[sdiy] Roland DCOs
ebrombaugh at earthlink.net
Tue Feb 26 00:08:03 CET 2008
> For a DCO I am planning to use a Phase Accumulator chip (Analog Devices
> AD9833) to drive a simple wave table, with a PIC reading control voltages
> and programming the Phase Accumulator frequency..in fact the AD9833 will
> create 10-bit triangle, sine and sqaure, but I'm not sure about it running
> slow at 20k rather than 12.5MHz...more fun to run through a wavtable and
> having all that waveform fun...
(I'm not Tom, but since you're posting to the list I'll jump in anyway)
Your oscillator idea sounds interesting. I've used the Analog Devices
AD98xx series parts for many years (usually in test equipment) and
they're fun parts. I expect you won't have any trouble with it.
A few things you might consider from someone who has done a lot of
* What you're proposing isn't exactly a DCO in the traditional
synthesizer terms. Yes it is digital, and yes it is an oscillator, but
in the world of synths, a DCO has a specific meaning: an analog sawtooth
integrator driven by a digital divider (see Tom's webpage for a nice
drawing of the signal flow). The differences between this structure and
an ordinary NCO are critical in that a DCO won't suffer from the type of
aliasing that an NCO does, due to the fact that the digital divider's
output edges don't jitter.
* Since the AD9833 is an NCO-based device, it will suffer from
jitter/phase noise. Keeping the NCO clock rate as high as possible will
mitigate that somewhat, so even though the chip may run at 20kHz (which
it probably will) you may find that doing so will result in fairly dirty
sounding waveforms. Of course, some folks like that sound and if that's
your goal then you'll be happy with the results. Just be aware of the
impact which NCO clock rate will have on your oscillator performance. If
pitch accuracy is a concern, note that the AD9833 gives you 0.047Hz
resolution when clocked at 12.5MHz - pretty good even in the low registers.
More information about the Synth-diy