Noise-efficiency factor (NEF)

The noise-efficiency factor is a dimensionless benchmark that measures how close an amplifier comes to the best possible noise for the current it burns. It normalises the amplifier's total input-referred noise-and-current against an ideal single bipolar transistor, which has NEF = 1 by definition. Because the underlying noise-versus-current tradeoff is a square-root law, NEF captures the brutal fact that improving noise is expensive: cutting input-referred noise in half at fixed bandwidth demands about four times the current, leaving NEF unchanged.

The power-efficiency factor (PEF) extends NEF by multiplying in the supply voltage (roughly NEF squared times the supply), so that designs on different supply rails can be compared on the metric that actually matters for an implant — power, not current alone. Practical neural amplifiers land somewhere around NEF of two to six; getting there uses current reuse, telescopic or inverter-based stages, and chopping.

NEF matters because in a high-channel-count implant the per-channel current sets the total power, and total power sets tissue heating. A design that shaves NEF lets you either add channels within the same thermal envelope or run cooler at the same channel count — which is why the metric, rather than raw noise, is the honest way to compare front-ends.