Skip to main content

2.4 Brackets, Shields & Heat Sinks

Brackets, shields, and heat sinks form the structural, electrical, and thermal backbone of an enclosure, and their installation sequence decides whether the system holds together cleanly or fights itself at every screw. Brackets define the geometry, locking in datums and clearances so that everything else aligns. Shields then establish low-resistance paths that keep electromagnetic noise under control, provided that conductive gaskets and bond lands are seated properly. Only after these foundations are secure can thermal hardware be installed, with TIMs and torque patterns tuned to spread heat without bending boards. Following this order ensures that mechanics, EMC, and cooling reinforce rather than undermine one another.

2.4.1 The idea (in one line)

Set mechanical datums first (brackets), close EMC paths second (shields/bonds), then lock thermal paths (heat sinks/TIMs)—in that order—so nothing fights the next step.



2.4.2 Sequencing that avoids rework

  1. Bracketry & frames: create the square, repeatable geometry the rest depends on.
  2. EMC bonds & shields: establish low-Ω paths before paint flakes or gaskets get trapped wrong.
  3. Thermal hardware (heat sinks/spreaders/pipes): apply TIMs and torque once, with final alignment set.
  4. Dress & verify: continuity (EMC), compression (seals), and temps (sanity).

If two parts compete for the same screw, the mechanical locator (bracket) wins the first pass; the shield or sink stacks under/over it per drawing with spacers/washers as designed.



2.4.3 Brackets & frames (make the chassis true)

Purpose: set position, carry load, define clearances.

  • Dry fit first: holes align without force; slots centered when possible.
  • Datum-first tightening: seat the bracket at the primary datum (pin/standoff), then work outward.
  • Pattern & torque: snug all → final torque in cross pattern (23.2).
  • Stack order: bracket → washer (if spec) → chassis; avoid “creative” shims.
  • Keepouts: confirm no bracket lip will rub a harness; add edge guards if route passes near.
  • Witness marks on critical fasteners after torque.

Acceptance cues

  • Bracket sits flush (no rocking); slots not hard against one end; reveal to neighbors consistent ±0.5 mm.



2.4.4 Shields & EMI gaskets (conductivity before cosmetics)

Types you’ll see: fingerstock, conductive foam, stamped cans, lid bonds, braid straps.

  • Bond lands clean (23.1): bare metal visible, no powder in path.
  • Seat shields square; no “springing” into place that preloads screws.
  • Gasket compression: conductive foams want 20–30%; fingerstock wants contact, not flattening.
  • Fastener pattern: short-to-long path to herd the gasket; final pass at spec torque.
  • Continuity check: seam-to-seam or shield-to-chassis < 0.1 Ω (log on first article).
  • Straps: 360° clamps on braid, or solder sleeves; pigtails ≤ 10 mm if unavoidable.

Reject if the gasket has gaps, is crushed flat, or bond Ω is high.



2.4.5 Heat sinks & spreaders (the clamp that carries heat)

TIM choice & handling

  • Pads/gap fillers (silicone/graphite/PCM): easy, sized pieces; avoid stretching; compress 10–30% depending on durometer.
  • Pastes/greases (thermal compound): thin, even film; bead size per kit; avoid pump-out paths.
  • Phase-change films: place cool; activate/hot torque once; watch for reflow temperature in SWI.
  • Insulating pads (mica/Kapton/ceramic): required where electrical isolation + conduction needed—don’t skip bushings on screws.

Mounting steps

  1. Dry align the sink/spreader to confirm standoff heights.
  2. Apply TIM: pads last-minute; paste with stencil/dispense (pattern = lines/X); keep off connectors.
  3. Set sink straight down; no slide on paste if avoidable.
  4. Torque in pattern: 30–50% pass → 100% pass; use spring screws if spec’d.
  5. Squeeze-out sanity: a minimal, uniform line at edges is good; puddles are not.

Starter torque (validate on your joint)

  • M2: 0.2–0.3 N·m; M2.5: 0.35–0.6 N·m; M3: 0.6–1.0 N·m (23.2.5).
  • For spring-loaded posts, run to stop height or angle spec per vendor.

2.4.6 Gap fillers & stack height (avoid bow and tilt)

  • Pick thickness so compressed in service hits target (e.g., 20%).
  • If multiple pads in one stack, their total spring must not bow the PCB; add mid-standoffs if needed.
  • Shims/spacers are part of the BOM—never substitute washers ad hoc.

Quick gauge: caliper pad pre vs post thickness or witness tape stack at first article.



2.4.7 TIM “recipes” (patterns that work)

Surface

Pattern

Why

Small die < 20×20 mm

Pea or thin X

Fast, avoids edge bleed

Long bar/VRM strip

Three parallel lines

Fills between components; reduces pump-out

Large plate

Cross-hatch or stencil

Controls volume; even coverage

Coverage target: 90–100% of mating area after clamp, with ≤1 mm squeeze-out and no voids over hotspots.



2.4.8 Order-of-operations examples (two common builds)

A) RF lid with shield + sink on CPU

  1. Install board brackets (standoffs/frame) → torque.
  2. Place EMI gasket around RF bay → drop RF lid → torque; verify <0.1 Ω.
  3. Apply CPU paste → mount CPU sink with cross torque; confirm minimal squeeze-out.
  4. Fit duct/air guide and fan bracket; spin test.

B) Power tray with spreader + braid bond

  1. Mount tray bracket to chassis; earth lug with serrated washer<0.1 Ω.
  2. Lay gap filler pads on MOSFET line; place spreader; spring-screw torque in pattern.
  3. Install braid strap 360° at gland; bond to tray; verify Ω.
  4. Close top shield over supplies; check clearances to harness.




2.4.9 Quick verifications (cheap, powerful)

  • Board bow near sink: ≤ 0.5 mm across card; visual straightedge.
  • EMC seam Ω: < 0.1 Ω across shield joints; record first article.
  • TIM witness: small, even bead around sink; peel-check one NPI unit to confirm spread pattern.
  • Fan/air path: no interference from brackets or shields; airflow arrow respected.
  • Screw head height: all seated; witness marks drawn.

2.4.10 Acceptance cues (what good looks like)

Item

Accept

Reject

Bracket fit

Flush, no rock; centered in slots

Forcing to align; slot maxed at one end

Shield/gasket

Continuous contact, corners tight

Gaps, crushed foam, peeling fingerstock

Bond

<0.1 Ω seam/strap

Paint under washer; high Ω

Heat sink

Even clamp; pattern torque done

One corner high; skew from sliding

TIM

Even thin spread; tiny edge bead

Dry spots; pumps onto connectors

PCB stress

No visible bow/whine

Bowed board; fan/pipe fouls parts



2.4.11 Common traps → smallest reliable fix

Trap

Symptom

First move

Sink used to “straighten” a warped board

Intermittents, cracked vias

Fix standoff heights; add mid support; then mount

Over-compressed foam gasket under lid

IP fails later, bent lid

Torque in two passes; verify 25–35% compression

Paint under star washer

High earth Ω

Scrape to bare, re-torque, measure

Too-thick gap filler

PCB bow, poor screw start

Drop one thickness grade; confirm compression target

Sliding sink on paste

Voids, misaligned holes

Place straight down; use alignment pins; minimal reposition

Shared screw stack in wrong order

Rattles, poor EMC/thermal

Follow BOM stack; bracket → shield → sink only if drawing says so

Spring screws torqued like rigid

Uneven pressure

Tighten to stop/angle spec; not N·m guess




2.4.12 Pocket checklists

Before install

  • Bracket holes free; standoffs at height; edges safe
  • Bond pads clean; gaskets cut to length; straps ready
  • TIM kit (pads/paste) at station; torque bits staged

Brackets & shields

  • Brackets datum-seated; cross-pattern torque; witness marks
  • Gasket in place; shield seated; seam <0.1 Ω measured

Heat sinks/spreaders

  • Pad/paste applied to recipe; no smear on connectors
  • Cross-pattern torque in two passes; spring screws by spec
  • Board bow within limit; small uniform squeeze-out

Final checks

  • Air paths clear; fans spin; no rubs
  • Harness clears edges; first clamp before first bend
  • Log Ω, torque tool IDs, and any pad thickness calls




By sequencing carefully and verifying fit, continuity, and compression, assemblies avoid warped boards, leaky shields, and overheated components. The result is hardware that stays square, quiet, and thermally stable, passing tests with fewer surprises and reworks.