Ok so often we can get a sense of the maximum spl capability of a driver simply by looking at it's displacement ability. Programs like WinISD use these formula's to calculate the max spl abilities of the driver, as well as power handling, etc. Is there any reason that a tweeter is any different, can I calculate the maximum spl of a tweeter in a similar fashion?
In other words, if I want to know, very roughly, how loud a tweeter will be able to play with a given crossover, can I simply treat it like a woofer, enter it into something like winisd (or any number of other programs with similar abilities), and add in the filter to find the max spl it will do with said filter. Again, I know this won't be exact, it's rough, but at least get me in the ball park?
I tried this, and it doesn't blow up, but I have no idea if it's accurate. I took known parameters for my Focal TC120 tweeters and for some Scan Speak models, and placed the tweeter into a very small enclosure, then added a filter, say a B-W 2nd order at 1800hz, apply a certain amount of power, look at the xmax, and see if I'm still in the linear range. When I did this, I came up with 100db's for the Scan-Speak with a 2nd order butter-worth filter at 1800hz and 50watts, putting it right around the xmax limit. I then put in a 6th order filter at 3600hz and found that I couldn't hit the xmax before hitting the power handling limits, but could push it to the xmax at around 2000 watts.
If I wanted to know the max spl, and had it crossed over in such a way that the tweeter couldn't hit it's mechanical limits, how much above the iec rating can you typically expect such a driver to handle for short transient bursts? I've read some articles indicating 2-3 times the iec is common.
In other words, if I want to know, very roughly, how loud a tweeter will be able to play with a given crossover, can I simply treat it like a woofer, enter it into something like winisd (or any number of other programs with similar abilities), and add in the filter to find the max spl it will do with said filter. Again, I know this won't be exact, it's rough, but at least get me in the ball park?
I tried this, and it doesn't blow up, but I have no idea if it's accurate. I took known parameters for my Focal TC120 tweeters and for some Scan Speak models, and placed the tweeter into a very small enclosure, then added a filter, say a B-W 2nd order at 1800hz, apply a certain amount of power, look at the xmax, and see if I'm still in the linear range. When I did this, I came up with 100db's for the Scan-Speak with a 2nd order butter-worth filter at 1800hz and 50watts, putting it right around the xmax limit. I then put in a 6th order filter at 3600hz and found that I couldn't hit the xmax before hitting the power handling limits, but could push it to the xmax at around 2000 watts.
If I wanted to know the max spl, and had it crossed over in such a way that the tweeter couldn't hit it's mechanical limits, how much above the iec rating can you typically expect such a driver to handle for short transient bursts? I've read some articles indicating 2-3 times the iec is common.
designing
figure about about 90 to 92 db range
look at the frequency response find the area the has a + - 3db or less, less is better but more expensive and difficult
double the Fs of the woofer and look at impedance chart 8 ohm speakers should be used in the 12 to 6 ohm range for amplifier stablilty
plan the crossover hopefully at full octave or 1/2 octaves closest to the point where the spl dips or rises going into second resonance.
you will also have to figure out the impedance of two speakers in parallel if you using a mtm or wtw arrangement and figure in a load resistor into the crossover.
the tweeter may be 4 ohm which will draw more amps so figure on a load resistor will be needed there too.
use the same rules for the woofer
do a spl analysis with speakerworks and other software that is out there and remember to look at the differnt crossover calculators 2nd orders will need to have the tweeter in the real speakers polarity reverse -180 Degree phase shift
use a different resistors in series and parallel to the tweeter to bring the spl down to around the same spl as the woofers just remember that this is just a starting point for you will have to fine tune it later.
figure about about 90 to 92 db range
look at the frequency response find the area the has a + - 3db or less, less is better but more expensive and difficult
double the Fs of the woofer and look at impedance chart 8 ohm speakers should be used in the 12 to 6 ohm range for amplifier stablilty
plan the crossover hopefully at full octave or 1/2 octaves closest to the point where the spl dips or rises going into second resonance.
you will also have to figure out the impedance of two speakers in parallel if you using a mtm or wtw arrangement and figure in a load resistor into the crossover.
the tweeter may be 4 ohm which will draw more amps so figure on a load resistor will be needed there too.
use the same rules for the woofer
do a spl analysis with speakerworks and other software that is out there and remember to look at the differnt crossover calculators 2nd orders will need to have the tweeter in the real speakers polarity reverse -180 Degree phase shift
use a different resistors in series and parallel to the tweeter to bring the spl down to around the same spl as the woofers just remember that this is just a starting point for you will have to fine tune it later.
dont go past a 4th order
no the program will not be able to tell you that because those measurement are done with an actual speaker if the manufacturer says the the max spl it is at 95db or - 3db of the highest peak for the actual sound will shift up -+ 3 db with burst for you have to remember that depending on the sound and the reaction of the speaker is fighting it self and the heat and if you exceed the current capablities of the wire of the voice coil you just burnt up that tweeter that is why they usually will put a load resistor in a crossover to help keep the current down and also to match the spl of the woofer and the spl is done at 1 watt and 1 meter. The spl db is used as a measure of effeciency
The spl will vary by distance and location in relation to the speaker.
High db does not always mean that is it the best way to go
1. Are you going to play it that loud all the time
2. The room acoustics will alter the sound
3. The more nth degree crossovers cause more problems
4 uses few components as possible
5 match speakers that compliment in frequency summimg
6 balance the load to the amplifier
7 the best polar dispersion capabilities and the more expensive area
8 the mininum wave defraction as possible
9 minimize cabinet resonance
there are great speakers that can be made but you dont have to spend an arm and a leg to get it
no the program will not be able to tell you that because those measurement are done with an actual speaker if the manufacturer says the the max spl it is at 95db or - 3db of the highest peak for the actual sound will shift up -+ 3 db with burst for you have to remember that depending on the sound and the reaction of the speaker is fighting it self and the heat and if you exceed the current capablities of the wire of the voice coil you just burnt up that tweeter that is why they usually will put a load resistor in a crossover to help keep the current down and also to match the spl of the woofer and the spl is done at 1 watt and 1 meter. The spl db is used as a measure of effeciency
The spl will vary by distance and location in relation to the speaker.
High db does not always mean that is it the best way to go
1. Are you going to play it that loud all the time
2. The room acoustics will alter the sound
3. The more nth degree crossovers cause more problems
4 uses few components as possible
5 match speakers that compliment in frequency summimg
6 balance the load to the amplifier
7 the best polar dispersion capabilities and the more expensive area
8 the mininum wave defraction as possible
9 minimize cabinet resonance
there are great speakers that can be made but you dont have to spend an arm and a leg to get it
You can model the LF rolloff and mechanical excursion characteristics of a tweeter with measured T/S data, but usually Vas is not specified, you can just assume something low (for guess purposes, note that a typical Vas in liters is just the square of the diameter of the speaker in inches) and and put it in a box at least 10x Vas for modeling purposes.
Don Keele (in Audio equipment reviews) used to do burst tests on tweeters and typically the tweeters would handle 1.5kW bursts as long as they weren't displacement limited. Look to a library for some old Audio speaker reviews for context.
Don Keele (in Audio equipment reviews) used to do burst tests on tweeters and typically the tweeters would handle 1.5kW bursts as long as they weren't displacement limited. Look to a library for some old Audio speaker reviews for context.
Agreeing with Ron, yes, you can model things that way if you know all the relevant information. If you are only interested in the excursion limit this works fine. If you are thinking about trade offs between power input and mechanical excursion, then the other things to keep in mind are power compression and driver non-linearities. The coil will heat up as you apply power. This increases the coil resistance causing less current to flow for a given input voltage. This is not modeled in typical box programs, so you will probably get a few dB less output in a steady state scenario than the box program would predict if you are looking at the thermal limit. You will get some compression even on bursts, but not as much. Also, driver suspensions typically stiffen with excursion requiring more force (current) than expected to push them further. This will also change the response of the driver around resonance. Also, BL typically drops as the coil moves away from center, so more current is required to produce the same force as the coil moves away from rest. Once again, if you know enough details, you can model everything, but most of the time it's easier to do some tests with the actual driver. Look at things like where the output stops increasing linearly with the input signal, where distortion starts to get too high (subjective, but maybe 10% 2nd + 3rd harmonics would be a place to start), and where you start to get close to the mechanical limit of the driver (typically the other two things would happen first).
Some lesser tweeters don't handle high power burts well but then again, many do.
Dynaudio use to tests their speakers with 1kW bursts.
A Swedish designer use 3kW bursts for dome tweeters and 10kW for bass/mids. One design is a four 8" 2.5 way system that can handle 40000W peak in the midrange.
If the bursts are short enough in time obviously the amount of energy is not very high and the VC won't burn. What you really get a grip on is the mechanical strength of the driver and short transient capability. Some are ripped apart while other well made designes can withstand extreme G-forces.
Thermal behaviour in dome tweeters are improved with ferrofluid in the gap and with non fabric domes as metal for ex.
/Peter
Dynaudio use to tests their speakers with 1kW bursts.
A Swedish designer use 3kW bursts for dome tweeters and 10kW for bass/mids. One design is a four 8" 2.5 way system that can handle 40000W peak in the midrange.
If the bursts are short enough in time obviously the amount of energy is not very high and the VC won't burn. What you really get a grip on is the mechanical strength of the driver and short transient capability. Some are ripped apart while other well made designes can withstand extreme G-forces.
Thermal behaviour in dome tweeters are improved with ferrofluid in the gap and with non fabric domes as metal for ex.
/Peter
as I said, I knew it would be rough, I knew it wouldn't take into account thermal compression, non-linearities other than xmax non-linearities, etc. My goal was to just be able to model very roughly about what the spl capability of the tweeter would be with a given amount of power with different crossovers.
Some of this was also to "check" my view that few speakers are capable of linear low compression output in excess of 100db's over a wide range, and thus far, this does seem to be the case.
For instance, if one was to be designing a speaker inwhich the goal was to achieve realistic sound pressure levels, and lets say that means being capable of at least 105db's with little or no compression or mechanical distortions, what would it take to do that? I know my speakers from Dr. Geddes can do that, but what about other more conventional designs. Can a 1" dome tweeter with .4mm of xmax peak to peak produce 100db's at say 1500hz (according to the formula Linkwitz provides, the answer is not even close). Can a 6" midbass produce 50hz at 100db's, and again, I'm finding that while not impossible, is uncommon. This seems to support my contention that a lot of conventional speakers on the market can't achieve realistic levels with suitably low levels of compressions and distortion. The midbass issue was more easy to model and test, but tweeters are harder. Listening to 10khz tones or even broad band tones from 1khz to 20khz is very obnoxious, especially that loud.
In thinking about the amount of reserve power one wants to meet these transients, it's also good to know just how much is safe and reasonable before you chance thermal problems. In other words, while a tweeter will likely blow rather quickly from hitting it's x-damage point, if the crossover is suitably high and of a steep enough order, it's possible that you will reach thermal limits before mechanical limits, and this brings about the issue of short term transient power handling. I knew that Dynaudio and Morel often stated short term peaks in excess of 1kw, but I wasn't sure if this was true of others.
Some of this was also to "check" my view that few speakers are capable of linear low compression output in excess of 100db's over a wide range, and thus far, this does seem to be the case.
For instance, if one was to be designing a speaker inwhich the goal was to achieve realistic sound pressure levels, and lets say that means being capable of at least 105db's with little or no compression or mechanical distortions, what would it take to do that? I know my speakers from Dr. Geddes can do that, but what about other more conventional designs. Can a 1" dome tweeter with .4mm of xmax peak to peak produce 100db's at say 1500hz (according to the formula Linkwitz provides, the answer is not even close). Can a 6" midbass produce 50hz at 100db's, and again, I'm finding that while not impossible, is uncommon. This seems to support my contention that a lot of conventional speakers on the market can't achieve realistic levels with suitably low levels of compressions and distortion. The midbass issue was more easy to model and test, but tweeters are harder. Listening to 10khz tones or even broad band tones from 1khz to 20khz is very obnoxious, especially that loud.
In thinking about the amount of reserve power one wants to meet these transients, it's also good to know just how much is safe and reasonable before you chance thermal problems. In other words, while a tweeter will likely blow rather quickly from hitting it's x-damage point, if the crossover is suitably high and of a steep enough order, it's possible that you will reach thermal limits before mechanical limits, and this brings about the issue of short term transient power handling. I knew that Dynaudio and Morel often stated short term peaks in excess of 1kw, but I wasn't sure if this was true of others.
IMO, tweeters are not necessarily the design problem when seeking a speaker that can produce 105dB average SPL. With typical Hifi tweeters you would be running them at 30-100W or so, which is out of their comfort range, so to speak - many are more designed for 1-10W input, whatever the specs may say.
So failing the HiFi tweeter route, you go the Pro route and you just gained 15dB of output capability by making that selection and going with a horn loaded driver.
That's not to say you can't make Loud direct radiator systems. It just takes bigger more sensitive drivers and higher crossover points.
So failing the HiFi tweeter route, you go the Pro route and you just gained 15dB of output capability by making that selection and going with a horn loaded driver.
That's not to say you can't make Loud direct radiator systems. It just takes bigger more sensitive drivers and higher crossover points.
well and if you've followed my thread my new speakers are Gedlee Abbeys, so obviously I did finally figure that out myself. However, when trying to discuss this with others, and explaining why speakers like these are not just best for this, but potentially the only reasonable approach, I need to best understand why other approaches can't or struggle to meet it. I had what I needed to explain the difficulties in something like a 6" driver or two producing 80hz at 105+db's, but I didn't really know how the tweeters faired by comparison.
A slight waveguide makes a lot of sense for a hifi style dome tweeter and crossed at 3k makes for a very potent speaker.
One must realize what the typical spectra is of music before deciding on what is needed. Obviously personal listening preferences and habits play a big role as well.
Small midranges is of no use so either one big midbass or several small/medium.
Pictured speaker has about 0.05%THD at mid frequencies at 100dB and no thermal compression. Ouput 130dB SPL and work over 80Hz with separate woofers down below to 16Hz or so. Handle 40kW peaks in midrange and 3kW peaks in tweeter range. Tweeter has about 0.05%THD at 90dB SPL in the low range.
/Peter
One must realize what the typical spectra is of music before deciding on what is needed. Obviously personal listening preferences and habits play a big role as well.
Small midranges is of no use so either one big midbass or several small/medium.
Pictured speaker has about 0.05%THD at mid frequencies at 100dB and no thermal compression. Ouput 130dB SPL and work over 80Hz with separate woofers down below to 16Hz or so. Handle 40kW peaks in midrange and 3kW peaks in tweeter range. Tweeter has about 0.05%THD at 90dB SPL in the low range.
/Peter
Attachments
Pipoes,
Almost all decent tweeters use an underhung motor design such that the BL versus coil displacement is perfectly constant as long as you don't exceed Xmax ((Top plate thickness-voice coil winding height)/2). Any tweeter over $10 OEM cost has at least a 0.5mm Xmax. Note that this is 2.5 times the 0.2mm Xmax that you brought up earlier.
Like Ron mentioned, one of the keys to getting a direct radiating tweeter to play loud is the crossover frequency. Drop the crossover by one octave and the driver displacement needs to increase by a factor of four to keep SPL constant. Conversely, raise the crossover frequency a little bit and the maximum SPL goes up a lot. There is no point in increasing the crossover slope beyond 2nd order from an excursion standpoint. A 2nd order HP filter will limit the driver to a constant excursion below the crossover frequency.
The "1kW" impulses that Dynaudio and Morel have used for years to brag about their power handling capabilities are completely bogus. There are a number of reasons for this.
1) Look closely at the chart recorder output on the Dynaudio impulse measurement. They usually have a small note indicating that a series capacitor was used. They fail to mention this. Since an impulse is comprised of the entire frequency spectrum, this capacitor reduces the current going to the tweeter quite a bit.
2) The 1kW power is calculated from the peak voltage at the output of the amplifier driving this system. It is not the actual power delivered to the driver from the voltage across it AND the current going through it.
3) Any system with a high frequency limit (the inductance of the tweeter in this example) will round off the peak of the impulse. This reduces the amount of current from the impulse that actually flows in the circuit.
Somewhere, I have photos of a real 1kW impulse applied to a Dynaudio tweeter. It only lasted a portion of the impulse.
Almost all decent tweeters use an underhung motor design such that the BL versus coil displacement is perfectly constant as long as you don't exceed Xmax ((Top plate thickness-voice coil winding height)/2). Any tweeter over $10 OEM cost has at least a 0.5mm Xmax. Note that this is 2.5 times the 0.2mm Xmax that you brought up earlier.
Like Ron mentioned, one of the keys to getting a direct radiating tweeter to play loud is the crossover frequency. Drop the crossover by one octave and the driver displacement needs to increase by a factor of four to keep SPL constant. Conversely, raise the crossover frequency a little bit and the maximum SPL goes up a lot. There is no point in increasing the crossover slope beyond 2nd order from an excursion standpoint. A 2nd order HP filter will limit the driver to a constant excursion below the crossover frequency.
The "1kW" impulses that Dynaudio and Morel have used for years to brag about their power handling capabilities are completely bogus. There are a number of reasons for this.
1) Look closely at the chart recorder output on the Dynaudio impulse measurement. They usually have a small note indicating that a series capacitor was used. They fail to mention this. Since an impulse is comprised of the entire frequency spectrum, this capacitor reduces the current going to the tweeter quite a bit.
2) The 1kW power is calculated from the peak voltage at the output of the amplifier driving this system. It is not the actual power delivered to the driver from the voltage across it AND the current going through it.
3) Any system with a high frequency limit (the inductance of the tweeter in this example) will round off the peak of the impulse. This reduces the amount of current from the impulse that actually flows in the circuit.
Somewhere, I have photos of a real 1kW impulse applied to a Dynaudio tweeter. It only lasted a portion of the impulse.
Matt
In theory a tweeter is just a small woofer and can be modeled as such. I do this all the time. What you find in practice is that a tweeter is thermally limited not excusrion limited. Its easy to see why this is.
If we look at the audio bandwidth it goes from lets say 100 Hz - 10 kHz (lower of course, but the argument still works). Its usual to consider 1 kHz as the mid point of the audio spectral bandwidth (which is a good rule of thumb from a pitch or percetual point of view). If we have a flat spectrum (which again isn't true of music, but it will make my point) then we need to consider that power (PSD, Power Spectral Density, is in volts^2 / Hz) is based on linear frequency NOT log frequency as our perception is. Now in linear frequency, for a flat spectrum from 100 Hz - 10 kHz. There is ten times as much power above 1 kHz (10,000 - 1,000 Hz = 9000 Hz) as there is below 1 kHz (1000 Hz - 100 Hz = 900 Hz). Thats a big difference, especially when you consider that the tweeter is usually a lot smaller (voice coil, etc.) than the speaker that operates below 1 kHz. This causes the tweeeter to virtually always be limited by its ability to disipate the heat from the input voltage. To the extent that it can't do this, it thermally modulates the sound, which I contend is a major factor.
In a test that I did using two two ways, one with a 1" dome tweeter and one with a 1" aperature compression driver, above 1 kHz, the dome tweeter had about 10 times as much thermal compression as the compression driver producing the same room SPL. Thats going to be a major audible difference.
In theory a tweeter is just a small woofer and can be modeled as such. I do this all the time. What you find in practice is that a tweeter is thermally limited not excusrion limited. Its easy to see why this is.
If we look at the audio bandwidth it goes from lets say 100 Hz - 10 kHz (lower of course, but the argument still works). Its usual to consider 1 kHz as the mid point of the audio spectral bandwidth (which is a good rule of thumb from a pitch or percetual point of view). If we have a flat spectrum (which again isn't true of music, but it will make my point) then we need to consider that power (PSD, Power Spectral Density, is in volts^2 / Hz) is based on linear frequency NOT log frequency as our perception is. Now in linear frequency, for a flat spectrum from 100 Hz - 10 kHz. There is ten times as much power above 1 kHz (10,000 - 1,000 Hz = 9000 Hz) as there is below 1 kHz (1000 Hz - 100 Hz = 900 Hz). Thats a big difference, especially when you consider that the tweeter is usually a lot smaller (voice coil, etc.) than the speaker that operates below 1 kHz. This causes the tweeeter to virtually always be limited by its ability to disipate the heat from the input voltage. To the extent that it can't do this, it thermally modulates the sound, which I contend is a major factor.
In a test that I did using two two ways, one with a 1" dome tweeter and one with a 1" aperature compression driver, above 1 kHz, the dome tweeter had about 10 times as much thermal compression as the compression driver producing the same room SPL. Thats going to be a major audible difference.
Jack Hidley said:
Hi!
Scan Speak 9800 = +/- 0.1mm.
Scan Speak ringradiator +/- 0.2mm
Scan Speak 9700/9900 +/- 0.4mm
Seas DXT overhung
Seas 27TDFC +/- 0.25mm underhung
Accuton C23 overhung
Absolutely!
Yes but that means that everything below Fs will cause serious cone travel and IMD. By crossing with a higher order slope you will significantly reduce movement.
Bursts, not impulses.
I don't remember if they mentioned it or not, it can't be both though.
Looking at the curves of the burst tests hould be enough to indicate a HP filter and they also state that power handling depends on crossoverpoint. gedlee said:
Agree!
But still we must consider spectra in music and typical listening habits. Most instruments roll off above 1kHz so the actual power of harmonics wil be small going to the tweeter. We have the occassional transient but as we have seen that can be handled well, it's the long term power handling that is poor but I for one don't have music that has constant high levels (+100dB) in the higher registers.
An SPL of 90dB is LOUD at in the tweeter range and that equals 1W into a typical tweeter and there will be no thermal compression to speak of at such level with a well designed dome tweeter.
There's a big difference between different tweeters. A soft dome without ferrofluid and a small magnet assembly will be much worse than a metal dome with ferro and a big magnet system (= large thermal mass).
/Peter
Here's a spectral analyzis of a grand piano. A piece played by both hands so we see a good represantation of a wide bandwith sound from the instrument. The sample is a 10s clip from a recording I did.
What can be seen is that the max level in the harmonic range (from 2-3k and up) is aprox- 20dB below the fundamentals.
This means that if we play back the piano at a realistic level of about 110dB we will have only 90dB from the tweeter and that is not constant. We will feed less than 1W into a typical tweeter and there is no thermal compression going on. Perhaps the mean power is something like 0.1-0.5W.
Of course we can come up with situations that is worse but for typical program material thermal compression from well designed dome tweeters is not an issue.
/Peter
What can be seen is that the max level in the harmonic range (from 2-3k and up) is aprox- 20dB below the fundamentals.
This means that if we play back the piano at a realistic level of about 110dB we will have only 90dB from the tweeter and that is not constant. We will feed less than 1W into a typical tweeter and there is no thermal compression going on. Perhaps the mean power is something like 0.1-0.5W.
Of course we can come up with situations that is worse but for typical program material thermal compression from well designed dome tweeters is not an issue.
/Peter
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.