I currently have a Vifa P21 and a Vifa D27TG in a 2-way speaker. I wish to implement a simple crossover to suit them. I have the attenation and impedance eq circuits worked out, but can I implement a 1st order slope for the woofer and a 2nd order slope for the tweeter? I will be crossing over at around 2.2kHz.
Is such a combination a good idea? If I use a first order slope for the woofer, why can't I use a capacitor in parralel instead of an inductor in series. Wouldn't this work the same way? I am not against inductors in the woofer circuit, I just have some Axon caps of the correct value on hand.
Using a combination of 1 st and 2nd orders is not covered anywhere. Can i use the formulas based on the same crossover frequency? Will I have phase problems?
Is such a combination a good idea? If I use a first order slope for the woofer, why can't I use a capacitor in parralel instead of an inductor in series. Wouldn't this work the same way? I am not against inductors in the woofer circuit, I just have some Axon caps of the correct value on hand.
Using a combination of 1 st and 2nd orders is not covered anywhere. Can i use the formulas based on the same crossover frequency? Will I have phase problems?
I thought everybody in australia has a copy of CALSOD (grin)
AFAIK it is not unusual to implement different orders to better match the natural rolloff of the driver in question, you do want to look at the whole system
not done it myself- wait for a better answer...
AFAIK it is not unusual to implement different orders to better match the natural rolloff of the driver in question, you do want to look at the whole system
not done it myself- wait for a better answer...
Nope, it is not unusual at all. With a 21cm driver, it will have a pretty steep roll-off of it's own. try a series coil and a zobel network.
As far as the tweeter, a series cap, and a parallel coil and a L-pad should work fine. You might have some problems with the phase working out quite right, but that is almost impossible to predict due to the Acoustic Center alignment and the phase response of each driver.
-Paul
P.S. Dont forget BSC
As far as the tweeter, a series cap, and a parallel coil and a L-pad should work fine. You might have some problems with the phase working out quite right, but that is almost impossible to predict due to the Acoustic Center alignment and the phase response of each driver.
-Paul
P.S. Dont forget BSC
You don't want to use a parallel cap unless you have a series resistor to control the impedance at high frequencies. Otherwise the cap will act as a short. The series resistor would drop sensitivity - probably not what you want...
You could make a series crossover....
You could make a series crossover....
do some XO order configurations have less phase issues than others?
I heard that 1st orders (as 3rd) are phase coherent, and 2nd (4rd) aren't
but how is it when we have a 1st with a 2nd? or a 1st with a 0th (no filter on the mid-bass)
I heard that 1st orders (as 3rd) are phase coherent, and 2nd (4rd) aren't
but how is it when we have a 1st with a 2nd? or a 1st with a 0th (no filter on the mid-bass)
This is a topic deeper than the pacific ocean itself. Really, there is no best way to do an XO, there are workarounds for everything, and nothing is perfect. The best way is to get a measurement mic and start expirmenting.
-Paul
-Paul
Bricolo said:
You have to realize that the "order" of a crossover is determined by the sum of the rolloff slopes of driver and crossover. If a driver rolls off at 12dB/oct and the crossover is 6dB/oct, the true result is 18dB/oct. This means that even crossovers with one component are not "phase coherent". One needs drivers with wide overlaps and carefuly designed zobels to be able to use a 1st order crossover and approximate a 1st order response.
Beginners seem to have real hangups about phase, and don't really understand what it is - just that the magazine reviewers seem to think it is important.
Ap
The term 'phase coherent' can be a bit misleading.
It comes from the fact that when using first order electrical xovers the drivers are both wired in positive polarity. With 2nd order you would invert the tweeter polarity ("out-of-phase").
Unfortunately a lot of speaker companies have used the phase coherent thing as a marketing tool and hence it has become a bit of a 'got to have it'.
Actual proponenets of the design methodology are Vandersteen, Dunlavy & Duntech - there are probably others.
I personally try to achieve three things -
freq. reponse smoothness , on & off axis.
Good phase tracking through the xover & for an octave above & below.
Smooth impedance with no low dips.
If these three graph well then I figure I'm onto a good thing.
I would recommend start with the woofer roll-off & add an inductor then try & get the hf to track it by using 2nd/3rd whatever. If you can then great, if not then back to the woofer and add a cap, then try the tweeter again, and so on.
ps textbook formulas don't really work, use design software.
The term 'phase coherent' can be a bit misleading.
It comes from the fact that when using first order electrical xovers the drivers are both wired in positive polarity. With 2nd order you would invert the tweeter polarity ("out-of-phase").
Unfortunately a lot of speaker companies have used the phase coherent thing as a marketing tool and hence it has become a bit of a 'got to have it'.
Actual proponenets of the design methodology are Vandersteen, Dunlavy & Duntech - there are probably others.
I personally try to achieve three things -
freq. reponse smoothness , on & off axis.
Good phase tracking through the xover & for an octave above & below.
Smooth impedance with no low dips.
If these three graph well then I figure I'm onto a good thing.
I would recommend start with the woofer roll-off & add an inductor then try & get the hf to track it by using 2nd/3rd whatever. If you can then great, if not then back to the woofer and add a cap, then try the tweeter again, and so on.
ps textbook formulas don't really work, use design software.
I don't have info for Vifa stuff on hand, but assuming they both have 8 ohm impedance; for a 1st order (butterworth) filter pair crossing (meaning each passband is -3dB I think...) at 2200Hz;
Low pass (woofer) will be a 0.58mH inductor in series,
High pass (tweeter) will be a 9uF capacitor in series.
If either are 4 ohm impedance, you would halve the inductor value and double the capacitor.
I think that 1st/2nd combinations are often used in an attempt to safely use the tweeter at lower crossover by using a 2nd order in the tweeter. Personally I don't like crossovers anywhere in approximately the 500Hz-1800Hz range because that's where human ears are most sensitive and therefore more likely to pick up any annoying crossover quirks.
If you go for 2nd order on both drivers I'd recommend linkwitz-riley rather than butterworth if you can find a calculator or formula, they sum together much more nicely, as long as you flip polarity of on of them.
Low pass (woofer) will be a 0.58mH inductor in series,
High pass (tweeter) will be a 9uF capacitor in series.
If either are 4 ohm impedance, you would halve the inductor value and double the capacitor.
I think that 1st/2nd combinations are often used in an attempt to safely use the tweeter at lower crossover by using a 2nd order in the tweeter. Personally I don't like crossovers anywhere in approximately the 500Hz-1800Hz range because that's where human ears are most sensitive and therefore more likely to pick up any annoying crossover quirks.
If you go for 2nd order on both drivers I'd recommend linkwitz-riley rather than butterworth if you can find a calculator or formula, they sum together much more nicely, as long as you flip polarity of on of them.
1st order xovers
Specs on vifa drivers are on www.d-s-t.com,
although 8ohm nominal drivers - this figure is pretty useless when determining xover values as no driver impedance is flat.
There is no such thing a s Linkwitz Riley 1st order.
There is only one formual for 1st order and that is Butterworth's ie. Q=0.707.
Don't use a cap only (1st order) on the tweeter as it always causes an irregular response which rises to a peak at xover f then dips then climbs again. You could eq the impedance peak & then use a cap but as you are adding more compenents than a 2nd order requires it defeats the purpose.
"I currently have a Vifa P21 and a Vifa D27TG in a 2-way speaker. I wish to implement a simple crossover to suit them. I have the attenation and impedance eq circuits worked out, but can I implement a 1st order slope for the woofer and a 2nd order slope for the tweeter? I will be crossing over at around 2.2kHz."
Yes do that.
Specs on vifa drivers are on www.d-s-t.com,
although 8ohm nominal drivers - this figure is pretty useless when determining xover values as no driver impedance is flat.
There is no such thing a s Linkwitz Riley 1st order.
There is only one formual for 1st order and that is Butterworth's ie. Q=0.707.
Don't use a cap only (1st order) on the tweeter as it always causes an irregular response which rises to a peak at xover f then dips then climbs again. You could eq the impedance peak & then use a cap but as you are adding more compenents than a 2nd order requires it defeats the purpose.
"I currently have a Vifa P21 and a Vifa D27TG in a 2-way speaker. I wish to implement a simple crossover to suit them. I have the attenation and impedance eq circuits worked out, but can I implement a 1st order slope for the woofer and a 2nd order slope for the tweeter? I will be crossing over at around 2.2kHz."
Yes do that.
1st and 2nd order crossover
Thankyou all for the replies
I have the details for the Vifa drivers, but I still have some questions. I mentioned that I have the numbers for the impedance equalistaion on the woofer. I do have that, but how do I calculate the new value of impedance of the woofer for the xover calculations?
If the resonance of the tweeter is at 650Hz, is 2,200 Hz too low for the Xover point when a 2nd order is used? I would think it is okay.
Are there any websites or free programs that show how to calculate the complete filter response curve for both high and low pass filters, so that I can simulate responses based on Xover freq and driver response curves. Is the theory worth it? I can't easily test anything myself.
At this point I am ussuming I will use a first order for the woofer and a 2nd order for the tweeter with the following results;
XOver point 2,200 Hz
Woofer impedance - assume 8Ohms at Xover freq (estimate)
Woofer series inductor - 0.58mH air core 14Ga
Impedance equal - 7.1 Ohms Resistor, 17.7uF Capacitor
Tweeter impedance 4.9Ohms at Xover freq (from the data curve)
2nd Order Linkwitz Riley
Tweeter wired in reverse phase
Series Cap - 7.38uF
Parallel Inductor - 0.71mH air core 14 or 16 Ga
Supposedly the sensitivities are within about 1dB of each other. I will implement an attenuation network only after listening.
I will initially mount the Xover externally. If I mount the Xover inside, should it be on the side wall, or on the back wall, behind the woofer. Or will it not matter as the vibration will be present everywhere? Does it really matter?
Thankyou all for the replies
I have the details for the Vifa drivers, but I still have some questions. I mentioned that I have the numbers for the impedance equalistaion on the woofer. I do have that, but how do I calculate the new value of impedance of the woofer for the xover calculations?
If the resonance of the tweeter is at 650Hz, is 2,200 Hz too low for the Xover point when a 2nd order is used? I would think it is okay.
Are there any websites or free programs that show how to calculate the complete filter response curve for both high and low pass filters, so that I can simulate responses based on Xover freq and driver response curves. Is the theory worth it? I can't easily test anything myself.
At this point I am ussuming I will use a first order for the woofer and a 2nd order for the tweeter with the following results;
XOver point 2,200 Hz
Woofer impedance - assume 8Ohms at Xover freq (estimate)
Woofer series inductor - 0.58mH air core 14Ga
Impedance equal - 7.1 Ohms Resistor, 17.7uF Capacitor
Tweeter impedance 4.9Ohms at Xover freq (from the data curve)
2nd Order Linkwitz Riley
Tweeter wired in reverse phase
Series Cap - 7.38uF
Parallel Inductor - 0.71mH air core 14 or 16 Ga
Supposedly the sensitivities are within about 1dB of each other. I will implement an attenuation network only after listening.
I will initially mount the Xover externally. If I mount the Xover inside, should it be on the side wall, or on the back wall, behind the woofer. Or will it not matter as the vibration will be present everywhere? Does it really matter?
Xover
OK - I did a bit of playing around & this is what I've come up with...
basically you want 2nd order LP/HP.
I set the xover f a bit lower (1.5k) as using an 8" which will start to beam about 2k.
For woofer (p21wo-39-08) - check the exact model as there are lots of versions of this driver.
Eq: Rp=5.5, Cp=33
Xover: Ls=1.0, Cp=10
For the Vifa D27tg-35-06 (again there are many versions)
It is worth measuring fs as this driver has reports of not meeting spec - check www.snippets.org / ldsg
It will require attenuation
Att: Rs=4.4, Rp=2.2
Xover: Lp=0.82, Cs=6
Now the system is about 3db down at xover so it will sound 'mellow' & I assumed you are using a vented box as this driver is intended. 3cu ft, fb = 25hz (f3 is about 20hz!)
ps the software I use is by Bullock and White called flexsys it's DOS and you have to enter all driver parameters.
good luck!
OK - I did a bit of playing around & this is what I've come up with...
basically you want 2nd order LP/HP.
I set the xover f a bit lower (1.5k) as using an 8" which will start to beam about 2k.
For woofer (p21wo-39-08) - check the exact model as there are lots of versions of this driver.
Eq: Rp=5.5, Cp=33
Xover: Ls=1.0, Cp=10
For the Vifa D27tg-35-06 (again there are many versions)
It is worth measuring fs as this driver has reports of not meeting spec - check www.snippets.org / ldsg
It will require attenuation
Att: Rs=4.4, Rp=2.2
Xover: Lp=0.82, Cs=6
Now the system is about 3db down at xover so it will sound 'mellow' & I assumed you are using a vented box as this driver is intended. 3cu ft, fb = 25hz (f3 is about 20hz!)
ps the software I use is by Bullock and White called flexsys it's DOS and you have to enter all driver parameters.
good luck!
Driver models
Ap,
Thanks for the information. I have the following:
Vifa P21WO-20-08
Vifa D27TG-45-06
42 litre vented box (1.48 cu feet), with -3dB at 43Hz. This is not all that low, but a nice size box with a smooth roll-off. Port is 86mm Dia (3.4") and 118mm long (4.65")
1.5kHZ seems like a low Xover point, but agree on the possibility of beaming.
Should the drivers be rebated into the baffle? Currently they are not, but it can be done. Should the sides of the baffle be radiused?
Ap,
Thanks for the information. I have the following:
Vifa P21WO-20-08
Vifa D27TG-45-06
42 litre vented box (1.48 cu feet), with -3dB at 43Hz. This is not all that low, but a nice size box with a smooth roll-off. Port is 86mm Dia (3.4") and 118mm long (4.65")
1.5kHZ seems like a low Xover point, but agree on the possibility of beaming.
Should the drivers be rebated into the baffle? Currently they are not, but it can be done. Should the sides of the baffle be radiused?
Drivers
That does change things a bit,
first off the enclosure seems a bit small according to the specs I have it needs about 56l, 2cu ft vented, I would tune at a ratio of h=1 which is Fs=Fb of 28hz.
With your port size I get Fb=50hz which results in a peak of 3db (not good). This will sound boomy & 'one-noted'.
If you have the enclosure built already then go sealed Qtc =.65 f3 is about 60hz, f10 about 30hz, with room gain you will have good bass response.
If you are going vented then oversize the enclosure (up to 3cu ft)& tune to Fs - this results in an EBP alignment which gives you low extension & 'tuneful' bass.
As for the tweeter the '45 I think is the horn laded version which will require more attenuation- maybe another 3db off.
If you haven't bought the tweeter already perhaps look at changing it.
Rebate the drivers if you have the tools (plunge router & jig)
I personally inset the baffle 5mm & use felt to fill the gap with cut-outs for drivers. Looks really nice & helps smooth out the top-end response.
Yes to radiusing the edges if you can, the larger the radius the better.
That does change things a bit,
first off the enclosure seems a bit small according to the specs I have it needs about 56l, 2cu ft vented, I would tune at a ratio of h=1 which is Fs=Fb of 28hz.
With your port size I get Fb=50hz which results in a peak of 3db (not good). This will sound boomy & 'one-noted'.
If you have the enclosure built already then go sealed Qtc =.65 f3 is about 60hz, f10 about 30hz, with room gain you will have good bass response.
If you are going vented then oversize the enclosure (up to 3cu ft)& tune to Fs - this results in an EBP alignment which gives you low extension & 'tuneful' bass.
As for the tweeter the '45 I think is the horn laded version which will require more attenuation- maybe another 3db off.
If you haven't bought the tweeter already perhaps look at changing it.
Rebate the drivers if you have the tools (plunge router & jig)
I personally inset the baffle 5mm & use felt to fill the gap with cut-outs for drivers. Looks really nice & helps smooth out the top-end response.
Yes to radiusing the edges if you can, the larger the radius the better.
There are some online (javascripted) crossover calculators here.
1.5k is low but it would be ok as long as you do use 2nd order filters. Since the Fs of the tweeter is also pretty low I don't think it'll matter. Maybe if you're worried about it go for 1800Hz or something?
I never suggested that 1st order Linkwitz Riley filters existed, but I'll repeat the recommendation to use their 2nd order variety.
According to "The Loudspeaker Design Cookbook" (Dickason), they have zero polar tilt, low sensitivity to horizontal driver separation, low sensitivity to driver Fs in filter stopband and minimally acceptable attenuation (whatever that is!). I use a 4th order LR between the mid and tweeter on a 3-way I built and they do a great job but I used to use a 2nd order and it was fine.
Calculating again without any data just at 8 ohm impedance at 1500Hz you'd need 6.6uF caps (series to tweeter, parallel to woofer) and 1.7mH inductors (series to woofer, parallel to tweeter.
For 1800Hz it's 5.5uF and 1.4mH,
for 2200Hz it's 4.5uF and 1.16mH.
The zobel is designed to stop excessive changes in driver impedance at higher frequencies caused by the reactive inductance of the voice coil and it's there so that the crossover actually "sees" an impedance closer to what it calculated for over these inductive frequencies. It should be calculated with the driver's voice coil resistance in mind and I don't think you need to recalculate the rest of the crossover accordingly.
1.5k is low but it would be ok as long as you do use 2nd order filters. Since the Fs of the tweeter is also pretty low I don't think it'll matter. Maybe if you're worried about it go for 1800Hz or something?
I never suggested that 1st order Linkwitz Riley filters existed, but I'll repeat the recommendation to use their 2nd order variety.
According to "The Loudspeaker Design Cookbook" (Dickason), they have zero polar tilt, low sensitivity to horizontal driver separation, low sensitivity to driver Fs in filter stopband and minimally acceptable attenuation (whatever that is!). I use a 4th order LR between the mid and tweeter on a 3-way I built and they do a great job but I used to use a 2nd order and it was fine.
Calculating again without any data just at 8 ohm impedance at 1500Hz you'd need 6.6uF caps (series to tweeter, parallel to woofer) and 1.7mH inductors (series to woofer, parallel to tweeter.
For 1800Hz it's 5.5uF and 1.4mH,
for 2200Hz it's 4.5uF and 1.16mH.
The zobel is designed to stop excessive changes in driver impedance at higher frequencies caused by the reactive inductance of the voice coil and it's there so that the crossover actually "sees" an impedance closer to what it calculated for over these inductive frequencies. It should be calculated with the driver's voice coil resistance in mind and I don't think you need to recalculate the rest of the crossover accordingly.
Polar tilt.... how should I explain this...
Using a basic woofer/tweeter 2-way example, they are obviously always separated by whatever distance apart on the baffle they are.
Since at the crossover point and either side of it there's an overlap in output and the sound wavefronts don't come from a single source, there is an effect called "lobing". This causes dips and peaks in the crossover overlap region, depending on the angle from which you are listening. Ideally this effect is minmum at "zero degrees" to vertical, i.e. the best place to sit is so that your ears are level with exactly halfway between the drivers, but the non-ideal situation is where you experiece polar tilt.
This is where a relative phase difference between the filter sections causes the lobing pattern to tilt up or down, so that best listening position is above or below the "zero degrees" point.
Since higher order filters have less overlap, this effect is at it's most noticable in 1st order filters, and 1st order butterworth filters actually have a tilt of 15 degrees below the listening plane. 3rd order butterworths tilt 15 degrees upwards but in much less of a frequency band.
Using a basic woofer/tweeter 2-way example, they are obviously always separated by whatever distance apart on the baffle they are.
Since at the crossover point and either side of it there's an overlap in output and the sound wavefronts don't come from a single source, there is an effect called "lobing". This causes dips and peaks in the crossover overlap region, depending on the angle from which you are listening. Ideally this effect is minmum at "zero degrees" to vertical, i.e. the best place to sit is so that your ears are level with exactly halfway between the drivers, but the non-ideal situation is where you experiece polar tilt.
This is where a relative phase difference between the filter sections causes the lobing pattern to tilt up or down, so that best listening position is above or below the "zero degrees" point.
Since higher order filters have less overlap, this effect is at it's most noticable in 1st order filters, and 1st order butterworth filters actually have a tilt of 15 degrees below the listening plane. 3rd order butterworths tilt 15 degrees upwards but in much less of a frequency band.
- Status
- Not open for further replies.