A laboratory equipment maker needed an acrylic housing that closed against a silicone gasket. The critical requirement was not the overall size but the flatness of one face — and the first prototype leaked.
The first housing was machined accurately and leaked anyway. The dimensions were within tolerance on delivery; the problem was that the sealing face was no longer flat by the time it was assembled.
Machining removes material from one side of a panel, which unbalances the internal stress that was locked into the sheet during manufacture. Over the following days and weeks, the panel relaxes toward a new equilibrium — and it bows. A face that measured flat on the machine was 0.25 mm out of flat a week later, which is more than enough to break a 3 mm gasket seal.
This is a general property of acrylic, not a defect in a particular sheet, and it is why the order of operations matters more than the nominal tolerance on the drawing.
1. Rough machine. Remove most of the material, leaving a finishing allowance on the sealing face.
2. Anneal. Heat the part to below its softening point, hold long enough for stress to relax, then cool on a controlled ramp. The cooling rate is the part that cannot be accelerated — a fast cool leaves the core stress-relieved and the surface stressed.
3. Finish machine the sealing face. Take the final cut after annealing, so the face is flat in the part's relaxed state rather than in its stressed state.
4. Verify flatness. Measure on a granite surface plate with a dial indicator, at the temperature the part will see in service, and record the reading on the inspection sheet.
The sequence costs about two extra days. Doing it in the other order is faster and produces a part that leaks.
A silicone gasket only seals if the face it closes against is flat and stays flat. On an acrylic housing that means the requirement is not the wall thickness or the clarity — it is the flatness of one machined land, held through assembly and through service.
The land is machined as a flat face 3 mm wide around the opening, and it is machined in the same setup as the mating face so the two are parallel by construction rather than by measurement. We hold it flat to 0.1 mm over its length. Anything less even and the gasket compresses unevenly: one side seals, the other leaks, and a leak test fails intermittently in a way that is hard to trace.
Machining a land into cast sheet introduces local stress, and stress relaxes over weeks. A land that measures flat on the day of inspection can be 0.15 mm out a month later, which is enough to break a seal. The housing was annealed after machining and before the final flatness check, so the measurement recorded on the inspection sheet is the measurement in service rather than the measurement on the bench.
| Requirement | How it was met | Why it matters |
|---|---|---|
| Land flatness | Machined in the same setup as the mating face, 0.1 mm over length | Even gasket compression; a repeatable leak test |
| Dimensional stability | Annealed after machining, before final check | Stops the land moving after inspection |
| Gasket retention | Bonded gasket land rather than a clamped edge | A clamped edge concentrates stress at the fixings |
| Traceability | Mill certificate number on the inspection sheet | Answers the question a device file will ask |
| Cleaning compatibility | Grade chosen against the stated wipe protocol | Daily IPA will craze a standard sheet over months |
Each batch shipped with an inspection sheet recording the sealing face flatness, the critical dimensions, and the ambient temperature at which they were measured. The sheet lot was recorded against the work order, so a part can be traced back to its material.
For a laboratory customer this is not bureaucracy. When an enclosure leaks in the field, the first question is whether it was made wrong or whether something changed after it left. A measurement record answers that question in an hour instead of a week.
The one change we would make is to have raised the annealing requirement at quotation stage rather than after the first prototype failed. At the time it looked like an unnecessary cost on a part that was, by the drawing, a simple machined box.
It is now a standard question on any part with a sealing face: does this have to stay flat after machining? If yes, the sequence includes annealing and the quote reflects it.
| Item | Typical range | Note |
|---|---|---|
| Sampling | 3–7 business days | Plus a fit test against the actual gasket |
| Production | 7–15 business days | Annealing and the flatness check run per batch |
| Documentation | 0–5 days added | Raw-material declaration is free; FAI report is not |
| Cost split | Sheet 20–30 %, machining 25–35 % | Inspection and paperwork 10–15 % on a documented order |
| Rejection risk | Drops sharply with a stated cleaning protocol | Most field failures are crazing, not dimensions |
The single most useful document a buyer can send with a medical or lab enquiry is the cleaning protocol. It costs nothing to write, and it decides the sheet grade, the edge treatment and the likely service life — three things that otherwise get discovered during a sample round.
On budget, the honest statement is that a documented enclosure costs more than an undocumented one by 10 to 15 percent, and almost all of that sits in inspection and paperwork rather than in material. Where a buyer needs to reduce it, the effective move is to specify a tolerance map instead of a blanket tolerance, which reduces both our machining time and their incoming inspection.
More projects
If a face has to stay flat, or a gasket has to seal, tell us at quotation time. The machining sequence changes, and so does the price — better to know both up front.
