The difference between polarity, phase shift, and time shift

Anything not covered elsewhere.
Post Reply
Message
Author
User avatar
BrentEvans
Posts: 3044
Joined: Thu Oct 09, 2008 10:38 am
Location: Salisbury, NC

The difference between polarity, phase shift, and time shift

#1 Post by BrentEvans »

This came up today in another forum and I stumbled upon an explanation that seemed to be well understood, so I thought I'd share it here.

Polarity shift, phase shift, and time shift, are related but not identical concepts. The terms are often confused and interchanged, but each has a specific definition.

Polarity is the basically direction of an AC signal. When you reverse polarity, you flip the direction of the AC signal. For audio, this can be accomplished by swapping the + and - leads on a wire. This swap is what the "phase" button (often marked with a Greek letter theta, the crossed out circle) does on audio gear.

Phase refers to relative position of amplitude. Phase is continously variable from 0 to 360 degrees. From 90 to 270 degrees, blending an out of phase signal results in destructive cancellation, with total cancellation at 180 degrees. From 0 to 90 and 270 to 360 degrees, the result is constructive cancellation with total summation at 0/360 degrees and no change in level at 90 and 270 degrees. Reversing the polarity of the wires introduces a 180 degree phase shift, so total destructive interference, provided the signals being mixed are identical except for the polarity reversal.

Time shift by nature introduces a phase shift, but the amount of phase shift introduced depends upon both the time delay introduced and the frequency. For example, a 100hz sine wave is at 180 degrees phase in 1/200 of a second, while a 200Hz sine wave is at 360 degrees phase in the same amount of time. Therefore, a 1/200 second time shift introduces 180 degrees of phase shift at 100hz, and (effectively) 0 degrees of phase shift at 200hz.

Phase shift alone does not introduce time delay, in theory. If you could introduce a phase shift without also introducing a time shift, the result would be that the waveform is altered to a relative position at all frequencies consistent with the phase position for that frequency. For instance, say you are going to introduce a +90 degree phase shift to the above signal at 1/200 second on the time line. The 100hz signal at -100% amplitude would be shifted to -50% amplitude, and the 200hz signal at +100% amplitude would be shifted to +50% amplitude. At first glance, this looks like a time shift if you only look at one frequency, but if you look at the all frequencies (which is really hard to visualize or demonstrate on a graph) you see that they have all shifted exactly the same amount. In a time shift, all frequencies are affected differently. This is the core of how flange effects work... they introduce and blend in a time delayed signal in some form. Sometimes there are other components, but that's the core of them. Blending true phase shifted signals results only in an level/amplitude change from the original signals.

Now, in practice, phase shift is hard to do. There is no way (that I know of) to do it in the analog world. High end DSPs can do a true phase shift, but with DSP there is always some latency involved, so you have to account for that. It's not a problem if all signals being summed are passing through the same DSP, but blending a processed phase shifted signal with an unprocessed signal will have a result other than simple blending.

I hope this helps someone. There are certainly plenty of applications for this information in the audio world, and thus in the speaker building world.
99% of the time, things that aren't already being done aren't being done because they don't work. The other 1% is split evenly between fools and geniuses.

Bruce Weldy
Posts: 8301
Joined: Fri Nov 27, 2009 11:37 am
Location: New Braunfels, TX

Re: The difference between polarity, phase shift, and time shift

#2 Post by Bruce Weldy »

Good info. Thanks.

6 - T39 3012LF
4 - OT12 2512
1 - T24
1 - SLA Pro
2 - XF210


"A system with a few knobs set up by someone who knows what they are doing is always better than one with a lot of knobs set up by someone who doesn't."

User avatar
Bill Fitzmaurice
Site Admin
Posts: 28619
Joined: Tue May 02, 2006 5:59 pm

Re: The difference between polarity, phase shift, and time shift

#3 Post by Bill Fitzmaurice »

BrentEvans wrote: Now, in practice, phase shift is hard to do. There is no way (that I know of) to do it in the analog world.
You can do it with an all-pass filter.
https://en.wikipedia.org/wiki/All-pass_filter

User avatar
BrentEvans
Posts: 3044
Joined: Thu Oct 09, 2008 10:38 am
Location: Salisbury, NC

Re: The difference between polarity, phase shift, and time shift

#4 Post by BrentEvans »

Bill Fitzmaurice wrote:You can do it with an all-pass filter.
https://en.wikipedia.org/wiki/All-pass_filter
Interesting. I've never seen that as an adjustable control on any piece of gear. It seems like it might be a fairly esoteric thing.

The VENU360 can do continously variable phase shift, as can other DSPs I've used, but I've never needed it. Did it exist on analog crossovers once upon a time?
99% of the time, things that aren't already being done aren't being done because they don't work. The other 1% is split evenly between fools and geniuses.

User avatar
Bill Fitzmaurice
Site Admin
Posts: 28619
Joined: Tue May 02, 2006 5:59 pm

Re: The difference between polarity, phase shift, and time shift

#5 Post by Bill Fitzmaurice »

I've seldom seen all pass filters used, other than in studio gear. Back when everything was analog most PA operators didn't know enough about phase etc. for there to be a market for it. Another method, rarely used, was analog 'bucket brigade' delays, but they weren't exactly high fidelity.

Post Reply