Elecraft KDSP2 Manual - Manual (Page 48)

48
Length
In DSP, filters consist of two blocks of memory. One holds the coefficients, and the other the sampled
signals. The size of these two memory blocks are the same, and are usually referred to by their length,
measured in taps. The longer the filter, the more taps it has, and the more precise it can be. Thus, longer
filters usually have steeper sides, less ripple, or more attenuation – or some combination of all three.
Longer filters also have greater delay, and require more processing power to implement.
LMS
Least Mean Squares. This is an algorithm that is widely used in adaptive filters. The two adaptive filters in
the KDSP2 are the autonotch and the denoiser.
Sampling
When signals are digitized, they are said to be sampled. In the KDSP2, analog signals from the K2 product
detector are sampled at a 16 kHz rate. This means that 16,000 times per second, the output voltage from
the product detector is converted to a 16-bit digital value by the ADC and fed to the DSP.
Taps
Each memory location used by a DSP filter is usually referred to as a tap. Thus, a filter using 100 memory
locations for its signal samples would be said to have 100 taps. The term is historic and comes from the
concept of a delay line, which was an L-C network that literally had “taps” along its “length.”