I'm working on an idea for a simple loudspeaker project that involves a small full range driver that has a peak at around 12khz of about 8-9db before going back to normal levels up to 20khz.
I'm a total noob when it comes to crossovers, but is it possible to reduce this peak without affecting frequency curve before and after too much? Or is it something you have to either live with or take care of with an equalizer?
Thanks!
I'm a total noob when it comes to crossovers, but is it possible to reduce this peak without affecting frequency curve before and after too much? Or is it something you have to either live with or take care of with an equalizer?
Thanks!
It's very easy to do a notch at line level with opamps, it just requires a bit of math.
Active Filters
You can build up one of these very cheap if you have the stuff lying around, and it can work easily off a single supply if you need to.
Active Filters
You can build up one of these very cheap if you have the stuff lying around, and it can work easily off a single supply if you need to.
It's not as simple as that. The peak comes from the eigenmode of the cone. Once you start equalising you are affecting the whole radiation pattern, affecting sound in the whole room, which is really what you listen to wherever you sit.
Not only will this do bad things to off-axis response, it will also introduce unexpected peaks and dips occurring in an almost binary fashion.
EQ is not bad, in fact its mandatory most of the time. But it should be done with incredible care. As much care as you put into the material parts of the setup.
I'd choose a reversible physical bandpass filter any time for a job like this.
In this specific case a slot lens made of two sheets of lose felt would be my first choice. Start with a slot the width of the wavelength you are trying to attenuate and experiment from there.
This will fan out that frequency, attenuating it to some degree.
That said, a bit aggressive 12Khz might not be a bad thing, unless you are absolutely direct on-axis which is very rare in real world listening.
Not only will this do bad things to off-axis response, it will also introduce unexpected peaks and dips occurring in an almost binary fashion.
EQ is not bad, in fact its mandatory most of the time. But it should be done with incredible care. As much care as you put into the material parts of the setup.
I'd choose a reversible physical bandpass filter any time for a job like this.
In this specific case a slot lens made of two sheets of lose felt would be my first choice. Start with a slot the width of the wavelength you are trying to attenuate and experiment from there.
This will fan out that frequency, attenuating it to some degree.
That said, a bit aggressive 12Khz might not be a bad thing, unless you are absolutely direct on-axis which is very rare in real world listening.
Last edited:
The posts above have omitted the second option, the less traveled road (both by diy-ers and manufacturers) - controlling the resonance that causes the peak, most likely a cone resonating mode (standing wave in the cone)... but I don't think there are dedicated methods = a step by step guide of doing it, plus there is the issue of permanent mods with no warranties they actually work.
Feel free to post the loudspeaker model or at least the cone material or a photograph of the cone, somebody may have some experience.
Feel free to post the loudspeaker model or at least the cone material or a photograph of the cone, somebody may have some experience.
Fullframe cones are fragile, otherwise they wouldn't be light enough to do what they should.
Therefore it's a risky and slow road to go down with actually modifying the cone.
If you don't have spares readily available and some grasp of material acoustics you could enter a long, fruitless and frustrating pursuit.
You could have very light and long metal or plastic springs hold clumps of felt or cotton wool very, very lightly to the cone and move them around. Then glue the damping material in place where it works best.
Dammar is popular for stiffening the cone and for light added weight but is non reversible. It's easy to think you have something going, and then realise you went too far.
Unless you have access to a laser analyser, you can get a rough idea of where the breakup modes are on the cone, by pouring a little sand onto a horizontal part of the cone and do a frequency sweep of the upper registers. Rotate the cone to get a complete picture.
Therefore it's a risky and slow road to go down with actually modifying the cone.
If you don't have spares readily available and some grasp of material acoustics you could enter a long, fruitless and frustrating pursuit.
You could have very light and long metal or plastic springs hold clumps of felt or cotton wool very, very lightly to the cone and move them around. Then glue the damping material in place where it works best.
Dammar is popular for stiffening the cone and for light added weight but is non reversible. It's easy to think you have something going, and then realise you went too far.
Unless you have access to a laser analyser, you can get a rough idea of where the breakup modes are on the cone, by pouring a little sand onto a horizontal part of the cone and do a frequency sweep of the upper registers. Rotate the cone to get a complete picture.
Last edited:
Tangband W4-657SD a 4" fullrange driver
+/- peak correction
Impedance
How it was done
The simple way to do something like this.
1.Use a online calculator to calculate the Zobel (the cap and resistor to the right, it flattens out the impedance. Values are not critical.
2. Use a online calculator to calculate the parallel notch filter
mh-audio.nl - Home
It will be far from perfect but compared to such a large peak it will be an improvement. The problem is getting coils small enough. In the US Parts-Express has several coils below 0.1 mH not so HifiKit and Ljudia in Sweden. Former Europe Audio now loudspeakerfreaks.com do have a lot of small coils/inductors for us in europe
The complicated way
Get tools to measure impedance and frequency response and get that into Xsim then you can tweak it a lot (I have better tweaks than those graphs but I did not images of those and I will save that for the final tweaking in a final box and with a baffle step correction as well
+/- peak correction
Impedance
How it was done
The simple way to do something like this.
1.Use a online calculator to calculate the Zobel (the cap and resistor to the right, it flattens out the impedance. Values are not critical.
2. Use a online calculator to calculate the parallel notch filter
mh-audio.nl - Home
It will be far from perfect but compared to such a large peak it will be an improvement. The problem is getting coils small enough. In the US Parts-Express has several coils below 0.1 mH not so HifiKit and Ljudia in Sweden. Former Europe Audio now loudspeakerfreaks.com do have a lot of small coils/inductors for us in europe
The complicated way
Get tools to measure impedance and frequency response and get that into Xsim then you can tweak it a lot (I have better tweaks than those graphs but I did not images of those and I will save that for the final tweaking in a final box and with a baffle step correction as well
Last edited:
The baffle design can alter the polar response drastically. Nasty peaks can be tamed by having a smooth physical transition away from the driver. This also improves the impulse response. Generally a good baffle design will broaden the coverage pattern allowing you to listen slightly off axis and not lose all the high frequency.
Filters are one way to go, but one has to worry about “sucking” the life out of the driver. I prefer trying to find and eradicate the issue mechanically — ie mod the cone/driver.
As krakatoa says, this is non-reversible, and it helps to have lots of experience and spare drivers. I have been modding drivers since the late ‘70s. I have ruined a few, made many quite a bit better.
What driver are you considering/have?
An on-axis 12kHz resonance is up high enuff that it may not be an issue. You don’t usually listen to a FR on-axis. Have you listened to them? And the resonance will move around due to factory deviations, you need to be able to check the specific drivers you are using.
dave
As krakatoa says, this is non-reversible, and it helps to have lots of experience and spare drivers. I have been modding drivers since the late ‘70s. I have ruined a few, made many quite a bit better.
What driver are you considering/have?
An on-axis 12kHz resonance is up high enuff that it may not be an issue. You don’t usually listen to a FR on-axis. Have you listened to them? And the resonance will move around due to factory deviations, you need to be able to check the specific drivers you are using.
dave
Stopped paying attention there.
Bob Brines chimes in on the second page of the thread though.
If I were me (and I usually am), I would first build the project without any “correction” built-in. Listen to it through at least the break-in period for the drivers (up to 300 hours depending on the manufacturer recommendation). Then decide if that 12kHz peak is bothersome. Why try solving a problem that may not exist?
Edit: You still haven’t told us which drivers you’re using. That would certainly help as others with direct experience could chime in with recommendations.
Edit: You still haven’t told us which drivers you’re using. That would certainly help as others with direct experience could chime in with recommendations.
Last edited:
Hi-Fi loudspeaker factories are so incompetent they never used this proven tweak which are used by car audio hobbist since 1990s ever with plenty of success, its even an old trick, but hi-fi ''engineers'' are completely unaware.Stopped paying attention there.
Bob Brines chimes in on the second page of the thread though.
I have a set of speakers here with a 12kHz peak 6dB high.
It's an issue. If you go off-axis, the top end gradually drops off as you'd expect, but the peak remains present in the off-axis response.
EQ kills it nicely and the results are quite listenable.
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.