We Had an IP67 Test Report, but Water Still Got Into the Production Speakers
Waterproof (IP Rating) Testing Guide - Why Samples Pass IPX4/IP67 but Mass-Production Units Leak, IP Rating Test Methods and Standards, QC for Waterproof Tape, Gaskets and O-Rings, and a Checklist for Catching Gaps Between Certification Reports and Actual Production

Hello, this is StartMade.

A Korean brand selling Bluetooth speakers for camping and outdoor use had 5,000 IP67 waterproof speakers made at an audio factory in Shenzhen, Guangdong. The factory sent over an IP67 test report from a third-party lab, and the two samples the brand received survived 30 minutes underwater in a bathtub. But two months after launch, customer messages started coming in: "The sound went muffled after I played it in the shower," and "I dropped it in a stream, fished it out, and now it won't turn on." Over three months, about 350 units came back as returns and exchanges, 7% of sales.

Tearing down the returned units revealed two causes. The first was the O-ring at the housing joint. Right before mass production the factory had switched O-ring suppliers, and the new O-rings were softer with a cross-section about 0.1mm smaller, so they were not compressed enough. The second was the waterproof foam tape on the inside of the speaker grille. The step where workers press the hand-applied tape down with a roller had been dropped, so some units had tape lifting at the corners. The test report had been issued a year earlier on two samples from a version made before the mold was modified, and the production line had no air leak test at all.

Waterproofing problems almost never show up in a visual pre-shipment inspection. They appear for the first time when a customer uses the product near water. This guide covers what an IP rating actually means, which tests each rating involves, how to control waterproof tape, gaskets and O-rings in mass production, and how to check that the test report and the production units are really the same product.

An IP test report only describes the sample that went into the lab. You can only confirm the waterproofing of production units by locking down the sealing-part specs, running air leak tests on the line, and then retesting units from a production lot.

1. Why Samples Pass but Production Units Leak

Waterproofing is not a property of a single part. It comes from the housing, the sealing parts and the assembly process working together. Samples are a handful of units assembled slowly by skilled engineers. Production units are assembled by line workers, thousands a day, against a takt time. Leaks come from that gap. The causes we see most often in consultations are these.

• Sealing-part changes: The material, hardness or dimensions of O-rings and gaskets differ from the sample. A new supplier or a cheaper substitute looks identical from the outside.
• Insufficient compression: O-rings and gaskets only seal when compressed by a certain amount. Small deviations in groove depth, housing dimensions or screw torque leave them under-compressed. Dimensional control was covered in Episode 134.
• Waterproof tape and bonding: Surface contamination, placement errors, or too little pressure and dwell time leave channels for water between the bonded surfaces.
• Flash and parting lines: When flash (毛刺) or a parting-line step on a molded mating surface crosses the sealing surface, water gets in right there.
• No retest after a design change: The button, charging-port cover or grille is changed, but the old test report keeps being used.

The operating environment adds to this. A lab IP test checks a new product in clean, room-temperature water, while real customers use it under hot shower water, in seawater and in soap suds. Often the product meets water only after a drop or months of temperature swings have already weakened the seals.

2. How to Read an IP Rating and How It Is Tested

IP ratings (Ingress Protection, 防护等级 in Chinese) are defined in IEC 60529 and in its Chinese equivalent, GB/T 4208. The first digit after "IP" covers solids such as dust, and the second covers water. Anything not tested is written as X, so IPX4 means dust protection was not tested and water protection is level 4.

IP67 Does Not Include IPX5 or IPX6

This trips up a lot of people. IPX7 is a test of sitting in still water, while IPX5 and IPX6 blast the product with strong jets. A product that is fine underwater can still leak under a high-pressure stream, which is why the standard does not treat an IPX7 product as also meeting IPX5 or IPX6. If the product has to survive showers or washing down, ask for both ratings, written as "IPX6/IPX7".

Terms used for watches, such as "water resistant" or "10 bar", are not IP ratings. They come from separate standards such as ISO 22810. Mixing IP ratings and watch-style water resistance on a product page easily misleads customers and can turn into a labeling and advertising problem.

3. QC for Waterproof Tape, Gaskets and O-Rings

The waterproofing of production units is decided by the sealing parts and the process that assembles them. Each part has its own check items.

Set Compression as a Number

O-rings and gaskets are usually designed to be compressed by about 15-30% of their original thickness. That figure depends on the O-ring cross-section, the groove depth and the dimensions of the mating part together. Locking down the O-ring spec alone is not enough; the sealing groove depth and mating dimensions also need to be marked as critical dimensions on the drawing. For screw-fastened designs, put the torque value and a cross-pattern tightening sequence in the work instructions too.

With Waterproof Tape, Half the Job Is Application

The same tape will not reach its rated adhesion if the surface is not cleaned or it is only pressed by hand. Make alcohol cleaning before application, a placement jig, a roller or press jig that applies consistent pressure, and the wait time before the next step part of the process standard, and variation drops noticeably. Write the tape brand and model in the spec, and require prior approval for any change. Change control was covered in Episode 119.

4. Air Leak Testing on the Production Line

An IP67 immersion test takes at least 30 minutes per unit, and tested units then have to be dried, so it is hard to apply to every production unit. Factories therefore usually use an air leak test (气密性测试) as the 100% inspection. The tester applies air pressure inside or outside the product and judges leakage by how much the pressure drops over a few seconds.

Results are recorded as numbers and each unit takes only tens of seconds, which suits 100% inspection. But the pass/fail threshold only means something if it is tied to the IP test. The usual approach is to run units that passed IP67 alongside units deliberately made to leak by leaving out a seal, then set a pressure-drop threshold that clearly separates the two groups. Also define the tester settings (test pressure, hold time, pass criteria) and a daily start-up check of the tester with good and leaking master samples.

Adding an actual immersion test on a few units from every lot makes it safer still. For example, pull 3 units from each day's output, immerse them at 1m for 30 minutes, then check function and tear them down. How to sequence this with other reliability processes such as aging was covered in Episode 133.

5. Checking That the Test Report and the Production Units Match

Chinese lab reports usually carry the line "本报告仅对来样负责", meaning the report is only responsible for the sample submitted to the lab. Having a report is not a guarantee that every production unit performs the same. When you receive one, compare the items below one by one.

The most reliable approach is to retest units drawn at random from a production lot by a third party. Fees vary by lab and rating, but a single IPX7 test typically costs a few thousand yuan. That is small compared with the cost of recalls and dead stock. For product certification in general, see Episode 10, and for third-party inspection, Episode 47.

6. Checks at Each Stage

Sample Stage

Run a simple immersion test yourself when samples arrive, but do not approve on that result alone. Tear down a few samples to confirm that the O-rings, gaskets and tape are the parts in the spec, and that there is no flash or step on the sealing surfaces. The lab IP test should be done on samples made with the final mold and final parts.

First Article and Production Stage

In the first production lot, check the supplier and lot numbers of the sealing parts, the air leak test criteria and the master-sample check records. The pre-production meeting and first article process were covered in Episode 123. Getting one IP retest on units drawn at random from the first lot gives you a baseline to compare against if problems come up later.

Pre-Shipment Inspection Stage

A standard AQL inspection does not include waterproofing, so you need to request it separately. Check the 100% air leak test records, and have the inspector run an immersion test or an on-site leak test on units they pick themselves. After immersion, check function and tear down one or two units to look for water marks inside. For pre-shipment inspection, see Episode 100.

7. Sample Request Wording for the Factory

With the wording below in your purchase order or spec, the factory will manage sealing parts and air leak testing as separate items, and it becomes easier to settle responsibility if leaks occur.

"本产品防护等级须达到IP67（按GB/T 4208-2017 / IEC 60529执行）。O型圈、密封垫、防水胶的材质、硬度、尺寸及供应商须按我方确认的规格书执行，如需变更须事先取得我方书面确认。量产须对每台产品进行气密性测试，测试参数及判定标准须与IP67测试结果对应并经我方确认，每日开线前须用合格品及泄漏标准样品点检测试仪并保存记录。首批量产须由我方指定的第三方从生产批次中随机抽取样品送CNAS认可实验室进行IP67测试。每批出货时须提供气密性测试记录。"
(This product must meet IP67 (per GB/T 4208-2017 / IEC 60529). The material, hardness, dimensions and supplier of the O-rings, gaskets and waterproof tape must follow the spec we have approved, and any change requires our prior written approval. Every production unit must undergo an air leak test, with test parameters and pass criteria tied to the IP67 test results and approved by us. Before the line starts each day, the tester must be checked with a good sample and a leak reference sample, and records kept. For the first production lot, a third party designated by us will draw samples at random from the production lot and send them to a CNAS-accredited lab for IP67 testing. Air leak test records must be provided with every shipment.)

8. Three Real Cases

These are anonymized versions of patterns we often see in consultations. Amounts and figures are examples.

Case 1 — IP67 Bluetooth Speaker

This is the case from the start of this article. Working with the factory, the brand went back to the O-rings from the sample-stage supplier and locked the O-ring hardness and cross-section in the spec. A placement jig and a press jig were added to the grille tape step, and an air leak tester was installed at the end of the line for 100% testing. For the next production run, IP67 was retested with samples drawn at random from the production lot. The leak tester and jigs cost about 20,000 yuan, but the exchange costs for the 350 units already returned and the damage left in product reviews were larger.

Case 2 — IP65 Solar Garden Light

Solar garden lights that had an IP65 test report kept drawing claims after the rainy season about water droplets forming inside the lamp cover. It turned out that each worker tightened the four screws holding down the gasket in a different order and with different force, so the gasket was compressed on one side only. As the lights heated up during the day and cooled at night, internal pressure dropped and moisture was drawn in through the less-compressed side. After fixing the electric screwdriver torque, adding a cross-pattern tightening sequence to the work instructions and fitting a waterproof breathable vent to reduce the pressure difference, condensation claims fell sharply. The physics of condensation was also covered in Episode 118.

Case 3 — A Different Product in the Report

A company planning to import an IPX7 wrist fitness band noticed, while reviewing the report from the factory, that the charging contacts on the product in the report photos were different from the production model. The factory confirmed that the report covered an earlier version under the same model name, and that the production model had switched to magnetic charging contacts without being retested. When samples drawn from a production lot were retested, 1 of 2 units failed the first test, and it only passed after the dispensing process around the contacts was fixed. Because it was caught before the order, there was no loss.

9. Common Mistakes

• Treating the factory's IP report as a guarantee for production units
• Getting IPX7 and claiming on the product page that it survives showers and washing down
• Not locking down the material and supplier of O-rings, gaskets and tape in the spec
• Not managing sealing groove depth and mating dimensions as critical dimensions
• No air leak test on the line, or a threshold with no basis
• No IP retest after design or part changes

10. Waterproof (IP Rating) Quality Checklist

Spec and Contract Stage

• The spec states the dust and water ratings and the standard (IEC 60529, GB/T 4208)
• If both jets and immersion must be survived, both ratings (e.g. IPX6/IPX7) are required
• Material, hardness, dimensions and supplier of O-rings, gaskets and waterproof tape are locked down
• Sealing groove depth and mating dimensions are marked as critical dimensions
• Prior approval and retest clauses for sealing-part or design changes are included

Sample and Report Stage

• Report model name, photos and test date match the production version
• The report number was verified with the lab
• Samples were torn down to confirm the sealing parts match the spec

Production and Shipment Stage

• 100% air leak testing is done on the line and records were received
• The air leak threshold is tied to IP test results
• An IP retest was done on units drawn at random from a production lot
• Pre-shipment inspection includes an immersion test or on-site leak test

Wrapping Up - Look at the Production Line Before the Report

Waterproofing defects surface weeks or months after sale, and when they do, reviews and returns arrive all at once. Rather than relying on a single test report, locking down sealing-part specs and running air leak tests on the line costs far less.

There is also something you can do right now. Open the IP report you have and check whether the product in the photos is the one you are selling today, and whether the test date is after the last part change. If either is off, get a retest on a production lot before your next order.

StartMade provides on-the-ground support for writing specs matched to the waterproof rating, locking down sealing-part specs, reviewing test reports for authenticity and consistency, random sampling from production lots and arranging IP retests, and on-site immersion and leak testing before shipment. If leak claims have already started, we begin with teardown analysis of returned units; if you are preparing a new product, we start with the waterproofing spec.