All right I did some searching here and couldn't find a definent answer, but let me tell you what I know and you guys correct me as needed.
I was leaning towards a 1st order crossover for its simplicity,but because of the target tweeters (Audax TM025F1) fs (1200hz) and freq resp (2500-20K) I was thinking it would be better to use a 2nd order for the steeper slope in the 2500-2800hz area (mating with a peerless 6 1/2 850467). However I keep hearing that the phase can change between the tweeter and the woofer. But the problem is sometimes poeple say it happens on the odd orders like 1st and 3rd, and some poeple say its the even orders. Now I've heard you can turn the polarity on the tweeter around and it will fix this, so is that true? I could just use the 12db second order and reverse the polarity?
Also is there a good website that shows the different wiring schemes of different crossovers and possible explain what a zobel (sp?) crossoever is and all that good stuff.
Man I need to dig out that loudspeaker cookbook!
I was leaning towards a 1st order crossover for its simplicity,but because of the target tweeters (Audax TM025F1) fs (1200hz) and freq resp (2500-20K) I was thinking it would be better to use a 2nd order for the steeper slope in the 2500-2800hz area (mating with a peerless 6 1/2 850467). However I keep hearing that the phase can change between the tweeter and the woofer. But the problem is sometimes poeple say it happens on the odd orders like 1st and 3rd, and some poeple say its the even orders. Now I've heard you can turn the polarity on the tweeter around and it will fix this, so is that true? I could just use the 12db second order and reverse the polarity?
Also is there a good website that shows the different wiring schemes of different crossovers and possible explain what a zobel (sp?) crossoever is and all that good stuff.
Man I need to dig out that loudspeaker cookbook!
For every pole (ie order) there is a 90 degrees phase shift. A 1st order XO is inherently phase correct (assuming everything is lined up & you have perfect drivers).
If you go to the FRD Group and download the passive crossover design calculator, it has many of the text book filters. There is a set of accompanying pictures to illustrate the topologies being calculated.
dave
If you go to the FRD Group and download the passive crossover design calculator, it has many of the text book filters. There is a set of accompanying pictures to illustrate the topologies being calculated.
dave
Phase and Delay in Crossovers
Hi Westrock2000,
One of the great misunderstandings in audio is that the phase of the crossover is not constant.
To give some examples:
A simple firstorder lowpass filter for the bass [a inductor] changes the phase from 0 degree at 20Hz to -90 degree at 20kHz. -45 degree at the crossover; 400Hz.
A fourth order Bessel highpass filter for the tweeter at 6000Hz has 0 degree phase at 20 Hz and -360 degree phase at 100kHz.
-180 degree at the crossover(6000Hz)
The delay is flat just up to the crossover point , then falls in a straight line on paper with double logaritmic scales (dB/frequency)
When combining all kinds of filters in a 3-way system the situation gets more complex. Also the above was for zero resistance of the coils. I real life the resistance of the coils is of course not zero and this complicates the situation even further. Also the driver does not have a constant impedance across its used frequency band. Overviewing all this the construction of a loudspeaker crossover is a work of Art and a lot of trial and error <B>and</B> listening!
Hi Westrock2000,
One of the great misunderstandings in audio is that the phase of the crossover is not constant.
To give some examples:
A simple firstorder lowpass filter for the bass [a inductor] changes the phase from 0 degree at 20Hz to -90 degree at 20kHz. -45 degree at the crossover; 400Hz.
A fourth order Bessel highpass filter for the tweeter at 6000Hz has 0 degree phase at 20 Hz and -360 degree phase at 100kHz.
-180 degree at the crossover(6000Hz)
The delay is flat just up to the crossover point , then falls in a straight line on paper with double logaritmic scales (dB/frequency)
When combining all kinds of filters in a 3-way system the situation gets more complex. Also the above was for zero resistance of the coils. I real life the resistance of the coils is of course not zero and this complicates the situation even further. Also the driver does not have a constant impedance across its used frequency band. Overviewing all this the construction of a loudspeaker crossover is a work of Art and a lot of trial and error <B>and</B> listening!
I have found this a very good website for designing cross-overs http://www.lalena.com/audio/ Also this link http://www.speakerbuilding.com/content/1044/ outlines the many problems to avoid with parrellel, those wired in parellel, cross-overs (otherwise known by me as the cross-overs most diy dudes use). Although he avoids phase shift problems by building a series cross-over, you probably won't want to do that. The link also has many other good projects.
To help you understand why people say it happens on 1st and 3rd cross-overs is because and correct if I'm wrong any one, is because inductors shilf the phase lets say, back 90 degrees for each inductor and capacitors shift it forward a little bit. So 1st and third ordet x-overs maybe maybe about 85 degrees and 255 degrees out of phase respectively but at you don't have almost total cancelation of frequencies like you do when they're 180 out-phase. That is to say, the wave peaks meet but one is high and one is low if you look at a graph or the ocean. Read the second link above it explains this part much better than I can.
To help you understand why people say it happens on 1st and 3rd cross-overs is because and correct if I'm wrong any one, is because inductors shilf the phase lets say, back 90 degrees for each inductor and capacitors shift it forward a little bit. So 1st and third ordet x-overs maybe maybe about 85 degrees and 255 degrees out of phase respectively but at you don't have almost total cancelation of frequencies like you do when they're 180 out-phase. That is to say, the wave peaks meet but one is high and one is low if you look at a graph or the ocean. Read the second link above it explains this part much better than I can.
Most of what we know about crossovers is wrong, not in
the theory sense, but in making real speakers.
It's a fact, and any manufacturer is welcome to argue with
me, that real crossovers are the result of a tremendous
amount of screwing around and listening and measuring.
Days. Weeks. Months.
The Audax tweeters are known for very forgiving characteristics
if you don't drive them too hard, and I have had a lot of
success with about 3 uF in series with them and that's all.
But this depends a lot on what your are mating them with.
If it's the Audax carbon fiber 8", you're in luck. It will work
with either no crossover or maybe .25 mH. The result can
be extremely good.
Some other driver? Back to paragraph 2.
the theory sense, but in making real speakers.
It's a fact, and any manufacturer is welcome to argue with
me, that real crossovers are the result of a tremendous
amount of screwing around and listening and measuring.
Days. Weeks. Months.
The Audax tweeters are known for very forgiving characteristics
if you don't drive them too hard, and I have had a lot of
success with about 3 uF in series with them and that's all.
But this depends a lot on what your are mating them with.
If it's the Audax carbon fiber 8", you're in luck. It will work
with either no crossover or maybe .25 mH. The result can
be extremely good.
Some other driver? Back to paragraph 2.
Dear Mr Pass,
I think what westrock want is a starting point. should he use 12db/octave, 6db, 12db LR or someting in between or other.
As a rule to make my life as simple as possible I prefer to use drivers that do not require steep XOs as that complicates things for a DIYer who does not have access to measuring instruments.
I have seen speaker manufactuers use driver that have all sorts of peaks and valleys in their freq. curves and than use passive XOs (notch filters, etc...) to compensate for all of this in the end having XOs with as many as 50 components. I find it hard to believe that this would result in a good sound but I am not one to judge. The market does that better than me.
given this and your vast expereince what would you suggest Westrock do? 6db high pass with 6db low pass or 12 db low pass or even a Low pass that is 9db or so? Where does he start?
I ask this as a DIYer and one in India where I do not have access to LDC or other books and drivers are hard to find leave alone measuring instruments I have to look to the DIYAUDIO forum for help, pointers, and guidance. I would like to know how to start too? My drivers are not the same as his but are similar (Vifa TC series 6" and 1").
I think what westrock want is a starting point. should he use 12db/octave, 6db, 12db LR or someting in between or other.
As a rule to make my life as simple as possible I prefer to use drivers that do not require steep XOs as that complicates things for a DIYer who does not have access to measuring instruments.
I have seen speaker manufactuers use driver that have all sorts of peaks and valleys in their freq. curves and than use passive XOs (notch filters, etc...) to compensate for all of this in the end having XOs with as many as 50 components. I find it hard to believe that this would result in a good sound but I am not one to judge. The market does that better than me.
given this and your vast expereince what would you suggest Westrock do? 6db high pass with 6db low pass or 12 db low pass or even a Low pass that is 9db or so? Where does he start?
I ask this as a DIYer and one in India where I do not have access to LDC or other books and drivers are hard to find leave alone measuring instruments I have to look to the DIYAUDIO forum for help, pointers, and guidance. I would like to know how to start too? My drivers are not the same as his but are similar (Vifa TC series 6" and 1").
Hi all
One point hasn't been mentioned so far: Not only does the driver's impedance influence the transfer function of the filter to which it is connected. The electrical transfer function of the filter/driver combination has to be added to the acoustic transfer function of the driver as well.
If one wants to be very exact then first order crossovers can't be built at all: They will be third order filters caused by a combination of an elctrical first order and an acoustical 2nd order filter (that's why the drivers on some Dynaudio first order crossovers are wired out of phase).
For this reason it can be dangerous to give general rules for the correct phase of the drivers. The steeper the filter's rolloff and the more it's f3 differs from the driver's f3 the more accurately the known rules may apply (i.e. in phase for 1st order, out of phase for 2nd, but only if....... there we go again ! ).
Regards
Charles
One point hasn't been mentioned so far: Not only does the driver's impedance influence the transfer function of the filter to which it is connected. The electrical transfer function of the filter/driver combination has to be added to the acoustic transfer function of the driver as well.
If one wants to be very exact then first order crossovers can't be built at all: They will be third order filters caused by a combination of an elctrical first order and an acoustical 2nd order filter (that's why the drivers on some Dynaudio first order crossovers are wired out of phase).
For this reason it can be dangerous to give general rules for the correct phase of the drivers. The steeper the filter's rolloff and the more it's f3 differs from the driver's f3 the more accurately the known rules may apply (i.e. in phase for 1st order, out of phase for 2nd, but only if....... there we go again
! ).Regards
Charles
while i agree with you in most parts I am also tring ot juggle 2 contradicting theories.
1. The F3 and solpe of the filter should act before the driver roll off and faster than the driver roll of
2. the KISS principle and the concept that every component howevr good in quality will veil the sound. Meaning that one should utillise as few components as possible.
If one were to build a XO based on theory 1 one would end up building a 4th or 6th order filter atleast one octave higher/lower than the driver's F3. Few drivers we use (SS, Vifa, Audax) roll off faster than 24db/octave. Most are around 12 db per octave of less.
If one were to build a XO using theory 2 one would have to stick to a single inductor and single cap XO or even and XO with no inductor (if one were building a 2 way). In practice we work within these "extremes".
In practice we try to employ higher order filters, impedance compensation, difraction step compensation, attenuation, notch filters, etc...
if one were to chance upon and choose drivers that have relatively smoooth roll ofs one could design a filter with the acoustical roll off of the driver as part of the total acoustical roll off. In most cases one could try and make do with a 6db/oct electrical roll off which would translate to a 12-18db/oct acoustical roll off even if the filter F3 was 1 octave higher/lower than the driver's F3. If the drivers (woofer and tweeter) roll of at 6db per octave of so then one can expriment with wiring the tweeter in phase and out of phase. That is my present goal. to find drivers that are wide range enough to make do with a simple XO.
1. The F3 and solpe of the filter should act before the driver roll off and faster than the driver roll of
2. the KISS principle and the concept that every component howevr good in quality will veil the sound. Meaning that one should utillise as few components as possible.
If one were to build a XO based on theory 1 one would end up building a 4th or 6th order filter atleast one octave higher/lower than the driver's F3. Few drivers we use (SS, Vifa, Audax) roll off faster than 24db/octave. Most are around 12 db per octave of less.
If one were to build a XO using theory 2 one would have to stick to a single inductor and single cap XO or even and XO with no inductor (if one were building a 2 way). In practice we work within these "extremes".
In practice we try to employ higher order filters, impedance compensation, difraction step compensation, attenuation, notch filters, etc...
if one were to chance upon and choose drivers that have relatively smoooth roll ofs one could design a filter with the acoustical roll off of the driver as part of the total acoustical roll off. In most cases one could try and make do with a 6db/oct electrical roll off which would translate to a 12-18db/oct acoustical roll off even if the filter F3 was 1 octave higher/lower than the driver's F3. If the drivers (woofer and tweeter) roll of at 6db per octave of so then one can expriment with wiring the tweeter in phase and out of phase. That is my present goal. to find drivers that are wide range enough to make do with a simple XO.
I always start with a cap to protect the mid or tweeter
from low frequencies and build from there. I add things
when they appear to improve the performance.
You can get away with a lot more doing it this way as long
as you don't abuse the drivers (and your ears) with excessive
power.
The Audax tweeters of the TW025 variety generally make it
down to about 2KHz or so, and I usually start there with a
3 uF or so cap in series. Then maybe an inductor on the
woofer, and so on......
from low frequencies and build from there. I add things
when they appear to improve the performance.
You can get away with a lot more doing it this way as long
as you don't abuse the drivers (and your ears) with excessive
power.
The Audax tweeters of the TW025 variety generally make it
down to about 2KHz or so, and I usually start there with a
3 uF or so cap in series. Then maybe an inductor on the
woofer, and so on......
Wow, thanks for adding so much to this thread. I will check out the web sites mentioned:
I do have another question. Say I come up with some preliminary crossover values, can I buy cheaper inductors and caps (would cheap bi-polar caps not be too wise?) and once I find the right crossover build a final one with the higher grade components?
I do have another question. Say I come up with some preliminary crossover values, can I buy cheaper inductors and caps (would cheap bi-polar caps not be too wise?) and once I find the right crossover build a final one with the higher grade components?
I agree with phase_accurate.
I have found that properly engineered loudspeaker/crossover design is pretty damn difficult.
The overall acoustic response is what is important and you really need to consider the magnitude and phase response of the drivers in their enclosures too. I think this is where a good measurement system would pay big. Bode Plots and phasor diagrams are the most helpful for me at this point to get a general idea of what needs to be done.
one day i'll get around to writing a matlab script to make it easier. The math right now is highly repetitive and fairly involved.
jt
I have found that properly engineered loudspeaker/crossover design is pretty damn difficult.
The overall acoustic response is what is important and you really need to consider the magnitude and phase response of the drivers in their enclosures too. I think this is where a good measurement system would pay big. Bode Plots and phasor diagrams are the most helpful for me at this point to get a general idea of what needs to be done.
one day i'll get around to writing a matlab script to make it easier. The math right now is highly repetitive and fairly involved.
jt
for anyone reading this who is planning on embarking on diy crossovers i have some advice (or maybe a warning?)
it's not all that difficult, with patience and a good collection of various component values and a simple piece of software to do the boring math - to come up with great 2-ways even up to my favourite (imho the king of passive crossovers) the 4-th order linkwitz riley.
BUT... if you are happy with your two way efforts and decide you want to try a three way please try to remember that i told you....
THAT PATH LEADS TO MADNESS!!!!
i think i probably added 10 years to my like in the approximately 3 weeks i once attempted a passive three way. it's like trying to juggle or something but the balls keep changing shape and size. in the end i gave up and decided to bi-amp between the woofer and mid and went to the land of bliss that is active crossovers. no impedance worries, easy compensation for driver sensitivity/efficiency differences, no loss to all those damn inductors, etc, etc. I still have a passive between mid and tweeter and it's one of the favourites i mentioned above, which as it happens, is also the spec of the active (that is 4-th l-r). I could never go back now! ACTIVE CROSSOVERS RULE!!!
it's not all that difficult, with patience and a good collection of various component values and a simple piece of software to do the boring math - to come up with great 2-ways even up to my favourite (imho the king of passive crossovers) the 4-th order linkwitz riley.
BUT... if you are happy with your two way efforts and decide you want to try a three way please try to remember that i told you....
THAT PATH LEADS TO MADNESS!!!!
i think i probably added 10 years to my like in the approximately 3 weeks i once attempted a passive three way. it's like trying to juggle or something but the balls keep changing shape and size. in the end i gave up and decided to bi-amp between the woofer and mid and went to the land of bliss that is active crossovers. no impedance worries, easy compensation for driver sensitivity/efficiency differences, no loss to all those damn inductors, etc, etc. I still have a passive between mid and tweeter and it's one of the favourites i mentioned above, which as it happens, is also the spec of the active (that is 4-th l-r). I could never go back now! ACTIVE CROSSOVERS RULE!!!
Musings from a "relatively new" DIYER (at least in terms of speakers)
I am a long time DIYER in terms of electronics (and heck, even some commercial stuff, mainly in the broadcast realm).
I played around with speaker design years ago, but was dragged kicked and screaming into it about 2 months ago when I purchased what should have been "excellent" speakers at a clearance only to come to the conclusion, after many hours of break in, that the cross-overs were woefully wrong.
And hence I embarked into a world of speaker design.... and along the way I found:
1) You can set up a measurement system really cheaply if you have a computer, and an amplifier. Just about any sound card will do.
- Some precision power resistors for impedance measurement
- Add in a Panasonic Electret microphone cartridge ($4 at Digikey)
- Add in some low cost software, AudioAnalyser if you want something easy to use, Speaker Workshop, if you want something that will allow you to do near professional work
2) Read, read, read, read, read.....
- Find out everything you can.
- Go to driver manufacturer's web sites and look at the schematics for cross-overs in their kits
- Read over projects on web-sites
- Read the Speaker Workshop tutorials, etc.
3) Start playing on paper. Speaker workshop will allow you to essentially test your finished speaker (at least the response), without ever touching a soldering iron. It is pretty darn accurate too! If you have access to spice, you can do some pretty good work too. (this is what I used initially)
4) Get as close as you can on paper, then build, test and listen.
I did some initial testing with simple cross-overs to confirm that my "assumptions" were correct. Then I worked my way towards more complex cross-overs.
While it is a nice thought to say you are going to use drivers that are "easy", the fact that they are easy to design with, does not mean they are good. Fast cars are not always easy to drive, but once you learn how to control them, they will do things that no other cars can do. I think you can do draw the same conclusions about drivers. The woofers in my speakers have a pretty nasty mass-limited break up peak around 4KHz. When that peak is not tamed (as in when I first bought them), they do not sound great. Once tamed, they sounded fantastic. A first order cross-over, or even a second order LR was not sufficient enough to reduce the peak. I ended up with a "modified" 4th order that is really a second order with a "notch" filter around the peak. I played around a lot on paper with both the woofer cross-over response and the tweeter response until I got the overlap I wanted and a conscious trade-off between keeping the cross-over frequency low so that system dispersion would be better, and not having the cross over frequency too low such that the tweeter is operating in a non-ideal range.
The resultant cross-over was very good and miles ahead of how the speakers came (second order LR on woofer\tweeter). I never modified the topology, but I made some minor component changes to shift the cross-over frequency slightly and to adjust the depth of the notch filter. The cross-over that I developed on the paper was more than acceptable. However, in fine DIY spirit, I modifed for perfection.
Good Luck
Alvaius!
I am a long time DIYER in terms of electronics (and heck, even some commercial stuff, mainly in the broadcast realm).
I played around with speaker design years ago, but was dragged kicked and screaming into it about 2 months ago when I purchased what should have been "excellent" speakers at a clearance only to come to the conclusion, after many hours of break in, that the cross-overs were woefully wrong.
And hence I embarked into a world of speaker design.... and along the way I found:
1) You can set up a measurement system really cheaply if you have a computer, and an amplifier. Just about any sound card will do.
- Some precision power resistors for impedance measurement
- Add in a Panasonic Electret microphone cartridge ($4 at Digikey)
- Add in some low cost software, AudioAnalyser if you want something easy to use, Speaker Workshop, if you want something that will allow you to do near professional work
2) Read, read, read, read, read.....
- Find out everything you can.
- Go to driver manufacturer's web sites and look at the schematics for cross-overs in their kits
- Read over projects on web-sites
- Read the Speaker Workshop tutorials, etc.
3) Start playing on paper. Speaker workshop will allow you to essentially test your finished speaker (at least the response), without ever touching a soldering iron. It is pretty darn accurate too! If you have access to spice, you can do some pretty good work too. (this is what I used initially)
4) Get as close as you can on paper, then build, test and listen.
I did some initial testing with simple cross-overs to confirm that my "assumptions" were correct. Then I worked my way towards more complex cross-overs.
While it is a nice thought to say you are going to use drivers that are "easy", the fact that they are easy to design with, does not mean they are good. Fast cars are not always easy to drive, but once you learn how to control them, they will do things that no other cars can do. I think you can do draw the same conclusions about drivers. The woofers in my speakers have a pretty nasty mass-limited break up peak around 4KHz. When that peak is not tamed (as in when I first bought them), they do not sound great. Once tamed, they sounded fantastic. A first order cross-over, or even a second order LR was not sufficient enough to reduce the peak. I ended up with a "modified" 4th order that is really a second order with a "notch" filter around the peak. I played around a lot on paper with both the woofer cross-over response and the tweeter response until I got the overlap I wanted and a conscious trade-off between keeping the cross-over frequency low so that system dispersion would be better, and not having the cross over frequency too low such that the tweeter is operating in a non-ideal range.
The resultant cross-over was very good and miles ahead of how the speakers came (second order LR on woofer\tweeter). I never modified the topology, but I made some minor component changes to shift the cross-over frequency slightly and to adjust the depth of the notch filter. The cross-over that I developed on the paper was more than acceptable. However, in fine DIY spirit, I modifed for perfection.
Good Luck
Alvaius!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.