Where the Cost of Going From IP67 to IP68 Really Comes From

The IP rating change itself is nearly free. What costs money is everything that has to hold the seal. The lines that actually move between IP67 and IP68 are the connector interface, the gasket material and the assembly time needed to control its compression, leak-test station capacity and first-pass yield, the NRE that comes back with every design or process change, the serviceability you give up and later pay for in warranty and field repair, and the program volume over which the sealing tooling and test fixtures get amortized. Two bids that name different ratings cannot be compared until both are pulled to the same test protocol, the same specimen count, and the same end-of-line leak test.

I was standing at the end of a supplier's leak test aisle on second shift, close enough to hear the compressed air charge the fixture and bleed off into the decay gauge. A pack sat clamped in the fixture with its lid on and its connectors capped. The needle held, drifted, held, then the station light went amber and then red. The technician unclamped the pack, dropped it in a gray bin, and reached for the next one without looking at it. The bin had a hand-written label on the side and a fair amount of company.

That bin is where the IP68 cost actually lives. I asked the line engineer what happens to a pack that fails. On the IP67 version of the same product, the lid came off, the gasket got reseated or replaced, and the unit went back through the station. On this one, the lid was bonded and the housing was designed to close once. To get in, you cut it. The housing is scrap, and the cells inside are scrap with it. He said it the way you would say the weather.

I took that back to my desk, where two quotes were open side by side. One supplier quoted IP67, one quoted IP68, and the gap between the two unit prices looked almost reasonable. It was not. Here is what I had to go find.

The first digit did not change, so start there

Both IP67 and IP68 carry the 6 for dust. Only the second digit moves. The dust test scope and its cost are not the difference, so anyone who quotes you a bigger environmental test package as the explanation for the delta is guessing.

The water digit is where it gets slippery. IPX7 is temporary immersion, defined at a fixed depth and time. IPX8 is continuous immersion under conditions the manufacturer specifies and the customer agrees to. That second clause does more work than any other sentence in this piece. IP68 by itself is not a test. It is a placeholder for a depth and a duration that somebody has to write down on a piece of paper.

If you do not write it down, your supplier will, and you will find out what they chose when the qualification report lands. I have held two datasheets that both said IP68 and were not remotely the same test.

The connector is a step, not a slope

An open header with a splash shield is often perfectly fine at IP67. IP68 pushes you toward a sealed connector: interfacial seal, individual wire seals, overmolded or potted backshell, sometimes a metal shell with a bonded gasket. That is a step change in component cost, not a percentage.

What buyers miss is that the step happens on both halves. If you are not buying the mating half, the cost does not disappear, it moves into your customer's harness. That shows up later as a commercial conversation, not a technical one, and it usually arrives with the word "unfair" in it.

The connector decision also drags a panel cutout change, a keying change, a bend radius problem inside the pack, and a housing revision that needs a new tool. One decision, four cost lines.

Gaskets are cheap; holding the compression is not

The gasket material line itself is modest. Silicone versus EPDM versus a dispensed foam, durometer, compression set, low-temperature behavior. Real differences, but not the ones that blow up a quote.

The expensive part is holding that gasket at a consistent compression. You need a positive stop so the lid cannot over-compress, because an over-compressed gasket takes a set and leaks later, at temperature, in the field. That means tighter housing tolerances, a fixture or press to close the lid evenly, a torque sequence if there are fasteners, and cycle time at the assembly station.

That money shows up as labor and capital, not as a bill of materials line. If your supplier's quote has a labor rate and a cycle time, you can see it. If the quote has neither, you cannot see it at all, and you should assume it is in there.

Dispensed gaskets add a cure or staging step. Molded gaskets add a part that has to be placed by a person, and placement is a defect opportunity. Both cost money, in different places.

Yield, and the stations behind it

First-pass yield on a seal falls as the requirement tightens. That is not a supplier failing you, it is geometry: more seal length, more interfaces, more chances for a pinched gasket or a contaminated bond line.

Whatever that yield is, you pay for the whole leak-test loop around it. Station time per unit, the second station if the test takes longer than takt, reject handling, rework labor, and scrap. The scrap is the one to hunt for, and the question you ask is simple: what happens to a unit that fails end-of-line leak test? Ask it again for a unit that fails after the lid is bonded.

On the IP67 pack, a leak failure costs a gasket and some labor. On the bonded IP68 pack, it can cost the housing, the cells, the BMS, and every assembly step that happened before the test. Same defect, completely different price, and it is almost never in a first-pass quote.

End-of-line leak test is not the ingress test

These get conflated constantly. End-of-line leak test is a production screen, per unit, ongoing. The IP qualification test is a design test, priced per configuration, run on a specimen quantity, and repeated whenever the seal changes. One is a rate, the other is a project.

There is also a component people forget to cost. A pack that is genuinely sealed for continuous immersion usually needs a pressure-equalization vent, because a fully closed housing will breathe through the seal under thermal cycling and pull moisture back in through the same path. The vent is its own part with its own cost. It also changes your leak test method, because you cannot pressure-decay test a housing with a hole in it, at least not without blocking the vent or testing the vent separately. Both workarounds cost station time.

Testing is priced per configuration

Any change to the sealing design reopens the IP qualification. New housing geometry, new gasket material, a different dispense pattern, a new connector. Each retest carries a specimen count, a lab slot, and a place in the schedule.

Suppliers handle this differently. Some absorb the first retest and say so. Some charge it and say so. Some say nothing and send a change order in month nine. That is a real difference between the two bids on your desk and it has nothing to do with unit price. While you are in there, look at how many specimens each supplier tested. If one tested three units and the other tested ten, the second one is not expensive, it is honest.

What you give up when the pack cannot be opened

A bonded lid, a potted assembly, a one-way connector: the repair path becomes whole-pack replacement. That is not automatically wrong. For some products it is the right engineering call. But it is a cost, and it lands in warranty and field repair, in a budget that is not the BOM.

If the warranty term runs five years, you carry that decision for five years. And if the customer's field technicians will open it with a knife anyway, you will find that out in year two, from a claim.

The connector has the same flavor. A sealed connector can be mated once, but the interfacial seal is a one-time item. Reconnecting in the field may not restore the rating, and the service procedure has to say so out loud or somebody will assume otherwise.

Volume multiplies everything above it

Tooling, fixtures, test stations, retest NRE: these are fixed costs wearing a per-unit costume. How much they actually add depends entirely on volume. The same sealing investment that rounds to nothing at high volume is a visible line item at low volume.

Ask whether the leak-test station is charged as a per-unit test fee or recovered through tooling, and ask what amortization volume sits inside the quote. If the answer is not your program's forecast, you are subsidizing something else. Ask whether the fixture is dedicated. If it is shared, you may be paying for someone else's launch.

How I normalize two bids that quote different ratings

I put both quotes on the same scope before I look at the price again:

• The IPX8 depth and duration in writing, plus the specimen count used to establish it.
• Whether the test was run on the pack as shipped, with connectors mated or unmated, vents present or blocked, cable exits in place.
• End-of-line leak test method, and whether it is 100 percent or sampled.
• Per-unit test time charged, retest charge, and who owns scrap on a failed bonded assembly.
• Tooling, fixture, and NRE as separate lines, with the amortization volume stated.
• The serviceability assumption and the warranty term attached to it.

Once both quotes sit on that common scope, the gap between them usually changes shape. Sometimes it survives. Sometimes you find that one quote was IP67 wearing an IP68 label.

I should be honest about the other direction, though. A higher rating is sometimes exactly right: a pack mounted low on a vehicle, an outdoor enclosure that sits in standing water, a product that gets hosed down. And sometimes IP67 is genuinely adequate and the money is better spent on thermal margin or a better BMS. The point is not that IP68 is overpriced. The point is that the number you are comparing across two bids is usually not the number you think it is.