There I was last night, playing with XoverSim and getting to an acceptable result. I had a pretty flat SPL response and nice looking impedance curve.
Then I thought, "let's simulate a xover from a good commercial enclosure using these drivers". Shock and horror. I wasted many hours tweaking a xover that was getting nowhere. The SPL response was way flatter in the commercial xover, but there's an pretty scary impedance rise around 5 KHz.
Is it me or even serious manufacturers don't focus too much on impedance correction?
After seeing how difficult it is to obtain an acceptable SPL response, I don't even dare to think about flattening the impedance curve.
Now I'm wondering, is it worth the effort, flattening the impedance?
How can I calculate how much attenuation I will have based on amp parameters and impedance?
Then I thought, "let's simulate a xover from a good commercial enclosure using these drivers". Shock and horror. I wasted many hours tweaking a xover that was getting nowhere. The SPL response was way flatter in the commercial xover, but there's an pretty scary impedance rise around 5 KHz.
Is it me or even serious manufacturers don't focus too much on impedance correction?
After seeing how difficult it is to obtain an acceptable SPL response, I don't even dare to think about flattening the impedance curve.
Now I'm wondering, is it worth the effort, flattening the impedance?
How can I calculate how much attenuation I will have based on amp parameters and impedance?
Manufacturers are very pragmatic when it comes to using parts that add to expense but don't necessarily help the sound, so instead of automatically using impedance correction they tend to be sure it is required. Amateurs tend to be more goal and less cost oriented when it comes to the search for perfection and will often use impedance correction before making sure it is either required or desired.
Also beware of crossover sims. They are quite useful for deciding on topology and potential component value ranges but, like box programs, are not 100% accurate; there is no substitute for measuring the finished product and tweaking it out.
Also beware of crossover sims. They are quite useful for deciding on topology and potential component value ranges but, like box programs, are not 100% accurate; there is no substitute for measuring the finished product and tweaking it out.
Impedance compensation is very important. It makes the amplifier's life easier since it sees a resistive load above 300Hz. This means lack of phase shift between voltage and current and the reproduction can be loud and clear up to maximum level without signs of compression.
Its not that difficult to implement, it just needs some trial and error after making your calculations with your favorite software. Crossover simulators are not that accurate in practice, even if its just about a zobel. Start with software simulation but dont count on it, measuring and fine tuning must follow but its quite easy if you find the trick.
Its not that difficult to implement, it just needs some trial and error after making your calculations with your favorite software. Crossover simulators are not that accurate in practice, even if its just about a zobel. Start with software simulation but dont count on it, measuring and fine tuning must follow but its quite easy if you find the trick.
Just to make myself clearer, I attached pictures of my xover + resulting SPL & impedance graphs, and the ones of a xover recommended by Visaton (after all I found one for my combination of drivers).
The pictures speak for themselves, and I hope you understand my dilemma now, so to speak.
The Visaton filter seems to be a lower Q one (forgive me if I'm making some confusions here and please correct me if so) and it flattens the SPL response an octave or so lower by attenuating it. My LP filter is a 3rd order Linkwitz alignment and it sure doesn't flatten the response that much, like it should, but it has a great looking impedance curve. But flat impedance is of no good if there's that early roll-off beginning at approx 1KHz, I think I won't have any bass!
Please tell me, how can I get rid of that impedance rise in the Visaton xover? I tried to add a zobel to it but it doesn't seem to do any good, in fact it gets worse.
The drivers are these:
woofer: http://www.visaton.com/english/artikel/art_457_1_19.html
tweeter: http://www.visaton.com/english/artikel/art_266_1_8.html
My xover:
Graphs for my xover:
Xover suggested by Visaton:
Graphs for my the Visaton-suggested xover:
The pictures speak for themselves, and I hope you understand my dilemma now, so to speak.
The Visaton filter seems to be a lower Q one (forgive me if I'm making some confusions here and please correct me if so) and it flattens the SPL response an octave or so lower by attenuating it. My LP filter is a 3rd order Linkwitz alignment and it sure doesn't flatten the response that much, like it should, but it has a great looking impedance curve. But flat impedance is of no good if there's that early roll-off beginning at approx 1KHz, I think I won't have any bass!
Please tell me, how can I get rid of that impedance rise in the Visaton xover? I tried to add a zobel to it but it doesn't seem to do any good, in fact it gets worse.
The drivers are these:
woofer: http://www.visaton.com/english/artikel/art_457_1_19.html
tweeter: http://www.visaton.com/english/artikel/art_266_1_8.html
My xover:
An externally hosted image should be here but it was not working when we last tested it.
Graphs for my xover:
An externally hosted image should be here but it was not working when we last tested it.
Xover suggested by Visaton:
An externally hosted image should be here but it was not working when we last tested it.
Graphs for my the Visaton-suggested xover:
An externally hosted image should be here but it was not working when we last tested it.
from here it looks like your woofer zoebel did a nice job of flattening the upper impedance rise and smoothing out the response in the crossover region.
As for the low end, are these simulations or your own measurements? Did you splice a nearfield measurement onto the farfield? what is the baffle width/enclosure alignment? It looks like you may either be tuned too low for that driver or in need of some baffle step compensation, or both.
As for the low end, are these simulations or your own measurements? Did you splice a nearfield measurement onto the farfield? what is the baffle width/enclosure alignment? It looks like you may either be tuned too low for that driver or in need of some baffle step compensation, or both.
mr_push_pull said:
The impedance peak can easily be corrected by a LCR network parallel to the inputs terminals of the total crossover, therefore correcting the system impedance of both drivers together. So for example the first parallel leg is the LCR-network, the second one is the woofer low-pass and the third parallel leg is the tweeter high-pass.
Nearly all two-way speakers have an impedance peak just below fx, it is more a coincidence that the first Visaton crossover doesn't show a peak.
I see two problems with your crossover apart from the the impedance problems. First of there is no BSC in your LP section. You should use bigger inductor instead of the 0,33mH, like in the Visaton one, so about 1mH. The capacitor then should be about 6,8uF to keep you XO point.
The second problem is with the HP filter. I think you should use a 3rd order filter as well, because of phase problems. Not necessarily, but IMO it gives better integration with the LP.
Try first modify the LP filter.
Zozo
The second problem is with the HP filter. I think you should use a 3rd order filter as well, because of phase problems. Not necessarily, but IMO it gives better integration with the LP.
Try first modify the LP filter.
Zozo
BobEllis said:
Those sims use the published data. I have been assured and reassured that it is trustworthy with Visaton as they have a very strict measurement procedure. Now I don't even have the drivers, my order should arrive each moment (I hope).
I once stumbled accross a page on the Visaton website where they describe the measurement procedure, but I can't seem to find it now, unfortunately. The data is far-field free air response in infinite baffle (this is what I remember from that page).
Zozo said:
I used 3rd order for LP because this way the impedance is flatter than with my 2nd order attempts. It has something to do with the second series inductor I think. Anyway, I'll try using the suggestion with the series RLC for the whole xover with 2nd order LP (the Visaton XO).
Visaton measurement procedure:
http://translate.google.com/transla...&hl=en&ie=UTF-8&oe=UTF-8&prev=/language_tools
(Google translation)
...and for those that know German:
http://www.visaton.de/vb/showthread.php?s=&threadid=5136
http://translate.google.com/transla...&hl=en&ie=UTF-8&oe=UTF-8&prev=/language_tools
(Google translation)
...and for those that know German:
http://www.visaton.de/vb/showthread.php?s=&threadid=5136
The only reason xover simulators are not accurate is if you fail to characterise correctly the components you are simulating or the impedance of the driver you are using!
This involves understanding the loss resistances of the components and whether the values are frequency dependant and voltage dependant.
With just minimal extra complication in the network simulation it is possible to get agreement between simulation and measurement better than 1dB first go around.
I understand that most people will not have the capability of characterising the components accurately or to know what a sufficiently accurate circuit representation of the component losses should be, but this is no reason to blame the software!
Andrew
This involves understanding the loss resistances of the components and whether the values are frequency dependant and voltage dependant.
With just minimal extra complication in the network simulation it is possible to get agreement between simulation and measurement better than 1dB first go around.
I understand that most people will not have the capability of characterising the components accurately or to know what a sufficiently accurate circuit representation of the component losses should be, but this is no reason to blame the software!
Andrew
AndrewJ said:
Andrew:
Doesn't speaker placement on the baffle board make a big difference in on axis response?
For instance, if the centers of mid and tweeter are one half wavelength apart at the crossover point, (4.5 inches at 1500 Hz), won't the response be different if we use the same crossover components and place the centers 9 inches apart, (one wavelength)?
I cannot help but notice that many manufacturers place the tweeter and midrange centers one wavelength apart at the crossover frequency, even if they could fit them slightly closer together.
I believe I have read somewhere that when two speakers are placed 1/2 wavelength apart on a baffle board, there is cancellation at that frequency, but when placed one wavelength apart, there is reinforcement.
True?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.