RF Shielding, Penetrations & Installer Methods

RF shielding is a CTTA-driven countermeasure, not a default ICD 705 requirement.

How a requirement arrives. The CTTA's TEMPEST Countermeasures Review may call for no shielding, foil on selected walls, shielded racks, filtered power only, or a full enclosure. It may state an attenuation level. One practitioner cites a CTTA review that specified "60dB of RF attenuation," with milestone inspections, testing and mock-ups. That's an example, not a standard.

See: RED/BLACK, TEMPEST & EMI Filters for the CTTA review and inspectable space.

Sources IC Tech Spec v1.5.1 (NAVFAC mirror) · ICS 705-1 · UFC 4-010-05 · SPG: De-mystifying TEMPEST · Cushman & Wakefield: POA&M rescission report

There's no single "SCIF shield." The CTTA's requirement, the attenuation level, the budget and the building set the type. Figures below are vendor or practitioner examples, not requirements.

Known weak point of foil systems. In Compliance Magazine notes that screws fastening the outer drywall layer penetrate the foil and create leakage. Foil systems generally can't reach the high-frequency performance of welded or modular steel rooms.

Sources Universal Shielding: Enclosure construction · Universal Modular: RF shielding in modular SCIFs · ETS-Lindgren architectural shielding · In Compliance: SCIF and RF secured facility design, part 2 · Wikipedia: Electromagnetic shielding

Attenuation (shielding effectiveness) is how much weaker a signal is on one side of the shield than the other, expressed in dB. Every 10 dB is a factor of ten in power, so 100 dB is a 10^10 power ratio. Three rules explain almost every shielding failure.

1. Low-frequency magnetic fields are the hardest. ETS-Lindgren publishes its steel honeycomb vent at "25 dB @ 1 KHz" for magnetic fields, but "120 dB" at higher frequencies. Its brass version is weaker at low frequency. Steel absorbs magnetic fields. Copper mostly reflects electric fields.

2. Holes must be much smaller than the wavelength. Wikipedia: "Any holes… must be significantly smaller than the wavelength." A seam gap, an unsealed conduit or a missing screw can leak like an open window. Higher frequencies have shorter wavelengths, so small gaps matter more as frequency rises.

3. A room is only as good as its worst seam, door or penetration. Holland Shielding: "a single weak point can affect the complete room."

Published component claims (vendor examples, not requirements):

The CTTA or the specification sets the required level. A 130 dB door does nothing for a room whose sprinkler pipe passes through an unbonded hole.

Sources ETS-Lindgren waveguide vent datasheet · ETS-Lindgren DKE door datasheet · Wikipedia: Electromagnetic shielding · Holland Shielding: 101 EMI shielding tips · Holland Shielding: Honeycomb ventilation panels · Tech-Etch glossary

Rule zero: every conductor, pipe, duct or opening that crosses the boundary is a potential leak. Tech Spec 3.G.1: "All penetrations of perimeter walls shall be kept to a minimum."

The right treatment depends on whether the boundary is a full RF shield or a non-shielded SCIF perimeter where the CTTA has directed specific countermeasures. The logic is often opposite: at a shield you bond metal to make it part of the shield; at a non-shielded perimeter you may break the metal path.

Also from Tech Spec Chapter 3.G:

• "Utilities servicing areas other than the SCIF shall not transit the SCIF unless mitigated with AO approval" (3.G.3).
• "Electrical Utilities should enter the SCIF at a single point" (3.G.4).
• Interior distribution on treated walls is surface-mounted, in raceway, or on an added furred wall (3.G.5).

UFC 4-010-05 adds that "All metallic perimeter penetrations present compromising emanations (CE) and pose TEMPEST hazards that must be addressed" per the TEMPEST Countermeasure Review.

See: Penetrations, Utilities & Life Safety for acoustic, man-bar and firestop rules that apply at the same openings.

Sources IC Tech Spec v1.5.1 (NAVFAC mirror) · UFC 4-010-05 · In Compliance: SCIF and RF secured facility design · NSTISSAM TEMPEST/2-95 (historical mirror)

Why one entry point. Every separate penetration needs its own bond, gasket, inspection and test point. Grouping them is cheaper, easier to inspect and far easier to test. It's also what the public criteria ask for:

• Tech Spec 3.G.4: "Electrical Utilities should enter the SCIF at a single point."
• UFC 4-010-05 §3-4.11.1: "Utilities (power and signal) should enter at a single point. Seal all utility penetrations to mitigate acoustic emanations and covert entry."
• UFC 4-010-05 §3-4.19: "Cabling must enter the protected area from a single location."

The penetration panel. On a shielded room, the shield vendor typically provides a penetration panel (or entry plate) where these are mounted and bonded together:

Spare capacity. Tech Spec 3.G.6 allows expansion conduit "filled with acoustic fill and capped." That closure is acoustic. On an RF shield, a spare must also be RF-sealed (for example, a bonded cap or a waveguide sized for future fiber) and tested with the room. This is an inference from shielding practice, so confirm the detail with the shield vendor.

The penetration schedule. Before the shield is fabricated, build a schedule listing every crossing system: power, lighting, fire alarm, sprinkler, HVAC, BMS, plumbing, IDS, ACS, CCTV, PA and mass notification, phone, LAN, RED circuits and fiber. For each, record the treatment, the size and the responsible trade. Then freeze it.

Sources IC Tech Spec v1.5.1 (NAVFAC mirror) · UFC 4-010-05 · Holland Shielding: Shielded rooms · Genisco: Filter panels

What it looks like. A dielectric union, an insulating flange kit, or a non-conductive flexible duct connector. The product must be listed for the system it's installed in.

Why it's used at a non-shielded SCIF perimeter.

• Metal that crosses the perimeter can carry signals picked up inside to the outside. Historical 2-95 warned that a conductive sleeve around lines leaving the inspectable space "can become a fortuitous conductor and could require isolation."
• Tech Spec 3.G.2: "Metallic penetrations may require TEMPEST countermeasures, to include dielectric breaks or grounding, when recommended by the CTTA."
• For DoD projects, UFC 4-010-05 §3-4.11.2 gives treatments that apply when TEMPEST countermeasures are required: non-conductive duct connections, non-conductive conduit unions, and grounding of conduit, sprinkler pipe and refrigerant lines near the penetration.

The opposite logic at an RF shield. A shield works by being continuous metal, so at a shield you bond every metal penetration so it becomes part of the shield. Any opening is then sized as a waveguide. A dielectric section may also appear outside a shield so building piping doesn't inject currents onto it, but that configuration isn't universal.

Code coordination. Non-metallic sections in sprinkler, gas or fire-rated assemblies must be listed for that use and firestopped. Coordinate with the fire protection engineer and the AHJ before ordering.

Sources IC Tech Spec v1.5.1 (NAVFAC mirror) · UFC 4-010-05 · NSTISSAM TEMPEST/2-95 (historical mirror)

The physics in one sentence. Below cutoff, the field "decays exponentially along the waveguide axis and the wave is thus evanescent" (Wikipedia). Above cutoff, the tube becomes a pipe for RF.

Two dimensions control performance:

Where waveguides are used:

• Honeycomb vents: hundreds of small cells in parallel pass air.
• Pipe waveguides: pass sprinkler water, plumbing or pneumatic lines.
• Fiber waveguide tubes: pass optical fiber.
• Access ports for filling, draining or inspection, where the shield design includes them.

The one rule you can't break: never put a conductor through a waveguide. A wire, a metallic cable, or a fiber cable with a steel strength member turns the tube into a coaxial line. A coaxial line has no cutoff, so RF passes straight through at every frequency. Historical 2-95 made the same point about metal strength members acting as "fortuitous conductors."

Sources Wikipedia: Cutoff frequency · Wikipedia: Waveguide filter · RF Cafe: Waveguide formulas · Holland Shielding: 101 EMI shielding tips · NSTISSAM TEMPEST/2-95 (historical mirror)

These are textbook formulas (RF Cafe tabulates the constant 1.841 for the circular TE₁₁ mode). They explain designs and let you sanity-check a submittal. The shield vendor's tested design governs, not a hand calculation.

Attenuation below cutoff (derived: α = (2π/λc)·√(1 − (f/fc)²) nepers per unit length, with 1 Np = 8.686 dB). For frequencies well below cutoff (f ≪ fc):

• Circular: A ≈ 32 × L/D dB, about 32 dB per diameter of length
• Rectangular: A ≈ 27.3 × L/a dB
• At f = 0.5 fc, multiply by about 0.87. Attenuation falls to zero as f approaches fc.

Worked example: a 1-inch tube, 5 inches long, air-filled

Same tube, filled with water: fc ≈ 6.9 / √80 ≈ 0.77 GHz. Frequencies above roughly 770 MHz now propagate through the water column. A design that looked excellent dry offers far less at UHF and microwave frequencies.

Lessons:

1. The math is ideal. Real limits usually come from the weld or bond where the tube meets the shield.
2. Rules of thumb for length-to-diameter ratio circulate in industry. Use the vendor's tested dimensions, not a rule of thumb.
3. Know what fills the tube before you size it.

Sources RF Cafe: Waveguide formulas · Wikipedia: Cutoff frequency · Wikipedia: Waveguide filter

The problem. An HVAC duct is a large hole in the shield. A 12-inch-wide rectangular opening has a cutoff near 0.5 GHz (5.9 / 12), so frequencies above that pass through it freely.

The solution. A honeycomb vent splits the opening into many small hexagonal cells. Each cell is a tiny waveguide with a high cutoff frequency, and together they pass enough air. Tech-Etch: hexagonal cells "permit smooth airflow… while also blocking electromagnetic radiation leakage."

Ideal vs measured. A 1/8-inch cell, 1/2 inch deep, has a theoretical cutoff around 47–55 GHz and about 128 dB per ideal cell. Measured panels are far lower. Tech-Etch publishes 40–75 dB for a chem-film 1/8-inch cell, 1/2-inch panel, and 60–105 dB cross-cell, over 100 kHz–10 GHz. Seams, cell joints, frame bonding and finish all cost performance.

Installation takeaways:

• Cross-cell construction uses a "minimal two layers… stepped and rotated 90 degrees" (Holland) for higher performance.
• Bond the vent frame by solder, braze, weld, or "RF gasket seals" with "Monel or tin-coated gaskets" (ETS-Lindgren). An unbonded frame leaks around the honeycomb.
• Steel beats brass at low-frequency magnetic fields.
• The mechanical engineer must account for pressure drop. Honeycomb adds resistance, and acoustic silencers or Z-ducts add more.
• Don't cut, drill or screw through the honeycomb, and don't let duct insulation pins pierce it.

Sources Tech-Etch: Vents and filters · ETS-Lindgren waveguide air vents · ETS-Lindgren waveguide vent datasheet · Holland Shielding: Honeycomb ventilation panels · IC Tech Spec v1.5.1 (NAVFAC mirror)

Fire protection is mandatory, and a sprinkler main is a metal pipe crossing the shield. The shield-side answer is an engineered pipe waveguide: a metal section welded or bonded to the shield, with a diameter and length chosen so the frequencies of concern are well below cutoff.

Why water changes the math. Water has a very high dielectric constant (εr ≈ 80 at low frequency, a textbook value). Filling a waveguide with it lowers the cutoff frequency by about √80, or roughly 9×.

A pipe waveguide sized for air can fail badly once the system is charged with water. Sprinkler, domestic water and drain waveguides must be engineered for the fluid they carry. Wet-pipe sprinklers are always full.

Coordination checklist:

Routing rules that come from the Tech Spec:

• A sprinkler main serving the SCIF may enter. A main feeding other tenant areas "shall not transit the SCIF unless mitigated with AO approval" (3.G.3).
• At a non-shielded perimeter, metallic sprinkler pipe may need a dielectric break or grounding when the CTTA recommends it (3.G.2). UFC 4-010-05 §3-4.11.2 gives the DoD grounding treatment.

Sources RF Cafe: Waveguide formulas · Wikipedia: Cutoff frequency · IC Tech Spec v1.5.1 (NAVFAC mirror) · UFC 4-010-05 · In Compliance: SCIF and RF secured facility design, part 2

Why fiber is the preferred data path. Holland Shielding describes the approach as "converting the signal to light and bringing the signal into the shielded room via a fiber optic cable through a waveguide. The fiber optic cable is non-conductive and will not bring in unwanted signals." Historical 2-95 noted that optical fibers "do not conduct or radiate radio frequency interference." In Compliance Magazine (2022) recommends fiber through an inexpensive waveguide over filtering copper data lines.

How it's done:

 OUTSIDE                            | SHIELD |                         INSIDE
 BLACK switch -> media converter -> all-dielectric fiber ==[waveguide]== fiber -> converter -> device
 UNCLASSIFIED CIRCUIT ONLY            tube welded/bonded to shield     (converter on FILTERED power)
 (building power)

Read the diagram carefully: it solves RF, not security. It shows an unclassified (BLACK) circuit entering a shielded space. The waveguide-and-fiber detail is the right answer to the RF problem, and it says nothing about the physical-security problem. Fiber is not an exemption from protected distribution. CNSSI No. 7003 covers "wire line and optical fiber distribution systems," and states that "PDS are used to protect all unencrypted NSI through areas of lesser classification or control." A classified circuit crossing the perimeter needs a PDS, NSA-approved (Type 1) encryption, or an approved CSfC solution — the fiber and the waveguide do not substitute for any of them.

The rules:

What still needs a decision:

• Whether BLACK fiber may share pathways with RED equipment is a CTTA call. See: RED/BLACK, TEMPEST & EMI Filters.
• Whether a filtered copper data path is acceptable instead of fiber is also a CTTA call.
• Converter placement inside the room counts as equipment placement for RED/BLACK purposes.
• No standard we could find names media converters explicitly. For an unclassified circuit their placement is an engineering choice; for a classified one, a converter handling plaintext outside the perimeter is a RED device in an area of lesser control, which CNSSI 7003 ¶22 does not permit without PDS or approved encryption. Confirm with the AO and CTTA.

Sources CNSSI No. 7003, Protected Distribution Systems (DCSA copy) · ICD 705 · Holland Shielding: 101 EMI shielding tips · NSTISSAM TEMPEST/2-95 (historical mirror) · In Compliance: SCIF and RF secured facility design · UFC 4-010-05

How an RF door seals. A normal door has gaps at every edge. An RF door closes them with metal-to-metal contact all the way around:

ETS-Lindgren's DKE door uses "a durable bronze double knife-edge extrusion" that mates with "four rows of beryllium copper finger-stock… replaceable with standard tools." Its vendor claim is "130 dB at frequencies up to 10 GHz, 100 dB up to 40 GHz, and 90 dB up to 100 GHz," tested per "MIL-STD-285, NSA 65-6/NSA 94-106, ITSG-02/IEEE 299 or EN 50147-1."

SCIF-specific conflicts:

• Acoustic vs RF. In Compliance Magazine reports that no single door on the market at the time met both typical SCIF acoustic ratings and top-tier RF performance. Designs use a vestibule, or a separate RF door plus an acoustic door.
• Security hardware. Tech Spec 3.E.5.g: "SCIF Perimeter doors shall meet TEMPEST requirements per CTTA guidance." The FF-L-2890 deadbolt, electrified lock, REX, balanced magnetic switch and card reader must all be installed without disturbing the knife-edge or finger stock. Their wiring goes through signal filters or fiber.
• Life safety. Egress hardware must work with the RF seal. Coordinate with the AHJ.

Installer rules:

1. Use only door hardware the RF door manufacturer approves, mounted with its templates.
2. Never drill the knife-edge, the finger-stock channel or the shielded door skin.
3. Route door-hardware conductors to the penetration panel, not through the frame.
4. Hang and adjust per the manufacturer. Historical NSA 94-106 required that "Door hinges shall support the door's weight without sagging… positive closure." That remains good practice.

See: Doors, Locks & Security Containers for FF-L-2890 and IDS door requirements.

Sources ETS-Lindgren DKE door datasheet · In Compliance: SCIF and RF secured facility design, part 2 · IC Tech Spec v1.5.1 (NAVFAC mirror) · NSA 94-106 (historical mirror)

A shield that passed its acceptance test degrades with use. Doors are the moving part, and they wear first.

RF door maintenance items:

Seams and joints. Holland Shielding notes that seam conductivity should be "more or less identical to that of the basic material." Seams in modular panels, foil laps and welded joints all need continuous electrical contact.

Corrosion. Dissimilar metals in contact corrode, and corroded joints lose conductivity:

• Holland: galvanic corrosion occurs above "0.3 volts in a salty environment, or 0.5 volts in an environment with just water."
• Historical NSA 94-106 required "noncorrosive material" and avoiding dissimilar metals.
• Humid mechanical spaces, condensate and cleaning chemicals speed up corrosion at gaskets and finger stock.

When to retest. Any change that affects the shield (a new penetration, a door replacement, a panel removed and reinstalled, a seam repair) calls for review and, as directed, a retest. The shield vendor and CTTA define what counts as a change for your facility.

Sources ETS-Lindgren DKE door datasheet · Holland Shielding: 101 EMI shielding tips · ETS-Lindgren EMI/EMP filter manual · NSA 94-106 (historical mirror)

Grounding a shield is not the same as grounding a building. Getting it wrong creates currents on the shield that can defeat it, or a safety hazard, or both.

Single-point shield ground. The shield connects to the building grounding system at one designed point, and is isolated from incidental contact everywhere else. This avoids ground loops that drive currents across the shield surface.

The safety ground is still required.

• Holland Shielding: the shield "should be connected to the common ground of the building, only for safety reasons."
• Filters need a continuous equipment grounding conductor (ETS-Lindgren).
• Don't create a separate, unbonded "clean" earth rod. The electrical engineer and AHJ confirm the electrode system and bonding under the electrical code.

Incidental contacts break single-point grounding:

RF bonds are short and wide. Historical 2-95: "Long pigtail and long ground wire shield terminations drastically reduce shielding effectiveness." Cable shields terminate "360 degrees around the cable" (Holland). Use flat braid or direct metal-to-metal contact, not long round wire.

RED/BLACK grounding (equipotential planes, single-point schemes, separated return paths) is designed by the engineer of record and the CTTA, not in the field. See: RED/BLACK, TEMPEST & EMI Filters.

Sources CNSSI No. 7003, Protected Distribution Systems (DCSA copy) · ICD 705 · Holland Shielding: 101 EMI shielding tips · ETS-Lindgren EMI/EMP filter manual · NSA 94-106 (historical mirror) · NSTISSAM TEMPEST/2-95 (historical mirror) · Premier: Filter installation best practices

What's tested. A shielding effectiveness test compares signal levels with and without the shield in the path, at the frequencies the specification names. It covers every door, seam, vent, filter and penetration.

IEEE 299 status:

Other methods on current datasheets: MIL-STD-285 (1954), NSA 65-6 and 94-106 (historical), ITSG-02, EN 50147-1 and ASTM E1851-15. Test labs and government customers still reference several of these. Confirm which method and edition your contract cites, the frequencies, and the pass/fail levels, before the shield is designed.

Sequencing rules:

1. Mock-ups and milestone inspections, if the CTTA or specification requires them (one practitioner describes CTTA-directed mock-ups and testing).
2. Seam and panel inspection before interior finishes cover the shield.
3. Final acceptance test only after all penetrations, filters, signal filters, waveguides, vents, doors and door hardware are installed.
4. Repair and retest any failures.
5. Any later penetration or modification means review, then retest.

Sources IEEE 299-2006 · IEEE P299 · Wikipedia: Tempest (codename) · ETS-Lindgren DKE door datasheet · SPG: De-mystifying TEMPEST · Keystone Compliance: NSA 94-106 testing

Every screw, anchor, powder-actuated pin or box that pierces the foil or steel is a hole in the shield. In Compliance Magazine notes that even the screws holding the outer drywall layer can penetrate architectural foil and create leakage. Tech Spec 3.G.5 therefore requires utility distribution on the interior of an acoustic- or RF-treated perimeter wall to be "surface mounted, contained in a raceway, or an additional wall… using furring strips as stand-off," so the treated layer is never punctured. The practical consequences are that devices can't be flush-mounted in the shielded wall, TVs and whiteboards can't be anchored straight through it, and hangers can't pass through the shielded ceiling. Installer mounting methods (Z-brackets, surface raceway, furred walls and stand-offs) are covered in a separate part of this module.

Sources IC Tech Spec v1.5.1 (NAVFAC mirror) · In Compliance: SCIF and RF secured facility design, part 2

Use this before the shield is ordered. Every "no" is a future change order.

Authority and scope

• [ ] The CTTA's written countermeasure recommendations are in hand before funds are committed.
• [ ] RF scope is confirmed: none, architectural foil, shielded racks, or a full enclosure.
• [ ] Test method, edition, frequencies and acceptance criteria are written into the contract.
• [ ] Responsibility for passing the test (shield vendor, GC or both) is assigned.

Shield vendor details

• [ ] Penetration panel layout, filter panel, waveguide and vent sizes are received.
• [ ] Door hardware interface and approved hardware list are received.
• [ ] Single-point ground location and isolation details are received.
• [ ] Seam, lap and fastening details are received for foil or modular systems.

Every crossing system is listed, with a single point of entry planned

• [ ] Power, lighting and emergency power
• [ ] Fire alarm, mass notification and sprinkler
• [ ] HVAC supply, return and exhaust; BMS
• [ ] Plumbing, drains, condensate and refrigerant
• [ ] IDS, ACS, CCTV, PA and intercom
• [ ] Telephone, LAN, RED circuits and fiber

Engineering reviews

• [ ] Electrical: filter leakage and GFCI/ground-fault coordination, generator and UPS kVAR, voltage drop.
• [ ] Mechanical: honeycomb pressure drop, duct break location, acoustic silencers.
• [ ] Fire protection: listed sprinkler waveguide and dielectric fittings, engineered for a water-filled pipe; AHJ acceptance.
• [ ] Architectural: furred walls or surface raceway so nothing fastens through the shield.

Information handling

• [ ] Drawings, penetration schedules, CTTA recommendations and test reports are controlled as sensitive information. They stay off public shares and open bid portals.

See: Construction Security & Build Sequence for the Construction Security Plan.

Sources IC Tech Spec v1.5.1 (NAVFAC mirror) · UFC 4-010-05 · Premier: Filters and generator startup (low kVAR) · NSTISSAM TEMPEST/2-95 (historical mirror) · UpGuard: Black Box, Red Disk (INSCOM)

Sources Premier: Filter installation best practices · ETS-Lindgren EMI/EMP filter manual · Holland Shielding: 101 EMI shielding tips · NSTISSAM TEMPEST/2-95 (historical mirror)

A shielded room is built from the outside in and tested last. Out-of-order work is the most expensive shielding mistake, because fixing it means opening finished work and retesting.

Notes for the schedule:

• ICS 705-1 wants RF shielding "planned for installation during initial construction." Build the test and any retest time into the baseline schedule.
• Coordinate the shield test with the AO's other acceptance activities (acoustic testing, IDS certification, TSCM) so a late shield repair doesn't re-open finished areas. See: Construction Security & Build Sequence.
• Construction security rules still apply throughout. The shield is part of the SCIF perimeter under construction.

Sources ICS 705-1 · IC Tech Spec v1.5.1 (NAVFAC mirror) · IEEE 299-2006 · NSA 94-106 (historical mirror) · Premier: Filter installation best practices

Why one screw matters. The fastener leaves a hole plus a metal stub that carries current through the skin and re-radiates it on the other side. Foil can tear or corrode around it over time. Holland Shielding adds that a wire not fully connected to the shield "will work as an antenna."

You won't see the leak. MIL-HDBK-1195 §5.1.1: "leaks in a shielded enclosure cannot be detected by visual inspection." Only instrument testing finds them.

The weld-over in UFGS is the shield vendor's detail, not a field fix. Only the shield designer decides whether a fastener is sealed by welding, and only on shields designed for it.

See: Penetrations, Utilities & Life Safety for acoustic sealing at the same openings.

Sources IC Tech Spec v1.5.1 · UFC 4-010-05 · UFGS 13 49 20.00 10 · MIL-HDBK-1195 (historical mirror) · MIL-HDBK-1195 §5.1 (historical mirror) · In Compliance: SCIF and RF secured facility design, part 2 (practitioner) · Holland Shielding: 101 EMI shielding tips

The door contact still has to meet every IDS rule. The mount only changes how it's attached.

Default for an RF-shielded door (design reasoning, not a published requirement):

1. Surface HSS on the protected side, on the switch maker's bracket, fastened to the door maker's hardware prep or to an interior strut. Never fasten through a frame that forms part of the shield.
2. Magnet on the inside face of the leaf, clear of the knife-edge, finger stock and seals.
3. Armored leads into surface raceway, then to the IDS panel inside.
4. Only filtered or fiber-converted circuits cross the penetration panel.
5. Install before the final shield test, so the test covers the room as it will be used.

On an STC-rated (acoustic-only) door, the same side logic applies. Screws must not reach acoustic fill or compress or bridge the seals.

Placement. Mount near the latch-side head corner so the 7.D.3 thickness test passes on heavy, thick doors. Keep brackets clear of door bottoms, gaskets, finger stock and knife-edges. A bracket that holds the door slightly open fails both the STC and RF tests.

Gap drift. Sag, cam-lift motion and seals taking a set all change the switch gap. Re-check at turnover and at each semiannual IDS test.

See: Intrusion Detection & UL 2050 / Extent 3 for walk-test and semiannual test rules.

Sources UFC 4-010-05 · IC Tech Spec, Ch. 5–end (v1.4 archive; text source) · IC Tech Spec v1.5.1 · Magnasphere HSS Level 2 CSI spec

Pick the method from the door and shield details, then confirm it with the shield vendor, the door maker and the switch maker. Rows marked derived are installer reasoning from the sources, not published approvals.

Sources Magnasphere HSS L2 installation accessories · Magnasphere surface mount accessories · IC Tech Spec v1.5.1 · UFC 4-010-05 · Pete Milham: ICD 705 construction (practitioner) · ETS-Lindgren architectural shielding datasheet

The installation sheet for the exact switch model governs. Magnasphere's UL 634 Level 2 documents were publicly fetchable, so they're the worked example here. Other makers' install sheets weren't reviewed.

Concealed model (HSS L2C):

• "installation must be in a metal frame" for UL compliance.
• "Prepare a deburred 1" hole directly opposite to the mounting site of the magnet module."
• "All wires must route within the metal frame."
• Maximum operating gap stated as 1/8 in (3.2 mm).
• "The tamper circuit (if equipped) must be wired to a 24-hr protection circuit."

Consequence: a concealed contact means drilling the frame and running wire inside it. Never field-drill a shielded or knife-edge frame for a concealed switch. Use a surface model on the protected side, or a frame the RF door maker prepared at the factory.

Surface models (L2S/L2D CSI specification):

• "Universal cross-hole mounting for in-swing or out-swing door installations," with #10 screw holes.
• An integrated removal tamper that alarms when the switch is removed from its mounting surface.
• Leads with "stainless steel armored jacket or conduit."
• Execution: "Per Manufacturer's Instructions."

Removal tamper depends on a mechanical mount. The tamper document says the tamper magnet "is inserted into this mounting hole" and capped. The switch is designed to be screwed down over it.

Magnet warning. The magnets are "extremely powerful: Take precautions to avoid the magnetic attraction between the magnet and ferrous metals." Steel doors and frames alter the field, which is why spacers exist.

Listed accessories (Magnasphere part numbers): HSS L-Bracket (1625), HSS Z-Bracket (1620), Adjustable Magnet Bracket (1229), plastic shims (SP062) and aluminum shims (SP062-FIRE), 1" magnet housing spacer (1102), 3" magnet module spacer (1731), 1/4" aluminum spacer (1578), Armored Cable Routing Clamps (1807), Tamper Test Magnet (1384), plus roll-up and overhead door bracket kits.

Nuisance alarms. NAVFAC design manual DM-13.02 notes that door movement from wind or pressure causes nuisance alarms on balanced magnetic switches. Pressurized vestibules and heavy acoustic doors are candidates.

Sources Magnasphere downloads · Magnasphere HSS L2C installation instructions · Magnasphere HSS L2C removal tamper · Magnasphere HSS L2X 800 series · Magnasphere HSS Level 2 CSI spec · Magnasphere HSS L2 installation accessories · NAVFAC DM-13.02

Double-sided tape and construction adhesive look like an easy way to avoid a hole. For a UL 634 Level II high-security switch on a SCIF door, they aren't a documented option.

What the public documents actually say:

• No public UL 634 Level II document describes adhesive-mounting the switch or its magnet.
• The surface-mount specification calls for screw mounting through cross holes, and execution "Per Manufacturer's Instructions."
• The removal tamper relies on a magnet "inserted into this mounting hole." A switch stuck to a surface has nothing for that tamper to detect in the way the design intends. An adhesive-only mount defeats that design.
• Adhesive bonds creep, soften with heat and let go as seals and doors move. A gap that drifts can fail the 7.D.3 test months after acceptance.

Where adhesive does appear. ETS-Lindgren's architectural shielding foil is itself set with "Vinyl-to-vinyl adhesive (or equivalent)." That's a shield material detail from the shield maker. It doesn't make adhesive an approved way to attach security devices, and adhesive compatibility with a particular foil or finish wasn't confirmed.

What to do instead:

• Screw the switch to a manufacturer bracket, spacer or strut that's fastened where the shield vendor and door maker allow.
• If a raceway clip or cable tie mount really must be adhered to a treated surface, get the shield vendor's written approval of the product and the surface first.
• Keep the switch's removal tamper functional and wired to a 24-hour zone.

Sources Magnasphere HSS Level 2 CSI spec · Magnasphere HSS L2C removal tamper · ETS-Lindgren architectural shielding datasheet · IC Tech Spec, Ch. 5–end (v1.4 archive; text source)

Once a device is mounted, its cable has to reach the single point of entry without touching or piercing the shield along the way.

Rules for the route:

How it works in the field (installer practice derived from the rules):

1. Device to raceway. Armored switch leads go into surface raceway or surface EMT on the protected side. Use the switch maker's armored cable clamps where supplied.
2. Raceway supports. Fasten to furring, strut or vendor-approved points, never through foil, sheet steel or shield panels. On acoustic-only walls, surface fastening into the finish board is the normal approach.
3. Keep off the skin. Interior conduit and strut don't touch the shield except at designed bonds.
4. At the panel. Low-voltage circuits cross only through signal filters, or convert to fiber and pass through a waveguide. Copper, PoE, armored fiber and steel fish tape never go through a waveguide.
5. Spares. On an RF shield, a spare also needs an RF closure the shield vendor details and the test covers. The acoustic cap alone isn't enough.

Non-shielded perimeter. The same single-entry, labeling and two-side sealing rules apply. Metallic conduit gets whatever break or grounding the TEMPEST Countermeasure Review directs.

See: RED/BLACK, TEMPEST & EMI Filters for signal filter placement.

Sources UFGS 13 49 20.00 10 · UFC 4-010-05 · IC Tech Spec, Ch. 5–end (v1.4 archive; text source) · IC Tech Spec v1.5.1 · Magnasphere surface mount accessories · Holland Shielding: 101 EMI shielding tips

Installers hear opposite advice about metal conduit where it meets a shield. Both views are published.

Why they aren't really contradictory. The handbook describes how a penetration that must be metal is made part of the shield: welded or brazed all the way around, by the shield builder. The practitioner warns against tying the shield to building conduit and piping in general, which creates unplanned current paths. Both reject casual clamp-on bonds.

The resolution:

1. Build each penetration exactly as the shield designer details it. Weld, braze, filter, waveguide or dielectric section: whatever the drawing shows, and nothing else.
2. Keep interior conduit and strut off the shield everywhere except the designed bonds.
3. Never improvise clamp or gasket bonds at the shield, even when an inspector or a code habit suggests a jumper.
4. Prefer fiber for data where the design allows, so fewer metallic lines cross at all.
5. Ask before bonding anything to a shield. An unrequested jumper can create a second ground path.

At a non-shielded perimeter the logic is different. UFC 4-010-05 §3-4.11.2, "Unless directed otherwise by the TEMPEST Countermeasure Review," gives metal conduit or pipe "a nonconductive union inside the perimeter adjacent to the penetration," or grounding near the penetration with a No. 4 wire to the building grounding system.

Sources MIL-HDBK-1195 §2.6.4 (historical mirror) · In Compliance: SCIF and RF secured facility design, part 2 (practitioner) · UFC 4-010-05 · UFGS 13 49 20.00 10

When a room needs many outlets, readers, speakers or panels on a treated perimeter wall, the cleanest answer is often a second wall in front of it.

What the rules allow:

• Tech Spec 3.G.5 permits "an additional wall… constructed using furring strips as stand-off from the existing wall assembly."
• UFC 4-010-05 §3-4.4.7 says to "construct a furred out wall for routing of utilities," which is what makes recessed outlet boxes and recessed panels acceptable.

Build rules (installer practice derived from the rules and shield details):

The inspectability tradeoff. Some AOs dislike furred walls because they hide the boundary (practitioner, Milham). UFC §3-4.1.1 asks designers to minimize "above ceiling obstructions on the controlled and uncontrolled side of the secure perimeter." Surface raceway versus furring is an AO preference, so settle it at design review.

Hardening layers. On walls with expanded metal or plywood for forced-entry resistance and visual evidence of penetration, don't cut or notch the layer for cable. Plan the entry point before the layer goes in, using an AO-approved detail. An unexplained patch is exactly what inspectors look for.

Interior (non-perimeter) sound-rated walls. Recessed boxes aren't allowed on treated perimeter walls. Practitioners report that interior compartment walls may carry them if the design keeps the rating; confirm with the AO, and note UFC 4-010-05 §3-4.4.8 still prohibits recessed fire extinguisher cabinets on sound-rated compartmented area walls. Offset boxes on opposite faces and use listed putty pads. Non-rated smoke and acoustic cable pathways exist, for example STI EZ-Path 44NEZ, described as "high-performance acoustical separation." Follow the rated-assembly listing and the acoustician.

Sources IC Tech Spec v1.5.1 · UFC 4-010-05 · Pete Milham: ICD 705 construction (practitioner) · STI EZ-Path 44NEZ

The primary entrance needs a reader outside the door, which puts a device and its cable on the uncontrolled side. The Tech Spec controls what can be reachable there.

Applying it (installer practice derived from the rules):

• Mount the reader over a small dedicated penetration so the cable goes straight into the SCIF.
• Nothing reachable outside: no exterior junction box, splice, exposed conduit or terminal strip.
• Use tamper-resistant fasteners and wire the reader's tamper output.
• Seal the penetration on both sides.
• Treat the reader data path as a line to outside equipment: encrypted or AO-approved. Don't describe a protocol as "FIPS AES certified" without the product's certificate.
• On an RF shield, the reader cable is a conductor crossing the shield. It goes through a signal filter or a fiber conversion at the penetration panel, as the shield designer details. Don't pass it through the door frame.

See: Access Control, Identity & Hirsch for head-end location and credential rules.

Sources IC Tech Spec, Ch. 5–end (v1.4 archive; text source) · UFC 4-010-05 · IC Tech Spec v1.5.1

An electrified lock or trim on the leaf needs power and signal carried from the frame. On an RF or heavy acoustic door, the path matters.

What the published sources say:

• UFGS 13 49 20.00 10 §2.6.1.5–.6: "Install electric connectivity for sensors, alarms, and electric interlocking devices in accordance with the door manufacturer's instructions, the approved drawings, and Section 26 20 00…" Vestibules with shielded doors at each end get electric interlocks.
• UFC 4-010-05 §3-4.6.5: electric strikes or electrified mortise locks with an ACS "must have a positive engagement, fail secure, and approved under UL 1034."
• UFC 4-010-05 §3-4.6.6: hinges "reinforced to a minimum of 7 gauge, and be cam-lift for acoustical door assemblies."
• ETS-Lindgren's DKE RF door uses "Three heavy-duty aluminum hinges… six thrust bearings" and lists options including "Interface with Security Systems," "Emergency Release" and "Battery Back-up Upgrade."

Comparison (design reasoning, not a published requirement):

Why the door maker controls this. Historical MIL-HDBK-1195 §2.6.2.1 notes that some manufacturers "refuse to guarantee performance of their door and door frames in shielding provided by others, unless they are in control of the door installation."

Steps:

1. Get the door maker's wiring path and interface diagram as a submittal before assigning IDS or ACS points.
2. Route lock and sensor conductors to the penetration panel, not through the frame into the uncontrolled side.
3. Confirm the fail-secure electrified hardware still allows mechanical free egress. See: Doors, Locks & Security Containers.

Sources UFGS 13 49 20.00 10 · UFC 4-010-05 · ETS-Lindgren DKE door datasheet · MIL-HDBK-1195 §2.6.2 (historical mirror)

The RF door is the most delicate part of the shield, and it's installed while the dirtiest trades are still working.

Protection during construction (installer practice):

• Cover the knife-edge and finger stock with the door maker's guards or rigid covers. Don't tape directly to finger stock.
• No welding, grinding or spray painting near an uncovered RF frame. Mask the contacts before painting the frame.
• Don't prop the leaf with wedges that crush fingers. Use the door's own hold-open provisions or keep the leaf closed.
• Keep carts and material off the threshold.
• Log any damage and let the door maker repair it.

Sequencing for low-voltage work:

1. Door hung, balanced and adjusted by the door maker.
2. HSS brackets, magnets, electrified hardware and raceway attachments installed, or at least mocked up at their final fastening points.
3. Confirm nothing holds the door off its seals.
4. Shield test.
5. Walk-test the HSS to Tech Spec 7.D.3 after seals have set.

Why it matters. A practitioner in the door and lock trade writes: "Most SCIF doors fail to pass accreditation as a result of poor installations, not defective products." A test run before hardware goes in describes a door that no longer exists.

Sources MIL-HDBK-1195 §2.6.2 (historical mirror) · MIL-HDBK-1195 §4.9 (historical mirror) · ETS-Lindgren DKE door datasheet · O'Haver: SCIF door tutorial and grandfathering (practitioner) · IC Tech Spec, Ch. 5–end (v1.4 archive; text source)

Mistakes happen: a drill walks, a hanger goes in the wrong place, a tech mounts a box before reading the drawing. What turns a mistake into a failed accreditation is hiding it.

The response:

Why reporting protects the installer. During construction, the SSM must "Document security violations or deviations from the CSP and notify the AO within 3 business days." A practitioner guide on accreditation adds that "Any deviations from the approved design, no matter how minor, must be documented and approved." A reported, repaired and retested hole is a record. A hidden one is a finding.

Sources MIL-HDBK-1195 §4.9 (historical mirror) · IC Tech Spec (NAVFAC mirror) · UFC 4-010-05 · ICS 705-1 · ICS 705-2 · ASI: SCIF accreditation process (practitioner)

Print this for the trailer wall. It covers shielded rooms and acoustic-only perimeters.

Two reminders behind the table:

• Access control notifications rank below IDS alarms (Tech Spec 7.A.2.i), which is why ACS signals never shunt the IDS door switch.
• UFC 4-010-05 §1-16.3: TEMPEST information is "classified at a minimum of CONFIDENTIAL when associated with a physical location."

Sources UFC 4-010-05 · IC Tech Spec v1.5.1 · IC Tech Spec, Ch. 5–end (v1.4 archive; text source) · UFGS 13 49 20.00 10 · Magnasphere HSS Level 2 CSI spec

Hold points stop the job until someone verifies conditions that will soon be hidden or tested. The Tech Spec requires periodic SSM inspections, but it doesn't list these specific points. They're recommended practice built from the sources cited.

Who attends. A practitioner build sequence recommends the SSM, and the AO or TSCM if they ask, inspect in-wall and above-ceiling conditions before board closes. UFC 4-010-05 §4-7 calls for "a photographic surveillance record documenting construction progress at critical areas such as perimeter penetrations, door installations, and wall assemblies."

Sources UFC 4-010-05 · IC Tech Spec v1.5.1 · MIL-HDBK-1195 §5 (historical mirror) · MIL-HDBK-1195 §5.2 (historical mirror) · MIL-HDBK-1195 index (historical mirror) · CenCore: SCIF construction Q&A (practitioner)

What it set out to do (§1.1): "This handbook provides guidance with the RED/BLACK concept for the engineering and installation of systems and facilities processing classified information." It covered power distribution, equipment installation, signal distribution, filters and isolators, grounding, bonding and shielding, physical security and administrative telephones.

Two ideas (§4.1). RED/BLACK means components "processing classified plain text information be kept separate from those that process encrypted or unclassified information." TEMPEST means controlling compromising emanations. The handbook teaches six techniques: grounding, bonding, shielding, physical separation, physical protection and filtering or isolation.

The honest limit (§4.1). The handbook assumes low-level balanced signaling. It acknowledges that facilities designed to RED/BLACK criteria may still have TEMPEST deficiencies, and that post-installation TEMPEST testing may require additional measures.

Lessons for installers:

• Following a guide doesn't guarantee a TEMPEST pass. Equipment, signaling and workmanship all matter.
• "Constant attention is required to ensure proper separation, isolation, and accountability" (§4.1 design verification).
• Bonds matter: a good bond means "the impedance through the connection is not appreciably greater than the impedance in the conductors themselves" (§4.6.2). Welding is preferred, and split bolts are "not recommended."
• Filters belong where the design puts them: "In no case should a line be multifiltered between two points" (§5.4.5).
• Build to the CTTA's review and the drawings, and expect that testing may add work.

See: History of TEMPEST & Shielding Specs for where this handbook sits in the document lineage.

Sources MIL-HDBK-232A index (historical mirror) · MIL-HDBK-232A §1 (historical mirror) · MIL-HDBK-232A §4 (historical mirror) · EverySpec: MIL-HDBK-232A · EverySpec: MIL-HDBK-232A Notice 3

Purpose. "This instruction establishes guidelines and procedures for determining the applicable countermeasures for national security systems. This instruction applies to all federal departments and agencies and their agents, which include, but are not limited to, contractors, consultants, and licensees."

Key concepts:

The teaching logic:

1. Define the inspectable space.
2. Weigh the factors.
3. The CTTA recommends the least-cost countermeasure that's adequate.
4. The agency's designated authority decides.

What it means on a jobsite today. Integrators and GCs build to the CTTA's current TEMPEST Countermeasures Review and the approved drawings. Don't add shielding, filters or breaks "to be safe." The Tech Spec says "Exceeding a standard, even when based upon risk, requires that a waiver be processed and approved," and a waiver can take the facility out of mandatory reciprocity. Don't advertise a room as "TEMPEST compliant."

See: RED/BLACK, TEMPEST & EMI Filters for the current CTTA review process.

Sources NSTISSI 7000 (historical FOIA mirror) · DoDI 8523.01 · IC Tech Spec v1.5.1 · ICS 705-1