I have put the TBs in 2 l sealed satellites. The sub is a an old 17 cm (7 inch) Eton Hexacone in a 15 l sealed box.
I wanted to cross over pretty highly and steeply to keep excursion stress from the TBs, maybe 250 Hz LR4.
How well do the things blend with an XO that high and steep?
Now, from the driver data, I get an acoustic hibh pass with Q=0.91 and fB=144.4 Hz, which is pretty close to the indended XO.
Now I could:
- keep the LR4 at 250 Hz and ignore the acoustic HP, after all it is almost an octave away
- move the box resonance down with a Linkwitz transform filter and keep the LR4 at 250 Hz
- put an LR2 at 250 Hz and somehow design a symmetrical 4th order filter for the sub??
- put an active 2nd order HP with Q=? at 144 Hz and build a symmetrical 4th order LP for the sub
What would work best?
I wanted to cross over pretty highly and steeply to keep excursion stress from the TBs, maybe 250 Hz LR4.
How well do the things blend with an XO that high and steep?
Now, from the driver data, I get an acoustic hibh pass with Q=0.91 and fB=144.4 Hz, which is pretty close to the indended XO.
Now I could:
- keep the LR4 at 250 Hz and ignore the acoustic HP, after all it is almost an octave away
- move the box resonance down with a Linkwitz transform filter and keep the LR4 at 250 Hz
- put an LR2 at 250 Hz and somehow design a symmetrical 4th order filter for the sub??
- put an active 2nd order HP with Q=? at 144 Hz and build a symmetrical 4th order LP for the sub
What would work best?
That sounds like a very good idea, thanks!
I've had another thought. In a satellite/sub configuration, I am probably not too concerned about the orientation of the central lobe, so why bother with an LR4.
So I could do it like the standard Manger XO, i.e. use only a first order electrical filter to obtain a third order response. I could place it at 144 Hz or (like Manger) a little higher than the acoustic rolloff to iron out the 1 dB hump caused by the high Q of the TBs. The sub would then have a standard third order filter (maybe even Q=0.7).
Blending would probably be better with lower XO point and flatter slopes.
The TB has +/- 0.5 mm linear excursion, so at 150 Hz and 32 cm² cone area this corresponds to about 85 dB. Well, I have a left and a right speaker and a sub, so I should add 18 dB, right? So 103 dB at the XO point seems ample enough...
I've had another thought. In a satellite/sub configuration, I am probably not too concerned about the orientation of the central lobe, so why bother with an LR4.
So I could do it like the standard Manger XO, i.e. use only a first order electrical filter to obtain a third order response. I could place it at 144 Hz or (like Manger) a little higher than the acoustic rolloff to iron out the 1 dB hump caused by the high Q of the TBs. The sub would then have a standard third order filter (maybe even Q=0.7).
Blending would probably be better with lower XO point and flatter slopes.
The TB has +/- 0.5 mm linear excursion, so at 150 Hz and 32 cm² cone area this corresponds to about 85 dB. Well, I have a left and a right speaker and a sub, so I should add 18 dB, right? So 103 dB at the XO point seems ample enough...
Hi Eric
The Manger crossover doesn't seem to be a bad idea either. When I was playing around with my MSWs I recognized that it allowed me (refering to the MSW of course which doesn't behave very well around it's fundamental resonance) to:
- use the lowest (acoustical) crossover frequency while
- getting the flattest frequency response
(compared to the asymmetrical subtractive crossover that I also tried).
It is quite tolerant to deviations of component values and it works even with a 2nd order lowpass. It was in fact quite difficult to get a really bad frequency respoinse using this topology ! It is however not fully phase-accurate due to the necessary phase inversion of the woofer but is still better than >95% of all speakers in this respect.
The LR4 otoh would be better regarding IMD etc. Maybe you should also try your fourth suggesstion starting with a 2nd order Butterworth highpass at 144 Hz ?
Regards
Charles
The Manger crossover doesn't seem to be a bad idea either. When I was playing around with my MSWs I recognized that it allowed me (refering to the MSW of course which doesn't behave very well around it's fundamental resonance) to:
- use the lowest (acoustical) crossover frequency while
- getting the flattest frequency response
(compared to the asymmetrical subtractive crossover that I also tried).
It is quite tolerant to deviations of component values and it works even with a 2nd order lowpass. It was in fact quite difficult to get a really bad frequency respoinse using this topology ! It is however not fully phase-accurate due to the necessary phase inversion of the woofer but is still better than >95% of all speakers in this respect.
The LR4 otoh would be better regarding IMD etc. Maybe you should also try your fourth suggesstion starting with a 2nd order Butterworth highpass at 144 Hz ?
Regards
Charles
When you say it was very insensitive to changes, how far could you move the pole of the 1st order HP?
Interesting to hear it worked with 4th order, too. Did you do that with the 3rd order LP? Where did you place the second pole? Was it a second RC filter or did you use a second order active filter instead? If so, what Q?
I had thought about using another pole, too, but placing at maybe 80 Hz where I believe it does little to screw up overall frequency response but helps to bring down excursion.
Cheers,
Eric
Interesting to hear it worked with 4th order, too. Did you do that with the 3rd order LP? Where did you place the second pole? Was it a second RC filter or did you use a second order active filter instead? If so, what Q?
I had thought about using another pole, too, but placing at maybe 80 Hz where I believe it does little to screw up overall frequency response but helps to bring down excursion.
Cheers,
Eric
Hi Eric
I didn't play around with the HP pole frequency. What I was fiddling around with was the lowpass cutoff frequency.
Furthermore I said that it worked with a second order LOWpass as well (instead of a 3rd order one) not a second order highpass nor any kind of 4th order filter.
I would try the second first order highpass pole at 80 Hz which could improve distortion figures - but don't be disappointed if it doesn't work (at 144 Hz it would already show a phaseshift of 30 degrees).
Regards
Charles
I didn't play around with the HP pole frequency. What I was fiddling around with was the lowpass cutoff frequency.
Furthermore I said that it worked with a second order LOWpass as well (instead of a 3rd order one) not a second order highpass nor any kind of 4th order filter.
I would try the second first order highpass pole at 80 Hz which could improve distortion figures - but don't be disappointed if it doesn't work (at 144 Hz it would already show a phaseshift of 30 degrees).
Regards
Charles
Sub is still WIP, but connected satellites to my amp yesterday. Sound was somewhat distorted at medium volume. Adding two 470 µF in series, i.e. a first order electrical XO at 112 Hz, solved that problem. I can listen pretty loudly, clarity and imaging is amazing, and there are surprising amounts of bass. Only when things get very dynamic, i.e. on Strange little girls by Tori Amos do these things sound challenged at higher volume.
I think I will try a first order active filter at 180-200 Hz, maybe supplemented by anther rolloff at 50-80 Hz which I can conveniently implement by using a smaller cap in the feedback network of the LM4766.
Never thought these simple things can sound so nice. For a small room, they are definitely enough.
I think I will try a first order active filter at 180-200 Hz, maybe supplemented by anther rolloff at 50-80 Hz which I can conveniently implement by using a smaller cap in the feedback network of the LM4766.
Never thought these simple things can sound so nice. For a small room, they are definitely enough.
Project has been dormant for a while, but now I have finally gotten my new soundcard to like me.
F_b of the finished speakers is 155 Hz, Q_t is 0.972. I would like to fiddle a while to simulate the combined response of simulated/measured response and an extra active filter, but maybe there are rules of thumb?
- First of all, at 150 Hz, both woofer and satellite would be pretty omnidirectional, so a Butterworth might be a better idea than a LR??
- I wanted to use a first order filter to generate an total third order response that approximates Q=.7 , at what frequency would I place it? If I placed it at 150 Hz, as a first order filter has Q=0.5 and 0.5 x 0.97, I couldn't achieve .7.
- The best idea might be to use an acoustic fourth order low pass. As the closed box already gives me Q=.972, I could do this by adding a second order Sallen-Key filter with Q=0.5 (for LR4) or .7 (for Butterworth). I gather there isn't any way to save an op amp stage and use two decoupled first order filters, as this would automatically put me at Q=0.25?
Regards,
Eric
F_b of the finished speakers is 155 Hz, Q_t is 0.972. I would like to fiddle a while to simulate the combined response of simulated/measured response and an extra active filter, but maybe there are rules of thumb?
- First of all, at 150 Hz, both woofer and satellite would be pretty omnidirectional, so a Butterworth might be a better idea than a LR??
- I wanted to use a first order filter to generate an total third order response that approximates Q=.7 , at what frequency would I place it? If I placed it at 150 Hz, as a first order filter has Q=0.5 and 0.5 x 0.97, I couldn't achieve .7.
- The best idea might be to use an acoustic fourth order low pass. As the closed box already gives me Q=.972, I could do this by adding a second order Sallen-Key filter with Q=0.5 (for LR4) or .7 (for Butterworth). I gather there isn't any way to save an op amp stage and use two decoupled first order filters, as this would automatically put me at Q=0.25?
Regards,
Eric
Hi Eric
First order filters don't have a Q.
If I were you I would try out a third-orer filter by adding a first order highpass whose f3 is between 140 Hz and 150 Hz.
If you have two decoupled first-order filters with the same f3 then you don't have a Q of 0.25 but 0.5 actually.
Regards
Charles
First order filters don't have a Q.
If I were you I would try out a third-orer filter by adding a first order highpass whose f3 is between 140 Hz and 150 Hz.
If you have two decoupled first-order filters with the same f3 then you don't have a Q of 0.25 but 0.5 actually.
Regards
Charles
Hi Charles,
this was the information I was looking for but didn't have handy! I should really go back and dig out a textbook on filter theory, as the loudspeaker books only give cookbook recipes...
So then using a first order filter on top of the acoustic rolloff would leave me at 0.97? Or is the Q definiton only valid for even order filters, as the others do not form a resonant circuit?
Putting a second order active filter on the acoustic rolloff, I assume I still get to multiply the Qs?
Regards,
Eric
this was the information I was looking for but didn't have handy! I should really go back and dig out a textbook on filter theory, as the loudspeaker books only give cookbook recipes...
So then using a first order filter on top of the acoustic rolloff would leave me at 0.97? Or is the Q definiton only valid for even order filters, as the others do not form a resonant circuit?
Putting a second order active filter on the acoustic rolloff, I assume I still get to multiply the Qs?
Regards,
Eric
I just consulted a filter table and it said that for a 3rd order butterworth filter you'd need a first order and a 2nd order filter of the same pole frequency with the 2nd order part having a Q of 1.
If you take a 150 Hz 1st order highpass you will get quite close to a 150 Hz Butterworth 3rd order highpass in total.
Q values are used for the 2ND ORDER SECTIONS of even order filters.
You cannot just multiply the Q values of the sections to get a desired filter characteristic. Determination of filter parameters is not trivial and is best done via a filter simulation program or filter tables.
Regards
Charles
If you take a 150 Hz 1st order highpass you will get quite close to a 150 Hz Butterworth 3rd order highpass in total.
Q values are used for the 2ND ORDER SECTIONS of even order filters.
You cannot just multiply the Q values of the sections to get a desired filter characteristic. Determination of filter parameters is not trivial and is best done via a filter simulation program or filter tables.
Regards
Charles
Thanks for bothering with this basic stuff!
So it is a coincidence that the two active filters that make up a LR4 have 0.7 each, giving .5 when multiplied?
What about my reasoning about Butterworth vs. LR at a low frequency, where both drivers would be omnidirectional? LR gives me on-axis constant SPL. As on-axis is also equal to off-axis in this case, then - contrary to what I assumed above - LR would give me the most even response?
Regards,
Eric
So it is a coincidence that the two active filters that make up a LR4 have 0.7 each, giving .5 when multiplied?
What about my reasoning about Butterworth vs. LR at a low frequency, where both drivers would be omnidirectional? LR gives me on-axis constant SPL. As on-axis is also equal to off-axis in this case, then - contrary to what I assumed above - LR would give me the most even response?
Regards,
Eric
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.