A Bluetooth module leaves the lab clean and returns from the field dead. Nine times out of ten the killer was invisible at the bench: a static zap at the connector, or RF noise the module itself generated and failed to contain. ESD and EMC are not test-lab afterthoughts. They are designed in, from the first footprint.
Where the energy comes from
A person walking a dry aisle builds 8 kilovolts before they touch a device — the IEC 61000-4-2 contact-discharge limit most products must survive. A conveyor or a robot arm, grounded through a rolling wheel, can dump more. The Bluetooth module sits downstream of that path, so whatever enters the connector reaches its pins unless something stops it.
Pick the TVS before you place it
The workhorse is a TVS diode, but a careless choice breaks the radio. A TVS with 30 picofarads of junction capacitance across an RF or high-speed line rounds off the very edges it should protect. For anything near 2.4 GHz, specify a low-capacitance part under 1 pF and place it within a few millimeters of the connector, on the outside of any series resistor. Distance is the enemy; a TVS halfway across the board protects nothing.
Harden the power entry
The supply line is the easiest path in. A Bluetooth module drawing sleep-and-burst current needs a ferrite bead to choke fast transients, a TVS to clamp the spike, and a local decoupling capacitor to absorb what leaks through. Put the TVS at the edge, the bead next, the capacitor last — in that order, so the clamp acts before the energy spreads. Get the order wrong and you clamp the capacitor instead of the pin.
Layout is half the battle
A solid ground plane under the Bluetooth module is not optional. Split or slot-riddled grounds turn every return current into a loop antenna. Keep the RF trace short and away from the DC-DC switcher, wrap the 32 kHz crystal in a guard ring so its oscillation does not leak, and give the antenna keep-out area its full clearance. Most “module is noisy” problems are really “board stole the ground” problems.
Protect the antenna port without killing it
The antenna feed is the most exposed net and the most fragile. A gas-discharge tube or a low-cap TVS at the feed clamps a zap, but every millimeter of protection adds insertion loss you pay for in range. A Bluetooth module with -6 dB of added loss talks to half the devices it should. Match the protector’s capacitance to the band and measure the loss on a VNA before you commit.
Contain the noise you make
ESD is incoming; EMC is outgoing. The Bluetooth module runs a DC-DC converter and a high-speed core, both of which radiate. A stamped shield can dropped over the module typically cuts radiated emissions by 10 to 15 decibels — often the difference between a pass and a failure at 30 meters. For the clock’s harmonics, spread-spectrum modulation spreads the energy into a lower, wider hump that the limit line forgives.
Tame the cables
Cables are antennas you did not mean to build. Any harness leaving the board — USB, serial, sensor leads — gets a common-mode choke at the exit, because the differential signal rides a shared return that radiates. A Bluetooth module on a clean board still fails certification if a loose cable turns the whole product into a transmitter of its own noise.
Pre-test before you pay
A formal EMC chamber bills by the hour and fails you anyway on the first pass. Build a pre-compliance setup: a spectrum analyzer, a makeshift shielded tent, and the real cables. Find the 240 MHz spur from the switcher, the 480 MHz from the clock, and fix them at the bench for a fraction of the cost. The teams that pass on attempt one pre-tested for a week first.
One failure, reconstructed
A customer’s Bluetooth module passed lab ESD but failed in the field within a month. The root cause was a TVS placed 4 centimeters from the connector behind a series resistor — the surge arrived at the pin before the clamp saw it. Moving the diode to the edge, outside the resistor, ended the returns. The part did not change. The placement did.
The trade you live with
Protection costs something on every axis. A lower-capacitance TVS clamps less hard; a shield adds height and a dollar; a ferrite steals a little signal. A Bluetooth module that survives the floor and passes certification is the product of those small, deliberate compromises — made on the layout, not discovered in the test report.
