First speaker to reproduce a 60hz sq wave

[fpitas]

this is what I like about ASR, one just needs to start a conversation just passing by and be gone

It's an interesting topic. But ultimately a scientific curiosity. As someone said earlier, you really just need to avoid the group delay variations going over audible thresholds to make a good speaker.

 

[NTK]

I searched the web and unfortunately found nothing that supports my point of view. However, with my approach I can directly recalculate Klippel's view of the phase shift as the temperature changes.

I found a web seminar by Kippel where he explains the influence of temperature on the phase frequency response in two minutes.
Please @Hayabusa @Zapper take a look at the video and explain to me exactly what he means (link starts at the topic):

KLIPPEL LIVE Series 1 - Part 2: Standard Acoustical Test Performed in Normal Rooms

In the video, he gives the example of "beam stearing" in speaker arrays. We can probably agree that if driver A is measured at 20°C and driver B on top at 22°C, then there will be a phase error due to the change in the speed of sound which will prevent correct beam stearing especially at high frequencies.

The sound from driver B arrives a little earlier (when assuming same Mic distance as A) at the mic and the higher the frequency, the greater the phase error in degrees - I showed how this can be calculated in post#98 and post#89.
Or do we disagree at this example?

I believe Dr Klippel was referring to the inherent nature of the temperature variations in the air when measuring using long distance measurements (the example he gave was 5 m for large speakers, and the rooms we have are usually not big enough and have to measure outdoors). In outdoors the ground temperature is usually different from the air temperature (e.g. due to from cooling of the night before, heating by the sun, etc.) and thus the air temperature is not very uniform. And then there are wind and breezes.

 

[audiofooled]

A real square wave is a mathematical abstraction. To actually make one you would need infinite bandwidth.

I would hate the Universe where 1+1=2 would be true for only one and single location. Move away a bit and all of a sudden it just may or may not be true, but you would anyway have to prove it.

Here's a 46Hz big one, in all of it's ugliness... Part true for just a single location in a room. Move a bit and you can make it more flat at the top, but round on the sides, more steep at the sides, but dipped at the top. I see no point. Luckily our hearing doesn't work that way. Phase compensation IMO is beneficial for low frequencies and that's about it.

 

[ctrl]

I believe Dr Klippel was referring to the inherent nature of the temperature variations in the air when measuring using long distance measurements (the example he gave was 5 m for large speakers, and the rooms we have are usually not big enough and have to measure outdoors). In outdoors the ground temperature is usually different from the air temperature (e.g. due to from cooling of the night before, heating by the sun, etc.) and thus the air temperature is not very uniform. And then there are wind and breezes.

The examples in the Klippel docs actually always refer to anechoic chambers.

Source: Training_8_Measurement_of_Loudspeaker_Directivity_en.pdf

Source: C8 Near Field Scanner System.pdf

 

[MAB]

I have no problems with squares and no problems with waves, just with square waves. Waves have no corners, squares do.
On a graph, a square wave doesn't look like wave to me, it looks like broken coordinates - anything BUT the wave. Shouldn't appear in nature. And it doesn't.

Actually there are conditions where odd amplitude harmonics are all aligned. Certain places are known for this due to currents, wind, and boundary conditions like profile of the shore.

 

[fpitas]

Actually there are conditions where odd amplitude harmonics are all aligned. Certain places are known for this due to currents, wind, and boundary conditions like profile of the shore.

 

[Hayabusa]

I searched the web and unfortunately found nothing that supports my point of view. However, with my approach I can directly recalculate Klippel's view of the phase shift as the temperature changes.

I found a web seminar by Kippel where he explains the influence of temperature on the phase frequency response in two minutes.
Please @Hayabusa @Zapper take a look at the video and explain to me exactly what he means (link starts at the topic):

KLIPPEL LIVE Series 1 - Part 2: Standard Acoustical Test Performed in Normal Rooms

In the video, he gives the example of "beam stearing" in speaker arrays. We can probably agree that if driver A is measured at 20°C and driver B on top at 22°C, then there will be a phase error due to the change in the speed of sound which will prevent correct beam stearing especially at high frequencies.

The sound from driver B arrives a little earlier (when assuming same Mic distance as A) at the mic and the higher the frequency, the greater the phase error in degrees - I showed how this can be calculated in post#98 and post#89.
Or do we disagree at this example?

he describes temperature variation of the air that results in different apparent distances.
These differences are the same in length as well as time for each different frequency.
As a constant time/distance difference gives a phase different that IS frequency dependent (phase = time*f),
you indeed get that "the higher the frequency, the greater the phase error in degrees".
I agree on that.

 

[René - Acculution.com]

I may well be wrong in my view, I am not an expert in such physical details.
The difference to your view is difficult to describe and very probably formulated "unphysical" or simply wrongly by me. Will try anyway

You are only looking at the time offset of the "wave front" at the temperature-related different sound velocities c1 (343 m/s at 20°C) and c2 (349 m/s at 30°C).

What also changes is the wavelength at a given frequency:
⁁1 = c1/f
⁁2 = c2/f
So something else is happening.

With the change in speed of sound, the time it takes for the sound wave to reach the microphone changes. This means that the "phasor for each frequency" is at a different "position" when the measuring microphone is reached after 5m at a different air temperature. This then causes the frequency dependent phase shift.

The phase shift ΔΦ (caused by change in propagation time) is then:

ΔΦ = 360° * Δt * f

Δt = d/c1 - d/c2 (d = distance)


Here again how Klippel describes the phenomenon:

View attachment 338310
Source: Klippel GmbH

If an expert like @René - Acculution.com is reading this, it would be nice if you could set the record straight about how wrong I am (before I write any more BS? and have to correct countless posts ).

I think you are just mixing up a couple of things here ;-)

The Fourier transform of a time delay T0 is exp(-i*w*T0), where w=2 pi f. If this comes from a distance R0 traveled, it can be written instead as exp(-i*w*R0/c), and so the associated phase is w*R0/c.

So you can have at least two issues:
a) Dispersion; c=c(f). So, if you have a source or system that can create square waves with a c=c0 that does not change with frequency, or frequency-dependent c=c(f) but known, you can have an issue if this speed of sound changes with temperature or humidity in way that you did not take into account.

b) Incorrect correction. What Klippel mentions is that the phase w*R0/c is very important at higher frequencies, because of the w-term (because the wavelengths are small). So if you measure a particular phase with c=c(f,T1,humidity1), you can measure a very different phase for another sound speed c=c2(f,T2,humidity2). This issue is not that one is more correct than the other, and so it is a bit of misnomer to say that there is a phase error. This issue is that typically you want to get rid of this phase all together, because it is trivial compared to the transducer phase, which is the interesting one, and that would typically be pressure phasor phase related back to voltage phasor phase. So you know the distance R0, but then you guess/assume a single value for c, and now you might see the remaining phase being more erratic than expected, due to this incorrect phase correction as c might be off compared to the actual physical setup.

 

[Hayabusa]

I think you are just mixing up a couple of things here ;-)

The Fourier transform of a time delay T0 is exp(-i*w*T0), where w=2 pi f. If this comes from a distance R0 traveled, it can be written instead as exp(-i*w*R0/c), and so the associated phase is w*R0/c.

So you can have at least two issues:
a) Dispersion; c=c(f). So, if you have a source or system that can create square waves with a c=c0 that does not change with frequency, or frequency-dependent c=c(f) but known, you can have an issue if this speed of sound changes with temperature or humidity in way that you did not take into account.

b) Incorrect correction. What Klippel mentions is that the phase w*R0/c is very important at higher frequencies, because of the w-term (because the wavelengths are small). So if you measure a particular phase with c=c(f,T1,humidity1), you can measure a very different phase for another sound speed c=c2(f,T2,humidity2). This issue is not that one is more correct than the other, and so it is a bit of misnomer to say that there is a phase error. This issue is that typically you want to get rid of this phase all together, because it is trivial compared to the transducer phase, which is the interesting one, and that would typically be pressure phasor phase related back to voltage phasor phase. So you know the distance R0, but then you guess/assume a single value for c, and now you might see the remaining phase being more erratic than expected, due to this incorrect phase correction as c might be off compared to the actual physical setup.

Not sure what your conclusion now is

 

[audiofooled]

The good old "Which way is up?" creating actual issues rather than the air... Happy New Year everyone!

 

[ctrl]

What Klippel mentions is that the phase w*R0/c is very important at higher frequencies, because of the w-term (because the wavelengths are small). So if you measure a particular phase with c=c(f,T1,humidity1), you can measure a very different phase for another sound speed c=c2(f,T2,humidity2). This issue is not that one is more correct than the other, and so it is a bit of misnomer to say that there is a phase error.

Not sure what your conclusion now is

I agree on that

@René - Acculution.com The question is (and keep in mind I'm an amateur ):

We have an ideal loudspeaker that is calibrated to a distance of 5m@20°C measuring with linear (perfect flat) FR and linear phase frequency response. So it reproduces a square wave, restricted by human frequency range, perfectly.

What happens with the reproduction of the square wave when the temperature increases e.g. to 30°C. Will the mic in 5m distance still record a perfect square wave or will the frequency depended phase shift "distort" the square wave?

 

[KSTR]

What happens with the reproduction of the square wave when the temperature increases e.g. to 30°C. Will the mic in 5m distance still record a perfect square wave or will the frequency depended phase shift "distort" the square wave?

There is only some very minor frequency dependence of the speed of sound.

Some tables from "Handbook of the Speed of Sound in Real Gases, VOLUME III Speed of Sound in Air" by Allan J. Zuckerwar, Academic Press 2002:

Increasing temperature by 10K does not make any significant relative change.

------:------

As for the thread topic, a 60Hz square is just opposite polarity step responses every 8.33ms, overlaid and added up.
For a minimum phases system it cannot get any better than the low frequency roll-off allows as that is the source of the skewed phase. A 4th order roll-off at 20Hz will affect the shape quite significantly and will also move "center of gravity" away from the center for a short sequence of just a few cycles. But that can be "fixed" by making the roll-off linear-phase with a phase pre-correction (DSP/FIR). Not free lunch though, and the cost is that the sequence's response now is spread out a bit farther in time, now both before (pre-ringing) and after the original sequence. With real music signals like kick-drums this sometimes creates audible artifacts because the "bass transient starts too early". The artificial leading portion of the wavefront -- an steep rise until the first roof of the square is reached -- typically sounds like a pink'ish-brown'ish noise burst quickly rising in level.

 

[mhardy6647]

The good old "Which way is up?" creating actual issues rather than the air... Happy New Year everyone!

 

[audiofooled]

View attachment 338763

Actually this is what I meant:

Which way is up? (Which way does a loudspeaker driver move?)

In a recent video by PS Audio/Paul, he mentions the sentiment that many probably share with him, that a loudspeaker driver works by compressing air as it moves outwards. This illustrates perfectly that you can work with loudspeakers for decades without ever understanding how a loudspeaker...

www.audiosciencereview.com

But if you must, I guess going West isn't always what it seems

 

[mhardy6647]

My father was wont to say:
How high is up?
Half the distance down, times two.

 

[audiofooled]

My father was wont to say:
How high is up?
Half the distance down, times two.

That's brilliant!

 

[fpitas]

My father was wont to say:
How high is up?
Half the distance down, times two.

It puts a vague, rhetorical question on a firm mathematical basis!

 

[mhardy6647]

It puts a vague, rhetorical question on a firm mathematical basis!

I am rolling on the floor now.
I wonder how he would've reacted if I said ever that to him in response?

 

[René - Acculution.com]

I agree on that

@René - Acculution.com The question is (and keep in mind I'm an amateur ):

We have an ideal loudspeaker that is calibrated to a distance of 5m@20°C measuring with linear (perfect flat) FR and linear phase frequency response. So it reproduces a square wave, restricted by human frequency range, perfectly.

What happens with the reproduction of the square wave when the temperature increases e.g. to 30°C. Will the mic in 5m distance still record a perfect square wave or will the frequency depended phase shift "distort" the square wave?

What is important is if the speed of sound changes the same at all frequencies or not with the change in temperature. Hence why I explained dispersion. If the speed of sound changes at some frequencies and not at others, the square is distorted. But if it changes from 340 to 344, the time of flight just changes accordingly, and so the square wave is seen, albeit with a slightly different overall phase related to the changed time delay, which is the second part explained.

 

[ctrl]

There is only some very minor frequency dependence of the speed of sound.

Thanks for the information. But that was not the core of the discussion and not what Prof. Klippel meant by the temperature-induced phase shift.

For a minimum phases system it cannot get any better than the low frequency roll-off allows as that is the source of the skewed phase. ... "fixed" by making the roll-off linear-phase with a phase pre-correction (DSP/FIR). Not free lunch though, and the cost is that the sequence's response now is spread out a bit farther in time, now both before (pre-ringing) and after the original sequence.

The initial question was whether the reproduction of a 60Hz square wave by a single driver is a decisive quality feature for a loudspeaker, as indicated in the initial post.

But this feature "only" means that the driver should have the most linear FR and phase frequency response possible (especially in the range above 60Hz) in order to be able to reproduce a square wave "distortion-free" - which should be achievable with any good active studio monitor.

Then I said the fateful sentence in post#73 that triggered the whole discussion between @Hayabusa and me:
If the phase frequency response was linearized in spring at 20°C (293K), then in summer at 30°C (303K) temperature dependence of the speed of sound leads to considerable phase shifts, so that the square wave will show significant distortion. At a distance of 5m, the phase shift would be around 11°@100Hz and around 110°@1000 Hz.

Which means when we have an ideal loudspeaker that is calibrated to a distance of 5m@20°C reproducing a square wave recorded with an mic.
What happens with the reproduction of the square wave when the temperature increases e.g. to 30°C. Will the mic in 5m distance still record a perfect square wave or will the frequency depended phase shift "distort" the square wave?

Update: This statement is wrong, phase shift occur but the square wave will not be distorted - the replies from others pointing out the error are correct.
Now comes the crucial point, if the measurement conditions are not changed (the mic and the measurement software are left completely unchanged), a phase shift occurs due to the time offset (because of the difference in the speed of sound at 20°C and 30°C) which "distorts" the plot of the square wave.
There may have been a misunderstanding between @Hayabusa and myself at this point.

Normally, the time of flight from the sound source at the driver and the microphone is cut out after the measurement. However, this is not what arrives at the ear/mic, as there is an additional phase shift due to the temperature change.

Rene pointed out this difference in his answer:

So if you measure a particular phase with c=c(f,T1,humidity1), you can measure a very different phase for another sound speed c=c2(f,T2,humidity2). This issue is not that one is more correct than the other, and so it is a bit of misnomer to say that there is a phase error. This issue is that typically you want to get rid of this phase all together, because it is trivial compared to the transducer phase, which is the interesting one, and that would typically be pressure phasor phase related back to voltage phasor phase.

Why does Klippel calls it a phase error:

A phase error can creep in during speaker design due to temperature changes.

As a rule, we do not know the exact point of sound generation in the driver (as a rule of thumb, we often take the position of the voice coil in the direction of the measurement axis). At high crossover frequencies, this can lead to large frequency response errors, as an incorrectly estimated offset of just a centimeter or so between the drivers leads to significant phase deviations.

To avoid this, a fixed time of flight (reference point) is assumed for all measurements (then the correct relative sound generation off-set between the drivers is in the measurements) - for example, the time of flight of the on-axis measurement of the tweeter (and tweeter voice coil as rotation axis for all measurements).
If measurements of the tweeter are carried out and then the measurements of the midrange driver, phase deviations between the tweeter and midrange driver may occur if the temperature of the measuring room increases, as the time of flight is a fixed value.
From this perspective, these are then temperature-related phase errors.