An externally hosted image should be here but it was not working when we last tested it.
Let me start by saying hello to everyone. I am new here and hope to share and learn.Let me start with what I am working with.
Audio Concepts AC-10 woofers (2)
Dynaudio d54af mids
Dynaudio d21af tweets
The Ac-10 woofers have a breakup frequency at about 1600hz. The d54 around 6200hz.
I have been using a 1st order crossover with zobel. Lately, I have attained a Dayton Omnimic, and have finally been able to measure whats going on.
My main question is this:
1) My first inclination was to try to build notch filters to take care of the cone breakups. As I understand things, these come in two versions. Parallel and series.
Looking at the wiring of each, its seems Series would be my preference given the circuitry is not in the signal path. But from what I have been able to understand so far, these are primarily used for resonate frequency issues.
Series Notch Filter Designer / Calculator Help
Parallel seems aimed at dealing with broad frequency peaks within the frequency response of the driver.
Parallel Notch Filter Designer / Calculator Help
Neither describes what I am trying to do. So which do I use? Why? And is there a calculator I can use to determine the parts I need based on the frequency and shape of the peak(s)?
Last edited:
Parallel notches are usually used to take the shoutiness out of single driver systems;
series notches are used with more conventional systems when required, e.g. metal cone drivers. It's better to use drivers so that cone breakup occur outside the passband, that way notch filters may not be required if the breakup isn't too severe. If you've been able to live with 1st order crossovers, it probably isn't too bad.
(The idea that series notches would be used on a woofer's fundamental resonance is laughable, but peddled on some car audio sites...)
Irregularities in an impedance graph is also a useful tool in dealing with cone resonances.
series notches are used with more conventional systems when required, e.g. metal cone drivers. It's better to use drivers so that cone breakup occur outside the passband, that way notch filters may not be required if the breakup isn't too severe. If you've been able to live with 1st order crossovers, it probably isn't too bad.
(The idea that series notches would be used on a woofer's fundamental resonance is laughable, but peddled on some car audio sites...)
Irregularities in an impedance graph is also a useful tool in dealing with cone resonances.
It's not so daft an idea really. Just very hard to find good capacitors of sufficient value. You do it with tweeters a lot when using low order filters.
Back on-topic, Jim1961, I have a feeling you are trying to make a silk purse out of a sow's ear here. You'd find it all a lot easier if you crossed over lower, perhaps at the bafflestep point of around 500 Hz to (say) a 5" midrange. Big high-inductance woofers are going to fight you all the way with a high crossover point. That's how it is.
A 5" mid would be easier to work with from the point of view of lending itself to a lower crossover frequency, yes.
But 2" domes is what I have, and what I am limited in working with.
Does 800hz seem too high for a 10" woofer in your opinion? Or anyone elses?
Other choices include going to a 2nd order network. Possibly a notch and a 2nd order network. But I wanted to explore notch filtering because my understanding of them isnt sufficient to know what and how much they can affect the problem of cone breakup.
As far as silk purses go, I think most of us give that a try if we can get away with it
But 2" domes is what I have, and what I am limited in working with.
Does 800hz seem too high for a 10" woofer in your opinion? Or anyone elses?
Other choices include going to a 2nd order network. Possibly a notch and a 2nd order network. But I wanted to explore notch filtering because my understanding of them isnt sufficient to know what and how much they can affect the problem of cone breakup.
As far as silk purses go, I think most of us give that a try if we can get away with it
Jim, are you using some modelling software? Like this which is useful even if it really only includes Visaton drivers, hopefully you can find similar:
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
My own feeling, without knowing the parameters of the bass unit, is that you need second order filtering here. maybe Likwitz-Riley slopes. First order makes some fairly extraordinary demands of drivers. Like smooth frequency response and good overlap. Have a look at these two Visaton designs (with crossovers) that use dome midranges:
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
Even notches are not easy. Close to crossover they drop the impedance and ring if the Q is high. Well beyond it they are more manageable.
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
My own feeling, without knowing the parameters of the bass unit, is that you need second order filtering here. maybe Likwitz-Riley slopes. First order makes some fairly extraordinary demands of drivers. Like smooth frequency response and good overlap. Have a look at these two Visaton designs (with crossovers) that use dome midranges:
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
Even notches are not easy. Close to crossover they drop the impedance and ring if the Q is high. Well beyond it they are more manageable.
To make sure I am understanding you correctly, are you saying that using a notch 1 octave above the crossover could cause as much harm as good?
I have not investigated modelling software. I was hoping, now that I have decent measuring equipment, to basically use test measurement results to allow me to try various passive crossover ideas and see which work out the best through basically trial and error.
But right now, i am at the information stage. I havent decided on any one approach as of yet.
Let me add that the speakers, just as they are, sound reasonably good to me. But I am also aware that there is room for improvement. So, I am making an effort to see if I can get a bit more out of them. And the cone breakup frequencies stand out as the areas of most needed attention.
I have not investigated modelling software. I was hoping, now that I have decent measuring equipment, to basically use test measurement results to allow me to try various passive crossover ideas and see which work out the best through basically trial and error.
But right now, i am at the information stage. I havent decided on any one approach as of yet.
Let me add that the speakers, just as they are, sound reasonably good to me. But I am also aware that there is room for improvement. So, I am making an effort to see if I can get a bit more out of them. And the cone breakup frequencies stand out as the areas of most needed attention.
Jim, the series bandstop (notch) filter (components in series, placed across the driver) is best used to make modifications to the impedance presented by the driver, and is commonly used to conjugate a tweeter's resonance impedance peak.
In this configuration it is the logical choice, because once it is partnered with the driver it's job is done so it is independent of all the other components. It is possible, though not normally recommended, to use other types instead.
Where you want to bandstop a portion of the passband, such as bringing down a peak in the frequency response, it is typical to use a parallel bandstop filter (placed in series with the driver as you would for a normal coil), but this will come down to a choice based on the circumstances.
The parallel filter will raise the total impedance against the load, whereas the series kind lowers the load impedance against the existing series impedance. So, in choosing which one to use......
in the first case in the middle of a drivers pass band there might be, say, one ohm of impedance due to the coil so if you used a series filter across the driver you'd need to bring the driver down to one ohm for a 6dB notch. No good, so go with the parallel. On the other hand if you can remove components in series with the driver and place them in parallel for the same effect then you'll have a tidier solution with some minor benefits.
As far as ringing is concerned I would initially just ensure that they don't cause any significant looking frequency response peaks.
In this configuration it is the logical choice, because once it is partnered with the driver it's job is done so it is independent of all the other components. It is possible, though not normally recommended, to use other types instead.
Where you want to bandstop a portion of the passband, such as bringing down a peak in the frequency response, it is typical to use a parallel bandstop filter (placed in series with the driver as you would for a normal coil), but this will come down to a choice based on the circumstances.
The parallel filter will raise the total impedance against the load, whereas the series kind lowers the load impedance against the existing series impedance. So, in choosing which one to use......
in the first case in the middle of a drivers pass band there might be, say, one ohm of impedance due to the coil so if you used a series filter across the driver you'd need to bring the driver down to one ohm for a 6dB notch. No good, so go with the parallel. On the other hand if you can remove components in series with the driver and place them in parallel for the same effect then you'll have a tidier solution with some minor benefits.
Yes, I used to have a series notch across my woofer's fs when using triode amplifiers so that I could tune the bass whenever I changed Zout.
No, as long as you're watching the directivity. Not all 10" woofers will play cleanly up here, although I use a 15" higher than this.
I sometimes use a series notch filter to tune the phase of a woofer at crossover. It is positioned above the crossover and has no significant effect on the frequency response. These things can certainly interact and you need to design carefully.
As far as ringing is concerned I would initially just ensure that they don't cause any significant looking frequency response peaks.
I don't think you can see what goes on without modelling software, Jim. The only thing tricky about the Visaton BoxSim program initially is placing the drivers on the front baffle of one of the standard box designs included. At least I didn't find it intuitive to start with. But I was up and running in an hour, and find it quite easy to use a Visaton driver equivalent of things that people use here.
What do Dynaudio do with first order crossovers? This is the Sirius design:
Looks to me like they put a larger than expected capacitor on the bass Zobel, maybe 2-4 times bigger. This creates a sort of 1.5 order rolloff on the bass without any resonant notches or impedance bumps and dips. Rather good, though you'd want to do something matching to the next driver in the chain to match it up nicely if you were fussy.
Oh, by way of explanation, the two networks on the RHS are impedance correction for valve amps, and the 1.2Uf and 6.8R is close to the tweeter Zobel values. 1.5uF in my calculation here. The trick is to get everything working together elegantly. Each component does several things when you get it just right.
What do Dynaudio do with first order crossovers? This is the Sirius design:
An externally hosted image should be here but it was not working when we last tested it.
Looks to me like they put a larger than expected capacitor on the bass Zobel, maybe 2-4 times bigger. This creates a sort of 1.5 order rolloff on the bass without any resonant notches or impedance bumps and dips. Rather good, though you'd want to do something matching to the next driver in the chain to match it up nicely if you were fussy.
Oh, by way of explanation, the two networks on the RHS are impedance correction for valve amps, and the 1.2Uf and 6.8R is close to the tweeter Zobel values. 1.5uF in my calculation here. The trick is to get everything working together elegantly. Each component does several things when you get it just right.
Last edited:
I don't think you can see what goes on without modelling software, Jim. The only thing tricky about the Visaton BoxSim program initially is placing the drivers on the front baffle of one of the standard box designs included. At least I didn't find it intuitive to start with. But I was up and running in an hour, and find it quite easy to use a Visaton driver equivalent of things that people use here.
What do Dynaudio do with first order crossovers? This is the Sirius design:
An externally hosted image should be here but it was not working when we last tested it.
Looks to me like they put a larger than expected capacitor on the bass Zobel, maybe 2-4 times bigger. This creates a sort of 1.5 order rolloff on the bass without any resonant notches or impedance bumps and dips. Rather good, though you'd want to do something matching to the next driver in the chain to match it up nicely if you were fussy.
Oh, by way of explanation, the two networks on the RHS are impedance correction for valve amps, and the 1.2Uf and 6.8R is close to the tweeter Zobel values. 1.5uF in my calculation here. The trick is to get everything working together elegantly. Each component does several things when you get it just right.
I wonder if there is a speaker Dynaudio built using the very drivers I have, i.e. d21af & d54af. If I could find such a speaker, it might be illuminating to see what they did.
Perhaps your resources are better than mine, but I have looked for such a speaker in the past and didnt have much luck. At least, not one where the crossover design was shown.
Just for grins, I downloaded the visaton software. I didnt get past the first driver, for it asked for data that I dont have. I am not going to ask you or anyone else here to go through it with me for it would be lengthy and complicated and a unrealistic thing to try to accomplish on a blog.
My brother is an EE with interests in audio. He might be able to accomplish what I can not in this regard.
No, Jim. It's easy. Just load a similar driver from file in drivers and enclosures, like this W 250 S 10".
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
I'd also look for any old bass driver with a roughly matching inductance, but that's just me.
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
I'd also look for any old bass driver with a roughly matching inductance, but that's just me.
No, Jim. It's easy. Just load a similar driver from file in drivers and enclosures, like this W 250 S 10".
Visaton - Lautsprecher und Zubehör, Loudspeakers and Accessories
I'd also look for any old bass driver with a roughly matching inductance, but that's just me.
I am a little doubtful, possibly do to my ignorance, that importing different driver(s) than I am actually using will yield results that are helpful.
But let me also say that I appreciate all your responses. I dont know, as said before, exactly how I am going to go about tackling the problem(s) at hand. But all information is good and potentially helpful
Cone breakup is the main thread topic. While my main question has to do notch filtering (because I needed more education on their uses, types and so forth), I meant it to be a call to all ideas that address dealing with the problem of cone breakup. it has been mentioned already, that steeper slopes (a higher order cross over design) might be a way to go, or at least help.
In my particular problem, and ill state it again so no one has to comb through thread to find the particulars , that I am crossing over at 800hz and need to deal with a cone breakup approximately 1 octave higher (around 1600hz).
One of the reasons I thought of a notch first, is because I dont have much room for a slope ( 1 octave) to get 12-20db down by 1600hz using merely a crossover slope. If you look at my graph, the peak is around 6-8db. So if I crossover at 800hz (-3db), at 1600hz I would be 15db down (using a 2nd order network) minus the peak resulting in the peak being 7-9db down (which isnt sufficient). The addition of a notch was the only way i could see not having to result to tricky and expensive 4th order networks.
If anyone sees errors in my calculating, please point it out.
In my particular problem, and ill state it again so no one has to comb through thread to find the particulars , that I am crossing over at 800hz and need to deal with a cone breakup approximately 1 octave higher (around 1600hz).
One of the reasons I thought of a notch first, is because I dont have much room for a slope ( 1 octave) to get 12-20db down by 1600hz using merely a crossover slope. If you look at my graph, the peak is around 6-8db. So if I crossover at 800hz (-3db), at 1600hz I would be 15db down (using a 2nd order network) minus the peak resulting in the peak being 7-9db down (which isnt sufficient). The addition of a notch was the only way i could see not having to result to tricky and expensive 4th order networks.
If anyone sees errors in my calculating, please point it out.
Last edited:
Breakup is hard to hear
I've had some success applying fiberglass mesh tape, the drywall patch stuff, to poly cones, and securing it with the acrylic glue intended for melamine. This plus (maybe) increasing the zobel capacitor value should tame the breakup which, once you notice it, you can't ignore. I cut triangles with a radius matching the surround pointing at the voice coil, so the original cone surface is about 1/2 to 2/3s damped. The midrange dome is a challenge I have not taken on. The tweeter is an all time favorite.
Don't believe those computer modelers. They're only right most of the time
I've had some success applying fiberglass mesh tape, the drywall patch stuff, to poly cones, and securing it with the acrylic glue intended for melamine. This plus (maybe) increasing the zobel capacitor value should tame the breakup which, once you notice it, you can't ignore. I cut triangles with a radius matching the surround pointing at the voice coil, so the original cone surface is about 1/2 to 2/3s damped. The midrange dome is a challenge I have not taken on. The tweeter is an all time favorite.
Don't believe those computer modelers. They're only right most of the time
Hi,
You say that and then say your mid-dome has break up issues, but I
doubt you'll be able to find any corroborating evidence that is the case.
I don't know you bass drivers either, other than they are polypropylene,
and they don't usually have serious break up issues either, typically.
I'd assume looking at your drivers a 3.5 way is the way to go, with
2nd order L/R acoustic targets at around 800Hz and 3KHz,
the 0.5 way driver somewhere around 300Hz roll-off.
I'm not going to waffle on about various possibilities, it just needs
to be done properly. The bass driver can either work well with the
mid-dome or is simply not suitable, that you need to identify.
For the above targets you'll probably need to conjugate the mid-domes
resonance and use a 1st order high pass on it, its low pass probably
a 1st order + zobel / 2nd order hybrid, similar on the tweeter.
Measurements can only be done gated nearfield typically, so the bass
end and BSC is usually simulated, further up you need to verify the
accuracy of the measurement process so not to be led astray.
Even with measurements you need a good modelling environment
to use them effectively, online calculators simply won't do the job.
Have a look at : Zaph|Audio - ZDT3.5 but note
I'm suggesting a different target for mid to treble.
rgds, sreten.
undefinition (see if nothing else, the excellent FAQs)
The Speaker Building Bible - Thread opened for edits/input. - Techtalk Speaker Building, Audio, Video, and Electronics Customer Discussion Forum From Parts-Express.com
Zaph|Audio
Zaph|Audio - ZA5 Speaker Designs with ZA14W08 woofer and Vifa DQ25SC16-04 tweeter
http://audio.claub.net/Simple Loudspeaker Design ver2.pdf
FRD Consortium tools guide
Designing Crossovers with Software Only
RJB Audio Projects
Jay's DIY Loudspeaker Projects
Speaker Design Works
HTGuide Forum - A Guide to HTguide.com Completed Speaker Designs.
A Speaker project
DIY Loudspeaker Projects Troels Gravesen
Humble Homemade Hifi
Quarter Wavelength Loudspeaker Design
The Frugal-Horns Site -- High Performance, Low Cost DIY Horn Designs
Linkwitz Lab - Loudspeaker Design
Music and Design
Great free SPICE Emulator : SPICE-Based Analog Simulation Program - TINA-TI - TI Tool Folder
Last edited:
Did you look at the graph on post #2 of this thread? Do you see the peak at 6200hz? That looks like cone breakup to me.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.