A modification to the typical accumulator based, direct-digital-synthesis (DDS) architectures has emerged in the form of a programmable-modulus DDS. This modulus’ goal is to alter the accumulator modulus. Yet that task is complicated because the angle-to-amplitude converter maps the entire P-bit input range (0 to 2RP) to 0 to 2p radians. This hurdle and an explanation of the main differences between conventional and programmable-modulus DDS can be found in a four-page application note from Analog Devices titled, “Direct Digital Synthesis with a Programmable Modulus.”

Thanks to the power-of-two mapping arrangement of the programmable-module DDS, the angle-to-amplitude converter can operate efficiently. Yet arbitrarily changing the accumulator modulus to something other than a power-of-two violates the mapping arrangement required by the angle-to-amplitude converter. The programmable modulus architecture thus uses a secondary accumulator to make the primary one appear to have an altered modulus. Another issue—that of spurious performance—arises if modification of the accumulator modulus is not properly done.

This application note is centered on the AD9913, which is implemented by programming three 32-b registers with values for N, Y, and X. The accumulator must cycle through M overflows to complete one unique output sequence and return to its original starting point. This output sequence repeats itself indefinitely. The note closes with different considerations, such as errors introduced by many computational programs. In doing so, it offers valuable information for engineers who want to extend the use of DDS to applications that require exact rational frequency synthesis.

Analog Devices, Inc., One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106; (781) 329-4700, FAX: (781) 461-3113, Internet: www.analog.com.