[sdiy] Phase correctors

[Magnus Danielson]

cheater cheater wrote:
> Hey guys,
> does anyone know of any variable phase correctors?
> 
> By phase correctors I mean the things usually put in EQs that correct
> the filters' phase response. Basically, the phase correctors have a
> flat frequency response, and the phase shift rises (or falls) as
> frequency rises (or falls).
> Unfortunately, I wasn't lucky enough to find one schematic ...
> 
> Those things look interesting, though:
> http://www.springerlink.com/content/g67w778950831075/
> http://www.springerlink.com/content/r5537667429tu61k/
> 
> http://sound.westhost.com/pcmm.htm Look at fig 2.5 - the phase delay
> drops (and becomes 'negative') with frequency rising, at a certain
> rate. The idea is to be able to control this rate.
> 
> I briefly thought about just taking such a simple all-pass filter and
> simply varying the 'wet' level, so the mix between the output of the
> APF and the original (dry) signal, but then figured that it would just
> create something to the effect of an EQ instead.

By doing that you have created a null, just as with a phaser... you 
don't want to do THAT.

> What do you guys think? Is this possible? Even by simply using
> different element values (and switching between them), so not being
> fully variable, but still allowing some degree of freedom.

You can do it. You can create one-pole and two-pole fully parametric 
phase correctors and they are not even hard to control. They are just as 
easy as a fully parametric EQ, but with the twist that the level control 
is at a fixed value. For one-pole you have only a frequency control, for 
the two-pole you have frequency and Q/width control. It is essentially a 
special case of an EQ. You do want a bypass-switch for each section.

> What are your favourite designs of all-pass filters? Maybe some of
> them would be more pliable for modifying the phase rolloff.

For one-pole, a very simple one-opamp design is possible, which is the 
standard of the shelf 1 pole AP found in many design examples.

For a two-pole, a state-variable design might be the best choice. Need 
to think about it.

The trouble with phase correctors is that they can be a bit hard to 
verify their tuning. If you don't have the right tools you might just 
fuck things up more than you repair them.

Well used full-parametric EQs can be used to "move" troublesome poles or 
zeros out of range. This requires very good tools and experience, but 
can help cleaning up responses. Using the propper tools could also with 
help of an imaginative mind and lots of experience help solving STRANGE 
resonances and cancelations. Start to think in terms of where poles and 
zeros are in frequency and damping and ways to move them or damp them 
out makes you look very differently on acoustics and transient 
responses. I encourage you to learn more about it.

Cheers,
Magnus