QI need to generate test tones from the sub-Hz range up to mid-audio (0.1Hz ..
4kHz) for which your AD8935 seems perfect although you market it primarily as
an RF device. For kHz freqs, AC-coupling the output is no problem, but around
1Hz and below this gets cumbersome. Is there a neat way to actively and
accurately cancel the DC offset seen at the output of the AD9835 with its
unipolar supply regime? I think of some combination of the voltage across the
300R output resistor and part of the voltage dropping across the Rset resistor
in the reference subcircuit.
AThe AD9835 contains a current output DAC. The output current range is between 0
and 4mA nominal assuming Rset = 3.9kOhm. That is, when the digital code loaded
into the DAC is 000H, the output current will be 0mA and when the digital code
loaded into the DAC is 3FFH (full-scale) the output current will be 4mA.
The COS ROM is what actually generates the cosine wave. The COS ROM applies the
appropriate code the DAC such that the output will be sinusoidal. The COSINE
waveform is therefore biased around the DAC mid-scale current. The nominal
mid-scale value is 2mA but this will scale with the full-scale output current
Note that as soon when you update the frequency register for the first time
after power-up, the DAC will begin from it's mid-scale value. Thereafter, the
DAC output is phase continuous.
I am presuming you wish to generate a signal which is biased around ground, in
this case the options are either to AC couple the output using a capacitor or
use an Op-amp to bias the output voltage.