[sdiy] A question about phasers/notch filters

[Tom Wiltshire]

I wasn't thinking only of allpass stages though. I understand that you need to pile a load of them in series to get the most phase shift you can and generate multiple notches. But that isn't the only way to make notches - probably just the cheapest/easiest.

Imagine I had a whole bank of voltage-controlled SVFs with notch outputs on each one. I could easily tune each filter to a different frequency, and then use an LFO to sweep the whole lot up and down. Lo and Behold! The world's most unnecessarily expensive phaser!

But practical considerations aside, is the only reason not to do it like that the fact that the notches aren't as deep? (As Tony confirmed) It's still a good reason, I guess.

The reason I ask is that I've been reading about the Buchla 297 Infinite phase shifter, and wondering how they did the barber's pole phasing. Seems to me the simplest way would be individual notches swept by ramps so that when they go off the top, they jump back down to the bottom. Feed  X stages from ramps shifted apart by 360/X degrees and bingo. However, this wouldn't work with the standard phase shifter series arrangement, since the position of the notches depends only on the overall phase shift, and you can't move individual notches separately, even if you do feed separate LFO signals to each stage. So they must have used some other, non-series arrangement of phase shifters. Does anyone know what the trick was?

Thanks,
Tom

On 7 Oct 2011, at 09:59, Mattias Rickardsson wrote:

> It's easy to believe that notch filters can be connected in parallel
> because we tend to think that a notch has a particular "sound". In
> fact it has not, and that's the whole point here. (-:
> 
> /mr
> 
> On 7 October 2011 09:14, Richie Burnett <rburnett at richieburnett.co.uk> wrote:
>> Hi Tom,
>> 
>> The allpass filter stages in the phaser need to go in series.  Each stage passes
>> all frequencies with equal amplitude but imparts a phase shift curve to the
>> signal as you go through it's "cutoff frequency".  As you said each time you
>> pass the signal through another allpass stage it imparts more phase shift.  All
>> of the allpass stages are typically of the same design so the phase shifts
>> contributed by each stage all add up.
>> 
>> The aim of the game with a phase if you like is to bugger about with the phase
>> of the incoming signal as much as you can by passing it through an allpass
>> filter many times.  Then mix it with the original signal again so that you get
>> lots of constructive and destructive interference.  You get notches when the
>> phase of the "buggered about signal" is 180 degrees relative to the original
>> signal because they cancel out when added.  However if you can shift the phase
>> of the signal by as much as 540 degrees or even 900 degrees you will get
>> destructive interference again and another produce another notch!
>> 
>> The more allpass stages you put in series the greater the total final phase
>> shift and the more chance that you will have several notches occuring somewhere
>> within the audio band.  Think of how a chorus or flanger works:  Here you delay
>> one version of the signal by lots of milliseconds before combining it with the
>> original and it introduces hundreds of peaks and notches!
>> 
>> If you did put lots of analogue allpass stages in PARALLEL and combined their
>> outputs, each stage would contribute at most 180 degrees of phase shift, so you
>> could only have a single notch at best.  Much better to cascade them to impart
>> the maximum phase shift to one signal, and leave the original signal un-shifted
>> before mixing the two together.
>> 
>> -Richie,
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at dropmix.xs4all.nl
>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy