Buying Injection Mould Tooling in China: Shot Life, Steel Grades and the Trial Report Nobody Reads

A tooling buyer I work with sent me a photo last month of a mould that had just finished a 12-month production run in Vietnam. Two of the six cavities had a visible parting-line step you could catch a fingernail on, and the ejector pins had punched about 0.3 mm into the base of every part. The factory blamed the press. The press was fine. The mould was simply made to a price that assumed nobody would ever ask it to run 400,000 shots, and nobody on the buying side ever wrote down how many shots it was supposed to survive.

Moulds are the one purchase in a B2B sourcing file where the real cost hides in a sentence nobody puts in the PO. Here is how I have learned to buy them, mostly by getting it wrong first.

Price per mould means nothing until you know the tool steel and the cavity count

A 4-cavity mould for a palm-sized ABS housing in the Ningbo area typically lands somewhere between USD 1,800 and USD 4,500. That range is not a negotiating range. It is mostly a materials range. A P20 pre-hardened block and a 718H block look identical from the outside and machine at nearly the same speed, so the difference in material cost is easy to absorb or easy to shave, and the toolmaker decides which one you get based on what you sign.

For anything over about 100,000 shots I want 718H, NAK80 or a proper S136 for the cavity inserts. For a short-run bridge tool or a market test, P20 is fine and I will happily pay half. For glass-filled nylon or anything with abrasive filler, the cavity steel needs hardening and the gate area needs a replaceable insert, otherwise you will be welding gates in eleven months.

Ask for the steel grade in writing on the drawing, then check it two ways: a small coupon of the same stock off the shelf, and a hardness reading on the finished cavity with a portable tester. The reading should sit around 30 HRC for P20, 33 to 36 for 718H, and 48 to 52 for hardened S136. If the toolmaker cannot produce a hardness reading, that tells you more than any certificate.

Who is going to run maintenance, and what happens to the mould if you leave

This is the part that costs the most and gets negotiated the least. A mould that runs 300,000 shots needs polishing, re-cutting of vents, replacement of ejector pins and ejection sleeves, and possibly re-working of the cooling channel if scale builds up. That is not a repair, it is a scheduled service, and it needs about 3 to 5 percent of the mould cost per year set aside.

Before signing, I ask three questions. Does the price include a spare set of fragile inserts? What is the promised surface polish standard and who pays for re-polishing after polish marks show in the part? And if we move to a second moulder, will you hand over the mould along with the 3D files of any inserts you manufactured yourself?

The last question is the one that ends badly. Some factories agree to "mould ownership rests with the customer" in the contract but keep the electrode set and the CNC programme for the inserts, so the tool physically leaves but cannot be maintained anywhere else. If you want real mobility, put it in the contract that electrodes, wire-cut programs and insert 3D models in native format are part of the deliverable. Yes, most factories will push back. The ones that agree without blinking and without extra cost are usually the ones who do this for European customers already.

The trial report is the actual deliverable

A mould without a trial report is a promise. What I want at T1, in writing, with photos:

First-article dimensional report on the critical dimensions only, with the drawing tolerance next to each measured value. Not a full CMM of everything, which hides the two dimensions that actually matter.
Process record from the trial: melt temperature, injection pressure, holding pressure, cooling time, cycle time. This is what your moulder will need when the mould is transferred, and it is also proof that the tool was actually run.
Cavity-to-cavity variation, measured on the same dimension across all cavities. If a 4-cavity tool has a 0.15 mm spread on a 0.1 mm tolerance, the tool is not done.
Water flow check on each cooling circuit, at least noting how each circuit flows in litres per minute. Blocked or short-circuited circuits are a common and invisible cause of long cycle times.

I have seen T1 reports where the trial was run at a 40-second cycle and the production quote assumed 26 seconds. Nobody lied. The trial was run slow to make the parts pass, and the production cycle was a different number. Compare cycle time on the T1 report against the cycle time in the quotation before you sign the tool off.

Where the money really leaks after the tool is approved

Three things go wrong repeatedly in the first six months.

The first is cooling layout. Toolmakers quote on cavity count and steel, and cooling lines are where they save time. A well-cooled tool gets channels about 12 to 15 mm from the cavity wall, with baffles or bubblers into deep cores. A poorly cooled tool has one straight line through the block. You cannot see the difference until you measure cycle time on your own moulder, and by then the tool has been paid for.

The second is gate finish. Gates that are hand-filed rather than EDM'd with a proper electrode look fine at T1 and leave a visible witness mark on every part for the next three years. Ask for gate photos at 20x magnification at T1 and at the 100,000-shot service point.

The third is spare parts. Ejector pins, springs, locating rings and a spare set of O-rings for the cooling circuit cost maybe USD 150 to USD 300 per mould. Buying them at the same time as the tool is cheap. Buying them eighteen months later from a Chinese toolmaker who has archived a four-year-old job is a month of emails.

What a good mould purchase looks like on one page

One page with the following on it: shot life target and the steel grade that matches it, cavity count, gate type and cut method, cooling circuit layout and flow record, ejection method and expected force, spare parts list included, maintenance intervals and who pays, ownership of electrodes and native CAD, and a T1 acceptance protocol with the measured dimensions listed.

If a toolmaker can sign that page without arguing about any of it, you have found a partner. If they start making phone calls, keep looking. The mould will outlive most of the people who negotiated it, and it will still be making parts long after everyone has forgotten what the price was.