S4000 Axial Insertion Machine
for Smart EMS THT PCB Assembly

Three suppliers quote the same axial inserter. The lowest number wins the order, and eighteen months later nobody can explain why the through-hole department still costs what it did.

Full articles are in English. The summary above carries the argument in your language.

Start with the moment the comparison actually happens. A purchasing engineer puts three quotations side by side. Machine A is fifteen percent cheaper than Machine B. The specification sheets look similar, the delivery times are similar, and the decision is made on the only column that is unambiguous. That is a rational decision made with the wrong input, because the number being compared is the price of an asset rather than the cost of an operation.

The operation is where the money is. An axial insertion machine is bought to replace a bench, and a bench has a cost per board that nobody has ever written down. It is composed of four things: the labour minutes to place the parts, the rework from mis-inserted leads, the changeover time between products, and the scrap from damaged components. None of those appear on any of the three quotations, and all four change when the machine changes.

So build the comparison the other way round. Take one representative board. Count the axial parts. Time one operator inserting ten of them, including the pick, the polarity check, the lead bend and the clinch. Multiply out to a labour cost per board. Then add the rework cost: how many boards come back to the bench for a lead that went in at the wrong span or a diode the wrong way round, multiplied by the minutes to fix one. Then add changeover: how many minutes to move the bench from product A to product B, divided by the batch size. Then add scrap: components destroyed per thousand insertions, multiplied by part cost.

That total is your baseline, and it is almost always larger than the department believes. It is also the only number a machine quotation should be measured against. A machine that costs fifteen percent more but inserts at twice the rate, or that holds span changes without a mechanical adjustment, or that reports a defect rate under 500 parts per million, can win that comparison comfortably while losing the one on the purchase order.

This is where the configuration detail starts to matter more than the price. A modular feeder bank changes how often the line stops for a reel change. A servo rotary table that inserts at any angle changes how many parts have to stay on the bench because the layout wanted them at 30 degrees. A dual jumper-wire feeder changes whether a wire gauge change is a changeover or a program line. Each of those is a per-board cost term, and each of them is invisible in a unit-price comparison.

Make the comparison repeatable rather than one-off. Keep the per-board model in a spreadsheet with the four terms named, and re-run it whenever the mix shifts. A board that was a thousand-off job last year may be a forty-off job now, and the labour arithmetic on it will have changed completely. The machine does not need to change; the decision about which parts run on it does.

Pick the three boards with the worst per-board through-hole cost and send us the axial part list for each. We will tell you what the insertion time looks like on an S4000, and if the answer is that those boards should stay on the bench, we will say so. Start with the part list, not the purchase order.

The number on the datasheet decides whether you get the budget. The number on the shift report decides whether you keep your credibility.

Full articles are in English. The summary above carries the argument in your language.

Every insertion machine is sold on a maximum rate, and every production manager has been burned by one. The pattern is so consistent that it has a shape: the machine is installed, it runs at the quoted rate on the first board, and then the mix changes and the realised rate settles somewhere between half and two-thirds of the number that justified the investment. The engineer who wrote the business case is not lying. They are quoting a ceiling as though it were an average.

The reason is mechanical, and it is worth understanding rather than resenting. An insertion rate is a function of the component, not only the machine. Parts with different lead spans force the width axis to open and close. Parts at different angles force the rotary table to turn. Parts at the extremes of the size envelope take longer to feed reliably than parts in the middle. A maximum rate is measured on the easiest possible combination of those variables, held constant for the duration of the test.

The practical fix is to stop using one number and start using three. The first is the ceiling: what the machine can do under ideal conditions, which is what the manufacturer publishes and what the brochure will say. The second is the planning rate: what you can expect across a realistic mix, including transfer time between boards, which is what you put in the capacity model. The third is the measured rate: what your board actually runs at, on your line, with your operators, which is the only number you should quote to anyone outside the project.

Most suppliers give you the first number and let you guess the second. The S4000 documentation is unusual in printing both: the 2024 brochure lists a top speed of 20,000 CPH and an actual speed of 10,000 CPH, and the series manual states 20,000 points per hour as a maximum. Two figures from the same manufacturer, one of them half the other, with the difference named rather than hidden. That is a better starting position than a single number, because it tells you which figure to use for which purpose.

The third number — yours — is the one you have to go and get. Do not accept a demonstration on the demonstrator's board. Send your own bare boards, with your own axial component list, and ask for the insertion count and the elapsed time to be reported separately so you can compute the rate yourself. Ask for the run to include a span change and an angle change, because those are the two events that separate a ceiling from a plan.

Then write the capacity model with the planning rate and keep the ceiling in a footnote. When the machine lands, measure the first week and replace the planning rate with the measured rate. When the mix changes — and it will — re-measure. A capacity model that gets re-measured quarterly is worth more than a better machine bought on a worse assumption.

Southern Machinery runs demo production free of charge with ten of your bare boards, streams the machine running live so you can watch it rather than take our word for it, and issues a written evaluation report afterwards. If you want the third number for your board, that is where it comes from.

Most axial machines are specified against the job that justified them, and then asked to run a product family that did not exist when the order was placed.

Full articles are in English. The summary above carries the argument in your language.

Here is how the specification meeting usually goes. The engineering team has one board in front of them, the one that has been causing the pain. It carries fourteen axial part numbers. The feeder bank is sized at fourteen stations plus a couple of spares, the quotation is approved, and everyone is satisfied. Two years later the same machine is expected to run the whole product family, which between them use fifty-one axial part numbers, and every changeover now involves physically swapping feeders in and out of a bank that was never designed for it.

The diagnosis is not that anyone was careless. It is that a feeder bank is a capital decision being made with a job-level input. The part numbers on the board in front of you are a snapshot; the part numbers across the family are the actual requirement, and the family is what the machine will be asked to serve for the next decade.

The framework is a three-year part-number count. Sit down with the product engineering list and count every distinct axial part number that any board in the family uses, at any volume, including the low runners. Do not filter by current volume — filter by whether the part exists. That count, plus a margin for parts not yet designed, is the station count you should be specifying. If the count is uncomfortable, that is information about the product family, not a reason to under-specify the machine.

The reason this is affordable rather than extravagant is that the feeder bank is modular. The S4000 is offered with 5, 10, 20, 30, 40, 50 and 60 station configurations in the product brochure, and the series manual lists 10, 20, 30 and 40 stations as configurable to need. A modular bank means the station count is a decision you make at order and revisit later, rather than a decision that is welded into the frame.

That changes the shape of the investment. Instead of buying a machine that matches today's job and replacing it when the family grows, you buy a frame with the growth already provided for and populate the stations as the family grows. The capital goes out in the same direction, but it stops being a bet on a single product staying in production.

There is a second, quieter benefit. Spare stations let you keep a feeder loaded and calibrated for a part that runs once a month. On an under-sized bank that part costs you a changeover every time it comes up, which means it gets scheduled in batches, which means work-in-progress and a longer lead time. On an adequately sized bank it costs nothing and it can be inserted in the same pass as everything else.

Review the station count once a year against the current part-number list, the same way you would review a tooling budget. The count only ever moves in one direction, and it is much cheaper to notice that in a review meeting than in a production stoppage.

Send us your axial part-number list — the whole family, not just the current board — and we will size the feeder bank against it and tell you what the next increment would cost if the family grows again. That conversation costs nothing and it usually changes the specification.

The machine handles diodes and resistors all day. Then it meets a jumper wire, and the shift stops being about throughput.

Full articles are in English. The summary above carries the argument in your language.

Ask anyone who has run an axial inserter what the worst part on the board is, and you will get the same answer more often than any other: the jumper. It is the simplest component on the board and the most difficult to feed, because it is not really a component at all — it is a length of tinned copper wire, and wire does not want to be oriented. It bends, it kinks, it twists in the feed path, and when it jams it jams in a place that takes a technician twenty minutes to reach.

The mistake most lines make is to treat jumper wire as an axial part with a different part number. It is not. It has a different feed mechanism, a different failure mode, a different changeover, and in many plants a different gauge. A 0.5 mm wire and a 0.6 mm wire cannot share a feed path without a changeover, so a board that uses both carries a hidden setup cost that never appears in the program time.

Treat it as its own subsystem. First, separate the wire feed from the component feed in your thinking: the questions you ask about it should be about straightening, length control and jam prevention, not about pitch and polarity. Second, count the wire gauges across the product family. Third, ask whether the machine can hold more than one gauge simultaneously, because that single answer decides whether a gauge change is a program line or a production stop.

On the S4000 the jumper has its own axis. A W-axis motor drives a gear pair that straightens the wire in the Z and Y directions and delivers a specified length to the lower cutter position, where the insertion shaft takes it. It is a dedicated mechanism rather than a mode of the component feeder, which is why the failure modes are separable: a wire jam and a component jam are different problems in different places.

The machine is also offered with dual jumper-wire feeders, which the 2024 brochure describes as handling two wire specifications — 0.5 mm and 0.6 mm — without a changeover, with direct insertion from barrel-fed wire and no taping step required. The brochure also states that the feeding arrangement handles both reel-style and drum-style jumper wire packages and is designed to prevent jumper-wire jams during production.

There is a cost argument attached to this that is easy to miss. Southern Machinery documents the insertion process as saving one third of the jumper wire. On a board with a high jumper count, a third of the wire is not a rounding error — it is a line item, and it compounds with the labour released by not cutting and forming those lengths by hand.

Build a wire audit into your product review. For each board, list the jumper lengths, the gauges, and the total metres consumed per thousand boards. When a gauge appears on a new board, ask whether it can be changed to match an existing one before the layout is frozen — it is much cheaper to standardise the wire than to add a feeder.

Send us a drum or a reel of each wire you use, with the gauge and the package type, and we will tell you whether the dual-feeder configuration covers all of them in one setup. Wire samples are small and the answer comes back quickly.

Insertion at any angle is usually sold as a machine capability. It is more usefully understood as permission for your designer to stop routing around the machine.

Full articles are in English. The summary above carries the argument in your language.

Look at a through-hole layout drawn by someone who has spent years working with a 0-and-90-degree insertion machine. You will see the constraint even where it is not written down. Components sit in rows and columns. Where a signal path wants to go diagonally, it goes in two orthogonal jumps instead. Where a part would naturally sit at an angle, it is rotated to the nearest right angle and a jumper wire is added to reach the point it needed to reach.

That is the real cost of a limited insertion angle, and it is not paid at the machine. It is paid in the layout: extra board area, extra jumper wires, extra insertion points, and extra hand assembly on the parts that could not be made to fit. A designer who is told the machine only inserts at 0 and 90 degrees will produce a board that respects that limit, and the board will be worse for it in ways nobody will ever attribute to the machine.

So run a layout test before you specify anything. Take your current family and count how many axial parts sit at angles that are not multiples of 90 degrees. Then count how many jumper wires exist purely to route around the constraint rather than to carry a signal. The second number is usually the larger one, and it is the one that a rotary-table machine reduces.

Mechanically, the mechanism is straightforward. The PCB is carried on a rotary table rather than a fixed fixture. The R axis turns that table to the angle programmed for each line of the insertion program, with the sign indicating clockwise or counter-clockwise. The X and Y axes bring the board to the insertion position, the head does its work, and the table turns again for the next part. There is no tooling change and no operator decision — the angle is a field in the program, alongside the coordinates, the span and the feeder number.

On the S4000 the rotary table sits on the Y worktable and is driven by the R-axis reducer through a timing belt, with the rotary disk constrained by a circular guide rail. The manual describes the arrangement as insertion at any angle in a prearranged sequence, and notes that the same servo-driven rotary table design is what allows the machine to insert both barrel jumpers and tape-and-reel components. Southern Machinery also documents the combination as saving one third of the jumper wire during insertion, which is exactly the routing cost described above.

The habit worth building is a design-for-insertion review that asks about angles explicitly. For every axial part, record the insertion angle. If the answer is only ever 0 or 90, either your layouts are already unconstrained or nobody has tested the constraint. Then ask the question the other way: if the angle were free, would this part move, and would this jumper disappear?

Run that review once on the family and you will usually find two or three parts per board that exist only because of the old constraint. Removing them is a design change, not a machine purchase — but you cannot make it until the machine allows it.

Send us one board drawing and its axial part list, with the insertion angles marked. We will tell you which parts could be simplified if the angle were free, and what that does to the insertion count.

An axial insertion happens in about a third of a second. Engineers who cannot name the stages inside that third of a second cannot diagnose a fault in it.

Full articles are in English. The summary above carries the argument in your language.

The scene is familiar. The machine is bending leads wrong, or cutting component bodies, or leaving parts standing proud of the board. Someone adjusts something, the fault moves, and now a different fault appears. An hour later the settings have drifted from where they started and nobody is sure which change caused which symptom. The root cause of that hour is not the fault. It is that nobody in the room could name the sequence of events that produced it.

So name them. An axial insertion machine performs six distinct operations per component, and each of them has a signature failure. Learn the six names and the fault diagnosis becomes a lookup rather than an experiment.

First, feed. The distribution head cuts a component off the tape and drops it into a link of the discharge chain. The signature failure is a component that arrives in the wrong orientation or not at all, and it is usually a feed-mechanism alignment problem — the three centrelines between the distribution head, the discharge chain links and the chain-track guard plates are supposed to be coincident, and when they drift apart the part does not fall cleanly.

Second, transfer. Pickup wheels move the component from the discharge chain into the insertion chain's clamps. The signature failure is a component that leaves the groove, which is what the hold-down plate above the pickup wheels exists to prevent.

Third, centring. The component is clamped at the centre of its body length before anything cuts it. The signature failure of a centring problem is asymmetric lead bending or a cut component body — and this is the stage most often confused with a cutting problem, because the symptom appears at the cutter. The centring clamps work on spring pressure, and they align the centreline of the component body with the centreline between the two chain clamps.

Fourth, cut. The upper cutter descends and works with the lower cutter to sever the leads from the chain clamp. Cutting length is adjustable from 1.2 to 2.2 mm. The signature failure is a lead that is too long or too short after clinching. One useful detail for diagnosis: the lead-cutting cylinders are insulated from each other and from the metal base specifically so the machine can detect insertion status electrically, which means a cut fault and a missing-component fault are electrically distinguishable.

Fifth, form. The forming tools bend the leads outward into the shape the manual calls the eight-character form, at an angle adjustable from 0 to 35 degrees. The signature failure is a part that does not sit flat, or leads that will not enter the holes square. If the angle is at an extreme, the part will insert but the clinch will not hold it against the board.

Sixth, insert and clinch. The pusher drives the formed leads into the holes and the cutter beneath the board trims and clinches them. The signature failure here is a lead that folds instead of clinching, which usually traces back to a hole that is out of position — and that is what the machine-vision alignment and multi-point mark correction exist to prevent.

Write those six names on a card and keep it at the machine. When the defect rate moves, walk the list in order and stop at the first stage whose signature matches. The published insertion defect rate for this class of machine is under 500 parts per million, which means defects are rare enough that pattern-matching from memory is unreliable — a written sequence is worth more than experience here.

Ask us for the mechanism drawing set for the S4000 and we will send it with the six stages marked on the relevant subassemblies. It is the same material our own engineers use when they commission a machine.

The machine arrives in good condition and works exactly as specified. Three weeks later the project is late, and the delay has nothing to do with the machine.

Full articles are in English. The summary above carries the argument in your language.

Installation delays on insertion equipment follow a recognisable pattern. The machine is set up, powered, and runs its first board. Then the first real product is loaded, and someone discovers that the program has to be created from scratch, that the panel is not the size anyone remembered, that the operator training was scheduled for a week when the trainer is somewhere else, and that the axis zeroing procedure has to be repeated after a timing belt is replaced. None of those is a machine fault. All of them are planning artefacts that did not exist before the machine landed.

The fix is to create five artefacts before the machine arrives, and to own them as documents rather than as knowledge in someone's head.

The first is a zeroing record. Every axis has its own zeroing procedure and they are not interchangeable: the X axis moves to the far right and releases the coupling, the Y axis moves to the topmost position, the H axis requires the timing belt to be removed, the U axis the timing belt, the R axis the timing pulley, and the C axis the reducer coupling. Write down the procedure for each axis, the limit positions, and the date each was last performed. This record is needed at commissioning, after any servo drive replacement, and after any coupling, belt or pulley change — which means it will be needed for the life of the machine.

The second is a program template. Creating a program from scratch on the machine means marking insertion positions at 0 degrees and then at 90 degrees, positioning the board on the middle rail, and stepping the camera to each hole to confirm coordinates. That is a perfectly good procedure, and it is a slow one. If your family has boards that share a layout, a template program with the common lines already entered turns a day of work into an hour.

The third is a panelisation plan. Where several small boards share one panel, the software generates the multi-board program from one board: complete the first board, enter the panel count in X and Y, acquire the top-left and bottom-right coordinates, calculate spacing, preview, choose the sorting method and execute. Knowing which of your products will be panelised, and in what arrangement, means the first production run does not discover it.

The fourth is a signal map. The equipment status page shows the live state of every input and output signal, and the feeder test page drives each station individually, one operation per click. Print the input and output signal tables and annotate them with what each signal means on your line. An operator with that sheet can tell a blocked sensor from a stuck cylinder without calling anyone.

The fifth is a data plan. The machine counts production information automatically and writes it to a record folder, and the software includes a folder for MES-compatible production data. Decide before installation what you want that data to do — feed an OEE dashboard, populate a maintenance schedule, or simply prove the rate to the finance team. A machine that reports data nobody has agreed to use is a machine with an unused feature.

Review those five artefacts annually rather than only at commissioning. The zeroing record especially: it is the document you will reach for on the day the line stops, and it is worth checking that it still matches the machine before you need it.

Southern Machinery includes free installation and training, and will supply the commissioning checklist for your configuration before the machine ships. Ask for it at order stage so your team can prepare the five artefacts while the machine is in transit.

The insertion head is the hardest-working assembly on the machine, and it is made of consumables nobody has inventoried.

Full articles are in English. The summary above carries the argument in your language.

The line stops on a Tuesday afternoon because a lead is not being cut cleanly. The cutter has worn, which is expected, and there is no spare on site, which is not. A technician finds the part number, raises a purchase order, and the machine sits for a week while a component that costs less than a lunch goes through a procurement process designed for capital equipment.

The diagnosis is that nobody built a consumable map. An axial inserter is not a monolithic machine; it is a set of tooling assemblies that wear at different rates, and the wear parts are concentrated in one place: the insertion head and the lead cutter beneath the board. Everything else on the machine — the axes, the servo drives, the frame — is designed to last. The tooling is designed to be replaced, and it is the tooling that stops the line.

The framework is to read the spare parts list as a maintenance schedule rather than as a shopping list. Southern Machinery publishes the spare parts list for the S4000 small-frame configuration, and reading it by subassembly tells you exactly where the consumables are concentrated. The insertion head alone carries a cross cutter block, left and right cutting blades, left and right outer forming blades, left and right pushers, right and left forming blade holders, right and left forming blades, and lower left and right cutters — fifteen wear items in one assembly.

The chain is the second concentration. Each insertion chain uses left and right material clamps, chain pins, three-port chain plates and E-clips in two sizes, supplied in sets of ten. The base carries the left and right lead cutter seats, the movable blade, the triangular blade, and a set of sealing rings in five different sizes. The centring mechanism contributes clamp springs, a material ejector plate and centring clamp plates. And the machine needs two different lubricants at intervals: a lithium-based grease for the insertion head, and an AFC grease specified for the lead screws and guide rails.

The habit to build is an annual consumable review with three columns: what is in the machine, what is on the shelf, and what the lead time is. The last column is the one that decides whether a worn blade is an afternoon or a week. Southern Machinery quotes a one-day lead time on spare parts, which is only useful if the order is placed on the day the wear is noticed rather than the day the line stops.

Two design details make the difference between a planned replacement and an unplanned one. The first is the drawer-style control box, which the brochure describes as being for easy access and simplified maintenance — it turns a panel inspection into a task an operator can do rather than one that needs an electrician. The second is the equipment diagnostics page, where the state of every input and output signal is visible and each feeder station can be driven individually. A worn cutter usually announces itself in the diagnostic data before it announces itself in the scrap bin, if anyone is looking.

Set the review on the calendar now, and put the spare parts list in it as the agenda. The list is specific to the configuration, so it needs to be regenerated whenever the feeder bank grows.

Ask us for the spare parts list for your configuration, with the consumable items grouped by subassembly and the wear-prone items marked. We will also tell you which items we recommend keeping on the shelf from day one, and which can be ordered when the diagnostic data starts to move.