Must an EMI Shielding Case always go to a third-party laboratory?

Answer: Send the first article out, then screen lots on site. Consistent method, state and record matter more than an anechoic room every time.

The report says 60 dB. Can we accept that as it stands?

Answer: No. Ask for test distance, turntable angles, whether feedthroughs were fitted, and latch torque; return any report missing one.

Out of a 50-unit batch, how many get the full test?

Answer: Three: one first article plus two at random. A first-article failure quarantines the batch; one failure in the pair doubles the sample.

A single bond reads 15 milliohms on site. Is that a failure?

Answer: Above 10 milliohms is a failure, but clean the surface and swap probes first; if it still reads 15, fit a new star washer.

Can a customer raise the limit as high as they like?

Answer: They can, but structure and process change with it: 40 dB to 60 dB means an extra gasket run, filtered connectors and a thicker coating.

Can a case that fails re-test stay in service?

Answer: Within 3 dB in one band, downgrade with a shorter interval; 6 dB or more means withdrawal and rework.

Why One Anechoic Report Cannot Sign Off an EMI Shielding Case

Quick answer: Acceptance of an EMI Shielding Case cannot rest on one anechoic report, which proves one band, one build state and one unit. It needs limits split by band and field type, a sample of at least three, eight recorded columns, and a re-test inside 12 months.

The Usual Mistake: Signing Off a Batch From One Report
Which Clause to Cite: Band, Field and State
What to Measure On Site: Four Instruments
Case Versus Bare Shell: Where 20 dB Goes
Failures: Rework, Downgrade or Reject
Records and Re-Test: The Eight Columns
FAQ

1. The Usual Mistake: Signing Off a Batch From One Report

One sample goes to a laboratory, a curve comes back, and the whole batch of an EMI Shielding Case is released on that sheet. The process looks strict and validates one unit, one band, one state.

Those curves usually come from a bare shell, feedthroughs unmounted and the lid at nominal torque. Installed, connector plates, filtered connectors and cable runs change everything, and a 20 dB gap is ordinary. A second mistake is writing the limit as one number, when the two sides may sit three orders of magnitude apart.

2. Which Clause to Cite: Band, Field and State

Every limit needs three parts: band, field type, state.

EMI Shielding Case closed on a worn wooden bench, parts shelving and coiled cables behind, camera and two lenses at right

State means bare or whole. The whole case carries every feedthrough, is latched at service torque and has cables at field length; a unit reading 80 dB bare drops to 30 dB with unfiltered connectors and one seam not seated.

Band	Field	Limit	Points
10 kHz to 150 kHz	Magnetic	20 dB minimum	One per axis
150 kHz to 30 MHz	Electric	40 dB minimum	One per axis
30 MHz to 1 GHz	Plane wave	40 dB minimum	Turntable every 15 degrees, 24 points
1 GHz to 18 GHz	Plane wave	30 dB minimum	Turntable every 15 degrees, 24 points

Each limit also needs an uncertainty statement, since shielding repeats poorly and 3 dB between lots is normal.

3. What to Measure On Site: Four Instruments

Site screening cannot reproduce an anechoic room, but four instruments catch most defects.

Four-point probe: five points inside the lid, worst at or below 0.1 ohm per square
Micro-ohmmeter: plate, lid bond, earth stud, each at or below 10 milliohms
Torque wrench: feedthrough screws at 1.2 N·m
Surface meter: one reading every 100 mm around the seam

Fasteners first, conductivity second, whole-case last. Log 23 ± 5 °C and 45 to 75 percent humidity; flag anything outside.

4. Case Versus Bare Shell: Where 20 dB Goes

Three places account for most of the gap, all in assembly.

EMI Shielding Case open on a pale wood bench, egg-crate foam in the lid, camera body and three lenses in cut-outs

The feedthrough plate comes first: every unfiltered pin is an antenna, and a plate held only by screws leaks from a few tens of megahertz upward. The lid seam is second, since four latches press unevenly. The harness is third, as a braid earthed at one end becomes a radiator.

Settle the bare-versus-whole dispute in the contract, and photograph the assembly at first-article acceptance.

5. Failures: Rework, Downgrade or Reject

Quarantine a failed unit first, so the evidence trail survives: batch, unit, item, reading.

Symptom	Usual cause	Disposition	Re-test
Local resistance high	Fast spray, thin film	Sand back, re-coat	Yes, five points
Whole seam high	Low compression	New gasket, re-set latches	Yes, whole case
Single bond high	Blunt washer, paint intact	New star washer	Yes, same point
Insertion loss low	Unfiltered connector	Filtered type or waveguide	Yes, port by port

Rework needs a count limit: after two recoats adhesion is unpredictable, so the unit is downgraded rather than repaired. Downgrade is not scrap, since strength and sealing still hold, but it must not carry a shielding label. Reject only for a critical item failing in two consecutive batches, or a supplier with no traceable records.

6. Records and Re-Test: The Eight Columns

The acceptance sheet is the baseline for the 12-month re-test. Eight columns:

Batch and serial, traceable to the mould
Band and limit, with method number and year
Build state, bare or whole, plus latch torque
Point layout, turntable angles, antenna distance
Ambient temperature, humidity, date
Raw reading per point, not an average
Verdict and margin, in decibels remaining
Disposition and due date, re-test within 12 months

Twelve months is the default interval, at once after a drop or a soaking. Records pay off as a trend: three re-tests each 2 dB lower justify pulling gasket renewal forward.

Conclusion

Acceptance of an EMI Shielding Case is not a signature but a repeatable verdict: limits in three parts, a sample of three, eight columns, a re-test inside 12 months. A report proves one day, one unit, one band; records, samples and disposition rules carry it forward. Put all three into the contract, and acceptance of an EMI Shielding Case holds.

This article discusses only the container (case body) side of specifications and design intent, and does not cover any handling, loading or use of the contents.