Document created by analog-archivist Employee on Feb 23, 2016
Version 1Show Document
  • View in full screen mode

A customer wants to know if it is realy possible to have the 12bit resolution
(4096steps) with an output signal of 200kHz ?

We didn´t find this realy in the D/S.


The ROM on the AD9830 has a 12-bit address. Therefore the 2^32 increments in
the phase accumulator are truncated into 4096 phase steps in the SINE look-up
table. In actual fact there are only 1024 values held in the sine look up
table, representing angles 0 to 90 degrees. The remaining 3 quadrants of the
sine wave are simply duplications of these values, and can be obtained by
mapping different phase inputs to the correct value  (if angle x is between 0
and 90 degrees, then input phases of x, 180-x will point to the same value, and
180+x and 360-x will point to the negative of that value).

If the phase word is 2^32/2^12 = 2^20 = 1,048,576 then the output will step
through all the values in the look up table in a single period of the sine
wave. If the phase word is greater that this then some of the values in the
look up table will be not be reached in every output wave (e.g. if the phase
word is 2 x 2^20 then every second address in the look up table will be
visited). There is a special case when the phase word is a power of 2 e.g.
2^20, 2^21, 2^22... In this case, the same addresses in the SINE lookup table
will be "visited" during each waveform and the output frequency will be an
integer fraction of the master clock frequency e.g. if the phase word is 2^21
then the output frequency will be MCLK/2048. You are not using the full
resolution of the SINE lookup table. The output can show worse spurious
spectral components in this case because the full resolution of the lookup
table (and the DAC) is not being used.

When this special case doesn't apply, then different locations in the sine
lookup table will be "visited" on every clock cycle, and over many cycles you
can expect to address every location in the SINE lookup table.

There are many variations in the "quality" of the output, depending on the
phase word employed. There are no particular rules which govern these, but
"sweet spots" exist usually where the phase word is not an even power of two.