Hi, i'm having some doubts on what kind of filter to use for audio and on if i made them right.
I need those filters in a preamplifier.
I designed three band pass filters:
1)30 - 125Hz (Subwoofer)
2)125 - 4000Hz (Midrange)
3)4000 - 20000Hz (Tweeter)
I designed them to be butterworth filters, but are they actually good for audio applications or they introduce too much distortion?
Every stage of the BPF is actually a 4th order filter. A 4th order HPF connected to a 4th order LPF.
For example, the first 30 - 125Hz BPF is made out of a 4th order 30Hz butterworth HPF connected to a 4th order 125Hz butterworth LPF.
Also, i decided to use a buffer OpAmp between the HPF and the LPF.
Is it going to make a difference or it's useless?
Is this going to work or it is better to change topology?
Here i posted a schematic that i found on the internet. It's 4th order butterworth LPF.
Imagine my design as one of those connected to another one of those but with the resistors and capacitors swapped (HPF). (Of course the two resistors for setting the gain are always in the same spot).
So, by swapping the resistors with the capacitors and the capacitors with the resistors, will it become a HIGH PASS filter with the same cutoff frequency formula, or the formula for calculating the cutoff frequency changes?
I mean, the formula for calculating the cutoff frequency for both LPF and HPF is the same?
I used this formula for designing both HPF and LPF:
Fc = 1 / (2 * 3.14 * R * C).
This is because the two filter resistors are the same value and same for the two capacitors. 3.14 of course is pi, i dont know how to type it.
Is this the right formula?
I also followed the rule that the first stage of a fourth order butterworth filter has to have a gain of around 0.152 and the second stage has to have a gain of around 1.234.
So both combined will have a gain of around 2.57.
This is only for a single LPF / HPF 4th order stage though.
So i think the total gain of a complete BPF is 2.57 * 2. (Because it's 4th order HPF + 4th order LPF combined).
Is this right?
For the best accuracy i'm going to use ±0.1% resistors and all capacitors are polypropilene for best audio performance.
But before placing the order, i wanted to ask...
Did I make some mistakes or those filters are going to work fine?
I'm basically a newbie in this kind of stuff, and this is a personal project.
I'm sorry if i made a lot of mistakes that might be ovbious to you.
If you need more information than what i provided for now, no problem at all, i will provide them.
I need those filters in a preamplifier.
I designed three band pass filters:
1)30 - 125Hz (Subwoofer)
2)125 - 4000Hz (Midrange)
3)4000 - 20000Hz (Tweeter)
I designed them to be butterworth filters, but are they actually good for audio applications or they introduce too much distortion?
Every stage of the BPF is actually a 4th order filter. A 4th order HPF connected to a 4th order LPF.
For example, the first 30 - 125Hz BPF is made out of a 4th order 30Hz butterworth HPF connected to a 4th order 125Hz butterworth LPF.
Also, i decided to use a buffer OpAmp between the HPF and the LPF.
Is it going to make a difference or it's useless?
Is this going to work or it is better to change topology?
Here i posted a schematic that i found on the internet. It's 4th order butterworth LPF.
Imagine my design as one of those connected to another one of those but with the resistors and capacitors swapped (HPF). (Of course the two resistors for setting the gain are always in the same spot).
So, by swapping the resistors with the capacitors and the capacitors with the resistors, will it become a HIGH PASS filter with the same cutoff frequency formula, or the formula for calculating the cutoff frequency changes?
I mean, the formula for calculating the cutoff frequency for both LPF and HPF is the same?
I used this formula for designing both HPF and LPF:
Fc = 1 / (2 * 3.14 * R * C).
This is because the two filter resistors are the same value and same for the two capacitors. 3.14 of course is pi, i dont know how to type it.
Is this the right formula?
I also followed the rule that the first stage of a fourth order butterworth filter has to have a gain of around 0.152 and the second stage has to have a gain of around 1.234.
So both combined will have a gain of around 2.57.
This is only for a single LPF / HPF 4th order stage though.
So i think the total gain of a complete BPF is 2.57 * 2. (Because it's 4th order HPF + 4th order LPF combined).
Is this right?
For the best accuracy i'm going to use ±0.1% resistors and all capacitors are polypropilene for best audio performance.
But before placing the order, i wanted to ask...
Did I make some mistakes or those filters are going to work fine?
I'm basically a newbie in this kind of stuff, and this is a personal project.
I'm sorry if i made a lot of mistakes that might be ovbious to you.
If you need more information than what i provided for now, no problem at all, i will provide them.
Attachments
Firstly, fourth order Butterworth can give good results.. if you can make it happen at the speaker. If the speaker varies then that will be more of a problem.
But this is just the preamplifier... I'm buffering the outputs of the three BPFs as well, so that their behaviour won't change when i load the signal more at the output of the buffer OpAmp. That signal is not going to be loaded much anyway, because it goes to the input of the amplifier that has a high impedance (47Kohm).
Like this is will work with any speaker... Right?
Of course my speakers frequency responses are in fact those.
30 - 125Hz for the subwoofer
125 - 4000Hz for the midrange
4000 - 20000Hz for the tweeter
Like this is will work with any speaker... Right?
Of course my speakers frequency responses are in fact those.
30 - 125Hz for the subwoofer
125 - 4000Hz for the midrange
4000 - 20000Hz for the tweeter
You still have to account for the variations between speakers. If you want, you can do this separately and keep the Butterworth filters.
But does the flatness of my 4th order butterworths change really a lot in comparation to the 2nd order ones?
If it is not huge, it is actually not a problem, because i know how to compensate for it. I was looking for a filter as sharp as possible.
I actually tested a 4th order LPF (100Hz) and it works really well, with acceptable flatness.
I'm just hoping that when i build the band pass filter it will function properly without the 4th order HPF and LPF stages somehow interfering with each other.
The first filter is a band pass because my subwoofer cant handle from 0 to 25Hz. So i decided to add a 30Hz HPF as a subsonic filter.
I could just use a high pass for the tweeter though. But there is just something that tells me that a band pass for the tweeter could reduce interference / noise. Like if there is a high frequency interference close to the amplifier, that frequency won't get in the tweeter and cause interference to the amplifier.
If it is not huge, it is actually not a problem, because i know how to compensate for it. I was looking for a filter as sharp as possible.
I actually tested a 4th order LPF (100Hz) and it works really well, with acceptable flatness.
I'm just hoping that when i build the band pass filter it will function properly without the 4th order HPF and LPF stages somehow interfering with each other.
The first filter is a band pass because my subwoofer cant handle from 0 to 25Hz. So i decided to add a 30Hz HPF as a subsonic filter.
I could just use a high pass for the tweeter though. But there is just something that tells me that a band pass for the tweeter could reduce interference / noise. Like if there is a high frequency interference close to the amplifier, that frequency won't get in the tweeter and cause interference to the amplifier.
This is a comparison. However sometimes Butterworth is preferred over Linkwitz-Riley. It's too early to say.
This is not a comparison between crossover flatness of Butterworth and Linkwitz-Riley.
Yes it is good idea to use input buffer to drive all filter sections
Also possible or often needed is individual gain control on outputs as well.
Since you will likely need to control levels. Since sub will likely need slight boost
and tweeter slight padding.
Your schematic shows common filter with gain.
If you do add individual gain controls for each output.
Then of course not always needed to add gain at filter stage but after
keeping filters unity gain.
I'll assume schematic is shown simplified for example only.
Full schematic of course would show power rail decoupling capacitors.
And likely not be using 741 opamp and using typical quad package low noise
audio types. Since 8 pin chip for every stage would be obvious nightmare for layout.
I would review some basic designs here for inspiration of 24 dB active crossovers
4th order Linkwitz and butterworth otherwise pretty standard normal stuff for crossover.
Q at cutoff could be fine tuned if tested in simulation depending on specific speakers used.
With simulation it would show more precise response and calculated values be more
precise and matched to exact speakers and baffle response to be expected.
If additional gain stages not wanted to simplify design. You could set individual gains
more accurate as well
Also possible or often needed is individual gain control on outputs as well.
Since you will likely need to control levels. Since sub will likely need slight boost
and tweeter slight padding.
Your schematic shows common filter with gain.
If you do add individual gain controls for each output.
Then of course not always needed to add gain at filter stage but after
keeping filters unity gain.
I'll assume schematic is shown simplified for example only.
Full schematic of course would show power rail decoupling capacitors.
And likely not be using 741 opamp and using typical quad package low noise
audio types. Since 8 pin chip for every stage would be obvious nightmare for layout.
I would review some basic designs here for inspiration of 24 dB active crossovers
4th order Linkwitz and butterworth otherwise pretty standard normal stuff for crossover.
Q at cutoff could be fine tuned if tested in simulation depending on specific speakers used.
With simulation it would show more precise response and calculated values be more
precise and matched to exact speakers and baffle response to be expected.
If additional gain stages not wanted to simplify design. You could set individual gains
more accurate as well
Last edited:
If the right choices were made, i.e. 4th-order Linkwitz-Reilly, there is nothing for 'global EQ' to correct. It is already maximally flat across the crossover region. That's my point.
I see, but I don't agree that it's as simple as that. You can force flat with other alignments but there will be fundamental differences besides the axial response.
In any case we're not up to that in this thread yet, I only make the point that neither alignment is necessarily incorrect.
In any case we're not up to that in this thread yet, I only make the point that neither alignment is necessarily incorrect.
Basically
4th order very good choice.
Far as fine tuning the exact speakers on exact baffle used.
Up to you to accurately sim or measure speakers used.
And you can fine tune the cut off and Q as needed.
Diy is Diy so we like to reinvent the wheel many times over
for the sake of the hobby.
You could obviously just buy a 3 way active crossover.
Or since we are talking typical Q of .5 to .7 you could make
Q adjustable. Neither is wrong or right, depends on what drivers
you trying to sum together.
4th order very good choice.
Far as fine tuning the exact speakers on exact baffle used.
Up to you to accurately sim or measure speakers used.
And you can fine tune the cut off and Q as needed.
Diy is Diy so we like to reinvent the wheel many times over
for the sake of the hobby.
You could obviously just buy a 3 way active crossover.
Or since we are talking typical Q of .5 to .7 you could make
Q adjustable. Neither is wrong or right, depends on what drivers
you trying to sum together.
Filters for loudspeakers, regardless of active or passive, need to be adjusted to the drivers amplitude, frequency and phase response.
As I see it you only make a standard filter which isn't adjusted for the drivers you want to use.
That will sound as a very bad attempt to make a crossover for your loudspeakers.
Making filters is an art that can't be compromised to just use a certain filter type.
It have to be adjusted to the specific drivers you want to design the filter for.
You need to have a proper measurement system that can handle both frequency and phase response, and you need to have the knowledge
how to change the filter topology to suite these drivers.
Standard filters will work to a degree, but it will work quite badly.
Dependes on the drivers amplitude, frequency and phase response.
Filters have to be adjusted to perfectly match the drivers in question to give a flat response in the room.
Making an active filter for a loudspeaker driver is just as complex as making a passive filter.
And a loudspeaker should not measure flat.
It must be adjusted for room gain and 2pi to 4pi conversion (baffle step).
That must be handled by the active filter too.
Even-order Butterworth gives a 3 dB peak on axis, while Linkwitz-Riley does not. If you want a flat response on axis and a flat response for the power averaged over all directions, you need odd-order Butterworth - but then the outputs of the crossover are not in phase (nor antiphase), causing a tilt of the main lobe and more sensitivity to small phase errors of the drivers than with Linkwitz-Riley.
@Gianluca_2000 , is the subwoofer to be placed such that it is at about the same distance from your ears as the mid and high, or do you put it wherever there is space in the room?
Sorry, yes. In practice it's fairly arbitrary and my point is that the variations that remain..
..after globally equalising (anything can be made flat that way), are what make the decision to go with one or the other.
Why would you start with a stupid choice and then equalize it instead of doing it properly from the start?