An RFQ That Just Says IP67 Will Cost You a Second Round of Quotes

An RFQ that asks for "IP67" and nothing else comes back with bids you cannot rank, which means a second round of quotes before you have a decision. A usable ingress-protection requirement carries four things: the standard and edition the test will be judged against — IEC 60529 for most packs, ISO 20653 if the pack goes into a vehicle — whether the specimen is a hand-built prototype or a production-representative unit, how many units are tested and in what sequence, and the post-test functional checks that actually define pass. Without those four, every supplier quietly invents their own test and their own definition of pass, and the price on the page stops meaning anything.

I was at my desk on a Thursday with three quote PDFs open in three windows, and every one of them had a line item that said something like "IP67 testing — included." The first one said the sample was a machined prototype housing. The second said "production sample, 1 pc." The third didn't say what was tested at all. Three prices, spread wide enough that I could justify any of them if I squinted.

I sent clarification questions to all three. One came back with an internal bench report where the specimen was described as "housing, sealed per drawing" — no BMS, no connector, no vent. One quoted a third-party lab as an add-on. The third said they'd need to build a fixture and get back to me. Two weeks later, I had three quotes that finally answered the same question, and the ranking had completely flipped from where it started.

That gap between Thursday and the following Thursday is the second round. It's not a supplier problem. It's a wording problem, and it starts in the RFQ.

The rating is a category, not a procedure

IP67 inside IEC 60529 means dust-tight plus immersion in water for a stated depth and duration. It tells you the outcome class. It tells you nothing about what was in the chamber.

Was it a bare housing with the lid bolted on? A full pack with the connector mated, the vent installed, and the BMS powered? Those are different tests with different failure modes, and both can honestly be reported as IP67. When a report describes the specimen as a housing sealed per drawing, that report is real evidence of something — just not the thing you're buying.

If the pack lands on a chassis, the standard is usually different. ISO 20653 covers road vehicles and uses a different code set, including high-pressure spray ratings. If a customer's own validation group is going to hold the pack to that standard later, quoting IEC 60529 numbers in the RFQ just moves the argument downstream.

The edition matters too. IEC 60529 has been amended over the years. A report citing an older edition isn't automatically wrong, but if you're comparing two reports side by side, you need to know which edition each one cites. And if a customer-specific pack specification overrides the general standard, that document belongs in the RFQ, not in the kickoff meeting after award.

Prototype or production representative

This is the field that costs the most to fix later.

A hand-built sample gets assembled by the engineer who designed it. Extra sealant on the gasket. Screws torqued by feel. Adhesive cured overnight on a bench. It passes, and it probably deserves to. Then the production line runs the same drawing with a compression-molded gasket sitting at the low end of its tolerance band, an operator torquing to spec on a timed cycle, and adhesive cured in an oven. Same drawing. Different seal.

So the RFQ says production-representative, and then it says what that means for this bid: units built on the intended production tooling, from production-intent parts, assembled in the production sequence with the production torque values, serialized so the report traces back to a specific build.

If a supplier can only test prototypes at this stage, that's a legitimate answer. It just has to be labeled as one. A prototype result is a design-stage data point. It is not the evidence you'll stand on at first article or PPAP, and if you let it sit in the same column as a production-representative result, you've built a comparison out of two different things.

How many units, and in what order

One unit is one roll of the dice. If the RFQ doesn't state a quantity, you'll get one, because one is the cheapest answer that still technically complies.

I usually ask for at least three, and I state whether those are three separate units or one unit run through both the dust and the water portion. Those are not the same test. A pack that goes through a dust chamber first and then into immersion is being asked a harder question, because dust sitting in a connector interface changes how the seal behaves under pressure. If your governing standard specifies the sequence, say so. If it doesn't, decide which evidence you actually want and write that down.

Then decide what happens to the units afterward. Any honest ingress test ends with someone opening the pack to look for moisture where a camera can't see — along the seal line, inside the connector shell, behind the vent. Once it's open, it's scrap. So the RFQ should state who owns the tested units and whether the report includes teardown photos.

The pass line is function, not "no water seen"

This is where most RFQs stop too early. "No water observed inside" is a visual observation made by the person who ran the test. It tells you nothing about the pack.

Here's what goes into my acceptance criteria. Insulation resistance measured after the test, with the threshold and test voltage stated — if you don't set them, you'll get a number with no verdict attached. BMS wake-up and communication over the pack's normal interface. A charge and discharge cycle if the fixture supports it. A visual inspection of the connector, the vent, the terminals, and every fastener, looking for corrosion and not just for standing water. And the teardown check for moisture tracking along the seal.

Then the report has to state pass or fail against the criteria in the RFQ. I want a verdict, not a data table. Reports that present readings and leave the judgment to the reader are the ones that turn into email threads in month nine, when everyone has forgotten the context and nobody wants to sign their name to the interpretation.

Ask who runs it, and whether they're accredited

Requiring the supplier to identify the test laboratory, and to state whether that lab holds accreditation for the specific method being cited, is the cheapest line in the whole document. It costs nothing to ask.

What you're checking is whether the accreditation scope actually covers the clause, or covers something adjacent to it. A certificate that lists a different method is not coverage for this one. ISO/IEC 17025 is the framework most buyers point to, but the framework isn't the point — the scope is. Ask for the scope, not the certificate.

I also spell out the report contents up front: unit serial numbers, build configuration, photos of the setup, the standard and edition cited, the date. If the report doesn't identify the unit, you cannot tie it to the build you're buying, and six months from now nobody can either.

Some suppliers will push back and want to run it on their internal bench. That's a real tradeoff, not automatically a red flag. An internal bench under the supplier's own quality system can be perfectly adequate for a design-stage screen. It's simply a different class of evidence, and it should be labeled that way in the bid so you're not putting it in the same column as an accredited third-party report from a competing quote.

Ask how the seal is made

The RFQ should ask whether the enclosure is sealed by gasket, dispensed adhesive, potting, or a welded seam. This isn't trivia for the engineering file. Each choice changes what happens after the test.

A gasket is serviceable. Someone can open the pack, replace it, and close it — which raises the question of what happens to the test evidence once a unit has been opened in the field. Adhesive bonding and welded seams aren't designed to come apart, which is fine for a genuinely sealed pack with no field service, but it changes your repair strategy and your spares plan. Potting is the most robust of the four and the least repairable, and it changes the thermal path through the enclosure as well.

I want that answer on the quote form, in a checkbox, not discovered during a factory audit.

The line that keeps the evidence alive

A test report is evidence of the thing that was tested. Nothing more. So the RFQ needs a line stating that any change to the gasket, the connector, the vent, or the assembly process invalidates the prior test and requires a retest before shipment.

Spell out the triggers: gasket material, hardness, or thickness. Connector family or its sealing method. Vent type or cracking pressure. Torque values, adhesive cure schedule, assembly sequence.

I watched a good report go stale when a supplier swapped a gasket compound because the original had a long lead time. Nobody hid it. Nobody thought to mention it either, because the drawing number never changed. What turned into a field return later traced back to that seal line, and the report everyone kept pointing at described a build that hadn't existed for more than a year.

How that failure gets handled in the commercial documents is not the RFQ's job — that's work for counsel, and it's a different conversation. The RFQ's job is narrower and simpler: make sure the evidence on the table describes the build you're actually buying.

The tradeoff, stated honestly

A full protocol makes the RFQ longer, and it will cost you something. Test units get consumed and never come back. An accredited lab adds cost and lead time. Some smaller suppliers will no-bid rather than commit to a third-party report, and a few will pad their price to cover the uncertainty. If you're on a tight schedule, that's a real friction point, and pretending otherwise would be dishonest.

It's still cheaper than the alternative. The second round isn't free — it's weeks of engineering and sourcing time, arriving exactly when you're trying to lock a build schedule. And there's a worse version of the same problem: no second round at all, award on the vague line, and then find out at first article that the IP67 report in the bid package came from a housing-only test with no connector and no vent.

So the protocol goes in the RFQ. Not in the follow-up questions, not in the clarification call, not in the kickoff meeting after award. If it's in the RFQ, every bidder answers the same question, and the only thing left to compare is whether they can actually do the work.