Mark both sides of the board.
Never touch it twice.

A traceability code is only as good as the worst process it passes through. Here is how to find out which process that is — before you buy a marking machine.

Articles are marketing commentary. Machine facts quoted inside them come from the S-400T product document and the published laser specification table.

Walk to the end of your line and pick up a finished board. Look at the code. Now walk backwards and find every step that board passed through after the code was applied. That list — not the marking equipment — is what decides whether the code is still there in two years.

Most factories treat marking as a printing step and then subject the print to a chemical and thermal gauntlet. Flux immersion. DI-water ultrasonic cleaning. Stencil wash cycles. Wave-soldering temperature. Fingerprints from handling. Each of those is designed to remove material from a surface. If your code is a film sitting on that surface, the process is doing exactly what it was built to do.

Here is a method you can run this week without buying anything. Take ten boards. Mark them the way you mark today. Send five through your normal process and keep five as controls. Then read all ten under the same scanner, at the same distance, by the same person. Photograph every code. Lay the photographs side by side. What you are looking for is not a failure — it is a change. A code that reads but has lost contrast is a code that will stop reading in a year.

If the change is in the wash or the soldering step, that tells you the code has to be part of the surface rather than on top of it. This is the argument for laser marking: the process changes the solder mask itself, so there is no film for the wash to lift and no ink for the cleaner to dissolve. The S-400T engraves the code and then reads it back with the same CCD, so the verification record is generated at the machine rather than at the customer's incoming inspection.

Run the ten-board test again after the machine is installed, with the same five-and-five split. Repeat it every time you change solder mask supplier, board finish or cleaning chemistry. The code is a process output, and processes drift. A marking step you can re-verify in an afternoon is worth more than one you can only trust.

If you want to run the test on the S-400T instead of on your own bench, send us a panel with the code content you actually use and we will mark it and send back the images.

The code scans at the marking machine and fails three stations later. The cause is usually one of four things, and three of them are not the marking machine's fault.

Machine facts quoted here come from the S-400T product document.

A code that reads at the marking station and fails downstream is a specific, diagnosable event. It is not bad luck and it is rarely the laser. Before you call the marking supplier, separate the four possible causes.

One: the code moved. If the code is placed within a few millimetres of a board edge or a large component, handling equipment can shift the panel relative to the scanner's expected position. The fix is in the placement rule, not the marking machine.

Two: the contrast decayed. A mark that was dark when it was fresh can lighten after coating, cleaning or thermal cycling if it was made with the wrong wavelength for that solder mask. This is why the laser table matters. The manufacturer's own guidance says green and UV suit essentially all PCB solder-mask layers on the substrate, that green gives the highest precision and adhesion with UV close behind, that CO2 suits larger boards where the marking requirement is not high and can yellow or blacken, and that fiber is only partially suitable and needs a sample test first.

Three: the code failed a grade, not a read. A scanner in "read" mode returns yes or no. A scanner in grade mode returns a quality score. A code can pass the first and fail the second — and it is the grade that tells you whether the code will survive the next six months. This is what the S-400T's optional bar code level analysis system is for: it grades the fresh code at the machine, so a slowly degrading marking process is caught before it produces a shipment.

Four: the optical path is dirty. Marking a solder mask produces fume and dust. If the extraction is not doing its job, the lens gathers a film and the mark gets lighter over a shift. The S-400T carries an automatic smoke purification system for exactly this reason — and a daily lens check is still worth the thirty seconds.

So the method is: locate the failure, then ask which of the four it is. In our experience the answer is placement, wavelength or grade about four times out of five, and the marking machine about one time in five. Diagnose it in that order and you will spend your money in the right place.

Marking both faces of a board usually costs you a second pass, a manual flip or a second machine. A built-in inverter collapses all three into one handling step.

Articles are marketing commentary. Machine facts quoted inside them come from the S-400T product document.

Ask a production engineer where boards get damaged and you will usually hear the same answer: between processes. Every time a panel is picked up, put down, flipped or re-staged, it accumulates handling damage, misalignment and labour. A marking step that requires a manual flip has added one of those events to every board you build.

The obvious alternative is two machines, one for each side. That works, and it costs twice the footprint, twice the power, twice the maintenance and a second alignment tolerance — the two sides are marked relative to two different datums, so a code on the top face and a code on the bottom face do not necessarily agree about where the board is.

A useful method when you are specifying any double-sided operation: count the handling events, then count the datums. Handling events are the opportunities for damage. Datums are the sources of error. A single station that marks both faces from one clamp wins on both counts, because the board is located once and turned inside the same fixture.

That is what the S-400T's built-in PCB reversal system does. The board is clamped, the fiducials are found, side one is marked and read, the inverter turns the panel inside the machine, and side two is marked and read against the same datum. The manufacturer documents the machine as a dual-side marking machine with the reversal system built in — not as a single-side marker with a handling accessory bolted on.

The second-order benefit is the one that shows up in the audit. Because both codes are read back by the same CCD in the same program, both sides of the traceability record are generated by the same station with the same grade analysis. When an auditor asks how you know the bottom code is good, the answer is not "an operator checked it". It is a stored read result from the machine that marked it.

If your product carries a code on both faces, tell us the board thickness, the component height on each side and the code position. Board thickness matters because the inverter has to handle 0.8 mm and 4 mm with the same reliability, and component clearance is documented as 18 mm above and 18 mm below the surface.

CO2, fiber, green or UV — the choice is not about power. It is about how the light couples into your specific solder mask, and that is testable.

Articles are marketing commentary. Machine facts quoted inside them come from the S-400T product document.

The most expensive mistake in laser marking is buying on wattage. Power tells you how fast the light can remove material. It tells you nothing about whether that material will absorb the light at all — and a solder mask that reflects your wavelength will mark badly no matter how many watts you point at it.

The S-400T accepts four sources, and the published table gives the wavelength of each: CO2 at 10.6 µm, fiber at 1064 nm, green at 535 nm and UV at 355 nm. Those are four very different interactions. The manufacturer's guidance is unusually direct about which to pick: green and UV are suitable for all PCB solder-mask layers on the substrate, green marks with high precision and strong adhesion, UV sits slightly behind green, CO2 suits larger boards where the requirement is not high and can yellow or blacken, and fiber is only partially suitable and requires a sample test.

Here is the method. Ask your solder mask supplier for the cure chemistry and the colour. Then ask for a marking sample on your own panel, in your own colour, from the machine supplier — not a demo panel in the colour that happens to be loaded. Mark the smallest code you actually intend to use, not a large one, because small marks are where wavelength differences show. Then send that sample through your real cleaning and soldering process and grade the code afterwards.

Two more variables belong in the decision. Focal length — 70, 110 or 190 mm on the S-400T — sets the working distance and the spot size, so a tall component beside the code decides which you can use. And the marking area you have to cover, documented as 1D 2 × 6 mm and 2D 1.5 × 1.5 mm per code, with an optional 70 × 70 mm field for panel-level work.

The acceptance evidence we can point to is a customer test on a UV source: 6 × 6 mm, 23-digit QR codes, marked directly from a program edited on site, at a 100 % success rate with no errors and no unclear codes — plus a reported minimum of 3 × 3 mm with 30 digits. That is a UV result, on that customer's board. It is evidence that UV works on that solder mask, not a claim about yours.

Send us three boards of the same product in the same colour and we will mark them with more than one source so you can compare the read results rather than the spec sheets.

Suppliers quote boards per hour. Your line only cares about seconds per board. Here is how to convert one into the other with your own numbers.

Articles are marketing commentary. Machine facts quoted inside them come from the S-400T product document.

A throughput figure is a division. It is one hour divided by the time one board takes. That means any throughput number you are quoted is really a statement about cycle time — and cycle time you can measure, add up and check.

The manufacturer of the S-400T publishes a worked example rather than a headline number, and it is worth copying the structure. A 330 × 250 mm panel carrying one 2D code: 3 to 4 seconds to load, 1 second for the fiducial mark, 0.5 seconds to engrave, 1 second to check. Total 5 to 7 seconds per board, about 600 boards per hour. Every one of those terms is a physical action you can watch and time.

So build your own version of that table, with four columns and one row per operation: load and transfer in, fiducial recognition, marking, read-back. Then multiply the marking and read-back rows by the number of codes per board — including the second side, if you are marking both. Add in the inverter step if the board is turned inside the machine. The sum is your cycle time.

Two honest corrections belong on the end. First, cycle time is not takt time. The marking station can be the fastest machine on the line and still be the constraint if it hands boards over in bursts; on a SMEMA line the station follows the pace of its neighbours. Second, the published numbers describe a machine in good condition. A contaminated lens and a worn extraction filter add seconds you will not find in the datasheet, which is why the daily lens check matters.

The S-400T also lets you trade cycle time against capability. Two codes instead of one roughly doubles the marking term. A larger marking field means more travel. A second side adds the inversion step and a second mark-and-read. That is not a weakness — it is the machine being honest about where the time goes, which makes it possible to design the marking step around your takt instead of hoping it fits.

Send us your operation list and your takt time and we will build the table with you. It takes one meeting and it removes the only number on the quotation that nobody can verify.

A marking machine that is faster than the line is not fast — it is a queue. Size the station to the pace of the process either side of it.

Articles are marketing commentary. Machine facts quoted inside them come from the S-400T product document.

There is a comfortable feeling in buying the fastest machine in a category. On a conveyor line it is usually the wrong instinct, and here is why: a marking station cannot run ahead of the boards arriving at it, and it cannot run behind the boards leaving it. Its useful speed is bounded by its neighbours.

So the first number to establish is not the machine's throughput. It is the takt of the process feeding it. If your wave-soldering line releases a panel every twenty seconds, a marking station that can do a panel in six seconds will spend fourteen seconds idle — every board, all shift. You have paid for capability that the line cannot use.

The method is to size against the constraint and leave the rest as headroom. Calculate the takt of the slowest machine on the line. Subtract your marking cycle time from it. If the result is comfortably positive, the marking station will never be the constraint and you should spend the remaining budget on capability instead of speed — a better laser source, the bar code grade analysis, or the bad-mark recognition that keeps the traceability record honest. If the result is negative, the board is too complex for one station and you need to look at splitting the marking load rather than buying a faster laser.

This is where the S-400T's configuration options become a sizing tool rather than a price list. The laser source determines the marking time. The focal length determines the working distance you can reach. The optional 70 × 70 mm marking field determines how much you can place in one pass. The second-side inverter determines whether you need one station or two. Choosing those four against your takt is a more useful exercise than comparing wattages.

One last sizing input that people forget: board thickness. The documented range is 0.8 to 4 mm, and both ends have to be handled reliably by the same conveyor and the same clamp. If your product mix spans that range, say so at specification stage, because that is a mechanical question and it is cheap to answer before the machine is built.

Bring us your line pitch and your board mix and we will size the station against the constraint rather than against the brochure.

A marking machine can be a printer on the line, or it can be the point where each board's identity enters your manufacturing system. The difference is an interface.

Articles are marketing commentary. Machine facts quoted inside them come from the S-400T product document.

Most factories can answer "how many boards did we build". Far fewer can answer "which board is this, what was it marked with, and did the mark read correctly". The second question is the one an auditor, a customer complaint or a field failure asks.

The gap is usually not the marking machine. It is that the marking step was never connected to anything. A code gets printed, an operator glances at it, and the only record is the board itself. If the code later turns out to be unreadable, there is no history to say when the process changed.

There is a straightforward method for closing that gap, and it works from either end. Start at the data: list the fields you actually want to hold against a board — the code content, the marking program version, the read result, the time, the operator or the line. Then work backwards to decide which of those the marking station must generate and which your MES already has. Only then look at interfaces. Factories that start with the interface usually end up logging the one field the machine happens to output.

The S-400T is documented with system access to SHOPFLOW, IMS and MES, plus customised intelligent management systems. It reads an existing code online before marking and reads the fresh code afterwards, and it supports the wrong-material checks — opposite, reversed and wrong board — that only make sense when the machine knows what it is supposed to be marking. The optional bad-mark recognition reads the "X" positions on a panel and skips them, which keeps the record free of boards that were never good.

One practical instruction: agree who owns the interface before commissioning week. If the machine is to talk to your MES, someone has to own the data mapping, and it is usually not the equipment supplier. Put that on the same checklist as the power feed and the network drop, and the marking step arrives as a data source rather than as a printer.

Labels, ink, ribbon and the operator time to apply them are a cost that repeats on every board you build. Laser marking replaces that with a capital cost and an electricity bill.

Articles are marketing commentary. Machine facts quoted inside them come from the S-400T product document.

Consumable costs are easy to underestimate because they are small and they never stop. A label is a few cents. Ink is a fraction of a cent per board. Someone applies them, someone replaces the roll, someone reprints the ones that go on crooked. Individually trivial; multiplied by every board you build for ten years, not trivial at all.

The manufacturer's own comparison between the marking methods is blunt about the trade. The traditional label has low capacity, the mark falls off easily and is not durable, and the total investment is high. The traditional inkjet code has low precision, a poor result, no permanence, high pollution and late investment. Against both, laser marking is described as clear and long-lasting, high efficiency, environmentally friendly and free of consumables — low input, high return.

Here is how to put a number on it. Multiply your consumable cost per board — label, ink, ribbon, reprints — by your monthly volume, and by twelve. Add the labour of application at a loaded hourly rate. Now compare that annual figure with the price of a marking station. That ratio is the part of the payback you can compute without assuming anything about quality. The calculator on this page does exactly that arithmetic with your numbers.

The second half of the return is the one that is harder to compute and usually larger. Every label that falls off in the field is a traceability gap, a possible recall of a wider batch than necessary, and sometimes a warranty claim on a board that was actually built correctly. Removing the adhesive film removes the failure mode. Laser marking changes the solder mask itself, so there is nothing to peel, nothing to wash off and nothing to reprint.

The honest counter-argument is capital. A marking station is a purchase, and the consumables were an operating cost. That is a real change in how the money leaves the business, and it needs to be a deliberate decision rather than a happy accident. Run the arithmetic on your own volumes, then decide whether you want the cost to be variable or fixed.

The calculator above will do it with your figures. If you send us your volume and your consumable cost per board, we will send back the ratio with the assumptions written out so your finance team can check them.