Is That PCB Factory Reliable? Signals Worth Checking

Every buyer of boards wants the same three things: the board that arrives matches the data that was sent, the date that was promised is the date that is met, and any problem is reported rather than discovered. Whether a particular shop delivers those three things is not visible in a price list, but it is visible in a set of behaviours that can be checked before the order and during the first one.

Documented Capability

The first signal is whether the shop publishes what it can build. A PCB factory that states its layer counts, its available materials and thicknesses, its minimum trace and spacing, its minimum drill and annular ring, and its tolerances has given the designer a way to check the design before the order. It has also committed to a set of numbers.

A capability list that is specific is a good sign. A list of superlatives is not. The useful questions are the ones that produce a number: what is the smallest finished via you plate, what is the standard tolerance on a finished board thickness, what impedance tolerance do you hold and how is it verified.

Evidence of Process Control

The second signal is the data the shop produces about its own work. Electrical test on every board, optical inspection of the artwork and of the finished panel, coupon measurements where the design has an impedance requirement, and plating thickness records are all normal outputs of a controlled process.

The test that separates a shop with control from one without is a request for a cross section. A cross section of a via or of the plating on a hole shows the copper thickness, the plating distribution and the registration, and a shop that will provide one on request has the capability to measure it. A shop that cannot produce one is not measuring it.

How Questions Are Answered

The third signal is the shape of the answer to a technical question. A shop with an engineer reading the data will respond with something specific: a query about a feature that is inside the tolerance but outside the usual practice, a suggestion to adjust a pad for the process, or a note that the specified finish is not the one the data implies elsewhere.

An answer that agrees with everything and quotes a price is not evidence of competence, because it means nothing was checked. The most valuable response a customer can receive on a first order is a question about something real.

How a Deviation Is Reported

The fourth signal, and the most predictive one, is what happens when something goes wrong. Measurements drift, a process runs at the edge of its window, and a batch may be built with a feature outside the specification.

The behaviour that distinguishes a reliable partner is a deviation report before shipment, stating which parameter is out of specification, by how much, and what the options are. The customer then decides whether to accept the batch, have it reworked or have it remade. A shop that ships the boards and waits for the customer to notice has decided that the customer will not notice.

Checking a Sample

A sample inspection of a delivered board is worth an hour with a bench microscope. The items to look at are the annular ring on the outer layers, the registration between the copper, the solder mask and the silkscreen, the clarity of the legend and the evenness of the surface finish.

The drilled holes are worth examining as well, because the quality of the hole wall shows in the plating. A hole with a rough wall or an uneven plating is a hole that may fail later. Where the design has an impedance requirement, the coupon data that came with the boards should be compared against the requirement, not merely filed.

One caution about the sample itself: the board a supplier chooses to show is the board the supplier is proud of. A cross section of a routine production board, or a board that was built for another customer and can be examined with the identification removed, is more informative.

Lead Time Behaviour

The fifth signal is whether the date moves. A quotation is an estimate until the order is accepted, and the point at which a shop commits is the confirmation. After that, a date that is revised for a reason the shop could have foreseen before accepting the order is a scheduling failure, and it will repeat.

The specific question worth asking is what the shop will do if the order cannot be met on the confirmed date: notify in advance with a new date, or ship what is available. Getting an answer before the order is placed is more useful than getting an apology after.

The Red Flags

The warning signs are consistent enough to be listed. A quotation that is far below the others without an explanation of why. A refusal to provide test data or coupon measurements. An engineering question that is always routed to a salesperson. A capability statement with no numbers in it. A sample board that is visibly different in construction from the design being discussed. And a promise of a lead time that is shorter than the shop’s own process requires, which is a promise that will be broken.

None of those rules out a cheap supplier, and none of them guarantees an expensive one. What they do is direct the questions towards the things that predict the outcome. Our PCB capabilities are published as numbers for exactly this reason, and the results that come back from a build are held in our quality management system so that a trend is visible before it becomes a defect.

How to Run the First Order as a Test

The cheapest way to establish whether a shop is reliable is to place a small order that contains one demanding feature: a controlled impedance layer, a fine drill, a heavy copper area or a small annular ring. Then check four things. Whether the data query arrives before the build rather than after it. Whether the delivery includes the test data and the coupon measurements, in a form that can be filed. Whether the boards match the drawings when they are examined on a bench. And whether the shop volunteers information it did not need to volunteer, such as a note that a batch ran at the edge of the process window. That last item is the strongest signal available, because it is the behaviour that only a confident shop exhibits. A production order then follows the same pattern at a larger scale, which is how our PCB manufacturing review runs at every volume.

FAQ

Is a factory audit necessary? For a first order of a simple board, no. For a product that will be built repeatedly, a visit or a documented audit is a reasonable investment.

What is the most reliable single indicator? Whether a deviation is reported before shipment. It shows that the shop is measuring its own process.

Should a sample be requested before the order? Yes, and it should be a production board rather than a demonstration piece.