A One-Set Trial Order: How to Test a Die Supplier for the Price of a Sample

Drill point dies are one of the few things you buy from China where a meaningful trial costs almost nothing. A single set, run against your own screw, on your own machine, with your own wire, produces numbers that no amount of correspondence can produce: drill time at the start of the set, drill time near the end of it, how far the point geometry drifted between those two moments, and how many thousand screws the set was still good for when it stopped. A drill point die trial order of one set is therefore not a courtesy sample. It is the cheapest instrument you will ever point at a supplier.

Most buyers waste it anyway. They order one set, hand it to whoever is running the machine that week, note that "it seemed fine", and place a program order on the strength of an impression. The set is consumed and no data survives. The protocol below costs one shift of moderate discipline and turns a sample into a measurement.

Why a die trial produces a number and a conversation cannot

Everything about a die that matters is a rate of change. A new die from almost any competent shop will make an acceptable screw on day one; the dimensions are checkable and the shop had every incentive to get the first article right. What separates a good set from a mediocre one is what happens between screw number one and screw number three hundred thousand — how quickly the drill flanks round over, how quickly the flute fills or the point diameter grows, and whether the failure at the end is gradual and predictable or sudden.

That trajectory cannot be conveyed in a quotation, and it cannot be inferred from a hardness figure alone. Two dies with identical drawings and identical stated hardness can differ by a large factor in service life because of heat-treat uniformity, finishing quality on the forming surfaces, and how well the die matches the wire you actually run. The only honest way to see the trajectory is to run the set and log it.

The second reason is commercial, and it follows on from vetting the supplier itself. Small tooling suppliers describe their track record in company-stated figures — years in the trade, peak monthly output, a customer who switched from a dearer source — and those claims are usually neither verifiable nor false. A controlled trial converts one of them into evidence you own, at close to the price of a sample.

Design it as a comparison, not a demonstration

The single most common design error is running the new set alone and judging it against memory. You need a control.

Reserve a control set. Before the trial set arrives, put aside a fresh, unused set of the same die code from your incumbent supplier — not a partly worn set, and not a different code because it happens to be in the cabinet. If your incumbent tooling varies batch to batch, note the batch and accept that you are comparing to one batch, not to a supplier average.

Lock the wire. This is the confound that ruins most tooling trials. Wire tensile strength, surface coating and drawing lubricant vary between heats and between coils, and those variables move drill time more than a modest difference in die quality does. Run the control set and the trial set on the same wire lot, ideally from the same coil, and record the lot number on the log sheet. If the trial spans a coil change, note where in the run it happened.

Lock the machine and the setup. Same pointing machine, same tooling holder and shut height, same forming speed, same coolant and coolant condition, same operator if you can manage it. If the sets must run on different shifts, run the control on both shifts so you can measure the shift effect and subtract it.

Check the fit before the trial, not during it. A set that does not mount correctly produces a bad result for a reason that has nothing to do with die quality, so settle the machine-fit dimensions at the order stage.

Choose the right die code. Trial your highest-volume standard code, not your most difficult one. A trial exists to predict routine production, and a difficult special tells you about the supplier's engineering ceiling rather than its consistency. The special is a second, separate trial.

The protocol

Run it over one continuous production period, not in scattered fragments.

1. Fix the measurement first. Before either set runs, agree on how drill time is measured and write it down: test plate material and thickness, the fixed load applied, the driver speed, and the number of screws per sample. Drive time only means something as a triplet of plate, load and speed — the standard drill-drive procedures used across the self-drilling screw trade all specify all three, and an in-house rule is fine as long as it is identical for both sets.
2. Run the control set first. Take a drill-time sample at the very start, then at fixed screw counts you choose in advance — a common shape is at start, then at even intervals across the expected life, then at end of set. Record every sample. Keep ten screws from each sample point in a labelled bag.
3. Section a screw at the start. Take one screw from the first sample point, mount and section it lengthwise, and photograph the point. You are recording point diameter, flute length, flute depth and web condition at the flute root while the die is new.
4. Run the set to the end. Do not stop early because the numbers look acceptable. The end of the set is the most informative moment in the whole exercise, and the reason the failure mode matters as much as the count.
5. Record how the set died, using the standard die wear and failure vocabulary. Wear on the drill flanks, chipping at an edge, a lap or fold appearing at the flute root, point diameter creeping past tolerance, or the operator simply rejecting the screws on drive time. Write down which one, and at what screw count.
6. Section a screw at the end. Same procedure as step 3, on a screw from the last good sample point. The pair of sections — start and end — is the clearest picture of dimensional drift you will get, and it is the exhibit to send back to the supplier.
7. Measure the die itself, not only the screws. If you have a profile projector or toolmaker's microscope, put the cavity under it before the set runs and after it is retired. Screw dimensions confound die wear with process drift; the cavity separates them.
8. Repeat all of the above with the trial set, in the same window, on the same wire lot.
9. Compute cost per thousand screws. Set price divided by good screws produced, plus the machine time lost to the changeover. This is the number that decides the sourcing question, and it regularly reverses the ranking produced by set price alone. The general framing sits in how tooling and mold costs actually work in China.

The log sheet

One page per set. Keep both pages together; they are worth more as a pair than separately.

What to send the supplier so the trial is fair

A die maker can only match what it is given, and there are three ways in. You can send a drawing; you can send a screw sample; or you can send the old die itself, or photographs of it. For a like-for-like trial, the old die is the strongest input, because it carries every dimension the drawing omitted and every correction the previous supplier made.

Alongside the physical input, send the machine make and model, the wire grade and diameter you run, your current set life if you know it, and your target volume. A quote and a die built without the machine named are both guesses.

ZLD Precision Mold is a workable example, because it publishes the entry points and the clock a trial needs. Custom dies are built to your drawing, screw sample or old die; the stated route runs engineering review, quote, sampling, trial production, adjustment, then stable output. Stated sampling is 3–4 days standard and 5–7 days custom, volume production about 10 working days, and trial orders are accepted from a single set. Dies are made in SKH high-speed steel (M2, M9, M51) or tungsten carbide, with optional PVD coating — specify which material you want tested, because a carbide set and an HSS set are two different trials with two different economics. The factory runs pre-shipment trial-run validation on in-house test equipment, a useful check that the set forms a point before it ships and no substitute for your own run data. It has stated it cannot supply measured trial-run figures, which is the honest position and the reason the numbers must come from your line.

For the wider mechanics of ordering and paying for a first sample from a Chinese supplier, the sample order process covers the ground this article assumes.

What one set cannot tell you

Be clear about the limits, because over-reading a good trial is how a promising supplier becomes an expensive one.

One set is one set. It proves that this shop can produce one good die set for this code. It does not prove batch-to-batch consistency, which is the property that actually matters over a year of reorders. The second and third sets are the real consistency test, and they should be logged the same way even after you are satisfied.

A trial does not test capacity. A shop that delivers one set in a week may or may not deliver twenty sets across six codes in a month. Capacity is verified separately, by walkthrough and by staged orders.

A trial does not test the response to failure. How a supplier behaves when a set dies early matters as much as how a good set performs, and you only learn it by having a set die. Agree in writing, in advance, what happens if the trial set fails below an agreed screw count.

A single-code trial does not transfer across the size range. A die that performs on a #12 roofing screw says little about a fine-point die for thin sheet. Trial each family you buy in volume.

Settle these commercial points before you order the trial

Four items are commonly left vague and all four are cheap to fix in an email:

• Trial-set price, and whether it differs from the production unit price at volume.
• Whether the sampling or trial cost is credited against a subsequent production order, and if so up to what value and within what window.
• Freight and duty on the trial set, which for a single set can be a large fraction of the total cost, and who arranges it.
• The remedy if the set fails early — replacement, credit, or nothing — and the screw count below which that remedy applies.

None of these are published by most die makers, ZLD included, and none of them are difficult to obtain in writing. Ask before the set is made, not after it is retired. If the trial goes well and you move to a program order, the same discipline applies to the transition itself; the sample-to-production timeline sets out where schedules usually slip.

Common questions

How many screws should a trial run cover?

Run to the natural end of the set rather than to a fixed count. A trial stopped at an arbitrary number tells you that both sets were still working, which you already assumed. If production planning will not allow a set to run to retirement, run both sets to the same count and compare drill-time drift and sectioned geometry instead of set life — a weaker result, but a real one.

Can I trial two suppliers at once instead of using my incumbent as control?

You can, and it is reasonable when you are replacing an incumbent entirely. Run all sets on the same wire lot and machine, and accept that you have a ranking rather than a baseline: you will know which set won without knowing how either compares to what you buy today.

What if I do not have equipment to section a screw?

Section by hand — a cut-off wheel, a flat file and fine paper will show the flute root and point form well enough to photograph, and any local materials lab will mount and etch a handful of screws cheaply. The sectioned pair is the highest-value artefact from the trial; skipping it leaves you with a set-life number and no explanation for it.

Should I tell the supplier I am running a controlled trial?

Yes. It changes nothing about the physics and a great deal about the response. A supplier that knows the set will be measured against a control, logged and sectioned will engage its engineering side rather than its sales side, and will usually ask better questions about your machine and wire — which is itself a useful signal about who you are dealing with.

What to ask the supplier next

Before you place the trial order, send one message containing all of it:

1. Your die code or screw drawing, plus photographs of the old die if one exists, including the ID stamp.
2. Wire grade and diameter, machine make and model, and your current set life if known.
3. The material you want the trial built in — HSS grade, carbide, or coated — and confirmation of the hardness range that will be delivered.
4. Price for one set, sampling time, production lead time, and whether the trial cost credits against a first production order.
5. Freight arrangement for a single set, and the remedy if the set fails below an agreed screw count.
6. Written confirmation of what pre-shipment checking the set will receive, and whether the used set can be returned for wear analysis after the trial.

ZLD Precision Mold is a fair candidate for this exercise: trial orders from one set, dies built to drawing, sample or old die across the L1–L7 range covering IFI #4–#14 and DIN ST2.9–ST6.3, HSS or carbide with an optional coating, and stated sampling of 3–4 days standard or 5–7 days custom. Send the specification, order one set of the code you buy most, and let the log sheet decide. That is a smaller commitment than a week of due-diligence email, and it is the only evidence in this category that is genuinely yours.