Five Months of Tooling: How the 311 Exit Device Was Made

The phone rang at eleven at night and somebody was still in the office to answer it. The caller wanted unit prices on exit devices we already make. We quoted, and then the conversation turned: he wanted to come and see the parts in his hands, the next day.

He arrived the following afternoon with a colleague. They spent the visit going through the range, and they found several products that suited their own factory, not one. Then they produced a sample of something we did not make, and asked whether we could produce it. We said yes. That sample became the 311.

What follows is not a marketing story about responsiveness. It is an account of what the yes actually committed us to, written down because buyers asking for a custom exit device are quoted a price and a lead time, and almost never told what sits underneath them.

THREE MOULDS, THREE DIFFERENT TRADES. An exit device is not one material, so it is not one mould. The 311 needed three: an ABS mould for the push bar body, a steel stamping die for the pressed components, and a zinc-alloy die-casting mould for the cast parts. Before any of them could be cut there was wire-cutting and drawing, the geometry has to be fixed on paper and in the electrode before it is fixed in steel.

Each of those three moulds belongs to a different craftsman. They are not interchangeable skills: the man who can hold a tolerance in a zinc die is not the man who knows how ABS shrinks as it cools, and neither of them sets up a stamping die. So every correction is a three-way conversation, and a change to one part propagates into the other two. When a problem appeared in trial, all three had to agree on what the fix was before anyone cut metal.

THE PAUSE. The development then stopped for more than a year for reasons outside the project (the international situation at the time) and was picked up again afterwards. That gap is worth stating plainly rather than editing out of the story. Tooling programs are not immune to the world around them, and a buyer who is told a tooling schedule should know that the schedule assumes nothing interrupts it.

THREE MONTHS TO CUT, FIVE MONTHS TO FINISH. Once the work resumed, the complete set of three moulds was finished in three months. That is the number people expect to hear, and it is the smaller half of the truth. What followed was trial shots, and trials find things. Where a trial showed something had to change, the mould itself was modified and shot again. Then the zinc-alloy and stamped-steel components went through polishing, grinding, electroplating and assembly.

From first trial to a tool that could be trusted to run took five months in total. Not five months of waiting: five months of making a change, shooting again, and looking at what came out.

Why correction is the long half, and cutting is the short one

Three months to cut, five months to finish. That ratio surprises people, and there is a physical reason for it rather than a scheduling one.

A mould can be opened, not closed: removing steel from a cavity is a morning's work. Putting it back means welding and re-cutting, which changes the hardness of that area and risks the surface finish for every shot afterwards. So tools are cut deliberately steel safe (with material left in) and the trials are what tell the toolmaker where to take it away. That is why the number of trials cannot be predicted at the start: each one removes the uncertainty it was run to find.

Each material misbehaves in its own direction: ABS shrinks as it cools, and not evenly, thick sections pull in more than thin ones, so a bar that came out of the cavity straight can be bowed an hour later. Zinc die-castings come out with flash at the parting line and can carry porosity where the metal froze before it was fully fed. Pressed steel springs back after the die opens, so a part formed to ninety degrees relaxes to something slightly over. None of these are faults. They are properties, and the tool has to be cut to anticipate them, which means cut, shoot, measure the error, and correct.

And the three corrections interact: a change to the zinc casting to cure a porosity problem moves a face that the pressed component registers against. Which is the point made above about three craftsmen: a correction is not a task, it is a negotiation between three tools that all have to agree afterwards.

This is also why a trial part is not evidence on its own. A trial component is usually hand-finished (flash filed off, a face lapped flat) to see whether the geometry works. It says the design is right. It does not say the tool is ready to produce that part a hundred thousand times without a hand touching it, and those are two separate approvals.

WHY ONE ANGLE DECIDES WHETHER IT WORKS. The requirement is not only that the part looks right. Every metal component has to fit every other one, and an exit device has to do a mechanical job while it does it. Get one angle wrong and the product does not work, it is not a cosmetic defect, it is a device that will not release. That is the difference between tooling a piece of hardware and tooling a piece of trim, and it is why trial-and-correction takes longer than cutting did.

Four questions to settle before a single mould is cut

Tooling is the one purchase in this trade where the commercial terms are harder than the engineering, and where almost nobody writes them down until there is a dispute. Four questions, settled in advance, remove nearly all of it.

Who pays, and how? Tooling is either invoiced as a cost or amortised into the unit price over an agreed quantity. Both are normal. They are not the same deal, and they behave very differently if the volume does not arrive: an amortised tool that never reaches its quantity leaves an unpaid balance somebody has to own.

Who owns the mould? Paying for a tool and owning a tool are two different things, and the difference decides whether you can move production elsewhere. Settle it in writing, in the same document as the price.

Is the geometry exclusive, and for how long? A distinct question from ownership. A buyer may fund a tool and still be happy for the factory to sell the product generally; another may need the shape restricted to their market. Both happen. The one arrangement that causes trouble is the one where neither party said.

What happens to the tool when it is idle, and when it is worn? A mould is a physical object that sits in a rack, rusts if it is not oiled, and has a finite life in shots before the cavity needs refurbishing. Who stores it, who maintains it, who pays for the refurbishment, and what notice is required to move it, these are boring until the day a buyer wants the tool shipped and discovers no one agreed the answer.

For the 311 specifically, we answered these by developing it ourselves: the moulds are ours, which is exactly why the geometry can be changed for the next customer who needs it changed. That is a statement about that product, not a set of standard terms, on a tool a customer funds, the answers are whatever the two parties agree, and our position is that they should be on the same page as the price rather than in somebody's memory.

WHAT THIS MEANS IF YOU ARE THE BUYER. Three things follow from this account, and they are the reason it is published rather than kept as an anecdote.

First, a tooled product and a catalog product are two different purchases with two different clocks. If a supplier quotes the same lead time for both, one of those numbers is wrong.

Second, the tooling is where a custom exit device is either right or wrong, and it is decided months before the first production order ships. That is an argument for visiting, for asking to see trial shots, and for approving a sample rather than a drawing.

Third (and this is the part we would rather say out loud) the 311 exists because a customer brought us a sample and we spent five months on it. The moulds are ours. That means the geometry can be changed for the next customer who needs it changed, including the center distance, which is why the 311 carries both 72mm (2-13/16") and 92mm (3-5/8") cylinder centers rather than one. A supplier who buys finished devices cannot offer that, because the mould is not theirs to alter.

The stages to approve, and what each one actually proves

A tooling program is not one approval at the end. It is a short series of them, and knowing what each stage can and cannot tell you is what stops a buyer signing off the wrong thing.

The drawing: proves the intent, and nothing else. It is the cheapest place to change anything and the only place where a change costs nothing, which is the argument for spending longer here than feels comfortable. Everything after this is steel.

The first trial shots: proves the geometry: does it come out of the cavity, does it assemble, does the mechanism move. Expect them to be imperfect, and expect them to be hand-finished. Ask specifically which surfaces were touched by hand, because those are the surfaces the tool has not yet proved it can produce.

The dimensional report: proves the tool against the drawing, feature by feature. This is the stage most often skipped on a hardware program and it is the one that finds the angle described above, a part can assemble on a bench and still be out of tolerance in a way that shows up on the hundredth unit.

A pilot run: proves repeatability, which is a different property from correctness. Parts taken from a continuous run, not selected from it, and preferably measured at the start and the end. A tool that drifts across a run is a tool that will produce a good first container and an argument on the second.

The first production sample, approved and retained: both parties keep one. It is the physical reference that settles a later dispute in an afternoon instead of a fortnight, and it costs one part.

None of this is exotic quality-management vocabulary. It is five checkpoints, and the reason to write them into the order at the start is that each one is cheap where it sits and expensive one stage later. That is the whole economics of tooling, and it is why five months of trial and correction is not a delay, it is the work.

Every product has a story behind it. This one took five months, three craftsmen and a phone call at eleven at night.

If you have a part of your own in mind, our guide to custom door hardware sets out the same steps for your project: what can be changed, what needs a new mould, and what is quoted per project.