ProSoundWeb Community

Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => Audio Measurement and Testing => Topic started by: Alex Thompson on February 25, 2012, 10:08:59 am

Title: Phase measurement strange result
Post by: Alex Thompson on February 25, 2012, 10:08:59 am
I got a strange result measuring my system.  The lines on the picture are left speaker and right speaker.  Test was done in the shop with 2 subs side by side and speakers on top of subs side by side, measurement mic on center with speakers.  If I'm interpreting the graphs right, it seems there is a 360 degree phase flip on one of the speakers at the crossover frequency but not the other. Does anyone have any idea what would cause this and should I be worried about it?

If it matters, system is Yorkville UCS1 subs, Yamaha S112IV mains, Ashly 3.24CL system controller, yamaha power amps, A&H mix wizard.

Thanks for any advice!



One other thing I forgot to measure.  Measurement software is Fuzzmeasure 3 on macbook pro.  Phase options are "group delay", "mixed phase response", "minimum phase response" and "excess phase response".  It is only on mixed and excess phase response that i see the 360 degree flip.
Title: Re: Phase measurement strange result
Post by: Mike Christy on February 25, 2012, 11:24:59 am
I got a strange result measuring my system.  The lines on the picture are left speaker and right speaker.  Test was done in the shop with 2 subs side by side and speakers on top of subs side by side, measurement mic on center with speakers.  If I'm interpreting the graphs right, it seems there is a 360 degree phase flip on one of the speakers at the crossover frequency but not the other. Does anyone have any idea what would cause this and should I be worried about it?

If it matters, system is Yorkville UCS1 subs, Yamaha S112IV mains, Ashly 3.24CL system controller, yamaha power amps, A&H mix wizard.

Thanks for any advice!



One other thing I forgot to measure.  Measurement software is Fuzzmeasure 3 on macbook pro.  Phase options are "group delay", "mixed phase response", "minimum phase response" and "excess phase response".  It is only on mixed and excess phase response that i see the 360 degree flip.

I have used SMAART, am not a top level expert, but can get around in it pretty well, but I would say it is just flipping over, it is essentially the same as the other top, 0 degree = 360 degree. If you saw 180 degrees, then youd have an issue.

Experts, please correct me if Im in error ( Im sure you will anwyays)

Mike
Title: Re: Phase measurement strange result
Post by: Ales Dravinec on February 25, 2012, 07:27:56 pm
As Mike said, it is all good and expected.

The thing I do not get is, how on earth is the phase trace so damn good ???

With respect
Title: Re: Phase measurement strange result
Post by: Langston Holland on February 26, 2012, 12:43:09 am
Hi Alex:

That's an impossible phase response for your system. It would only be possible with sophisticated FIR processing or hundreds of all-pass filters, either of which would require a latency on the order of an audible echo to the original source.

Those phase traces look like an extremely well behaved magnitude measurement flipped upside down and plotted on a phase graph in error. Another possibility is that the numbers on the Y-axis scale are missing a zero.

To put this in perspective, a perfect measurement of an ideal loudspeaker with no crossovers would yield a "minimum phase" response. That means that the arrival time at each frequency of the magnitude response is as short as possible.

Many moons ago Tom Danley posted an ideal minimum phase passband with a 2nd order Butterworth (12dB per octave) high pass at 100Hz and a 2nd order BW low pass at 5kHz. 90˚ Of phase error is added per order per filter, which totals 360˚ in this case. The roll offs of real loudspeaker systems are never this smooth and rarely this gradual:
Title: Re: Phase measurement strange result
Post by: Paul Tucci on February 26, 2012, 11:04:55 am
Lang,

Yes, that's a very well-behaved phase trace except for where it flips 360. That is a similar behavior to Smaart's "unwrap phase trace" display parameter. That would often cause instability in the phase trace display. Remember, this is not Smaart, but some other platform with undefined display options. I susupect it's a display issue too.

Tucci

Title: Re: Phase measurement strange result
Post by: drew gandy on February 26, 2012, 12:06:58 pm
I think you should just disregard the phase response in FuzzMeasure altogether.  I have found that it does some kind of idealized guessing about the phase.  I discovered this while taking measurements with Spectrafoo and FuzzMeasure using the exact same setup.  Spectrafoo showed the phase anomalies that I expected whereas Fuzz showed a totally unreasonable phase graph like what you show here.  Magnitude responses were basically the same between the 2 platforms.  I've been planning to bring this up with the Fuzzmeasure people but haven't yet.  FWIW, I like a few things that the software does but the phase thing is a deal breaker for the vast majority of uses. 
Title: Re: Phase measurement strange result
Post by: Ales Dravinec on February 26, 2012, 12:26:34 pm
... That's an impossible phase response for your system. It would only be possible with sophisticated FIR processing or hundreds of all-pass filters, either of which would require a latency on the order of an audible echo to the original source....

Lang,

Thank you for valid explanation.
However, equipment listed by OP could not possibly produce the phase TF as shown.

I like to think that I have an in depth understanding of 'phase' and graph(s) as shown tickle my curiosity.

w/r
Title: Re: Phase measurement strange result
Post by: Langston Holland on February 26, 2012, 02:48:05 pm
Hi Aleš:

I was wondering if some confusion might happen given the similarity of names - my reply was aimed at the OP.

In about 10 years I'd like to think I'll have half the depth of understanding in these areas as you do now. :)