Boxcar averager

The main purpose of this measurement technique is to improve signal to noise ratio in pulsed experiments with often low duty cycle by the following three mechanisms: 1) signal integration acts as a first averaging step that strongly suppresses noise components with a frequency of the reciprocal gate width and higher, 2) time-domain based selection of signal parts that actually carry information of interest and neglect of all signal parts where only noise is present, and 3) averaging over a defined number of periods provides low-pass filtering and convenient adjustment of time resolution.

The principle of operation can be understood as a two-step process: signal integration over the desired gate width and averaging the integrated signal over a defined amount of periods/trigger events Considering a simple implementation of the core circuitry looks like regular RC low-pass filter that can be gated by a switch S. Provided the filter time constant τ = RC is set to sufficiently large values with respect to the gate width, the output voltage is to a good approximation the integral of the input signal with a signal bandwidth of B = 1/(4RC).

The time it takes for this reset is one of the major speed limitations for analog implementations, where maximum trigger rates of a few 10 kHz are typical even though the gate width itself can be as low as a few ten picoseconds and delay is set to zero.

The origin of the boxcar averager dates back to as early as 1950 where the technique helped to improve signal quality in experiments investigating nuclear magnetic resonances with pulsed schemes.[1]).

[2][3] In their 1955 paper Holcomb and Norberg credit the invention of the “boxcar integrator” to a large extent to L. S. Kypta and H. W.