Telecom & 5G Stamping: RF Shields, Heat Sinks and Contacts

Open a 5G base station or a Massive MIMO active antenna unit (AAU) and you will find the same three failure modes showing up in field returns: passive intermodulation (PIM) drifting past the -150 dBc ceiling and poisoning the uplink, heat sinking that falls short so the power amplifier throttles or dies, and contact interfaces that lose spring force after a few hundred thermal cycles and start dropping bits. None of these are software problems. They start on the stamping floor - in the alloy you pick, the plating you specify, and the tolerance you hold on a 0.15 mm contact beam. The base station is only as reliable as its smallest stamped part.

5G radios are the densest metal assemblies in telecom. A single AAU carries an antenna array, dozens of RF transmit and receive chains, a power amplifier section, and a baseband board, all bolted into one housing that sits outdoors and thermal-cycles between -40 °C and +85 °C for a decade. Between the semiconductor dies and the air, the signal path runs through stamped metal: RF shield cavities, spring contacts, finger stock, heat sink fins, board-to-board connector terminals, and mounting hardware. Every one of those parts is made by a progressive die, and every one carries an electrical or thermal specification that most people treat as a mechanical drawing.

This guide walks the engineering decisions behind 5G stampings: what the frequency bands actually ask of geometry and surface finish, how PIM is created and killed on the stamping floor, why beryllium copper and aluminum 6063 dominate the two most failure-prone part families, how plating is specified for RF duty, which tolerances matter and how a production line holds them, and finally how to spec and quote a 5G stamping program so the first article and the millionth part behave identically.

What a 5G Radio Asks of a Stamped Part

5G New Radio spans two very different regimes. Sub-6 GHz bands - n77 at 3.3-4.2 GHz, n78 at 3.3-3.8 GHz, n79 at 4.4-5.0 GHz - carry wide-area coverage, while mmWave bands - n257 at 26.5-29.5 GHz, n258 at 24.25-27.5 GHz, n260 at 37-40 GHz - carry dense urban capacity. The frequency changes everything downstream. At radio frequencies, current crowds to the surface of a conductor: the skin depth of copper is about 2.1 µm at 1 GHz, roughly 1.1 µm at 3.5 GHz, and below 0.4 µm at 28 GHz and above.

5G band	Frequency range	Copper skin depth	Dominant stamped part
---	---	---	---
n78 (sub-6)	3.3-3.8 GHz	~1.1 µm	RF shields, filter cavities, BTB contacts
n79 (sub-6)	4.4-5.0 GHz	~0.9-1.0 µm	Shield cans, spring contacts, phase shifter parts
n258 (mmWave)	24.25-27.5 GHz	~0.42 µm	Thin-wall shields, mmWave contact arrays
n257 (mmWave)	26.5-29.5 GHz	~0.40 µm	Antenna module frames, micro shields
n260 (mmWave)	37-40 GHz	~0.33 µm	AoP module shields, precision springs

Because the signal travels on the surface, a scratch, a burr, a plating void, or a 0.02 mm dimensional error becomes an electrical defect, not a cosmetic one. At 28 GHz a burr only a few hundredths of a millimeter tall is a significant fraction of the current-carrying skin, and any discontinuity acts as a nonlinear junction that generates interference. Geometry and finish are electrical specifications, and that single fact drives every material, plating, and tolerance decision in this guide.

The mechanical side is just as demanding. A radio module dissipates hundreds of watts through stamped heat sinks into a housing that must survive vibration, salt air, and decades of thermal cycling. Springs must hold contact force from first insertion to end of life. The stamping supplier is therefore not a commodity vendor - it is the party that turns the electrical design into repeatable metal. The progressive die high speed stamping lines that make these parts run copper alloys, aluminum, and specialty strip at up to 300 SPM with positioning accuracy of ±0.005 mm, which is the precision budget the whole radio depends on.

RF Shield Cavities: Where PIM Is Won or Lost

The RF shield cavity is a one-piece drawn or progressive-die stamped enclosure that separates the transmitter path from the receiver path and from the digital board. Its job is deceptively simple - keep RF energy where it belongs. In practice it is where PIM is born.

PIM comes from nonlinear junctions. A loose seam, a ferromagnetic contaminant, a sharp burr, or a plating void acts like a diode under high RF power and generates mixing products at 2f1 - f2 and 2f2 - f1. With two carriers at 2.1 GHz and 2.11 GHz, the third-order product lands right inside the receive band. Operators spec PIM at -150 dBc as a floor and push toward -160 dBc for 5G NR bands where the noise floor is brutally tight. A shield that ships at -143 dBc looks fine on the bench and fails in the field.

Stamping controls PIM three ways. First, seam continuity: a progressive-die shield with a continuous folded seam beats a two-piece welded can because there is no discontinuous weld nugget to act as a nonlinear junction. Second, material cleanliness: ferrous tooling wear debris and steel or brass contamination are kept out of the part - the cavity is 100% non-ferrous, typically aluminum 6061 or 5052 for the structural can, with a conductive finish. Third, burr control: edge burrs under 0.03 mm and a consistent seam compression force keep the joint linear. Seam flatness is held to ±0.05 mm across a 120 mm wall so the gasket sees even pressure.

Filter cavities add a fourth requirement. The bandpass filters and duplexers inside a radio use stamped cavity walls whose internal dimensions set the resonant frequency. A 0.05 mm drift on a cavity wall shifts the passband, and a plated surface with voids turns into a lossy, PIM-generating surface. These parts are stamped, then plated with a uniform conductive finish, and every cavity is dimensionally verified - often on a CMM or optical measuring system - before the filter is tuned. For a deeper look at how shield cans are designed for manufacturing, the metal enclosure and shield can guide covers wall thickness, draw ratios, and seam design in detail.

PIM Drivers and the Stamping Controls That Kill Them

PIM failures are rarely exotic. They trace back to a short list of stamping and finishing details, and each one has a known control and a known verification method.

PIM driver	Stamping control	Verification
---	---	---
Loose or discontinuous seam	One-piece folded seam, no weld	PIM test per IEC 62037, seam crush test
Ferrous contamination	Non-ferrous die materials, strip cleaning, segregated tooling	Magnet sweep, XRF spot check, PIM retest
Edge burr acting as a nonlinear junction	Burr control under 0.03 mm, deburring pass	Optical burr measurement on sampled parts
Plating void or corrosion at the joint	Minimum plating thickness as a floor, adhesion test	Salt spray (ASTM B117), cross-section, PIM retest
Lubricant or process residue	Cleaning step before plating and final assembly	Surface cleanliness test, PIM retest after cleaning
Uneven gasket compression	Seam flatness ±0.05 mm over 120 mm	Flatness gauge, gasket compression check

The verification loop matters as much as the control. A PIM test is a two-tone test: two high-power carriers are injected, and the receiver looks for intermodulation products in the receive band, typically at -150 dBc or better. Because PIM changes with temperature, humidity, and mechanical stress, a serious program tests parts not just as-stamped but after thermal cycling and after assembly, and it retests when the die is reworked. PIM that passes on a fresh part and fails after 100 thermal cycles is a stamping or plating problem that belongs back at the die, not in the field.

Spring Contacts and Finger Stock: Why C17200

Spring contacts, finger stock, and board-to-board contact beams live or die on two numbers: fatigue life and contact resistance. For a base station that thermal-cycles for a decade, the material must keep its spring force after 500+ cycles and stay conductive doing it.

That is why beryllium copper C17200 (UNS C17200) is the default for RF spring contacts. The spec gives 1.80-2.00% beryllium with cobalt plus nickel as the age-hardening agent. In the TF00 condition - solution annealed then precipitation hardened - C17200 delivers yield strength around 1000-1240 MPa (145-180 ksi) with electrical conductivity near 22% IACS and Rockwell C hardness of 38-42. Compare that to phosphor bronze at roughly half the yield, and the reason C17200 carries the beam that must hold 0.5-1.0 N of normal force across its life becomes clear.

The fatigue argument is the real one. A C17200 contact beam 0.12 mm thick, deflected 0.4 mm, survives well past the 500-cycle mark with force drop under 15%. The same geometry in 301 stainless would either relax or corrode. These beams are stamped at ±0.01 mm on the deflected length because contact force scales with the cube of the beam thickness - a 10% thickness error swings force by 33%, and that is how intermittent contacts appear in the field. For the design rules behind cantilever and finger contacts, the spring contacts stamping design guide covers beam geometry, grain direction, and bend radii in depth.

Material	Typical yield strength	Electrical conductivity	Strength at temperature	Typical 5G use
---	---	---	---	---
C17200 BeCu (TF00)	1000-1240 MPa	~22% IACS	Excellent to 150 °C+	Contact beams, finger stock, BTB springs
C5191 phosphor bronze	~500-600 MPa	~13-15% IACS	Good, relaxes earlier	Lower-force springs, cost-sensitive contacts
301 stainless	~1200 MPa (full hard)	~2-3% IACS	Good, corrosion issues	Shields and clips, not signal contacts
C7701 nickel silver	~500-600 MPa	~5-6% IACS	Moderate	EMI gasket springs, non-magnetic parts

Two process details decide whether a spring performs in production. First, bending direction: the beam should be formed across the strip grain, not along it, or the bend zone becomes a fatigue crack starter. Second, burr placement: the burr from the stamping operation must sit on the inside of the bend, never on the tension face, and burr height is held under 10% of material thickness. A stress-relief bake at 150-200 °C after forming stabilizes the beam; above roughly 250 °C the age-hardened temper collapses and contact force drifts. None of this shows up in a first-article dimensional check, which is why spring programs need force measurement on a population of parts, not a single sample.

Heat Sink Fins and the Thermal Path

The power amplifier is the hottest thing in the AAU, and the stamped heat sink is what keeps it alive. For finned sinks, aluminum 6063 is the workhorse because it extrudes and stamps cleanly and carries heat well.

The numbers: 6063-T5 gives a thermal conductivity of 209 W/(m·K), a coefficient of thermal expansion (CTE) of 23.4 ppm/°C, density 2.70 g/cm³, and yield strength near 145 MPa. A fin spacing of 4-6 mm with 0.8-1.0 mm wall thickness is the sweet spot - tighter fins choke the airflow, thicker walls add mass without helping the convection area. A typical PA spreader targets thermal resistance under 0.5 °C/W by balancing fin height against pressure drop.

Stamped fins differ from extruded and skived sinks in a way that matters at 5G volumes. Extrusion is efficient at very high volume but forces a constant cross-section. Skiving is precise but slow. Stamping gives freedom to vary wall thickness, add mounting feet, and integrate clips and standoffs into the same blank, at speeds that suit multi-hundred-thousand-piece programs. That is why stamped sinks dominate the mid-volume AAU market, and why the trade-off between tooling cost and per-part cost has to be priced at the quoting stage.

Here is the trap most people miss: the sink is aluminum, the PA package base is often copper or AlSiC, and the PCB is FR-4. Aluminum's CTE of 23.4 ppm/°C against copper's 16.5 ppm/°C and FR-4's in-plane 14-16 ppm/°C means a 105 °C swing drives roughly 0.8 µm/mm of relative growth between the sink and the board. Get the mounting scheme wrong and the sink walks relative to the board, cracking solder joints. The fix is to design the sink's mounting feet as compliant features stamped into the same blank, absorbing the mismatch instead of fighting it with rigid bolts. For the broader thermal design trade-offs, the heat sink stamping guide walks fin geometry, alloy choice, and attachment methods.

Aluminum alloy	Thermal conductivity	CTE	Yield strength	Best role in an AAU
---	---	---	---	---
6063-T5	209 W/(m·K)	23.4 ppm/°C	~145 MPa	Finned heat sinks, spreaders
6061-T6	167 W/(m·K)	23.6 ppm/°C	~276 MPa	Structural brackets and housings
5052-H32	138 W/(m·K)	23.8 ppm/°C	~193 MPa	Shield cans, drawn cavities, ductile parts

Board-to-Board and RF Connector Contacts

Board-to-board (BTB) connectors in a 5G radio stack link the RF board to the baseband and power boards. They are the highest-density stamping in the unit, with contact pitches now down to 0.4 mm and signal pins carrying several amps alongside RF lines.

CTE matching is the design constraint that kills BTB connectors that look fine on paper. The connector housing is liquid-crystal polymer or nylon with a CTE near 15-20 ppm/°C, the stamped contacts are copper alloy at 16.5 ppm/°C, and the mating board is FR-4 at 14-16 ppm/°C in-plane but 45-70 ppm/°C through-thickness. When the stack heats, the housing and the board expand differently and the contact can lose normal force or, worse, short adjacent 0.4 mm pitches. The standard answer is a copper-nickel-silicon or C17200 contact so the metal tracks the board, with molded-in contact position held to ±0.02 mm so the pitch survives reflow at 260 °C. For the material and plating logic behind stamped connector terminals, the connector terminal stamping guide is the companion reference.

Two more numbers matter on these parts. Coplanarity - the height spread of SMT tails across a connector - must stay tight, typically under 0.1 mm, or reflow opens shorts and voids across the pitch; it is controlled by die flatness, strip flatness, and careful forming in the final stations. Contact resistance on signal contacts is targeted below 10 mΩ and below 5 mΩ on power pins, measured with a four-wire micro-ohmmeter on mated samples, because a drifting interface shows up as bit errors long before it shows up on a multimeter.

IPC-2221 is the generic printed board design standard used to size the current-carrying capacity of every stamped power contact. Its conductor charts show that a 1 oz (35 µm) external trace carrying 1 A rises about 10-20 °C depending on width; that translates into the stamped contact cross-section so a power pin rated 3 A does not run hot enough to anneal the spring temper. For insulation, the creepage and clearance tables set the minimum air and surface distances for the 48 V and 12 V rails inside the radio - stamped terminals stay at least 0.8 mm apart with no burr bridges, because a single plating whisker across that gap is a field failure.

Plating for RF Reliability

Plating is where corrosion resistance, solderability, and contact reliability meet. For telecom stampings the workhorse is matte tin over a nickel underplate.

Tin per ASTM B545 at 3-8 µm gives solderability and a soft, gas-tight contact surface; matte tin is preferred over bright tin because it grows fewer whiskers, and whiskers are the silent killer of fine-pitch RF contacts. Under the tin sits nickel per ASTM B689 at 1.25-2.5 µm. Nickel is the diffusion barrier that stops the tin from dissolving the copper alloy base during the 260 °C reflow and stops copper from migrating to the surface and oxidizing. Nickel also adds hardness - 500+ HV - so the contact resists fretting wear at the micro-ampere signal pins.

For RF shield cavities that must stay conductive at the seam, bulk plating is often skipped in favor of a selective conductive finish or a tin flash at 0.5-1.0 µm on the mating surfaces only, leaving the rest of the aluminum bare to save weight and cost. The principle is to put the plating exactly where the current and the corrosion risk are, not everywhere. That is what our managed selective reel-to-reel plating lines do: gold, silver, tin, and nickel at 2-8 µm and zinc at 5-12 µm, applied only where the drawing calls for them, with salt spray validation to ASTM B117.

Finish	Typical thickness	Where it is used in a radio	Key property
---	---	---	---
Matte tin over nickel	Sn 3-8 µm, Ni 1.25-2.5 µm	Soldered contacts, shield seams	Solderability, whisker resistance
Gold (selective)	2-8 µm zones	Edge connectors, high-cycle contacts	Low contact resistance, wear life
Silver	2-8 µm	High-current RF contacts, busbars	Highest conductivity of common finishes
Nickel	2-8 µm	Barrier layer, hard-wearing surfaces	Diffusion barrier, hardness
Zinc	5-12 µm	Outdoor steel brackets and hardware	Salt spray resistance

Plating thickness is specified as a floor, not an average. A spec that says "gold 2-8 µm" means the thinnest acceptable point on the contact zone is 2 µm; plating that thins mid-reel under a defective mask is the classic hidden failure, and it is caught only by cross-sectioning sampled parts and by salt-spray testing. Buyers who want a decade of field reliability should ask for plating thickness data per lot, not per quote.

The Tolerance Budget and How It Is Held

A precision stamping house lives on the tolerance budget. For these parts the numbers that matter are: contact beam thickness ±0.01 mm, seam flatness ±0.05 mm over 120 mm, fin wall ±0.03 mm, and BTB contact position ±0.02 mm. These are held with progressive dies running hardened powder-metal inserts, in-die detection, and SPC on the first and last shot of every coil. A 0.02 mm drift on a contact position is invisible to the eye and fatal to a 0.4 mm pitch.

The capability stack behind those numbers is real equipment, not brochure language: 21 presses including Aida high-speed machines from 25 to 80 tons running up to 300 SPM, dedicated EV terminal presses with automatic feeding, and heavy presses from 45 to 110 tons for larger enclosures and brackets. Positioning accuracy sits at ±0.005 mm on the high-speed lines, strip runs from 0.05 mm to 3.0 mm thick and up to 650 mm wide, and the tool room - wire EDM, CNC, and grinding - builds and reworks the dies that hold the tolerances. Quality control adds CMM and optical measurement plus online vision inspection, so dimensional drift is caught in the line, not in the field.

The other half is traceability. Every coil of C17200 and 6063 carries a mill cert mapped to the lot, and plating is salt-spray tested to 48-96 hours before a new tool runs production. Telecom buyers do not buy a part; they buy a decade of field reliability, and the cert package is the proof.

Cost Economics of 5G Stampings

Cost on a 5G stamping program is decided by four levers, and they interact. Tooling is the first: a multi-station progressive die is a six-figure-class investment that amortizes over the program life, so the annual volume forecast sets how much die sophistication is justified. Material utilization is the second - progressive die strip layouts typically land at 60-80% utilization, and at telecom volumes material is a large share of part price, so every millimeter of web saved is real money.

Speed is the third lever: multi-up tooling on a 300 SPM line makes terminals and shields economical at multi-million-piece volumes, but only if the die holds tolerance at speed. Plating is the fourth: selective reel-to-reel plating plates only the contact zone, which keeps finish cost controlled compared to full-strip precious metal. Secondary operations - insert molding, overmolding, riveting, and assembly - are increasingly done in-house so the tolerance stack and the freight move stay in one place.

Cost driver	Lever	Buyer action
---	---	---
Tooling	Die sophistication vs program volume	Give a realistic annual volume range, not a heroic one
Material	Strip utilization 60-80%	Let the stamper optimize the strip layout before freezing the drawing
Speed	Multi-up tooling, up to 300 SPM	Ask for the sustained rate, not the datasheet rate
Plating	Selective zones vs full-strip	Define plating zones on the drawing

How to Spec Your Next 5G Stamping

A 5G stamping RFQ fails when it leaves out the operating environment. A drawing with dimensions but no temperature range, current per pin, PIM target, or board material forces the stamper to guess - and the guess surfaces as a field failure years later. The RFQ package below closes the loop:

• Drawing with GD&T, material grade and temper, strip thickness and tolerance
• Plating spec per zone, in micrometers, with the required floor
• Operating envelope: temperature range, current per pin, voltage rails
• PIM target (for example -150 dBc) and whether the part is PIM-tested
• Annual volume range and program length, so tooling can be amortized honestly
• Mechanical requirements: normal force target, mating cycles, insertion force
• Quality requirements: PPAP level, SPC expectations, mill certs and plating certs per lot
• Test methods: four-wire resistance, salt spray hours, force gauge setup

If you are building RF shields, spring contacts, heat sink fins, or board-to-board connectors for a 5G radio and you need PIM, the thermal path, and the contact life designed in from the first article rather than discovered in field returns, send your drawing and your operating envelope to our team. We will quote the alloy, the plating, and the tolerance budget that gets you there, and we will ship first articles with the mill certs and the test data attached. Start the conversation on the request a quote page, and tell us what band you are fighting.