Phase and Polarity, I thought I understood!

[Tenson]

Hi,

I'm a bit confused about this one..

I was playing around with my DCX2496 to try align a tweeter and mid for a nice step and frequency response, and I found something a bit odd IMO.

I was under the impression that a phase shift of 180degrees (just a phase shift, not a delay) is exactly the same as a polarity inversion. Unless there is a DC offset.

However, yesterday I tried measuring the frequency response between tweet and mid... if the polarity of the driver is 'normal' I get a smooth response with the Xover summing as it should. If I switch the polarity to 'inverted' I get a pretty deep null, as one would also expect. But if I switch it to 'inverted' and put a 180degree phase shift on it... well I would have expected it to be the same as with the polarity 'normal' and to get a smooth response again but I don't, it stays as a null! Whats going on? Surely inverted polarity with 180degree of phase shift is the same as a non-inverted signal, isn't it?

[pinkmouse]

Only with a constant unchanging signal like sine, square or triangle waves.

[Tenson]

Where can I find a graphical example? I have just been toying in CoolEdit and I can't do much that makes an inversion look any different to a 180degree phase shift on a sample of white noise, which should have a random waveform.

Maybe CoolEdit actually does an inversion instead of a 180 shift.

Here is a sample of an asymmetric waveform. I have even added a DC bias.



Now with an inversion.

And now a 180 degree shift... it looks just like the inversion to me!

[pinkmouse]

I really don't know what cooledit is doing there, but you could use something like Photoshop layers to overlay the adjusted waveforms with the first graph.

Excel's graphing functions might be another better way of simulating the process.

[Eva]

The phase shifters in the DCX2496 actually simulate all-pass filters, which don't act in the same way as polarity inversion would do. Learn about all-pass filters and play with different phase shift values.

[pinkmouse]

Of course, Eva, you are correct. I was answering the question that was asked, and failed to see the big picture.

[Davey]

To be more specific, the DCX phase shift function implements a second-order all-pass filter. So, you'll get a frequency-dependent phase shift that varies between 0 and 180 degrees. You can shift/slide the turnover frequency up or down depending upon the setting of the crossover filter.

An inverted polarity setting in the DCX yields a frequency-independent shift of 180 degrees.

Cheers,

Davey.

[poptart]

a frequency-independent shift of 180 degrees.

I might be on stupid pills today, but I can't tell what that would mean. How could phase have any meaning without reference to a particular frequency?

[jnb]

It simply means that all frequencies are inverted...180degrees regardless of frequency.

[poptart]

Okay I can see how in something like an inverting amplifier stage the input and output are 180 degrees out of phase at all frequencies. I was getting caught up by thinking of a phase "shift". If you delayed a signal so that a 1kHz component was shifted by 180 degrees then a 500Hz component would only be shifted by 90 degrees right? I work on phase modulated transmitters all day, I guess I get stuck thinking of advancing and retarding phase.