If you use the term 'corrected' loosely, there's no absolute correct anyway. You can run it with the reduced impedance.. besides, the capacitor will bring it back up anyway. I'd still be careful of where in a circuit I was wasting power..
Anyway, the ideal electrical treatment for the peak is a series band-pass filter in parallel with the tweeter to augment the current flowing at these frequencies. They should be used normally, their use is better suited to a measured system but they can be estimated. And yes, if you are a couple of octaves above you can ignore the peak, but note that this is dependent on the peak and the application, it may be one octave it may be three..
You don't need to do this unless you are using passive components because the filter is not using the speaker as its load. Furthermore an amplifier usually has a low output impedance and so rejects the varying impedance on the speaker itself.
I will still compensate a speaker that uses external (eg active or passive line level) filtering, sometimes because it is being fed from an amp with a higher output impedance, and sometimes because it can make a minor improvement for amps in general to see a resistive load. Sometimes not, but it doesn't hurt. This would also mean compensating between the amp and a passive crossover if it might help.
I will still compensate a speaker that uses external (eg active or passive line level) filtering, sometimes because it is being fed from an amp with a higher output impedance, and sometimes because it can make a minor improvement for amps in general to see a resistive load. Sometimes not, but it doesn't hurt. This would also mean compensating between the amp and a passive crossover if it might help.
I was pointed to this thread by Scottmoose from a thread I had started seeking advice on what I had assumed would be a fairly straightforward crossover (LINK).
Thank you very much for a wonderful, digestible thread Allen. I can't claim to have taken it all in yet but I have "screen grabbed" your posts to make it easier to follow and will keep working on it.
You are a hero and even more importantly, a teacher!
Thank you very much for a wonderful, digestible thread Allen. I can't claim to have taken it all in yet but I have "screen grabbed" your posts to make it easier to follow and will keep working on it.
You are a hero and even more importantly, a teacher!
I'll preface this with a huge thanks to AllenB and all others that have contributed here.
Observation with a thinly veiled question
L-pads (2 resistor variety).
In a tweeter circuit, I've noticed that lowering the series resistor tends to increase the output as the frequency gets higher.
Question.
Would raising the parallel resistor, then raise the entire tweeter level uniformly ?
Next question.
In a mid drivers series notch filter (parallel to driver), if the peak is still to high after removal of the lcr resistor (thus its now an lc filter), obviously youd swap the inductor for one with lower dcr.
Is their a point at which you have to ask if theres something else at work creating the peak ?
Thanks...........
Observation with a thinly veiled question
L-pads (2 resistor variety).
In a tweeter circuit, I've noticed that lowering the series resistor tends to increase the output as the frequency gets higher.
Question.
Would raising the parallel resistor, then raise the entire tweeter level uniformly ?
Next question.
In a mid drivers series notch filter (parallel to driver), if the peak is still to high after removal of the lcr resistor (thus its now an lc filter), obviously youd swap the inductor for one with lower dcr.
Is their a point at which you have to ask if theres something else at work creating the peak ?
Thanks...........
It looks like the L-pad increases the high frequencies when you make the series resistor smaller. It increases the level and cuts the lows, as the capacitor sees a smaller load impedance.
Stray resistance would normally be seen as low when it is one tenth of the load. Damping in a series resonant circuit, I guess it depends. I don't like needing to adjust the inductor gauge for it. There should be at least some series impedance in this band before these, to make them effective. These filters can sometimes be replaced by a common second order filter, or combined with something similar to adjust level and phase, and series impedance. Otherwise there is the parallel bandstop filter in series with the driver
Stray resistance would normally be seen as low when it is one tenth of the load. Damping in a series resonant circuit, I guess it depends. I don't like needing to adjust the inductor gauge for it. There should be at least some series impedance in this band before these, to make them effective. These filters can sometimes be replaced by a common second order filter, or combined with something similar to adjust level and phase, and series impedance. Otherwise there is the parallel bandstop filter in series with the driver
Ok, so the relationship between the series and parallel resistor has an effect on the series cap, thus affects tuning frequency..
Yes, I should've seen that before, but was surprised at the amount of change from such a small resistor change..
notch. Mmm ok. more to think about.
Many thanks Allen
Yes, I should've seen that before, but was surprised at the amount of change from such a small resistor change..
notch. Mmm ok. more to think about.
Many thanks Allen
You're welcome. I do prefer the series kind of notch filter, but then I usually use them around cone breakup or tweeter resonance, both being where there is plenty of series crossover action at the same time to make them work well.
At a pinch you could try using a lower Q notch and make it up with a higher Q LP filter, for example, yes I've gone that way.
At a pinch you could try using a lower Q notch and make it up with a higher Q LP filter, for example, yes I've gone that way.
Yes, the speakers I'm working with (3 way horn) I'm trying to extract as much out of them as possible.
No small task considering my pitiful knowledge in the area of crossovers
My readings suggest series is going to be better as its, (obviously) not directly in the signal path.
with that, I've been working on 'fixing' a single 5db peak (3.5k) with a series filter.
down to just a cap and inductor now, and it would be nice to pull that last 2ish db down.
current inductor is 2.3 ohms dcr. hence why I was wondering if I swapped it for one with say half ohm was such a smart idea or a potential problem...
I'm putting this notch right at the mid driver, so the signal flow is, amp, crossover, l-pad, notch, driver.
I'm trying to isolate the the notch from everything as much as possible.
Hopefully I'm doing the right thing.....
My biggest problem is the speakers reside a loooong way away from me.
So the owner sends me a sweep, I try to work out what he can do with what he has, or mail him something, he does another sweep, and so on.
No small task considering my pitiful knowledge in the area of crossovers
My readings suggest series is going to be better as its, (obviously) not directly in the signal path.
with that, I've been working on 'fixing' a single 5db peak (3.5k) with a series filter.
down to just a cap and inductor now, and it would be nice to pull that last 2ish db down.
current inductor is 2.3 ohms dcr. hence why I was wondering if I swapped it for one with say half ohm was such a smart idea or a potential problem...
I'm putting this notch right at the mid driver, so the signal flow is, amp, crossover, l-pad, notch, driver.
I'm trying to isolate the the notch from everything as much as possible.
Hopefully I'm doing the right thing.....
My biggest problem is the speakers reside a loooong way away from me.
So the owner sends me a sweep, I try to work out what he can do with what he has, or mail him something, he does another sweep, and so on.
Yes, that could be a problem.
These can be less predictable with low resistance, however in the order of 2 ohms isn't unusual. I assumed your inductor would have a DCR in the order of an ohm. Yes, I'd probably try lower in this case.
If you have a simulator it might be worth a minute to look at the effect of impedance compensation for the driver between it and the notch, just in case. Otherwise you might achieve adjustment of the normal filters, in the case of a midrange that make the notch work better but might affect sensitivity by a dB or so. Maybe using different shaped filters would do it, whatever works.
These can be less predictable with low resistance, however in the order of 2 ohms isn't unusual. I assumed your inductor would have a DCR in the order of an ohm. Yes, I'd probably try lower in this case.
If you have a simulator it might be worth a minute to look at the effect of impedance compensation for the driver between it and the notch, just in case. Otherwise you might achieve adjustment of the normal filters, in the case of a midrange that make the notch work better but might affect sensitivity by a dB or so. Maybe using different shaped filters would do it, whatever works.
Thanks for your help and advise Allen.
Hes gotten a new horn for the tweeter that has had a massive smoothing effect on the tweeter and upper mid response.
Hes also trying physically moving the tweeter below the mid.
So until hes tried these and reported back, I can do no more.
Cant 'fix' something when the posts are moving..
But, none the less, its looking really promising...
Hes gotten a new horn for the tweeter that has had a massive smoothing effect on the tweeter and upper mid response.
Hes also trying physically moving the tweeter below the mid.
So until hes tried these and reported back, I can do no more.
Cant 'fix' something when the posts are moving..
But, none the less, its looking really promising...
3-4 kHz is a sensitive region. A good crossover shouldn't be a problem but no crossover is perfect so crossing away from this band can help. It just happens to be practical place to cross for many spekers. The drivers and the overall response will set that point.
Crossing high can reduce distortion produced by the tweeter. It can increase the power handling and keep the voice coil cooler. Ordinary harmonic distortion isn't always a problem with speakers, and crossing high has disadvantages.
The speaker should preferrably output sound that is overall as balanced as the sound that is just on the listening axis, but a woofer at it's top end will have a narrowing directivity that means when you balance the sound at the listening axis it won't be balanced overall relative to a dome tweeter.
Crossing high also increases the distance between the drivers relative to a wavelength at the crossover, increasing lobing which is where this distance results in cancellations above and below the listening axis.
Woofers should also avoid the breakup region where their response somewhat loses control.
Crossing high can reduce distortion produced by the tweeter. It can increase the power handling and keep the voice coil cooler. Ordinary harmonic distortion isn't always a problem with speakers, and crossing high has disadvantages.
The speaker should preferrably output sound that is overall as balanced as the sound that is just on the listening axis, but a woofer at it's top end will have a narrowing directivity that means when you balance the sound at the listening axis it won't be balanced overall relative to a dome tweeter.
Crossing high also increases the distance between the drivers relative to a wavelength at the crossover, increasing lobing which is where this distance results in cancellations above and below the listening axis.
Woofers should also avoid the breakup region where their response somewhat loses control.
Last edited:
Thanks for the response.
I think I could understand about disadvantages you mentioned.
I have a b&w 601 that has a xo point at 4khz. And I should spent a lot of time to find an on axis listening position to make it sound right.
My listening room is 5,5x6 meters. I realize it's too big for this speaker.
What I have in mind is, what if the drivers like tweeter and midrange have the capability to be crossed properly at 4khz. Would it be any advantage in sound character to do so? (beside increased power handling and tweeter distortion).
Are there any certain sound characters that you could explain when speaker crossed at 4khz and crossed below.
B&w seems to develop a paticular habit of cutting their xo at 4k with a 6" mid. Risking their on axis performance for some reason.
At stereophile, sony ss ar1 also reviewed to have less snap and sparkle because of the 5" paper driver with the same xo point.
Is there any specific sound character they after?
I think I could understand about disadvantages you mentioned.
I have a b&w 601 that has a xo point at 4khz. And I should spent a lot of time to find an on axis listening position to make it sound right.
My listening room is 5,5x6 meters. I realize it's too big for this speaker.
What I have in mind is, what if the drivers like tweeter and midrange have the capability to be crossed properly at 4khz. Would it be any advantage in sound character to do so? (beside increased power handling and tweeter distortion).
Are there any certain sound characters that you could explain when speaker crossed at 4khz and crossed below.
B&w seems to develop a paticular habit of cutting their xo at 4k with a 6" mid. Risking their on axis performance for some reason.
At stereophile, sony ss ar1 also reviewed to have less snap and sparkle because of the 5" paper driver with the same xo point.
Is there any specific sound character they after?