Phase Shift Through Analog Gear (less than 180)...

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Ian MacGregor

Well-known member
Joined
Jun 3, 2004
Messages
280
Location
Echo Park, Los Angeles, CA, USA
Yo,
I've been doing more and more engineering these days and have been running into weird phase issues more often that have me thinking. For example, this has nothing to do with "polarity"; I fully know how to work with a multiple mic setup (like a drum kit) and get everything working with proper polarity.

What I'm experiencing is more subtle phase shifts that still cause issues with low end in multiple mic situations. For example, a studio I did a mix at had a Calrec stereo comp (sort of Neve-ish) that I used for parallel drum compression (mixing analog). When bringing the compressed channels up on faders, the low end would start to get phasy. Not "out of polarity" though. I could flip the polarity and it wouldn't get better, just different.

I've experienced this same phenomenon occasionally with similar arrangements. Kick in & kick out mics, bass DI & bass amp mic...

My EE instinct thinks that somehow there is a less than 180 degree phase shift going on and it's due to the inevitable low roll off frequency through the processing path. For example, if a transformer has a bit too high LF rolloff point (say 40Hz), there will be 45 degrees shift @ 40. Right?

Am I headed in the right direction? Am I hearing bad caps and their respective low freq rollof's moving higher? I just acquired a AP system one with all the bells and whistles and I'm going to start putting some gear under the microscope to see what's going on.

ian
 
Hi,


  I guess i  don't need to point out that 2 mics at different distances to a source will have differing phase, dependent on the distance . . .



        you'd be surprised(maybe not!) how many "engineers" pay scant attention to this whilst multi-micing a kit, for instance.




      Kindest regards,


      ANdyP
 
Yes! Well aware of that... I'm hearing this with multiple mic situations as well as parallel compression while mixing. I've engineered enough to know when something is a bit out of ordinary while, say, micing a kit with an inside mic and an outside mic.

I was having weird issues with the last kick chain that I used:

Kick in: D112 - BAE 312 - API 560 - Summit Audio Pultec-ish EQ - PT in
Kick Out: Geffel LDC - BAE 312 - API 560 - PT in

The only difference being the Summit. When I bypassed the Summit, things seemed to be a lot better all of a sudden... Obviously, EQ adds phase shift as well...

ian
 
Well phase is also dependent on frequency... 2 mics that are in phase in the mid range might be a little out of phase in the bass. but I think your totally right on with phase shift due to unusual high pass. Or are you EQing one half of the parallel comps while leaving the other side unprocessed?  Eq could do it too. I think that maybe all or some of the issues could easily add up to a noticeable effect.
  This, i think, could be one of the main over looked issues between equipment and op amps and stuff. We always look at THD and other obvious things like noise, but what about that graph of phase shift that all opamps have? but to me low end equipment (especially mic pres) sounds like they are out of phase with themselves... like how your explaining not out of phase but not in phase....just weird
AC
 
What you're experiencing is a classic case of phase shift -- real phase shift, not polarity inversion. And yes, it could come from something in the signal chain having a rolloff near the lowest frequencies.

When I tested the Little Labs IBP Phase Corrector, the manual noted that this is not an uncommon occurrence when, for example, combining a DI'd bass and a mike on the bass cab. Not only is there a time offset from the time the signal takes to get from the cone to the capsule, but there's a phase shift in the cab itself from its bass rolloff (and another one from the mic's rolloff). Just time-shifting the signal takes care of the delay in the mic, but it doesn't fix the cabinet or mic's phase shift. For that, the IBP does remarkable things. It's one of the most useful tools I've run across.

You can now get a plug-in from Little Labs that does the IBP thing in software.

Peace,
Paul
 
I am not a fan of parallel compression, but whenever there are multiple streams of the exact same signal being processed differently and recombined; polarity, relative delay, and/or phase shift can (in theory) cause them to interfere constructively or destructively (comb, or notch).

Delay is more of an issue for high frequencies than low, and I wouldn't expect similar processing paths (two comps) to generate 180' of phase shift between them, especially at the bottom end. Even if one was DC coupled and the other not, by the time the phase shift from a HPF in the one generated 180' of phase shift relative to the other, the amplitudes wouldn't be remotely similar so a null still wouldn't occur.

But it's always something, so keep looking...Phunny phase response in a single path that doesn't cause audible problems alone, could be a different story when in parallel with the same source material.   

Note: as a practical matter you could roll off the low bass in one of the two paths a few dB and reduce potential interference.

JR
 
One other thing to check -

Rule out the mixing path itself by running a familiar commercial CD through it and compare to your tracks.

If it is still going on to any degree in the mixing chain check the internal wiring on the transformers where applicable.  If they were mistakenly miswired in relation to each other or to themselves this can easily cause some phase weirdness. 

FWIW - over the years of attempting multiple mic set ups on guitars I have normally had to settle for a compromise on correct phasing.  Drums have seemed a little easier to manage but It's probably going to still be there to some degree.

Just curious, do your phase issues sound like 'linear' low end roll offs or is it weirdness in the stereo imaging (head bending effect) ?  Does it sound as if you're approaching a null or resonance as the faders move?
 
lassoharp said:
FWIW - over the years of attempting multiple mic set ups on guitars I have normally had to settle for a compromise on correct phasing.  Drums have seemed a little easier to manage but It's probably going to still be there to some degree.

Just curious, do your phase issues sound like 'linear' low end roll offs or is it weirdness in the stereo imaging (head bending effect) ?   Does it sound as if you're approaching a null or resonance as the faders move?

I'm not getting the classic "out of polarity" type issues with the head bending (when in stereo) and the complete low end loss that is fixible by flipping polarity. I'm getting more of what might be a resonance that shifts up or down depending on where the phase switch (polarity) is on the mic pre. Neither option sounds "right." I'm definitely going to check out the little labs IBP.

Another thing I have been pondering is that instead of multiple chains with processing, the trick might be to mix signals together early and then process that mixed signal.

So instead of having a kick in that gets EQ and then a kick out that gets eq -> just mix kick in and out together and eq that blended signal.

I like getting sounds that sound mixed (or at least headed in that direction) while tracking, which usually requires some EQ and a bit of compression to tape/PT. Everyone's feedback has been good... At least I'm not crazy and it's worth looking into. At least it seems that there are some designs that lend themselves better to parallel processing and others that don't. 

ian
 
Ian MacGregor said:
Another thing I have been pondering is that instead of multiple chains with processing, the trick might be to mix signals together early and then process that mixed signal.

This is how I liked to approach multiple sources when I did more rock recording.  Typical examples would be a 441 and 57 on elec, kick in and out mic or bass DI and mic.  I absolutely found that finding the appropriate blend of the sources and then processing that if necessary made for a more cohesive and direct sound.  Given the types of productions I worked on I would rarely get to see or hear the whole picture when tracking, this made it difficult to call exactly how I would shape the sound at tracking time.

The IBP is a clever little box.  I had one and sold it when times got tough during a prolonged illness.  The thing to bear in mind is that it can only really align phase at one frequency, often though this can make a big difference to how two sounds gel.  I loved it for tracking bass more than anything.  I'm rambling, sorry!

Cheers,
Ruairi
 
I had this with a solo singer in an orchestra last time around and her mic just had a funny phase relationship
to the rest of stuff, getting her in there was pretty hard...my Halos have a phase rotating DSP thingy but it
was a pretty elusive thing in general.

Cool thing with combining unprocessed stuff and then hitting the processors with...makes perfect sense to
think about. I can imagine this applies to compression and effects more than it does to EQ?
 
A few caps & xfmrs , doesn't take much
to create a Phase shift
I'd be curious how many " degrees " of shift
there has to be before you start to hear it
[ guess that also depends on the level ]
 
okgb said:
A few caps & xfmrs , doesn't take much
to create a Phase shift
I'd be curious how many " degrees " of shift
there has to be before you start to hear it
[ guess that also depends on the level ]
The answer is not an easy one, because phase shift is not audible on its own, it's audible when two signals have different phase. It depends on the relative levels and the way these signals are mixed. 
Assuming an all-analog signal path, phase is tightly dependant on the frequency response, in particular the steepness of the variations of the frequency response.
When two signals of equal amplitude are mixed, a phase difference of 45° will create an attenuation of 0.7 dB; it may be audible.
45° of phase-shift from an EQ is not common: as an example, a KT DN360 (which is known as one with the sharper curves) produces 37° of phase-shift when the fader is at its maximum (+ or - 12 dB) and 55° with one fader at +12 and the next at -12.
A 2nd-order high-pass filter, such as can be found on many mixers, produces 90° phase-shift at its cut-off frequency; if the high-passed signal is summed with the unfiltered signal, there will be an attenuation of ca. 7 dB at half the cut-off frequency. This is undoubtly audible, but this is more or less what one expects when engaging the HPF...
A well designed piece of equipment should not produce any undue phase-shift other than what is inherent to whatever EQ or Hi or Lo-pass filtering is applied.
So-called "phasyness", which is basically the result of comb-filtering is more audible at mid-frequencies than high or low. So it seems that the problem here is created by the differences in acoustical path between mics.
Still the issue with the Summit EQ suggests there is something to investigate; fortunately, it is much easier to do measurements of electronic units than microphones and air.
 
abbey road d enfer said:
Assuming an all-analog signal path, phase is tightly dependant on the frequency response, in particular the steepness of the variations of the frequency response.

Ok... so maybe this is a tangent, but it is then possible to "correct" for phase anomalies with EQ? Somewhere along the way I heard an engineer say this...
 
This is why I will eq or compress as little as possible when recording. Mixing is a whole other story though. ;D

To quote the late, great, Tom Dowd "If you grab an eq while we are recording, your fired!!! If you don't like the sound move the mic around until you find a sound you like. If that doesn't happen the use a different microphone."

The your fired part was the exact quote the rest was recalled to the best of my memory but the idea is there in full
 
Ian MacGregor said:
abbey road d enfer said:
Assuming an all-analog signal path, phase is tightly dependant on the frequency response, in particular the steepness of the variations of the frequency response.

Ok... so maybe this is a tangent, but it is then possible to "correct" for phase anomalies with EQ? Somewhere along the way I heard an engineer say this...
It may be possible, but the problem is that you don't hear phase directly, you hear the tonal changes.
 
pucho812 said:
This is why I will eq or compress as little as possible when recording. Mixing is a whole other story though. ;D

To quote the late, great, Tom Dowd "If you grab an eq while we are recording, your fired!!! If you don't like the sound move the mic around until you find a sound you like. If that doesn't happen the use a different microphone."

The your fired part was the exact quote the rest was recalled to the best of my memory but the idea is there in full
I don't subscribe to this purist attitude; very often the sound I want to hear is not achievable by just moving or swapping mics.
Does anybody care about the phase shifts introduced by using a cardioid mic in close micing, or by an omni placed near a reflecting surface?
It seems phase is a dirty word; it shouldn't be. Phase is just the other side of frequency response, a very natural thing.
 
well at one point in time eq's were known as correctional devices as used in a sparse manor.  But I do agree about your phase comment. sometimes slight phase variation is a good thing and can be a nice effect. Sometimes it's not...
 
Pucho, I think you're missing the meaning of 'natural' slightly.

In analog circuitry there is normally a correlation between RATE OF CHANGE of amplitude response and phase response. I believe that this is what was meant by 'natural'.

Also, the original quote was "if you dont liek the sound while you're recording... and the FIRST thing you reach for is an Equalizer... you're fired".

This is a very different meaning; specifically addressing a person's laziness, NOT blasting equalizers.

Equalizers were also ORIGINALLY preset devices, designed to correct for known, quantified response anomalies... hence their name.

Keith
 

Latest posts

Back
Top