[sdiy] A question about phasers/notch filters

[Tom Wiltshire]

On 8 Oct 2011, at 15:50, Richie Burnett wrote:

> Hi Tom,
> 
> I don't know if this is the effect you are looking for, but I've uploaded a couple of examples of "infinite phasing" I knocked up here.
> 
> Orignal audio file: http://www.richieburnett.co.uk/temp/orig.mp3
> Slowly falling notches: http://www.richieburnett.co.uk/temp/phased1.mp3
> Faster rising notches: http://www.richieburnett.co.uk/temp/phased2.mp3
> 
> These examples were made using a digital implementation of a frequency shifter and then mixing the output with the original dry audio to introduce notches.
> I used two lines of six cascaded digital allpass filters, pinching the "dome filter" time constants from what Jurgen Haible used here for his hardware implementation:
> 
> http://www.jhaible.de/fs1a/fs1a.html
> 
> -Richie,

Yep, that's the effect. Reach for the lasers!

I suspect you've made the same effect a different way. Whilst Buchla could have done it like that, I've seen one oblique reference to the method used in the 297 that suggests that it was two separate analogue phase shifters both fed quadrature ramp waves, with VCAs to fade them out whilst they jumped back down. Whether this is true or not, I have no idea. But it's possible, at least.

There are some things I don't understand about the frequency shifting method. I think I understand the frequency shifting itself. You take a frequency A, ring mod it with a signal B, which gives you frequencies A+B and A-B. If you simultaneously ring modulate A with a shifted version of B, you can presumably cancel either A-B or A+B to leave you with just the 'single sideband' you mentioned. Is that right?

The bit I don't get is why adding that shifted signal back to the original would give *moving* notches. I suppose it's to do with the beat frequencies between the two signals, in the same way that two close frequencies sound like one frequency modulated by an LFO (which isn't really there). I'll have to think about that a bit more until it becomes clearer.

Given the simplicity of generating quadrature sinewaves of exact frequency in the digital domain, coupled with the ease of generating perfect four-quadrant modulation with no feedthrough, it does rather scream out for a digital implementation.

Incidentally, there's another infinite-phase example  on the Modcan frequency shifter page:

	http://www.modcan.com/bseries/freqshift.html
	http://www.modcan.com/mp3/newmodules/barberpole.mp3

All this is a long way from an old fashioned analogue phaser. Sounds a bit like they're not going to do barber's pole phasing without a lot of extra mucking about. Pity.

Thanks,
Tom