[sdiy] Novation peak NCOs

[Eric Brombaugh]

Tom's analysis below is exactly why this is such an attractive option. 
I've done some experiments where I implemented the NCO and waveshaping 
at the high sample rate and then used bandlimited decimation (basically 
filter and discard samples) to reduce the sample rate down to something 
that a normal audio DAC could handle so it's not even necessary to use 
fancy RF-grade converters.

Lattice has some nice, low-cost FPGAs in small DIY-friendly packages. 
The ice40 Ultra 1K has a couple of MAC cores and comes in a 48-pin QFN 
for less than $4. There are sufficient resources on this part to do a 
fair bit of high-rate DSP and it would be ideal for something like this.

https://www.digikey.com/product-detail/en/lattice-semiconductor-corporation/ICE5LP1K-SG48ITR50/220-2064-1-ND/5358105

I did a little demo board for the 4K version of this chip (unfortunately 
no longer easily available but pin-compatible with the 1K part) and 
open-sourced it here:

http://ebrombaugh.studionebula.com/embedded/f303_ice5/index.html

Eric

On 04/20/2017 04:13 PM, Tom Wiltshire wrote:
> Ok, so let's just have a look at that for a moment.
>
> Assume a highest output frequency of 20KHz (I wouldn't even be able to hear this). We've got a nyquist frequency of 12MHz. That means that only anything that goes far enough over 12MHz to alias all the way back down below 20KHz is actually going to be audible. So that's 23.98MHz before we get audible aliasing. That's equivalent to the 1199th harmonic of  20KHz ramp wave.
>
> It's a high enough frequency they can ignore aliasing, even for naive waveforms, and even for tough cases like ramps and squares.
>
> It does seem like overkill, but FPGAs are cheap these days, and running a few (or a lot of) NCOs isn't even going to make one break a sweat.
>
> Yeah, why not? I like it. Simple, direct, does the job.
>
> Tom
>
> On 20 Apr 2017, at 18:20, Richie Burnett <rburnett at richieburnett.co.uk> wrote:
>
>> An interesting direction Novation are taking with the use of FPGA based NCO oscillators on their new hybrid polysynth. There's a video from Superbooth about it on sonicstate. It uses a sample rate of 24MHz to generate classic analog saw, pulse, tri waves, etc, then feeds them through conventional analog VCF, VCA, etc.
>>
>> Seems like a bit of a brute force way to crack the old aliasing nut, but I guess it gives them the ability to do some wacky audio rate modulation stuff with reduced aliasing too.
>>
>> -Richie,
>>
>> Sent from my Xperia SP on O2
>>
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at synth-diy.org
>> http://synth-diy.org/mailman/listinfo/synth-diy
>
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
>