Phase response inquiry

[omni]

Attached please see my latest phase response chart.....I have no idea how to interpret it and very little knowledge about phase response and how it effects sound........Will some of you please help me interpret this chart and answer a few questions for me?.....What does this chart represent?.....How does this relate to how the speakers actually sound?.........What in crossover design effects phase response?.......What are the chart characteristics of a good phase response? ....What am I looking for to achieve a good phase response?.....And any other information you can give me that will help me in understanding phase response..........Thank You for any help I can get......Omni

[omni]

Here is another chart which has phase information for the same design .....I am also submitting this for your review, and again, the same questions I posted above apply to this chart as well.....Omni

[Rybaudio]

What does this chart represent?

If you put a sine wave into a speaker you get one out of the same frequency but amplitude scaled and phase shifted. A frequency response plot shows, as a function of frequency, what the amplitude scaling is, in dB. A phase plot shows, as a function of frequency, what the phase shift is. Thus, if a plot says 120 degrees at 1 kHz, when you put a 1 kHz tone into the system, the output is a 1 kHz tone shifted 120 degrees.

How does this relate to how the speakers actually sound?

If you're talking about the phase of a given completed system, then my answer is that it doesn't matter much. It is possible to screw up phase in such a way that it will be audible, but for anything reasonable you might do, the phase response of the end system isn't much of a concern.

When I say that, I mean that phase isn't inherently a concern. Phase is a very important thing to keep track of when you are designing and tuning a system! If you have two sources that are playing at the same frequency (drivers at crosspoint), then the phase response of each of them tells you how they will sum. For example, if they are shifted by 180 degrees and of the same amplitude, they will cancel each other.

What in crossover design effects phase response?

In the system, the drivers have phase shift, the electronics have phase shift, and there is phase shift from the propagation time. In the passive components and the speaker, the phase shift is essentially from elements (springs, masses, inductors, capacitors, acoustic masses, acoustic springs) storing energy and then releasing it later.

What are the chart characteristics of a good phase response?

For the most part, it isn't as simple as good and bad. It's more about being able to optimize given a situation. For example, in a crossover you most likely want the phase responses of the two sections in the crossover region to be the same so they sum perfectly.

What am I looking for to achieve a good phase response?

Again, the curve isn't inherently worth much (as opposed to say, a FR graph), but it tells you what the system is doing. That allows you to manipulate other things to optmize things that do matter inherently.

[Tom Danley]

Hi

An acoustic phase measurement shows when in phase the pressure is produced relative to the driving signal. It is the systems magnitude and phase, which govern the systems response to any signal for example its impulse response.
Your phase plot seems to show the acoustic phase plus the phase shift caused by time delay. To see actual acoustic phase, one must remove all Time delay related phase rotation. In other words, a loudspeakers acoustic phase response is not affected by distance to microphone other than all frequencies being delayed equally.
Thus, when you remove the phase shift related to time, one has acoustic phase.
Also then the distance to the ear has no effect on the speakers ability to preserve the input waveshape again because all frequencies are delayed equally.
Waveshape is preserved if the phase is near zero or –180 from an inverting system,in normal multiway speakers the phase shift prevents the speaker from preserving waveshape. Acoustic phase can be thought of as a frequency dependent change in time, a difference between when one frequency is produced and another.
Simple but sort of and described first by Richard Heyser, but many popular measurement systems have a plot for phase but it is not an actual measurement of acoustic phase.

That can be tested if you have a DSP speaker controller; the controller is set up with a high pass and low pass filter slopes plus a time delay to act like the mic to speaker distance. Then the controller is wired in as a perfect speaker in a noise free environment.
If the indicated phase is the same / similar to the actual phase from the filter set, then it measures phase.
Here are some posts with some graphs that talk about this.

http://srforums.prosoundweb.com/inde...92/0/40/16795/

If your measurement system has some kind of time setting, try looking slightly earlier in time than your current reference.
Hope that helps,
Tom Danley

[omni]

Thank you both for the replies as I will probably have to study them further to digest your insights.........Can you provide any interpretations of those particular graphs that I posted?.......I have studied various aspects of crossover design and am now in the process of tweaking my current crossover via listening and the use of Jeff Bagbys' passive crossover designer.......My Frequency Response charts seem to pretty much reflect what I am hearing during listening tests, but I guess I was sort of wondering whether I am missing something or need to be concerned with the Phase charts as I change component values during my tweaking process...........Respectfully.......Omni

[JLC7]

For you, all I'd worry about is getting the phase of tweeter and woofer at the crossover point to be the same or at least very close.

Basically just get the phases for each driver to 'cross' each other at or very near the crossover point. Phase of the individual drivers isn't what you're concerned about. It's the phases relative to each other that you're concerned with.

How the frequency response of each driver will combine to create the total frequency response of your speaker depends on phases of the drivers relative to each other. Like Rybaudio said, if they're 180 degrees out of phase (basically going the opposite direction) then any sound at that frequency will cancel.

To remove the time of flight phase shift you need to perform a hilbert transform on the frequency response data. This will get you the minimum phase of the driver which you can use in crossover optimization. You can use the FRD spreadsheet from the FRC to do that.

Once you put the driver in a crossover, its minimum phase shifts again. To get the phases of each driver to line up at the crossover point, you need to tweak the values of the components in the crossover. You can also use asymmetrical crossover slopes or change the frequency of the crossover to try and get the phases to line up better at the crossover point.

[djarchow]

QUOTE]Originally posted by omni
Attached please see my latest phase response chart.....I have no idea how to interpret it and very little knowledge about phase response and how it effects sound........Will some of you please help me interpret this chart and answer a few questions for me?.....What does this chart represent?
[/QUOTE]

It is hard to say, is this the acoustic response of the woofer and XO? The plot says LP HP and BP but I don't see the frequency response for the tother two drivers. Your second graph is clearly the XO transfer functions and their phase response. This doesn't matter at all for what you are asking. All the matters is the frequency response of the driver and XO together, the acoustic response.

Originally posted by omni
.....How does this relate to how the speakers actually sound?
[/B]

As other have said phase impacts how the drivers sum together. For drivers to integrate properly in the XO region their acoustic frequency and phase response (driver + XO) has to match the target response frequency response and phase for the drivers to sum properly.

The target spl and phase response are the acoustic XO slopes you are shooting for in your design. For example, a 2500 Hz 4th order Linkwitz Riley XO as your target implies that both your drivers will be in phase at the design point. So if the measured phase of your woofer and XO should be the same as that of the tweeter. In fact for best summation, they should be in phase well on either side of the XO point as well. Just to be clear though, the target phase difference between the drivers, either 0, 90, or 180 degrees is determined by the target slopes. So in phase is only correct for those XOs that sum properly when they drivers are in phase. Others such as the 1st order Butterworth require that the drivers be 90 degrees apart at the XO point and so on.

Originally posted by omni
.........What in crossover design effects phase response?
[/B]

A XO and driver combination is, in most cases minimum phase which means that there is a direct relationship between the frequency response and phase. If you XO changes the frequency response in any way other than just changing the overall level (turning the volume down), the phase will also change to match.

Originally posted by omni
.......What are the chart characteristics of a good phase response? ....What am I looking for to achieve a good phase response?
[/B]

Good phase response means that your acoustic freq and phase for each driver/XO matches the target response over a wide range. It also means that for the drivers to sum properly they must also have the proper relative phase offset as the target.

Originally posted by omni
.....And any other information you can give me that will help me in understanding phase response..........Thank You for any help I can get......Omni [/B]

Here is a great paper on phase response and it's effects by John Kreskovsky

http://www.geocities.com/kreskovs/Phase-B.html

Regards,

Dennis

[pjpoes]

I don't fully agree with everything that has been said here. First, there are things that can make a phase response bad, in that, strong phase shifts coupled with a difficult load impedance (4 ohms) can make a speaker difficult to drive. Additionally, these strong shifts when coupled with certain amplifiers, such as tube amplifiers, can actually cause the amplifier to interact with the speaker in such a way as to change its response. This is why you commonly see speaker designers refer to a speaker as being a good or bad match with tube amplifiers, or a difficult load.

The second issue I take with what is said is that it doesn't address a theoretical ideal important in crossover design. In a perfect way, you would have a speaker which has all the drivers hand off to the next driver with no impact on amplitude and NO CHANGE IN PHASE. However, that isn't possible, but, you can come close. So while for the average designer only interested in a flat amplitude response you can ignore phase, it is important if you are also trying to build a speaker which can cleanly pass the input signal unchanged. Again, this is theoretical as it isn't possible in the real world, but again, you can come close. The name for such crossover topologies are transient perfect, quasi-transient perfect, minimum phase, or I have even heard phase delayed. These all use methods to minimize or eliminate the phase shifts which cause the incoming signal to also be shifted. One commonly referred test for this is a speakers ability to cleanly pass a square wave. Elsinore project The Elsinore project is a good place to read a bit on an attempt to have minimum phase shifts in the crossover as to pass transient signals perfectly.

Now, having said all that, most speaker designers do ignore the phase response and do not attempt transient perfect crossover topologies because doing so often must be at the expense of the amplitude response. It's also a theoretical ideal that can not be achieved, merely approximated, and I think most designers think the average consumer will find a smooth amplitude response more important. I think that a good speaker designer takes into account all aspects of a speakers design, including its transient response (i.e. step response) which exists in the time domain, as well as its amplitude response. It's also very important to look at the phase response in relation to the impedance plot, as again, if you have a strong negative phase angle (say -120 degrees) coupled to an impedance of 3 or 4 ohms, many amplifiers will have trouble with that. Such a design would more than likely cause the average consumer receiver to shut down, and probably give a good many separate amplifiers a bit of trouble too.

Oh I realized that I was talking about both acoustic and electric phase almost interchangeably, and that might be confusing. Acoustic phase is what really impacts the transient performance, and can be affected with delay (be it physical or electrical), where as electrical phase is what causes the load problems for amplifiers, and can not be physically changed, only electrically. Moving a driver in or out in relation to the tweeter, or changing its angle will not have any impact on its electrical phase, but will have an impact on its acoustical phase. Acoustical phase is what is important for a good pass band and good driver integration, electrical phase is what is important for a speakers drivability. But...to make things a little more confusing, they are directly related to each other.

[djarchow]

Originally posted by pjpoes
I don't fully agree with everything that has been said here. First, there are things that can make a phase response bad, in that, strong phase shifts coupled with a difficult load impedance (4 ohms) can make a speaker difficult to drive. Additionally, these strong shifts when coupled with certain amplifiers, such as tube amplifiers, can actually cause the amplifier to interact with the speaker in such a way as to change its response. This is why you commonly see speaker designers refer to a speaker as being a good or bad match with tube amplifiers, or a difficult load.

I understand what you are saying, but for Omni and others, the phase you are talking about is the impdeance phase derived from the impedance of the drivers and XO. The further the system impdance deviates from a resistive load (and a 0 degree impedance phase angle) toward an inductive or capacative load, especially into a low impedance, the more difficult the load is for an amplifier to drive.

Originally posted by pjpoes
The second issue I take with what is said is that it doesn't address a theoretical ideal important in crossover design. In a perfect way, you would have a speaker which has all the drivers hand off to the next driver with no impact on amplitude and NO CHANGE IN PHASE. However, that isn't possible, but, you can come close. So while for the average designer only interested in a flat amplitude response you can ignore phase, it is important if you are also trying to build a speaker which can cleanly pass the input signal unchanged. Again, this is theoretical as it isn't possible in the real world, but again, you can come close. The name for such crossover topologies are transient perfect, quasi-transient perfect, minimum phase, or I have even heard phase delayed. These all use methods to minimize or eliminate the phase shifts which cause the incoming signal to also be shifted. One commonly referred test for this is a speakers ability to cleanly pass a square wave. Elsinore project The Elsinore project is a good place to read a bit on an attempt to have minimum phase shifts in the crossover as to pass transient signals perfectly.
[/B]

Well for Omni who is just trying to understand how acoustic phase affects his system, transient perfect speaker design is probably a bit more than he needs right now. In fact John's paper I linked to does touch on how the phase response impacts the transient response. This is in fact more than he needs right now but the paper does a good job of explaining how the acoustic phase impacts the driver summation which is why I linked to it.

Also to clarify your point your point that a person not interested in transient perfect and wants flat frequency response can ignore phase, this is generally true if you are talking about deviation from zero phase shift in the system acoustic phase. The designer however must be very aware of the phase response of each single driver and it's XO in relation to the target phase and the phase of the other driver. If you don't, you may end up with a flat response (on axis) but the sound will not be that great especially off axis.

Again, not disagreeing with what you have said, just trying to clarify a few things for others.

Regards,

Dennis

[pjpoes]

Thanks Dennis, that is well said. I agree that it appears to be outside what omni said he was looking for, but sometimes it can be posts like these that become misconstrued later on. Someone will say what you have said, someone else will quote it or search it, and suddenly those of us who know better are being told that phase is unimportant, and suddenly this whole area of crossover science is being neglected all from a misunderstanding.

On a tangent here, I'm not sure where it came from, but you here that transient perfect crossovers are easily achieved through simple 1st order crossovers. Having tried that as part of a series of experiments I was doing, I can say without a doubt that they are quasi, not true. At first I thought it must have been my test rig, but I sent my results to some well known designers, and they told me these were in fact consistent with their own experimental results, as well as those of most of the work in this area. I'm starting to think that the only way to get a true transient perfect crossover is going to be with the use of subtractive delay. Even then, I bet it will only be transient perfect at the crossover itself. Then you have the drivers which can be ever so slightly off in placement, problems or inconsistencies with the drivers, etc. which all could cause the system not to pass transients perfectly.