The impact of EQ on phase

DrewPeterson7

DrewPeterson7

Sage of the Order
Ok, I've never entirely gotten this...

I've seen a ton of people discuss this around here, how using too much EQ can cause phase issues in a mix. However, I don't think I've heard anyone actually explain what it is they mean by that.

Are we talking about something as simple as EQing the hell out of a bunch of tracks can make things start to sound out of phase with each other/potentially cancel out when collapsed into mono, something like a "phasing" or flanging sound in the mix, or something else?

I'm not a huge EQ abuser, I generally try to get things fundamentally where I want them while tracking and then just clean up a little here and there, so it's possible I've just never run into an over-EQd mix, but for whatever reason I'm having a hard time imagining what that could mean.

Thanks!
 
It is in the nature of standard filter design (an EQ is nothing but a filter) to cause varying (but very small) amounts of time delay in the signal, with the amount of time delay based upon the frequency. This means that for a complex wave (like virtually any real-world audio signal outside of a simple synth sine wave oscillator), when it goes through a standard EQ will find it's component frequencies slightly shifted in time by different amounts, changing the phase relationship between those frequencies within that signal.

Just how audible this effect actually is, or when it becomes unpleasant if it is audible, is one of those subjects up for endless debate between the geeks and gearsluts of this business - kind of like those boring and tiring endless arguments about other esoterica such as jitter, harmonic distortion, sample rate and so forth.

The simple answer that avoids all that is, "If the EQ sounds good, use it. If it sounds bad, don't use it." :)

G.
 
Interesting... Thanks guys! So it's made out to be a much bigger deal than it actually is, and the fact I've never really understood what the problem was isn't a huge issue?

That's relieving. :)
 
Phase shift is not particularly a problem until you have something to compare it to, like splitting a signal, applying a filter to one copy and mixing them back together. Crossovers are probably the most common example of this and speaker designers spend a lot of time trying to realign phase at the crossover points. Phase shift on a lone signal is a non-issue.
 
bouldersoundguy said:
speaker designers spend a lot of time trying to realign phase at the crossover points.
Click to expand...

Yes, excellent point. Again, with crossovers the problem is not the phase shift itself either, but the resulting frequency response change when the woofer and tweeter outputs combine acoustically in the air.

--Ethan
 
Ethan Winer said:
Yes, excellent point. Again, with crossovers the problem is not the phase shift itself either, but the resulting frequency response change when the woofer and tweeter outputs combine acoustically in the air.

--Ethan
Click to expand...

It's a marriage made in heaven. :D









:cool:
 
SouthSIDE Glen said:
It is in the nature of standard filter design (an EQ is nothing but a filter) to cause varying (but very small) amounts of time delay in the signal, with the amount of time delay based upon the frequency.
Click to expand...

Glen,

I knew this about analog filters. Does the same hold true for digital filters?

Thanks,
Jim
 
diggy_dude said:
I knew this about analog filters. Does the same hold true for digital filters?
Click to expand...
I have to admit that this question is on the bleeding edge of my knowledge. I *think* that it depends upon the algorithms used for the digital filters; that the shift is mathematically built into the way in which the filtering is done, and is not strictly dependent upon whether it's a physical device or a virtual one. IOW, a digital EQ that EQs using the same principles as a standard (non-linear phase) analog EQ to do it's EQing will still introduce the the same internal issues that an analog one would. But I honestly am not yet fully confident of that answer.

But I still say, who cares? It's an academic argument only. The only thing that matters is how it sounds, and I don't know about you, but I've been using non-linear-phase EQ for my entire life and have no complaints about using it ;).

G.
 
SouthSIDE Glen said:
IOW, a digital EQ that EQs using the same principles as a standard (non-linear phase) analog EQ to do it's EQing will still introduce the the same internal issues that an analog one would.
Click to expand...

In the somewhat outdated Computer Music Tutorial, Curtis Roads does say that digital filter algorithms use delay in some fashion, if I remember correctly. I don't know how far digital filter design has progressed in the last 15 to 20 years though, hence my question.

Glen, I've read many of your posts in the Mastering forum and have come to respect your professional opinion. Would you care to advise me on the Klark Teknik question I posted in the Rack forum?

Thanks,
Jim
 
SouthSIDE Glen said:
I have to admit that this question is on the bleeding edge of my knowledge.
Click to expand...

You want a Band-Aid? ;) :D

Hey...your closer to that edge than I am! :cool:
Analog is comprehensible...but digital algorithm talk gives me a headache.
Not that I couldn't understand it...I just don't want to burn up the remaining few active brains cells left in my head! :)
 
miroslav said:
You want a Band-Aid? ;) :D

Hey...your closer to that edge than I am! :cool:
Analog is comprehensible...but digital algorithm talk gives me a headache.
Not that I couldn't understand it...I just don't want to burn up the remaining few active brains cells left in my head! :)
Click to expand...

I know what you mean. All things analog have what Burroughs would call "mythic resonance". Digital processing concepts are purely mathematical and clinical. It's not surprising that they would leave an artist cold.
 
diggy_dude said:
Would you care to advise me on the Klark Teknik question I posted in the Rack forum?
Click to expand...
You're probably asking the wrong guy, Jim, as I have no hands-on experience with the Klark. They do have a very good reputation, but you probably already know that. I am not a pro mastering engineer myself (I'm more into the mixing and producing sides if it); you'd probably get better responses from the real mastering guys here who may possibly have had actual exposure to or experience with the Klark.

reading your post over there though, my comment is that if you're looking at something top shelf to add to your rack, side chaining it on a dbx 166 is kind of a mismatch IMHO. I'd consider upgrading that compressor as a valid option to look at as well.

IMHO, YMMV, 1080P, ETC.

G.
 
SouthSIDE Glen said:
reading your post over there though, my comment is that if you're looking at something top shelf to add to your rack, side chaining it on a dbx 166 is kind of a mismatch IMHO. I'd consider upgrading that compressor as a valid option to look at as well.
Click to expand...

That seems reasonable advice. My further thinking about the Klark is that having 10 independent high-quality filters would solve all of my equalization needs for life. So, for example, if I were to eventually upgrade to a better compressor, there would be no need to shop for another equalizer as well. In fact, it would continue to be useful if I gave up on music production altogether. I could, for instance, slap it in the tape loop of my stereo amp and enjoy it there.

All of this may be a moot point by now anyway, as the seller has just made me an offer that I'd be a fool to pass up.
 
diggy_dude said:
I know what you mean. All things analog have what Burroughs would call "mythic resonance". Digital processing concepts are purely mathematical and clinical. It's not surprising that they would leave an artist cold.
Click to expand...

:rolleyes: :rolleyes: :rolleyes: :rolleyes: :rolleyes: :rolleyes:

The answer to this question:

In the somewhat outdated Computer Music Tutorial, Curtis Roads does say that digital filter algorithms use delay in some fashion, if I remember correctly. I don't know how far digital filter design has progressed in the last 15 to 20 years though, hence my question.
Click to expand...

The short answer is the theory hasn't progressed at all because it didn't need to. Filter theory is entirely based on math and has been fully described for . . . I dunno, a whole lot of years. It's like EE 101 stuff.

If analog components were ideal, or stated another way, if DSP resources were unlimited, then either would be a perfect realization of the mathematical concepts involved.

Failing that, the preference for analog vs. digital EQ is based upon your feelings about their particular nonlinearities. I can design a complete crap analog or digital filter for you if you like. Designing a good filter is somewhat more difficult, but not really if one knows what one is doing, which doesn't really include me at the highest level but both are fun to play with anyway.

There are certain types of filters that are not possible to execute in analogland. Whether or not those are useful is a reasonable question. They are absolutely essential for high-quality digital audio, even if EQ is not required, but that's a post for another day. They were never needed in analog and that arguably still remains the case, although radio broadcasters are happy to have them, I would think as an example.

In general, digital filters have improved because resources have increased. Even so, there were plenty good DSP filters 10 years or even 15 years ago, but they would have been hardware DSP solutions and typically expensive. Now there is no reason for any VST to lack high-quality algorithms.
 
You must log in or register to reply here.
Back
Top