dbuckley
Senior Member
- Location
- Canterbury, New Zealand
Following from the (now closed) Battle of the Phases thread, in context of the difference between polarity and phase, I made the throw-away comment:
After Bob closed the thread, I got a number of PMs saying basically that mains power and audio were different, and sine waves are different to non-repititive waveforms such as pink noise. Rather than shuttle PMs back and forth, I thought I'd post a considered response here.
Lets start with a picture, which I hope will be non-controversial(!)
We have a perfectly ordinary transformer, the sort that feeds many residential installations, with a MV primary, and a 120/240 secondary. I've illustrated where the oscilloscope probes go, and the assumption is that we are going to feed this transformer with one complete cycle of power from the 60Hz MV source. The oscilloscopes are set to display one complete cycle starting from when the primary starts its ascent.
Now, one school of thought states that the difference between waveforms A and B is that they are phase shifted 180 degrees, and another says that they are polarity inverted without phase difference.
Lets look more closely.
The underlying question is "what transform can I apply to waveform A to get waveform B?".
The polarity school says that you take the value of any point in waveform A, change it's sign, and plot it and you then get waveform B. The phase shift school says you need to apply, well, phase shift. "Phase shift" is another word that means time, so if you delay waveform A by 180 degrees, you get waveform B. 180 degrees is an odd thing to delay something by, but doing some simple maths (the only kind I can do), at 60Hz, each full cycle of the waveform (which is 360 degrees) takes 16.666 milliseconds. So delaying by 180 degrees is half that, so delay waveform A by 8.333 milliseconds and you get waveform B.
Which is almost true, lets look at this pictorially.
Whats wrong with this picture? Well, with the phase shift way, you lose the first half cycle, and the second half cycle isn't seen on the scope at all, becasue as noted at the top, the scope only displays one full cycle.
Can this really be what is happening.
I say not, and to those that say I'm wrong, consider this:
What we have here is nine transformers, each wired with swapover wires, and as there is an odd number of transformers, the whole thing will end up inverting the incoming power.
Now if each transformer introduces 180 degrees of phase shift, or a delay of 8.333 milliseconds, then nine of the things gives (180 x 9) 1620 degrees of phase shift. In time, thats 9 x 8.333ms, or about 75ms. If that is the case then the two bulbs will come on 75ms apart, and thus it will be visible that electricity is being "delayed" through those transformers.
Just for completeness, you cant argue that the phase shift clock resets every 360 degrees, as they either (a) time goes backwards, or (b) the transformers know they are undoing a previous transformers work.
Nah.
Polarity inversion is polarity inversion, and phase shift is phase shift.
Now all the above is real world, based on transformers, oscilloscopes and other stuff electrical people deal with. In the world of maths then maybe there are scenarios where there is no difference between polarity and phase shift, and you can substitue a polarity inversion with a 180 degree phase shift. What that tells us is that the math model is not dealing with the full reality, but for some sorts of problems that may not matter. But, in the real world, polarity inversion and phase shift are never the same.
(Still editing this post for spelling mistakes!)
We did all this stuff in the audio world eons ago, I thought everyone knew this stuff...
After Bob closed the thread, I got a number of PMs saying basically that mains power and audio were different, and sine waves are different to non-repititive waveforms such as pink noise. Rather than shuttle PMs back and forth, I thought I'd post a considered response here.
Lets start with a picture, which I hope will be non-controversial(!)
We have a perfectly ordinary transformer, the sort that feeds many residential installations, with a MV primary, and a 120/240 secondary. I've illustrated where the oscilloscope probes go, and the assumption is that we are going to feed this transformer with one complete cycle of power from the 60Hz MV source. The oscilloscopes are set to display one complete cycle starting from when the primary starts its ascent.
Now, one school of thought states that the difference between waveforms A and B is that they are phase shifted 180 degrees, and another says that they are polarity inverted without phase difference.
Lets look more closely.
The underlying question is "what transform can I apply to waveform A to get waveform B?".
The polarity school says that you take the value of any point in waveform A, change it's sign, and plot it and you then get waveform B. The phase shift school says you need to apply, well, phase shift. "Phase shift" is another word that means time, so if you delay waveform A by 180 degrees, you get waveform B. 180 degrees is an odd thing to delay something by, but doing some simple maths (the only kind I can do), at 60Hz, each full cycle of the waveform (which is 360 degrees) takes 16.666 milliseconds. So delaying by 180 degrees is half that, so delay waveform A by 8.333 milliseconds and you get waveform B.
Which is almost true, lets look at this pictorially.
Whats wrong with this picture? Well, with the phase shift way, you lose the first half cycle, and the second half cycle isn't seen on the scope at all, becasue as noted at the top, the scope only displays one full cycle.
Can this really be what is happening.
I say not, and to those that say I'm wrong, consider this:
What we have here is nine transformers, each wired with swapover wires, and as there is an odd number of transformers, the whole thing will end up inverting the incoming power.
Now if each transformer introduces 180 degrees of phase shift, or a delay of 8.333 milliseconds, then nine of the things gives (180 x 9) 1620 degrees of phase shift. In time, thats 9 x 8.333ms, or about 75ms. If that is the case then the two bulbs will come on 75ms apart, and thus it will be visible that electricity is being "delayed" through those transformers.
Just for completeness, you cant argue that the phase shift clock resets every 360 degrees, as they either (a) time goes backwards, or (b) the transformers know they are undoing a previous transformers work.
Nah.
Polarity inversion is polarity inversion, and phase shift is phase shift.
Now all the above is real world, based on transformers, oscilloscopes and other stuff electrical people deal with. In the world of maths then maybe there are scenarios where there is no difference between polarity and phase shift, and you can substitue a polarity inversion with a 180 degree phase shift. What that tells us is that the math model is not dealing with the full reality, but for some sorts of problems that may not matter. But, in the real world, polarity inversion and phase shift are never the same.
(Still editing this post for spelling mistakes!)
Last edited: