Tweeter XO Freq. and Xlim SPL

[robert]

It occurred to me that figuring out excursion-limited SPL for a tweeter might be an interesting exercise.

I used the following figures for a 1 1/8" dome tweeter. Sd = 7cm2, Xmax = 0.2mm. These might not be exact, but I got some surprising results for the 2Khz area where a lot of these large-caliber dome tweeters get crossed over. All of the below are without Baffle Step.

97dB at 2Khz
96dB at 1.8Khz
92dB at 1.5Khz

It seems to me that "pulling" a tweeter down to operate at these frequencies is quite the compromise, since most small quality woofers will Xlim much higher than this above 100Hz. And then distortion, since at high volume a tweet with a low crossover will be operating near it's mechanical limits. Any thoughts/comments would be appreciated, especially if I am off-base here.

Cheers/Robert

[billfitzmaurice]

> It occurred to me that figuring out
> excursion-limited SPL for a tweeter might be
> an interesting exercise.

I'd be curious as to how you arrived at the results. For instance, did you calculate excursion using voltage or power?

[paul_n_oneal]

> I'd be curious as to how you arrived at the
> results. For instance, did you calculate
> excursion using voltage or power?

yes, more infformation please.

[Pete Schumacher ®]

> I'd be curious as to how you arrived at the
> results. For instance, did you calculate
> excursion using voltage or power?

Voltage and power have nothing to do with it Bill. As in woofers, calculating SPL is purely a function of displacement, so if you have Sd and Xmax, you know SPL capability at a specific frequency.

I do have something to add here though. Most tweeters I know that are used to below 2KHz are crossed LR4, which means they are down 6dB already at the XO frequency. Also, most of them have Xmax values more than double the .2mm quoted, more like .5mm.

So, keep in mind that you have double the Xmax available as well as being down 6dB at the XO point.

There's nothing inherently wrong with that lower XO point, as long as you are aware that you're SPL limited.

[paul_n_oneal]

> Voltage and power have nothing to do with it
> Bill. As in woofers, calculating SPL is
> purely a function of displacement, so if you
> have Sd and Xmax, you know SPL capability at
> a specific frequency.

> I do have something to add here though. Most
> tweeters I know that are used to below 2KHz
> are crossed LR4, which means they are down
> 6dB already at the XO frequency. Also, most
> of them have Xmax values more than double
> the .2mm quoted, more like .5mm.

> So, keep in mind that you have double the
> Xmax available as well as being down 6dB at
> the XO point.

> There's nothing inherently wrong with that
> lower XO point, as long as you are aware
> that you're SPL limited.

OK, Pete, what's the max SPL on the rear mount neo crossed LR4 at ~10,000hz?

Paul

[Pete Schumacher ®]

> OK, Pete, what's the max SPL on the rear
> mount neo crossed LR4 at ~10,000hz?

> Paul

Really, really high, limited only by max power input. Xmax is not a factor here as the VC will likely melt before you can get much excursion.

[paul_n_oneal]

> Really, really high, limited only by max
> power input. Xmax is not a factor here as
> the VC will likely melt before you can get
> much excursion.

As long as it can do 110db I'm a happy camper . Of course, at $5 a pop, I may just try to fry one for the fun of it.

Paul

[billfitzmaurice]

> Voltage and power have nothing to do with it
> Bill.
They can have everything to do with it. If you use a program that calculates excursion as a product of voltage applied you'll get a far different result than if you use one that uses power, as power and voltage aren't linear with respect to each other. The difference is especially critical in the area near fs where higher impedance can drastically reduce current, and therefore power, and likewise excursion. OTOH, using a program like WinISD to calculate excursion as a function of power wouldn't be accurate, as WinISD assumes a full-bandwidth signal that would be grossly higher than the actual power content above the crossover. For these reasons knowing the methodology used is critical to understanding the results obtained.

[robert]

I used the linkwitz labs spreadsheet, just Sd and Xmax, no others. With the 6dB down from the crossover, additional padding likely, etc., it makes sense to me.
Since most mfr's don't list these for their tweets except for Morel, the Xmax figure was a SWAG, and the Sd was from Morel. Another "learn-a-lot" post. 0.5 is better than 0.2
Thanks all/Robert

[Pete Schumacher ®]

> They can have everything to do with it. If
> you use a program that calculates excursion
> as a product of voltage applied you'll get a
> far different result than if you use one
> that uses power, as power and voltage aren't
> linear with respect to each other. The
> difference is especially critical in the
> area near fs where higher impedance can
> drastically reduce current, and therefore
> power, and likewise excursion. OTOH, using a
> program like WinISD to calculate excursion
> as a function of power wouldn't be accurate,
> as WinISD assumes a full-bandwidth signal
> that would be grossly higher than the actual
> power content above the crossover. For these
> reasons knowing the methodology used is
> critical to understanding the results
> obtained.

That's only important for determining sensitivity Bill. Absolute peak SPL is purely limited by displacement, or Sd and Xmax. You don't need to know power to calculate it.

You do need power or voltage to guage sensitivity and where the input power limit will be reached.

[jonpike]

> I used the linkwitz labs spreadsheet, just
> Sd and Xmax, no others. With the 6dB down
> from the crossover, additional padding
> likely, etc., it makes sense to me.
> Since most mfr's don't list these for their
> tweets except for Morel, the Xmax figure was
> a SWAG, and the Sd was from Morel. Another
> "learn-a-lot" post. 0.5 is better
> than 0.2
> Thanks all/Robert

There are more Xmax numbers out there to be found.. Seas posts them for most, Peerless/Vifa now seems to be making it a regular thing.. Digging up the manufacturer's specs rather than the "short form" on distributor's sites often turns them up.

A quick check showed most Seas tweeters were .5mm, excepting the Milleneum, with a mighty 1.0mm Xmax. (guess you're getting something for your $170!)

[curt_c]

While max SPL can be determined solely by Sd and Xmax., the impedance of the driver will have a significant effect on the resultant plot of excursion vs. frequency, and the plots will be different if a constant power or a constant voltage is applied.

The Linkwitz spreadsheet, WinISD, and most ‘box’ programs do not consider impedance, or assume it’s a constant value in their calculations. This is too simplistic for tweeter modeling as the fs will almost always be in the stop band of the system transfer function.

In most drivers, the higher magnitude of the impedance bump will cause a significant increase in excursion. Ferrofluid goes a long way to mitigate the impedance variation, but a shunt conjugate can be used in either case to linearize the impedance around fs. This, IME will allow crossing a tweeter far lower than would be otherwise possible, irrespective of other concerns such as harmonic distortion, etc.

C

[Pete Schumacher ®]

> While max SPL can be determined solely by Sd
> and Xmax., the impedance of the driver will
> have a significant effect on the resultant
> plot of excursion vs. frequency, and the
> plots will be different if a constant power
> or a constant voltage is applied.

There's no arguement here. But we were talking about max SPL delivered by a specific area given a specific linear displacement. Impedance does not come into the equation at all when determining max SPL based on linear displacement.

> The Linkwitz spreadsheet, WinISD, and most
> ‘box’ programs do not consider impedance, or
> assume it’s a constant value in their
> calculations. This is too simplistic for
> tweeter modeling as the fs will almost
> always be in the stop band of the system
> transfer function.

> In most drivers, the higher magnitude of the
> impedance bump will cause a significant
> increase in excursion. Ferrofluid goes a
> long way to mitigate the impedance
> variation, but a shunt conjugate can be used
> in either case to linearize the impedance
> around fs. This, IME will allow crossing a
> tweeter far lower than would be otherwise
> possible, irrespective of other concerns
> such as harmonic distortion, etc.

These types of transducers are voltage driven and have their responses plotted accordingly. The increase in excursion at resonance will show up in the frequency response, even if only a mild bump on the downward slope. The conjugate filter will certainly flatten the impedance, which allows reasonable values for the HP components. Otherwise, it has no effect on the performance of the tweeter. If the tweeter with Fs of 2KHz is a poor performer at 2KHz from a distortion standpoint, then no amount of impedance compensation will improve that.

The target acoustic slope is all that matters in the end. You just need to reach that within the range where the tweeter has good performance.

[Pete Schumacher ®]

> There are more Xmax numbers out there to be
> found.. Seas posts them for most,
> Peerless/Vifa now seems to be making it a
> regular thing.. Digging up the
> manufacturer's specs rather than the
> "short form" on distributor's
> sites often turns them up.

> A quick check showed most Seas tweeters were
> .5mm, excepting the Milleneum, with a mighty
> 1.0mm Xmax. (guess you're getting something
> for your $170!)

I do believe the Morel MDT33 also sports that mighty 1mm. There's no doubt which of the two is suited to sub 2000Hz performance better. Just check Mark K's site for a test of the MDT33.

[Jeff B.]

> There's no arguement here. But we were
> talking about max SPL delivered by a
> specific area given a specific linear
> displacement. Impedance does not come into
> the equation at all when determining max SPL
> based on linear displacement.

> These types of transducers are voltage
> driven and have their responses plotted
> accordingly. The increase in excursion at
> resonance will show up in the frequency
> response, even if only a mild bump on the
> downward slope. The conjugate filter will
> certainly flatten the impedance, which
> allows reasonable values for the HP
> components. Otherwise, it has no effect on
> the performance of the tweeter. If the
> tweeter with Fs of 2KHz is a poor performer
> at 2KHz from a distortion standpoint, then
> no amount of impedance compensation will
> improve that.

> The target acoustic slope is all that
> matters in the end. You just need to reach
> that within the range where the tweeter has
> good performance.

I am lining up more with Pete here, but if you guys want to get picky you need to include the crossover slope (and the impedance curve's effect on it) in the excursion plot.

I did this when I modeled my PHL mids. I imported the frd and zma files into the PCD, set-up the crossover, and exported the crossover's transfer function. Then I imported the transfer function into my Box and Circuit Designer (Unibox can be used too) and modeled the midrange in it and looked at how much power I could give it before it reached it's xmax with the crossover on it.

I don't see why we can't do the same for a tweeter. You just have to model it as a small sealed system. You may have to guess at some parameters, but you can get it pretty close. In my Box and Woofer Designer you can change the frequency range too. (Although I am not sure you can import the transfer function into the one that is released. I'll need to check that out).

Otherwise, the way Pete is describing it is the way it is typically done to determine tweeter max output.

[billfitzmaurice]

> you need to
> include the crossover slope (and the
> impedance curve's effect on it) in the
> excursion plot.

Of course. When calculating the displacement limited output of a tweeter, or any driver for that matter, a factor that must be considered is excursion below the passband, which is directly affected by the HP filter slope and driver impedance at and below the corner frequency. Sd, Pe and Xmax are only three parts of an equation that has at least five variables.

[robert]

> Of course. When calculating the displacement
> limited output of a tweeter, or any driver
> for that matter, a factor that must be
> considered is excursion below the passband,
> which is directly affected by the HP filter
> slope and driver impedance at and below the
> corner frequency. Sd, Pe and Xmax are only
> three parts of an equation that has at least
> five variables.

Once again, lotsa knowledge gained. Thanks All/Robert