Time-division multiplexing (TDM)

Time-division multiplexing shares one downstream resource — an amplifier, an ADC, or an output wire — among many channels by visiting them in a repeating sequence, giving each a brief slot. It is the mechanism that lets a probe with thousands of electrode sites read out over a handful of lines: instead of one dedicated chain per site, a multiplexer cycles through channels fast enough that each is sampled at its required rate.

The cost is bandwidth arithmetic. The shared resource must run at the aggregate rate, roughly the number of multiplexed channels times each channel's sample rate, so multiplexing more channels either raises the switching and settling speed demanded of the buffer and ADC, or lowers each channel's effective sampling. Within each slot the multiplexer plus buffer must settle to full accuracy, which sets a hard settling-time constraint.

The penalties are switching charge injection, kT/C sampling noise, and crosstalk between channels that share the path if settling is incomplete. Time multiplexing (share in time) contrasts with frequency-division multiplexing (share in frequency by modulating channels onto different carriers); TDM dominates in integrated neural front-ends because it maps cleanly onto a switch matrix and a shared ADC.