[sdiy] Simple DSP as VCO substitute (was Re: CEM3340 reissue)

[sleepy_dog at gmx.de]

Aliasing - but, considering that this aliasing starts from very low, not 
yet audible frequencies, with typical PWM synth sounds in mind (< 20 Hz ?),
how significant would the amplitude of aliasing actually be as it gets 
more in the audible range?

Just to make sure I'm viewing this correctly:
Say I have a Pulse signal generated by a timer output, switching at 
1kHz. The signal as such would not be any different from a pulse signal 
switched by a VCO comparator - it's basically an anlog signal, right?
Now if the pulse width is changed on the fly in a stepped way, say with 
a least 7 bits resolution, with a 10 Hz triangle LFO, this is where the 
aliasing creeps in - correct?
Is there a calculation of how much it actually is at *really* audible 
range (including speakers and such), say at 40 Hz or something?
I wouldn't know how to calculate that, alas.
But If I use OTA chips which have 70 dB or so SNR, and use 12 bit DAC 
for the other channels which is close to that, I guess I wouldn't need 
to aim at super HiFi with that?

Steve


Am 14.06.2016 um 21:22 schrieb Vladimir Pantelic:
> On 14.06.2016 20:39, sleepy_dog at gmx.de wrote:
>>
>> Square wave on the DAC?
>
> yes :)
>
>> Would it not be enough to just do that with a harwdare PWM channel?
>> It did sound nice enough at first "glance" when I tried it, but I'm 
>> not sitting
>> on a pile of signal theory to know what I'm doing there ^^
>
> there is the issue of aliasing, unless you run the PWM at an insanely 
> high rate so that you hit the exact spots where the signal changes 
> happen. with the DAC you can do something like PTR or BLEP to 
> calculate the inbetween samples at the transitions in order to reduce 
> aliasing.
>
>
>