Variable Phase?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

MikoKensington

Well-known member
Joined
Jun 3, 2004
Messages
1,371
Location
Detroit
Has anyone tried to implement a continuously variable phase control into a preamp? Sometimes it seems a mere in/out switch isn't enough.
 
There's a way to do this using a pot and cap and a phase splitter. I'm pretty sure I've seen it in 'the art of electronics'.

I'll try and find it....

cheers,

K
 
The problem is getting equal shift at all frequencies.
Maybe this isn't necessary, but I would think you would want to shift everthing by the same angle.
I remember this question. I think Scenaria mentioned mic placement, and somebody wanted an electronic solution, at which time the thread sank.
cj
 
I've been messing around with simulating all-pass circuits with Electronics Workbench (piece of crap...). Like CJ said, the hard part is getting the same phase shift at all frequencies (which is physically impossible, right?)

But the allpass filters definately do SOMETHING to the phase of the signal without affecting the frequency response. I never had the time :evil: but I have a feeling that to really find out how to do this, I need to breadboard some stuff up and have a go.

Ian
 
Here is some info on phase shifting circuit, I have tried to simulate it on Spice but never been able to achive phase shift equally at all frequencies.....

http://img.villagephotos.com/p/2004-7/769916/laglead.jpg

http://img.villagephotos.com/p/2004-7/769916/laglead2.jpg

http://img.villagephotos.com/p/2004-7/769916/laglead3.jpg

http://img.villagephotos.com/p/2004-7/769916/laglead4.jpg

http://img.villagephotos.com/p/2004-7/769916/laglead5.jpg
 
indeed...and if you chain enough of these circuit elements together with carefully chosen frequencies you can achieve an overall time shift, which equates to an equal phase shift at all frequencies ( within an audio bandwidth ). Focusrite did this on one of their compressors to delay the main signal path with respect to the sidechain to achieve a true 0mS attack time. But...who wants to poke their audio through 20 odd op-amps?
 
miko-

you probably know this trick, but in case you dont, take the two mics you are setting up on a source, flip the phase on one and move it around until it completely cancels the source and you cant hear much of anything. When you flip the phase again, you'll have a coherent signal. That trick might get you by until you can figure out an electronic solution.

dave
 
> continuously variable phase control into a preamp? Sometimes it seems a mere in/out switch isn't enough.

Then you probably want a delay line, to equalize the air path length among several mikes all hearing the same source. Moving the mike is by far the cheapest and most-correct answer, but sometimes that conflicts with other choices. Digital delay lines are now not-crappy, and may be better than fighting phases.

> getting equal shift at all frequencies.

Not the same thing at all. A 1 foot path difference will be nearly zero phase at 50Hz, but many cycles out of phase at 15KHz.

> physically impossible, right?

I don't think there is an elegant way to get arbitary phase shift, constant over frequency, on complex waveforms.

But in real life, you cascade several phase shifters at overlapping frequencies. The number of phas shifters is determined by how wide a frequency range you need and how far off the mark you can let the phase get. SSB radio transmitters used to use "90 degree" phase shifters, which were really +/-10 degrees 500Hz-3KHz. That was good enough for voice communications. They were not simple, and even DIY radiomen often bought a canned phase-shift network. Op-amps make it easier, but trying to get 5 degree precision over hi-fi bandwidth probably takes an awful lot of parts, fairly well matched, and potential smearing from many buffer amplifiers.
 
Don't crucify me for this one, but since I've been using protools almost exclusively these days I've been having a blast sliding regions to line up phase discrepancies. Works like a charm on drums and stuff that you don't really have much leeway with in mic placement. Or if you're just too damn tired to get up and move the mic :oops: .

Zach
 
There's no real reason to want a constant phase shift across all frequencies. You use phase to fight against time domain problems. Delays are just constant (exponential) phase variation with no magnitude change. By using a well thought out phase shifting network you can probably minimize the comb filtering caused by combining a signal with a slightly delayed version of itself.

Where would this be handy...well, how about a bass that you're tracking through a DI and by micing a nice ampeg B15. The mic'd track will necessarily be delayed versus the DI as a result of the distance from the mic to the speaker (not to mention the transfer function of the amp and speaker itself). Combine the two in an attempt to get a super 'phat' bass sound and maybe there's a little comb filtering around 3kHz or so. Here's where that phase lag circuit comes in handy...lag the DI track so that the comb filtering is minimized in the band where our ears are most sensitive (1kHz-3khz). Should wok quite nicely.

Oh, and I don't think that an arbitrary phase response is possible...just like an arbitrary magnitude response is not possible (our filters must be increments of 6dB/8ve, 20dB/decade).

Cheers,

Kris
 
> I don't think that an arbitrary phase response is possible...just like an arbitrary magnitude response is not possible (our filters must be increments of 6dB/8ve, 20dB/decade).

Possible, not elegant.

3dB/oct is the standard pink-noise filter. A 9dB/oct slope high-cut is a great hiss filter. Cascade several overlapping 6dB/oct shelf filters. Completely arbitrary magnitude is what graphic and parametric EQs aim for. It always takes a lot of parts compared to the "natural" n6dB/oct filters. Arbitrary phase curves are also possible, same way.
 
Completely arbitrary magnitude is what graphic and parametric EQs aim for

It's what they aim for, but can't ever possibly really acheive. There's only so much you can do with one pole in a filter. Even with a modern 'state space' approach only certain combinations are causal. Sure, you can almost cancel poles with zeros and place them real close together, but in the end you can only be *so* accurate with your placement due to tolerances in components. Basically you end up with a wiggly approximation to an arbitrary magnitude response that's made up of a bunch of 6dB/8ve increment sections arranged in such a fashion as to emulate a particular filter response.

Same situation exists with Linear FIR EQ's like the waves lin EQ....you try to draw in an arbitrary curve, but it prevents you from doing certain things (probably doesn't have enough filter taps to let you specify any curve you want).

okay....maybe given an infinite number of poles and zeros, or filter taps an arbitrary response is possible, but it's certainly not practical.

Cheers,

Kris
 
i've never had the need for such a device when a fine resolution delay is about. about the only part of the sony dmx r100 is that is has one on ever channel.
 
Back
Top