the Fixture
Testing is the last chance to catch a fault, and the fixture is the part buyers forget. A fixture is a frame with pins that touch the board.
Ask for the test coverage as a list. Which nets are probed, which functions are run, and what the pass mark is for each.
Then ask who owns the fixture. If the shop pays for it, the fixture follows the shop. Move to another supplier and you pay again.
This is the same ownership question that comes up with moulds, and it belongs in the order for the same reason. The clause that keeps a tool with the buyer takes two lines and saves a fixture quote.
| Method | What it proves | What it needs |
|---|---|---|
| Bed of nails | Every net on the board is connected and correct | A fixture built for this board |
| Flying probe | The same checks, without a fixture | More time per board |
| Functional | The board works as a product | A test program and a known good board |
| Visual only | That a person looked at it | Nothing, and it proves almost nothing |
Ask for the test log from the last batch too. Every board, its result and the time it ran. A log with no failures at all over a long run is possible, and worth a second look.
Proof Seven: The Lot Record
If a board fails in the field, the lot record is what lets a shop find its siblings. Without it, every board from that month is suspect.
Ask what is recorded per board. A serial number, the date, the line, and the lot numbers of the parts that went into it.
The part lot is the line most shops cannot produce. It needs a scan at the feeder, and a scan at the feeder needs the discipline to do it every time.
Ask where the scan happens. A record built from a paper sheet at the end of the shift is a summary. A record built from a scan at the feeder is evidence.
Ask to see one record from a past job, with the customer name removed. A shop that cannot show a redacted record does not keep records.
Proof Eight: Static Control
Static damage does not always kill a board. It can leave a part weaker, and the failure arrives weeks later at the customer.
Ask for the static control programme, not a certificate. What is earthed, what people wear, and how the floor and the benches are checked.
The standard behind the programme is written by the EOS/ESD Association, and the body that writes it publishes what a programme has to cover. Ask for the measurement from the last check, on the line that will run your board.
This is where a visit earns its fee. A programme on paper and a wrist strap in a drawer are two different shops.
Proof Nine: Capacity, and How Your Order Fits
Capacity is not a number on a website. It is the answer to one question. What runs on the line in the week my order is planned.
Ask how many changeovers happen per day. A line that stops often for small orders places parts less accurately at the start of each run.
Ask what happens when a part is late. A shop with a buffer answers with a plan. A shop without one answers with a new date.
Panelisation belongs in the same conversation. The panel is the larger sheet the line handles. The number of boards per panel decides how much of your material becomes rails and gaps.
Ask who designs the panel and who pays for the spare boards on it. A panel designed for the shop’s convenience is not always the cheapest panel for you.
Small orders have a real place here. A shop that runs many small jobs is set up for them, and a shop built for long runs will treat yours as an interruption.
Proof Ten: The Certificate, and Its Edition
A quality certificate is the document most buyers ask for first, and the one they read least. Three things on it decide whether it means anything.
The first is the edition. The 2015 edition of the quality management standard was withdrawn on 16 September 2026 and replaced by the 2026 edition. The record on the ISO site shows both dates, so a certificate naming the older edition is a shop mid transition.
That is not a reason to end the conversation. It is a reason to ask when the transition is planned, and to see the answer written down.
The second is the scope. A certificate covers a named site and a set of activities. If the scope describes trading or sales, it says nothing about the workshop that builds your board.
The third is the expiry and the body that issued it. A certificate from a body nobody recognises is a framed page.
Two more documents sit alongside it. The safety listing file number for the parts on your board, and the substance statements from the part makers. Both come from the supply chain, and the shop only passes them on.
The Ten Proofs in One Table
| Proof | What to ask for | What it hides when missing |
|---|---|---|
| Line list | Machine models, feeder count, service date | Which line will actually run the board |
| Placement report | Reject count and cycle time from a similar job | A worn feeder or the wrong nozzle |
| Paste record | Storage, age, and the end of shift rule | Dried paste carried into the next run |
| Oven profile | Board name, date, oven number, peak | A profile borrowed from another product |
| Inspection log | Sample size, setting, false call rate | Joints no machine and no person checked |
| Test coverage | Nets probed, functions run, fixture owner | A fixture that stays behind when you move |
| Lot record | Serial, date, line, part lot numbers | No way to find the siblings of a failed board |
| Static check | Last measurement, date, and the line it covers | A programme on paper only |
| Capacity plan | Changeovers per day, buffer on late parts | A delivery date that moves after the order |
| Certificate | Edition, scope wording, expiry, issuing body | An old edition covering a sales office |
Ten documents sound like a lot to collect. In practice a shop that runs properly sends seven of them the same day, because they already exist.
The three that take longer are the ones that need a person. The lot record, the static check from your line, and the capacity answer for your week.
Those three are also the ones that predict how the order will go.
Questions Buyers Ask About Assembly Shops
How many shops should I ask for the ten proofs?
Three is enough to see a pattern. Two shops can agree by accident. Three show whether a gap is common in the trade or specific to one supplier.
Is a bigger shop safer?
Not on its own. A bigger shop has backups and treats your order as a slot. A smaller shop has attention and fewer spare lines. The line list tells you more than the headcount.
Do I have to visit?
Not for nine of the ten. The static check and the paste rule are the two where a visit adds most, because both are habits rather than documents.
What if the shop will not share a lot record?
Treat it as a no. A redacted record from another job costs the shop nothing. A refusal usually means the scan was never done.
Does a quality certificate guarantee a good board?
No. It shows a system exists at a site. The board comes from a line, on a day, with a set of settings, and those are the other two layers.
How long should the documents take to arrive?
A week is reasonable for a first enquiry. If the first three take a month, the shop is assembling the paperwork for you, and it will assemble it again for the next buyer.
The Bottom Line
Price compares shops that look alike. Proof separates them.
Ask for ten documents in three layers. Paper, machine settings and records. Then place a small order and check whether the records match what you were shown.
Ask for the Ten
Send the board file, the batch size and the test you need. We will put the ten proofs to three shops, compare what comes back, and tell you which answers were written for the enquiry and which ones already existed.
Send the Files
Questions & Comments
We read every one and reply within 24 hoursAsked three shops for a line list and got three brochures? Send us what they gave you. We will say which of the ten documents it actually answers, and which one the reply quietly avoids.