Rational Acoustics



Dr. J
May 31st, 2012, 02:13 PM
Hey guys -- How many wraps is considered normal? Since the measurement mic is placed a certain distance from the speaker -- I know it is impossible to have a starting position of ZERO degrees. What is it that you look for to mark a spot at Zero degrees or does it matter?

Thanks!

Rasmus Rosenberg
May 31st, 2012, 03:08 PM
Hey J,
Im not sure what you mean, but your measuring relative so Zero is zero relatively. A quick look at the last years onaxis measurements generalizes/average about 3 phase wraps, but don't think you get much out of that generalization.

PaulTucci
May 31st, 2012, 04:14 PM
Hey guys -- How many wraps is considered normal? Since the measurement mic is placed a certain distance from the speaker -- I know it is impossible to have a starting position of ZERO degrees. What is it that you look for to mark a spot at Zero degrees or does it matter?

Thanks!

Dr J,

Firstly, it is not impossible to have a starting position of zero degrees. In fact, it's necessary. To "slow down" the Reference signal compared to the measurement signal (typically acoustic) an offset time is inserted. Now a valid comparison can be made. Some part of the phase trace will be flat, or at zero. This is where the offset is perfectly timed to the arriving signal. Where it is ahead or behind is where the phase trace is leaning up or down.
Do this, put the same signal on both channels. Look at the TF. It's flat. Look at the phase trace, it's flat. No speakers do that. The low end tends to lag behind, visible by the downward to the right angle as frequency increases.

If all the frequencies arrived simultaneously, the phase trace would be flat. That requires superb engineering and money. we live with a few wraps in the low end. This time smear would be unaccepatable in a visual sense. It would be as if the red colors lagged behind the blue colors by a few frames of video. Imagine that.

PT

Dr. J
June 1st, 2012, 12:14 PM
Dr J,

Firstly, it is not impossible to have a starting position of zero degrees. In fact, it's necessary. To "slow down" the Reference signal compared to the measurement signal (typically acoustic) an offset time is inserted. Now a valid comparison can be made. Some part of the phase trace will be flat, or at zero. This is where the offset is perfectly timed to the arriving signal. Where it is ahead or behind is where the phase trace is leaning up or down.
Do this, put the same signal on both channels. Look at the TF. It's flat. Look at the phase trace, it's flat. No speakers do that. The low end tends to lag behind, visible by the downward to the right angle as frequency increases.

If all the frequencies arrived simultaneously, the phase trace would be flat. That requires superb engineering and money. we live with a few wraps in the low end. This time smear would be unaccepatable in a visual sense. It would be as if the red colors lagged behind the blue colors by a few frames of video. Imagine that.

PT

Perhaps I should re-phrase. My curiosity is coming from an article (trying to locate again) that explained the difference in degrees is due to the measurement mic having to be out front rather than pushed up against the cone of the speaker. I guess what I am asking is -- do you get to choose where ZERO degrees starts at or are you just stuck with whatever frequency area lands on zero? Does it even matter?

I wondered about the number of wraps because it seems that there is a acceptable amount versus an excessive amount. So is it common for a system to have about 1,000 degrees of rotation? I maybe getting in over my head here so please excuse me....

Langston Holland
June 1st, 2012, 05:23 PM
Hi Doc:

I don't understand your question about phase... Maybe you're interested in understanding how and why to correctly set the reference delay in Smaart?

On the other question about a loudspeaker system having 1,000˚ of phase rotation (or more); yes - it's more common than not. Most acoustic crossovers in loudspeaker systems use approximately 4th order slopes. Each "order" accumulates 90˚ of phase rotation, thus a 4th order type will be 4 times that much (360˚) at each crossover point.

A triamped system with a sub supplementing the lows will of course have 3 crossovers, for a total of 3 x 360˚ = 1,080˚. There will be another 4th order (or higher) slope on the high pass of the sub which will add yet another 360˚ and this is an ideal situation for the vast majority of systems that do not use linear phase filters and/or flat phase tuning for the higher passbands.

Examples follow of an idealized (electrical output of a processor) and a real loudspeaker system. The latter does not include the subwoofer, thus looks a bit better than it should. The real loudspeaker is an on-axis measurement of a triamped EAW KF650z using EAW's recommended processing (red traces) and something I came up with (blue traces) back in the day when I was getting started and a guy named Chuck McGregor was politely thrashing me. My curves look much nicer, but it's on-axis only and made a mess of the downward vertical off-axis response. Oddly, it turns out that most folks aren't directly on-axis to the loudspeakers. :)

518
517

Kip Conner
June 1st, 2012, 09:41 PM
You can manipulate the point at which the phase begins by adding a low pass filter set to the live trace. This is easiest to see with a nice linear response measurement such as a direct box. Generally the frequency and phase response will be flat depending on the quality of the direct box. The phase response will start at zero and as you add the filters the phase to the 45 and 90 degree positions depending on the filter. It's a great experiment that can tell you a lot about your own gear. I found out one of my three countryman di's was wired wrong to be pin2 hot. Not a big deal, it just helps to know that one is not functioning the way you would expect. (Countryman DIs are wired in the factory to be pin 3 hot)

As for "normal" it's hard to say... I think. The more delay you need to add to a particular pass band for your phase alignment the more wraps that would occur. I tend to worry more about the individual pass bands responses and how they relate to the adjacent band and less about the overall number of wraps.

merlijnv
June 2nd, 2012, 03:53 AM
Also be aware off severe comb filtering. Reflections of near identical level that arrive half a wavelength later, combined with the first or direct arrival. Produce a phase response, for that specific range, that might be mistaken for a wraparound. A hole in both transfer and coherence are indicators that something's off. All the more reason to make a reflection "free" reference trace up close, where the direct sound dominates. This will help a lot in the far field to distinguish wraparound from comb filtering.

PaulTucci
June 2nd, 2012, 05:23 PM
Perhaps I should re-phrase. My curiosity is coming from an article (trying to locate again) that explained the difference in degrees is due to the measurement mic having to be out front rather than pushed up against the cone of the speaker. I guess what I am asking is -- do you get to choose where ZERO degrees starts at or are you just stuck with whatever frequency area lands on zero? Does it even matter?

I wondered about the number of wraps because it seems that there is a acceptable amount versus an excessive amount. So is it common for a system to have about 1,000 degrees of rotation? I maybe getting in over my head here so please excuse me....

Often times the answers you get are more sophisticated than the question asked but we are trying to help.
So the article probably confused you a little. True, it takes a measurable amount of time for the sound to get off the speaker to the measurement mic. That finite, measurable time can be used as the "delay" for the reference side of a TF to synchronize the two things being compared. What we're trying to convey is that the phase wraps that occur (because the measurement signal happens AFTER the reference signal) for the early arriving stuff can be undone by synchronizing the arrival times via the offset. YES, you can purposefully choose to manage the offset time so that a particular area of interest results in a flat phase trace. That indicates synchronous arrival times. It's not a beginners' technique, but it opens up possibilities into audio surgery.

Yea, a couple 3 or 4 wraps seems to be common.
I had a situation where the low end of a PA du jour was so late arriving that the PA sounded drunk. It slurred every word. I would call that an excessive amount of lagging phase.

PT

Rasmus Rosenberg
June 3rd, 2012, 05:22 AM
A few pic. What i think throws most people off, and thats forgetting to turn off the auto delay track. First a pic of a virtual top box, same measurement but with two different ref delay times to show that the phase trace change when you change the Ref delay. Also you can see the lagging low freqvence as Paul mentioned due to the high pass filter, in the top trace.

519


Next two image is emulating trying to align a sub with a top speaker, And how you can fool your self if you try to compare traces with different Ref delay times. It shows the 2 individual and the sum overlaid.
520521

And lastly a pic of what Merjlinjv, wrote about where comb filter from a reflection (0,5ms) causes phase wraps. (same "top speaker" as above). The "reflection" is 3 db louder just to pump the visual.

522

Dr.J did you find the article, you where talking about?

mhv
R

Dr. J
June 3rd, 2012, 12:55 PM
Thanks everyone for your responses! Sorry for the delay in my responses ...... been out the last three nights in a row doing sound. I will try to locate the article today. I do know it was on the Prosoundweb site written by one of the many authors (cannot remember).

Ok -- so there can be upwards of 3 or 4 wraps and over a 1,000+ degrees of rotation. Seems to be right around what I have seen as well....... just not as flat. More like some wraps in the lower spectrum that eventually looks like a giant smile by the time it gets to the upper spectrum. I think that is pretty normal. I have NO Allpass cabability on my DSP unit.

Paul -- I laughed when you said the PA sounds Drunk (LOL). I have actually heard that from systems before....... usually it is the band that is drunk or the sound guy.....

Anyways..... I will try to find the article and attach a link so you all can read it. Just that one part caught my attention and sparked my curiosity. Don't know if it is even applicable but just good FYI....

Dan Currie
November 24th, 2012, 08:26 PM
Playing around at the shop with a single box and am looking for some input. Thought this thread would be a good place to place to put it. Sorry but I don't have MS paint so there are pictures, no screen shots. My first measurement is with the factory preset. Second measurement I played with phase and output delays. Any thoughts on which would have better intelligibility? Does anything look 'wrong'? As you've noticed the second measurement has one less phase wrap.

Rasmus Rosenberg
November 26th, 2012, 03:54 AM
Hey Dan,
Sorry but you post too little information to give answer to any of your questions. General we can try to comment on traces that compares to another, and its easier if they are overlaid. Its really hard to say anything about "final responses" that people post. What are you trying to measure? What output delays did you adjust? Did you re adjust the Ref delay? Why is the top of the response hovering at -180 deg (out of polarity or maybe its meant to be like that?) Etc etc

Rasmus

Arthur Skudra
November 26th, 2012, 10:07 AM
In addition to what Rasmus said, what are your ears telling you?

luigichelli
November 26th, 2012, 08:09 PM
is that a box with a ribbon tweeter?

Dan Currie
December 4th, 2012, 01:13 PM
Thanks for the responses. I was in the shop working on another rig and thought I'd pull out a XLC while rush hour passed. Not the most ideal environment. The single box was on the floor and my purpose was to have some fun with it. While tinkering with the factory preset I noticed inverting the polarity of the lows on the dx38 processor and adding a little delay got rid of a phase wrap. Yes, I changed the reference delay after the polarity reversal. I wasn't so much interested in the transfer function, only that it didn't change much from the original program. Bringing the phase of the highs to 0deg did a whole bunch of ugly stuff to the phase trace. The processor doesn't have all pass filters so it looks like some other tweaking will have to be done. As for the ultimate test of listening to the box well I didn't. I need to find some more time to get into the shop for that. My goal is to find if a flatter phase trace has a detrimental effect on intelligibility when using conventional filters.

Harry Brill Jr.
May 12th, 2013, 02:30 AM
I could swear the dx38 had allpass filters. Oh well been 15 years.

The main issue with second guessing the factory is that you are measuring on axis and they measure all around the box both horizontally and vertically. Sometimes you have to compromise in one area to get things to work better overall. Looks like those settings are designed for 4 boxes minimum. I would hate to use them as front fills with those settings.

Keep experimenting.


Thanks for the responses. I was in the shop working on another rig and thought I'd pull out a XLC while rush hour passed. Not the most ideal environment. The single box was on the floor and my purpose was to have some fun with it. While tinkering with the factory preset I noticed inverting the polarity of the lows on the dx38 processor and adding a little delay got rid of a phase wrap. Yes, I changed the reference delay after the polarity reversal. I wasn't so much interested in the transfer function, only that it didn't change much from the original program. Bringing the phase of the highs to 0deg did a whole bunch of ugly stuff to the phase trace. The processor doesn't have all pass filters so it looks like some other tweaking will have to be done. As for the ultimate test of listening to the box well I didn't. I need to find some more time to get into the shop for that. My goal is to find if a flatter phase trace has a detrimental effect on intelligibility when using conventional filters.

Dan Currie
June 6th, 2013, 09:00 PM
XLD's are used for front fills. Experiments have come to a halt since all the processing got upgraded to Lake.


I could swear the dx38 had allpass filters. Oh well been 15 years.

The main issue with second guessing the factory is that you are measuring on axis and they measure all around the box both horizontally and vertically. Sometimes you have to compromise in one area to get things to work better overall. Looks like those settings are designed for 4 boxes minimum. I would hate to use them as front fills with those settings.

Keep experimenting.