I am designing a satellite/sub combo using a pair of Peerless HDS 164 drivers that I have. Using a 8.7 L box gives a Qtc of 0.707 and an F3 of 76.96 Hz. I wanted to crossover the speakers to a sub at less than 120Hz for imaging reasons, but would prefer this frequency to be nearer 80Hz. However, I am am concerned the F3 will then be too close to the XOver frequency... can anyone advise?
Cheers,
chris
Cheers,
chris
your satellites will roll off at 80Hz with a classic Butterworth
second order roll off.
Further electrical crossover is difficult to achieve passively,
the box resonance thoroughly interfering with the filter.
for 120Hz you'll need an active arrangement, or just
use the bass management feature in an AV amplifier.
sreten.
second order roll off.
Further electrical crossover is difficult to achieve passively,
the box resonance thoroughly interfering with the filter.
for 120Hz you'll need an active arrangement, or just
use the bass management feature in an AV amplifier.
sreten.
Passive is exaclty sreten said it creates problems.
I just ran a few things thru LSPcad and came up with a passive network you could use however it would cost, due to large caps and inductors.
With just a simple 2nd order added you get the target 24db roll off but high Q so you get a substantial peak prior to this. So to compensate I added a series notch filter (L+C+R).
This worked like a charm and you get the desired response, 24db roll off linkwitz riley topology at 80hz
Xover as follows
I just ran a few things thru LSPcad and came up with a passive network you could use however it would cost, due to large caps and inductors.
With just a simple 2nd order added you get the target 24db roll off but high Q so you get a substantial peak prior to this. So to compensate I added a series notch filter (L+C+R).
This worked like a charm and you get the desired response, 24db roll off linkwitz riley topology at 80hz
Xover as follows
Attachments
Don't worry, you've got yorself a theoreticaly perfect cabinet tuning for the peerless. When the speaker/cabinet has a rolloff at 77Hz with a Q of 0.707 just add an electronic filter in front off it with also a rolloff at 77Hz with a Q of 0.707 and you will get an acoustic rollof at 77Hz with a Q of 0.5
In Other words, you've created a second order Butterworth at 77Hz (Q=0.707) So you have to add a second order butterworth filter (Q=0.707) And you will get a perfect fourth order (24dB/oct)Linkwitz Crossover (Q=0.5)
Q= 0.707 * 0.707 = 0.5, just multiply the Q factors off speaker and crossover, that's some basics about filtering everyone should know, but lots of people don't.
This can also be done passivly but it's a lot more difficult. Firts you have to compensate the impedance peak of the driver's resonance and second it requires big capacitance and inductance to do this.
In Other words, you've created a second order Butterworth at 77Hz (Q=0.707) So you have to add a second order butterworth filter (Q=0.707) And you will get a perfect fourth order (24dB/oct)Linkwitz Crossover (Q=0.5)
Q= 0.707 * 0.707 = 0.5, just multiply the Q factors off speaker and crossover, that's some basics about filtering everyone should know, but lots of people don't.
This can also be done passivly but it's a lot more difficult. Firts you have to compensate the impedance peak of the driver's resonance and second it requires big capacitance and inductance to do this.
I think you guys are missing the point of the original post.
What Chris is saying is that he is concerned (rightfully so) that the second order HP slope due to box/speaker combination when combined with his 4th order LR active crossover will result in a 6th order HP and 4th order LP at 80Hz and therefore sum improperly causing a dip in response.
This might indeed be the case, or not. The system would have to be modelled including proper distances between drivers. For a quick setup my advice would be to let the midbass run full range (unfiltered on the bottom) and find a best complementary frequency for the sub crossover.
I simulated a 2nd order HP with Q of 0.707 at 80Hz and found a best 4th order LR filter for a driver that is positioned 1 meter below and 0.5 meter behind the midbass. It turned out to be at 74Hz with inverted polarity and -1dB gain. The resulting ripple is about 1 dB and is most likely to be swamped by the room modes. Again this is only a simulation.
In real world tuning the sub by ear will most likely result in something different. This would be a good starting point though.
See the image
What Chris is saying is that he is concerned (rightfully so) that the second order HP slope due to box/speaker combination when combined with his 4th order LR active crossover will result in a 6th order HP and 4th order LP at 80Hz and therefore sum improperly causing a dip in response.
This might indeed be the case, or not. The system would have to be modelled including proper distances between drivers. For a quick setup my advice would be to let the midbass run full range (unfiltered on the bottom) and find a best complementary frequency for the sub crossover.
I simulated a 2nd order HP with Q of 0.707 at 80Hz and found a best 4th order LR filter for a driver that is positioned 1 meter below and 0.5 meter behind the midbass. It turned out to be at 74Hz with inverted polarity and -1dB gain. The resulting ripple is about 1 dB and is most likely to be swamped by the room modes. Again this is only a simulation.
In real world tuning the sub by ear will most likely result in something different. This would be a good starting point though.
See the image
An externally hosted image should be here but it was not working when we last tested it.
And here is one with your actual HP at 77Hz.
The LP is still a LR4 at 74Hz, inverted polarity, -1.8dB gain this time. Same position of 1 meter below and .5 meter behind the midbas.
The LP is still a LR4 at 74Hz, inverted polarity, -1.8dB gain this time. Same position of 1 meter below and .5 meter behind the midbas.
An externally hosted image should be here but it was not working when we last tested it.
Sorry, mist that point indeed.
That's exactly the reason why a of the shelve active crossover almost nevers works properly. It's the same as bying a allready soldered textboox passive crossover, everyone knows that aint right, so why expect wonders from a textbook active crossover ?
I have tried it that way, many times, but in the end it simply didn't work out.
moving the midspeaker 0.5 behind the woofer only compensates for a wrong behaviour in phase and amplitude response for a very small bandwith. The impulse response is still totally wrong. I'm sorry. Please think about that before spending any money on it.
That's exactly the reason why a of the shelve active crossover almost nevers works properly. It's the same as bying a allready soldered textboox passive crossover, everyone knows that aint right, so why expect wonders from a textbook active crossover ?
I have tried it that way, many times, but in the end it simply didn't work out.
moving the midspeaker 0.5 behind the woofer only compensates for a wrong behaviour in phase and amplitude response for a very small bandwith. The impulse response is still totally wrong. I'm sorry. Please think about that before spending any money on it.
Sjef said:
Yes, a textbook electrical crossover seldom works properly, especially without the help of parametric EQs and delay lines. Even then it takes a skilled operator to dial in a proper setting. That's why I'm not crazy about a lot of the DSP crossovers that are so popular these days. They are generally a good thing but not the ultimate tool by far.
I moved the sub 0.5m back in the simulation, not the other way around. That's how most people have theirs set up. Sub against the back wall, the satellites maybe half a meter in to the room. It is the reason why I had to invert polarity on the sub in the simulation.
Impulse response is wrong with any high order crossover, except for a few particular analog crossover filters and some digital filters (mostly of the FIR variety). If you are interested in pursuing transient perfect filters do a search for John Kreskovsky on the Web. He has a few papers on his site dealing with the subject.
birkinshawc said:
how adjustable is this crossover ? What points ?
Can the low pass and high pass have different frequencies ?
(i.e. can you have an offset crossover)
Is Q of the 24/dB slopes adjustable, Butterworth or Linkwitz/Riley ?
Can the slope be changed to 12dB per octave ?
Do you have only one subwoofer ?
The satellites have not been built yet ?
sreten.Isn't it the intrinsic way filters work, to filter and so to adjust the step response? either active or passive....
The symmetrical 1x 12dB butterworth HP filter and the 24 dB LR at 77 Hz should work fine for you.... and of course with the displacement of the sub you have to adjust for time (difficult with analog, easy with digital) and in your analog case phase reverse should work fine.
About the impulse behaviour:
To me the impulse response is about phase lag around the crossover. Theoretically, around 77Hz, top phase will be +180 degrees, and sub -180..... So acoustically they are summing 0 degrees (theory of a symmetrical filter, phases are compensating each other) so the filter impulse response problem is gone, as long as both drivers are active for another octave above/under their cross, wich with the top off course is not true....
But phase reversing the sub with the time offset is actually about choosing wich part of the low frequencies you want to acoustically sum better: under the 77Hz crossover point or above that point. But it is true that this gives an impulse response disturbance, because of the above mentioned. You will get a "phase step" however its doubtful (read: not audible) if you can hear this. The real problem of course is the time offset between sub/top.
To me the impulse response of a good designed/crossed loudspeaker is mainly disturbed by the driver(s) itself (their mass/stiffness etc), and the room reflections.
FIR filters are correcting for time (phase), so this makes it possible to create crossovers with for instance overlap.... or make a highpass without the groupdelay, but this means that the whole filter has a longer delay to compensate, and so all frequencies come out later.
To me the "tekstbook" active crossovers (IIR filters) work fine, I don't know how they could not except for making mistakes yourself... Active crossing needs less EQ than passive because you can always change driver levels to start with.
Active crossing is the way to go, with the benefit to change driver levels, easy configuring of the crosspoints, no driver impedance problems, higher voltage effciency, because no passive components inbetween. But thats really off topic....
The symmetrical 1x 12dB butterworth HP filter and the 24 dB LR at 77 Hz should work fine for you.... and of course with the displacement of the sub you have to adjust for time (difficult with analog, easy with digital) and in your analog case phase reverse should work fine.
About the impulse behaviour:
To me the impulse response is about phase lag around the crossover. Theoretically, around 77Hz, top phase will be +180 degrees, and sub -180..... So acoustically they are summing 0 degrees (theory of a symmetrical filter, phases are compensating each other) so the filter impulse response problem is gone, as long as both drivers are active for another octave above/under their cross, wich with the top off course is not true....
But phase reversing the sub with the time offset is actually about choosing wich part of the low frequencies you want to acoustically sum better: under the 77Hz crossover point or above that point. But it is true that this gives an impulse response disturbance, because of the above mentioned. You will get a "phase step" however its doubtful (read: not audible) if you can hear this. The real problem of course is the time offset between sub/top.
To me the impulse response of a good designed/crossed loudspeaker is mainly disturbed by the driver(s) itself (their mass/stiffness etc), and the room reflections.
FIR filters are correcting for time (phase), so this makes it possible to create crossovers with for instance overlap.... or make a highpass without the groupdelay, but this means that the whole filter has a longer delay to compensate, and so all frequencies come out later.
To me the "tekstbook" active crossovers (IIR filters) work fine, I don't know how they could not except for making mistakes yourself... Active crossing needs less EQ than passive because you can always change driver levels to start with.
Active crossing is the way to go, with the benefit to change driver levels, easy configuring of the crosspoints, no driver impedance problems, higher voltage effciency, because no passive components inbetween. But thats really off topic....
The crossover unit only has adjustment of the middle frequency of the xover, and is a 24db LR type. I have the drivers but have not built the boxes yet.
How much above the F3 should an electronic crossover be set so that it does not combine with the effects of the acoustical high pass as the box bass response rolls off?
How much above the F3 should an electronic crossover be set so that it does not combine with the effects of the acoustical high pass as the box bass response rolls off?
birkinshawc said:
For a really good match you should be well over an octave above the acoustical HP.
When designing crossovers the rule of thumb is to keep the slopes on target in a 30dB window. You have a 24dB/octave electronic crossover. Your speaker is flat to about 100Hz. That would dictate a crossover point of over 200Hz. Not a good idea for a subwoofer.
If you insist on applying the HP to your satelites start with a low crossover point, 50-60Hz, and play with the subwoofer gain and phase. You should be able to arrive at a good sounding system by ear. The effects of the room modes will be in the same ballpark as the error due to mismatched filters. You'll just have to experiment.
Thunau said:
What if I were to turn off the sub output on the xover I have, using it only for the mid/hi xover, and then construct a separate active low pass filter for the subwoofer? I presume I'd use the 12db/oct high pass roll off of the mid in a sealed box and a 12db/oct low pass filter on the sub to do this?
My knowledge in this area is not great so if anyone could clarify this then I'd appreciate, ie. what measurements to make, if this will work etc.
Cheers for the help so far!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.