Crossover too near resonance?
Everything I have read says you have to keep the crossover frequency far above the driver’s resonant frequency, like 1-2 octaves.
But in Vance Dickason’s speaker cookbook he mentions an article by Joe D’Appolito in Speaker Builder 4/84,
(Which I don’t have, has anybody read the article?) Which apparently says that a second-order Butterworth network “...can be used in conjunction with the 6dB/octave acoustic high-pass response [of the driver] to derive a combined 18dB/octave Butterworth response... “
In other words, you put the crossover frequency right on top of the driver’s resonant frequency, instead of far above it.
First, wait a minute, is that a misprint? I thought the acoustic high-pass response of a driver was 12 db/octave, not 6dB /octave?
Second, if that technique really worked, wouldn’t everybody use it? If you could really cross over your tweeter AT resonance, instead of an octave or two above resonance, you would get a lot more extension out of the tweeter. If you could really cross over your mids AT resonance, instead of an octave or two higher, you would get a lot more extension out of the mids.
It seems too good to be true. Has anybody tried this, or read the article, or if not, what are the drawbacks to this technique?
Thanks,
Niacin
Everything I have read says you have to keep the crossover frequency far above the driver’s resonant frequency, like 1-2 octaves.
But in Vance Dickason’s speaker cookbook he mentions an article by Joe D’Appolito in Speaker Builder 4/84,
(Which I don’t have, has anybody read the article?) Which apparently says that a second-order Butterworth network “...can be used in conjunction with the 6dB/octave acoustic high-pass response [of the driver] to derive a combined 18dB/octave Butterworth response... “
In other words, you put the crossover frequency right on top of the driver’s resonant frequency, instead of far above it.
First, wait a minute, is that a misprint? I thought the acoustic high-pass response of a driver was 12 db/octave, not 6dB /octave?
Second, if that technique really worked, wouldn’t everybody use it? If you could really cross over your tweeter AT resonance, instead of an octave or two above resonance, you would get a lot more extension out of the tweeter. If you could really cross over your mids AT resonance, instead of an octave or two higher, you would get a lot more extension out of the mids.
It seems too good to be true. Has anybody tried this, or read the article, or if not, what are the drawbacks to this technique?
Thanks,
Niacin
niacin said:
I haven't read Dickason's book, so I cannot say what he means.
However, I would point out one thing. My understanding is that a speaker with a crossover slope of 18 dB/octave should be 3 dB down at the crossover point.
As the chart below shows, a speaker in a box to yield a combo of Qtc of 0.5, (or a speaker in an open baffle with a Qts of 0.5), will be 3 dB down at 1.4 times resonance. Which is half an octave above resonance.
An octave below that, at 0.7 times resonance, the speaker will still be only about 9 dB down. So from the crossover point to an octave below it, a speaker/box combo with a Qtc of 0.5-or slightly above-will have a rolloff of 6 dB/octave.
Below that, the rolloff rate begins to increase up to 12 dB/octave. However, once the speaker hits -12 dB compared to the midband, it is pretty much "out of the loop" anyway, as far as being heard. At any rate, the conversion from -6 dB rolloff rate to -12 dB rolloff rate is a gradual one, so by the time a speaker with a Qtc of 0.5 really is declining at the rate of 12 dB/octave, it is far, far below the midband and cannot be heard.
I can't tell you step-by-step how to build this kind of electrical/acoustic filter, only that from the crossover point to an octave below it, a box with a Qtc of 0.5 or so will have a slope of 6 dB/octave.
Attachments
This thread here deals with something similar to this:
http://www.diyaudio.com/forums/showthread.php?threadid=24230&highlight=
http://www.diyaudio.com/forums/showthread.php?threadid=24230&highlight=
This is WRT the mid-range and tweeter drivers.niacin said:
This is an effective way of getting a faster roll-off with a far less complex x-over when using a mid-bass driver. This is usually done in the 50Hz to 150Hz range and than crossing over to a sub below that.niacin said:
As KW indicates, this also allows you to run a lower Q closed box alignment on the mid-bass to maintain good transient response while still having the benefit of a fast roll-off in the stop band.
6dB per octave high pass is not a general high pass characteristic
of drivers though if it was the the statement would be true.
A much more realistic statement would be
" the "2nd order high pass function of a driver, e.g. a sealed
midrange driver can be combined with a 2nd order hign pass
electrical filter to form an overall acoustic 4th order high pass
function.
Using active filters this is easy to arrange however using
passive filters the impedance of the driver at resonance
complicates matters considerably.
If the driver is a tweeter power handling issues often
become a major factor preventing this approach "
What is common is combining a 6dB/12dB low pass roll-off
of a mid or bass mid unit with a 6dB/ 12dB electrical c/o to
form acoustic 12DB/18dB/24dB low pass alignments.
I suspect in original context "high pass" should be "low pass".
sreten.
of drivers though if it was the the statement would be true.
A much more realistic statement would be
" the "2nd order high pass function of a driver, e.g. a sealed
midrange driver can be combined with a 2nd order hign pass
electrical filter to form an overall acoustic 4th order high pass
function.
Using active filters this is easy to arrange however using
passive filters the impedance of the driver at resonance
complicates matters considerably.
If the driver is a tweeter power handling issues often
become a major factor preventing this approach "
What is common is combining a 6dB/12dB low pass roll-off
of a mid or bass mid unit with a 6dB/ 12dB electrical c/o to
form acoustic 12DB/18dB/24dB low pass alignments.
I suspect in original context "high pass" should be "low pass".
sreten.Tweeter Crossover at Resonance
Jeff Macaulay did this in one of his designs, with a Morel tweeter.
He combined a 2nd order filter with that of the tweeter to give a 4th order response, crossing over at the tweeter's resonant frequency. The filter's Q was chosen so that when combined with the tweeter's Q, it gave the desired 4th order Q. He said it worked well.
I may depend on which tweeter you use, because it might be too distorted, depending on the construction of the tweeter. Power handling may be an issue too.
Jeff Macaulay did this in one of his designs, with a Morel tweeter.
He combined a 2nd order filter with that of the tweeter to give a 4th order response, crossing over at the tweeter's resonant frequency. The filter's Q was chosen so that when combined with the tweeter's Q, it gave the desired 4th order Q. He said it worked well.
I may depend on which tweeter you use, because it might be too distorted, depending on the construction of the tweeter. Power handling may be an issue too.
The problem is not power handling but limited excursion. If you cross a common tweeter (Fs= 1000 hz, x max = 0.25mm) at 1000Hz, the tweeter will exceed it's linear excursion before anything else, whatever the slope used.
Surely, the power handling may be OK but not the excursion, even at moderate levels.
There are ways to calculate the excursion for a given frequency, at a given spl, for a specific tweeter of known dome area. This is the way to determine the correct highpass function.
F
Surely, the power handling may be OK but not the excursion, even at moderate levels.
There are ways to calculate the excursion for a given frequency, at a given spl, for a specific tweeter of known dome area. This is the way to determine the correct highpass function.
F
I think I get it..
I think I get it... if you try to cross over at resonance using a passive crossover it will be very tricky due to the impedanace peaks, phase shifts, etc. , that's why it is not a common design.
But, if you cross over at resonance using an active crossover it will be easier.
Here is what I want to do. I want to use a sealed box with a Q of 1.0 (for maximum SPL and power handling) and a f3 of about 100Hz. That will give me ( I think) a rolloff of 12dB/octave below cutoff. Then I want to add a single-pole active highpass filter, 6db/octave, so I end up with a 18db/octave Butterworth response.
My question is: what frequency do I want to set the active crossover to, and what is my final f3?
I don't have the fancy software to model active/passive responses together, can anybody help me?
thanks, niacin
I think I get it... if you try to cross over at resonance using a passive crossover it will be very tricky due to the impedanace peaks, phase shifts, etc. , that's why it is not a common design.
But, if you cross over at resonance using an active crossover it will be easier.
Here is what I want to do. I want to use a sealed box with a Q of 1.0 (for maximum SPL and power handling) and a f3 of about 100Hz. That will give me ( I think) a rolloff of 12dB/octave below cutoff. Then I want to add a single-pole active highpass filter, 6db/octave, so I end up with a 18db/octave Butterworth response.
My question is: what frequency do I want to set the active crossover to, and what is my final f3?
I don't have the fancy software to model active/passive responses together, can anybody help me?
thanks, niacin
Gary:
Here is the displacement versus SPL chart for woofers in both normal and Metric measurements.
http://www.diyaudio.com/forums/showthread.php?threadid=5668&highlight=
Since a one inch tweeter, crossing over at 1000 Hz or so, will be reproducing wavelengths far greater than it's diameter, I would think we could take this woofer guide and extend it upward to find out how much excursion a tweeter would need to produce which SPl at which frequency,
I have seen tweeters with resonances around 1200. Let's take one of those. Let's also assume that at 1200 Hz, the tweeter will never be called upon to go over 100 SPL, since in real life music, the requirements for high SPL fall off as you enter the treble region. I admit this requirement might be open to debate.
From the chart, we see that it takes about 2 cu inches to produce 100 dB at 100 Hz. Since you need only one quarter the displacement for every octave you go up, (and 4 times the displacement for every octave you go down), let's count this out. This is for a tweeter that is expected to cross over at 1200 Hz.
Two hundred Hz will require 0.5 cu inches. 400 Hz will require 0.125 cu inches. 800 Hz will require .03125 cu inches. 1600 Hz will require .0078 cu inches.
Since 1200 Hz is roughly halfway between 800 Hz and 1200 Hz, let's call it .0156 cu inches necessary to put out 100 dB at 1200 Hz.
One more thing. If you are going to cross this over at 1200 Hz, at a third order crossover, that means out speaker only has to be 3 dB down at the crossover point. That is 25% less displacement required-remember,the higher the SPL, the more displacement you need. So instead of 0.156 cu inches, we really only need 0.0112 cu inches.
A Peerless tweeter, 811582, has a gap of 2.5 mm and a voice coil length of 1.6 mm, for a +/- excursion of 0.45 mm. That is .018 inches. It has a cone area of 6.3 sq cm. That is 1 sq inch, (I suspect this tweeter is really 1.25 cu inch diameter). So it has a +/- displacement of 0.018 cu inches, and that is enough to produce 100 dB at 1200 Hz, with some to spare. This is assuming the tweeter is required to be -3 dB at the crossover frequency. Which a tweeter with a final Qtc, (remember, these are enclosed speakers, albeit tiny little enclosed speakers), of 0.7 will give. Many tweeters seem to have a final Qtc of 0.8 or so, so things might not be so badly off.
So it looks like if you can find a tweeter with a resonance of around 1200 Hz, you can do this. I have seen tweeters with that resonance.
I freely admit I did this math on the fly, so if I got it wrong, my apologies. The link I gave also gives the SPL's in Metric, so if anyone wants to work it out that way, be my guest. But it dies seem to open up some possiblilties here, I would think.
Here is the displacement versus SPL chart for woofers in both normal and Metric measurements.
http://www.diyaudio.com/forums/showthread.php?threadid=5668&highlight=
Since a one inch tweeter, crossing over at 1000 Hz or so, will be reproducing wavelengths far greater than it's diameter, I would think we could take this woofer guide and extend it upward to find out how much excursion a tweeter would need to produce which SPl at which frequency,
I have seen tweeters with resonances around 1200. Let's take one of those. Let's also assume that at 1200 Hz, the tweeter will never be called upon to go over 100 SPL, since in real life music, the requirements for high SPL fall off as you enter the treble region. I admit this requirement might be open to debate.
From the chart, we see that it takes about 2 cu inches to produce 100 dB at 100 Hz. Since you need only one quarter the displacement for every octave you go up, (and 4 times the displacement for every octave you go down), let's count this out. This is for a tweeter that is expected to cross over at 1200 Hz.
Two hundred Hz will require 0.5 cu inches. 400 Hz will require 0.125 cu inches. 800 Hz will require .03125 cu inches. 1600 Hz will require .0078 cu inches.
Since 1200 Hz is roughly halfway between 800 Hz and 1200 Hz, let's call it .0156 cu inches necessary to put out 100 dB at 1200 Hz.
One more thing. If you are going to cross this over at 1200 Hz, at a third order crossover, that means out speaker only has to be 3 dB down at the crossover point. That is 25% less displacement required-remember,the higher the SPL, the more displacement you need. So instead of 0.156 cu inches, we really only need 0.0112 cu inches.
A Peerless tweeter, 811582, has a gap of 2.5 mm and a voice coil length of 1.6 mm, for a +/- excursion of 0.45 mm. That is .018 inches. It has a cone area of 6.3 sq cm. That is 1 sq inch, (I suspect this tweeter is really 1.25 cu inch diameter). So it has a +/- displacement of 0.018 cu inches, and that is enough to produce 100 dB at 1200 Hz, with some to spare. This is assuming the tweeter is required to be -3 dB at the crossover frequency. Which a tweeter with a final Qtc, (remember, these are enclosed speakers, albeit tiny little enclosed speakers), of 0.7 will give. Many tweeters seem to have a final Qtc of 0.8 or so, so things might not be so badly off.
So it looks like if you can find a tweeter with a resonance of around 1200 Hz, you can do this. I have seen tweeters with that resonance.
I freely admit I did this math on the fly, so if I got it wrong, my apologies. The link I gave also gives the SPL's in Metric, so if anyone wants to work it out that way, be my guest. But it dies seem to open up some possiblilties here, I would think.
Kelticwizard, i have checked a few things using Linkwitz SPL-excursion calculator..., you were not far from the truth
"standard" Seas type tweeter or others: Sd=7.5 cm^2, X=0.25mm
SPL MAX at 1000 hz= 88 dB
SPL MAX at 1500 hz=95 dB
SPL Max at 2000 Hz=100dB
Scan Speak 9300 or 9500 tweeter: Sd=8.5, x=0.4mm
SPL MAX at 1000Hz=93 dB
SPL MAX at 1500 Hz= 100dB
Peerless tweeter you described: Sd=6.3, X=0.45
SPL MAX at 1500 hz= 99dB
well, it is true that tweeter can handle a lot. Still, i would be hesitant to cross a tweeter at 1500 or lower....
F
"standard" Seas type tweeter or others: Sd=7.5 cm^2, X=0.25mm
SPL MAX at 1000 hz= 88 dB
SPL MAX at 1500 hz=95 dB
SPL Max at 2000 Hz=100dB
Scan Speak 9300 or 9500 tweeter: Sd=8.5, x=0.4mm
SPL MAX at 1000Hz=93 dB
SPL MAX at 1500 Hz= 100dB
Peerless tweeter you described: Sd=6.3, X=0.45
SPL MAX at 1500 hz= 99dB
well, it is true that tweeter can handle a lot. Still, i would be hesitant to cross a tweeter at 1500 or lower....
F
Re: I think I get it..
As far as I understand it a Q=1 of frequency F (its f3 is not
relevant) cascaded with a first order filter of frequency F
will be a 3rd order Butterworth -3dB at F.
If going for active first order a line level passive filter is easy,
you can just set the input coupling capacitor of the amplifier.
sreten.
niacin said:
As far as I understand it a Q=1 of frequency F (its f3 is not
relevant) cascaded with a first order filter of frequency F
will be a 3rd order Butterworth -3dB at F.
If going for active first order a line level passive filter is easy,
you can just set the input coupling capacitor of the amplifier.
sreten.
Re: Re: I think I get it..
If my box is down 3dB at F, and my crossover is ALSO down 3dB at F, then my compound response is going to be down 6dB at F. Right? So the new f3 of the system is going to be found a little higher up. How much exactly is my question, because now I need to design a matching 3rd-order electronic lowpass for my subs.
Maybe I can just draw an 18dB/octave curve and measure it with a ruler.
sreten said:
As far as I understand it a Q=1 of frequency F (its f3 is not
relevant) cascaded with a first order filter of frequency F
will be a 3rd order Butterworth -3dB at F.
If going for active first order a line level passive filter is easy,
you can just set the input coupling capacitor of the amplifier.
If my box is down 3dB at F, and my crossover is ALSO down 3dB at F, then my compound response is going to be down 6dB at F. Right? So the new f3 of the system is going to be found a little higher up. How much exactly is my question, because now I need to design a matching 3rd-order electronic lowpass for my subs.
Maybe I can just draw an 18dB/octave curve and measure it with a ruler.
Re: Re: Re: I think I get it..
True enough. Please review Post #1 of this thread. As you can see, a closed box with a Qtc of 1.0, at resonance, is exactly at the midband. In fact, it is falling off from being +1.3 dB slightly above midband.
So if your 18 dB crossover is at -3 at resonance, then it will be where it is supposed to be.
Sounds nifty, but there is a problem, which Sretan alluded to earlier. At resonance, your impedance is way, way high up there. So the ohmage to cut the speaker's SPL at resonance will be much higher as the speaker's ohmage falls as it gets away from resonance. This is why Sretan was talking about active filters.
There might be a way to still do what you want with passive filters, though. That would be to build an aperiodic box. I have never built one, but I can model one, and they have a real nice flat impedance curve that fits in with a crossover at the low end.
By the way, what was the midbass you were planning to use that you wanted an 18 dB/octave rolloff on?
niacin said:
True enough. Please review Post #1 of this thread. As you can see, a closed box with a Qtc of 1.0, at resonance, is exactly at the midband. In fact, it is falling off from being +1.3 dB slightly above midband.
So if your 18 dB crossover is at -3 at resonance, then it will be where it is supposed to be.
Sounds nifty, but there is a problem, which Sretan alluded to earlier. At resonance, your impedance is way, way high up there. So the ohmage to cut the speaker's SPL at resonance will be much higher as the speaker's ohmage falls as it gets away from resonance. This is why Sretan was talking about active filters.
There might be a way to still do what you want with passive filters, though. That would be to build an aperiodic box. I have never built one, but I can model one, and they have a real nice flat impedance curve that fits in with a crossover at the low end.
By the way, what was the midbass you were planning to use that you wanted an 18 dB/octave rolloff on?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.