[sdiy] frequency shifter

Richie Burnett rburnett at richieburnett.co.uk
Sat Mar 14 19:00:04 CET 2020

If you want good performance, the tolerances on the phase shift networks are quite stringent. So it's actually easier to make a decent frequency shifter using digital filters than it is using analogue filters. With IIR digital filters you can get the poles exactly where you want them and they won't drift with age or changes in ambient temperature.

Functions like multiplication are also usually mathematically perfect to the precision of the ALU when preformed digitally, but are often an approximation with unwanted distortion, noise and feed-through when implemented with analogue electronics.

Likewise, even simple addition and subtraction of signals are limited by the tolerance of resistors in analogue mixing circuits, or have to be manually trimmed. In the digital domain, if you add together two numbers you get a precisely even weight of the two values, without any tolerance, trimming or temperature drift issues. 


Sent from my Xperia SP on O2

---- Eric Brombaugh wrote ----

>On 3/14/20 5:26 AM, ColinMuirDorward wrote:
>> Has anyone seen this implemented in a digital FX unit?
>> Here <https://www.fullbucket.de/music/freqshifter.html>is a vst I'm 
>> looking forward to trying out.
>There are lots of digital implementations of frequency shifters out 
>there - the quadrature oscillator / multiplier is a fundamental 
>operation of most digital radios and the dome filter (aka Hilbert 
>Transform, aka analytic signal converter) is straightforward to 
>construct with allpass networks.
>There are a few sites out there that discuss this. I think the oldest 
>one I know of is from C Sound:
>Olli Niemitallo gave a good backgrounder on construction of a Hilbert 
>Transform filter:
>I've implemented the Hilbert approach in several synth modules:
>The algorithm is simple enough to run in a low-end ARM processor.
>Synth-diy mailing list
>Synth-diy at synth-diy.org

More information about the Synth-diy mailing list