[sdiy] Digital accumulator VCO core?

[Richie Burnett]

Ok, thanks for clarifying.

Your additive synth project sounds fantastic BTW.  Still reading all the 
details!

-Richie,



-----Original Message----- 
From: Mike Bryant
Sent: Friday, February 12, 2021 12:12 PM
To: Richie Burnett ; Brian Willoughby ; cheater cheater
Cc: SYNTH DIY
Subject: RE: [sdiy] Digital accumulator VCO core?

Sorry yes I wasn't clear.  No not using a perfect mathematical ramp but 
using a table lookup of a sum of sinewaves or some other band-limited 
waveform, but then filtering off higher frequencies when playing higher 
notes on the keyboard.  There still be some aliasing but I've never found it 
at audible levels.

-----Original Message-----
From: Richie Burnett [mailto:rburnett at richieburnett.co.uk]
Sent: 12 February 2021 11:56
To: Mike Bryant; Brian Willoughby; cheater cheater
Cc: SYNTH DIY
Subject: Re: [sdiy] Digital accumulator VCO core?

> I'm also not sure why people use BLEP or other techniques - just use
> the accumulators to create a perfect sawtooth, square, triangle or
> indeed noise...

I'm not sure what you mean here?  Generating a perfect mathematical ramp for 
the sawtooth using an accumulator?  Or are you meaning generate a sawtooth 
through the process of additive synthesis?

>...and then put it through a simple digital filter before the DAC to
>get rid of anything that could alias.

If you are generating a mathematical ramp with reset (like what you get from 
the phase-accumulator in an NCO) then this has an infinite spectrum, and 
those frequency components that "could alias", have *already* aliased.  They 
are baked into the signal as soon as you synthesise it using such a basic 
method.  You can't just filter them out before feeding the signal to the 
DAC!  In practice you need to synthesise a waveform that is mathematically 
not a trivial ramp, but instead is the shape of a sawtooth that is 
band-limited.  This is what BLIT, BLEP, minBLEP, Poly-BLEP all do to various 
degrees and in different ways.

> If you run at 96kHz sampling this becomes pretty trivial in
> computational terms - a few shift rights or multiplies and some adds.

If you generate an "A" at 3.52 kHz, you can only include the harmonics up to 
the 13th harmonic (45.76 kHz) if you don't want anything to alias.  If you 
don't mind harmonics aliasing in the ultra-sound region where you can't hear 
them you could include all the harmonics up to the 21st harmonic (73.92 
kHz).  The the next harmonic is the 22nd that would be at 77.44 kHz but 
would alias back to 18.56 kHz and be audible with a 96 kHz sample rate.  In 
practice you'd probably stop at the 5th harmonic (17.6 kHz) with additive 
synthesis because anything over 20kHz is inaudible anyway.

Regardless of whether your synthesis is via additive, wavetable or a simple 
phase accumulator you still need to curtail the otherwise infinite spectrum 
of the basic waveforms (sawtooth, Triangle, Square, Pulse, etc) in order to 
prevent aliasing.

-Richie,


--
This email has been checked for viruses by AVG.
https://www.avg.com