I recently made a post trying to get some information on how to design a crossover as I had previously no knowledge on how to and wanted to renovate some speaker cabinets and remake a crossover to do this. To inform my design I used frequency and impedance graphs from the manufacturers. I know this doesn't deliver the best results as the boxes used for these tests aren't the same as the one that I'll be using for my speakers. Nonetheless I decided to use them as I don't have the equipment to make my own graphs. Here are the data sheets for the drivers that I used; Bass driver: https://www.farnell.com/datasheets/2623802.pdf , Horn driver: https://www.parts-express.com/pedocs/specs/290-5634--eminence-f110m-8-spec-sheet.pdf and tweeters: https://doc.soundimports.nl/pdf/brands/GRS/DTM-8/DTM-8 Specifications.pdf. Here is the resulting circuit
And the resulting graphs from this
I can tell from the frequency graph that the tweeters aren't doing a lot, now I'm not sure if this is from how I've tried to link them in parallel or how my attempt at a Zobel network has affected the level of these drivers. Or whether it is a good idea to have them connected at all or if I should just use the horn and bass drivers to make the speakers. Any advice on how I can better improve this design or use this design to inform another design would be greatly appreciated!
And the resulting graphs from this
I can tell from the frequency graph that the tweeters aren't doing a lot, now I'm not sure if this is from how I've tried to link them in parallel or how my attempt at a Zobel network has affected the level of these drivers. Or whether it is a good idea to have them connected at all or if I should just use the horn and bass drivers to make the speakers. Any advice on how I can better improve this design or use this design to inform another design would be greatly appreciated!
Why are there so many tweeters, I'd generally advise against this as you will encounter serious comb filtering effects (driver interference)
How high can the horn driver go frequency wise?
Your impedance is going incredibly low towards 20kHz, that would effectively be shorting the amp.
You are also crossing your tweeters and horn above 20kHz, why bother adding them when they are doing so little?
How high can the horn driver go frequency wise?
Your impedance is going incredibly low towards 20kHz, that would effectively be shorting the amp.
You are also crossing your tweeters and horn above 20kHz, why bother adding them when they are doing so little?
This has a different effect depending on where you place it. I suspect it's not where it should be. What was it's intended purpose?
8.2uH is a very small value, try increasing it.
The source of the shorting is the resistor parallel to your horn driver, its value is 68 miliOhms which is tiny, I'd increase that as well.
The capacitor value on your horn driver zoble network is 2.7 miliFarads which is probably 1000x larger than intended.
You are also using some very large capacitor values on your bass driver for its crossover, see if you can decrease those while maintaining the desired crossover frequency, take steps changing the capacitor and inductor values.
The capacitor value on your horn driver zoble network is 2.7 miliFarads which is probably 1000x larger than intended.
You are also using some very large capacitor values on your bass driver for its crossover, see if you can decrease those while maintaining the desired crossover frequency, take steps changing the capacitor and inductor values.
Definitely a learning exercise to learn the software.
You would have to model the drivers on the baffle and what position
they will be at. Since how phase interacts and baffle step comes into play.
It is just a horn driver and does not represent the frequency response with
a horn.
Plus as mentioned the impedance curve is almost dead short.
No amplifier would be happy with that load.
I would try to find drivers that can be modeled realistically.
And model the baffle. On positive note the woofer looks like
a fun unit to start with. Find a single mid and tweet that will work
and model the baffle. So you can see off axis behavior and baffle
step. And also why such a large array of tweeters would not
work very well. Luckily there is a pre designed 3 way in the software
including common compensation for impedance if needed.
You can scroll the values and makes the learning curve easier.
You would have to model the drivers on the baffle and what position
they will be at. Since how phase interacts and baffle step comes into play.
It is just a horn driver and does not represent the frequency response with
a horn.
Plus as mentioned the impedance curve is almost dead short.
No amplifier would be happy with that load.
I would try to find drivers that can be modeled realistically.
And model the baffle. On positive note the woofer looks like
a fun unit to start with. Find a single mid and tweet that will work
and model the baffle. So you can see off axis behavior and baffle
step. And also why such a large array of tweeters would not
work very well. Luckily there is a pre designed 3 way in the software
including common compensation for impedance if needed.
You can scroll the values and makes the learning curve easier.
Hi everyone thanks for the responses, I’m going to try address some of the points mentioned to hopefully further my learning surrounding this build.
This photo is for reference so that the excess of tweeters makes more sense. The only reason I had 5 was to fill the spaces in this box which I’m rebuilding. I would be very happy to drop down the number of tweeters to say 2-3 or completely ignore them and just go for the horn driver to fill the middle section. I also just thought that maybe instead of tweeters two smaller “full range” drivers could be used but this might not be practical in actual design. The range of the horn driver is 2khz-20khz would I want to look for one that deals with more of the mid range to make it more valid in the build? The low impedance at 20khz is due to the zoble network I tried to implement flattening the impedance incorrectly so that’ll need correcting. To address the talk about baffles, I’m assuming this is about the size of the box and spacing of drivers within said box. Can this be changed within the room settings of vituix CAD? As far as the horn driver yes it is just a screw in horn driver I’m still yet to find a horn for it. Will the shape and design of the horn effect how I design the crossover?
I’m going to have a fiddle around with the capacitor and resistor values that some of you mentioned to hopefully sort some of the issues. Thanks for all the help so far!
This photo is for reference so that the excess of tweeters makes more sense. The only reason I had 5 was to fill the spaces in this box which I’m rebuilding. I would be very happy to drop down the number of tweeters to say 2-3 or completely ignore them and just go for the horn driver to fill the middle section. I also just thought that maybe instead of tweeters two smaller “full range” drivers could be used but this might not be practical in actual design. The range of the horn driver is 2khz-20khz would I want to look for one that deals with more of the mid range to make it more valid in the build? The low impedance at 20khz is due to the zoble network I tried to implement flattening the impedance incorrectly so that’ll need correcting. To address the talk about baffles, I’m assuming this is about the size of the box and spacing of drivers within said box. Can this be changed within the room settings of vituix CAD? As far as the horn driver yes it is just a screw in horn driver I’m still yet to find a horn for it. Will the shape and design of the horn effect how I design the crossover?
I’m going to have a fiddle around with the capacitor and resistor values that some of you mentioned to hopefully sort some of the issues. Thanks for all the help so far!
After making my first post, I decided to try and improve on the design by using some of the inbuilt circuit designs to base my own design off. I have also decided to use just the horn and bass driver for now as adding in the 5 tweeters in the previous circuit seemed to complicate things, and from doing research it seems that if I wanted I could get 5 piezo tweeters and not need to run a complex crossover circuit to use them. Anyway, here is the result of my experiments with these inbuilt circuits.
I'm quite happy with the SPL response as I was aiming for a crossover frequency around 2khz as the horn I intend to use has a cut off frequency of 1.5khz ( and the drive a usefull range down to 2k) and my woofer being able to go to 4k. However, I'm still struggling quite a bit to flatten out the impedance load on the generator with the use of zobel networks in the circuit. Does anyone know a better way to flatten the impedance?? Any help and pointers on how to do this and with the general structure of the circuit would be greatly appreciated!!
I'm quite happy with the SPL response as I was aiming for a crossover frequency around 2khz as the horn I intend to use has a cut off frequency of 1.5khz ( and the drive a usefull range down to 2k) and my woofer being able to go to 4k. However, I'm still struggling quite a bit to flatten out the impedance load on the generator with the use of zobel networks in the circuit. Does anyone know a better way to flatten the impedance?? Any help and pointers on how to do this and with the general structure of the circuit would be greatly appreciated!!
The 39 mF cap (bipolar! not sure where you will source this!?) with 12 ohm resistor in series will just burn amp power.
The other component values also seem quite a bit off. 22 mH prallel inductance for the tweeter is huge ...
47 uF parallel after only 270 uH series for the woofer pushes the impedance down to 1 ohm at 1.5 kHz.
you could try using boxsim and it's crossover optimizer defining your desired impedance to find suitable values.
Edit: by the way, you should have continued in your other thread (maybe you can ask moderators to connect them), it would make your project easier to understand.
The other component values also seem quite a bit off. 22 mH prallel inductance for the tweeter is huge ...
47 uF parallel after only 270 uH series for the woofer pushes the impedance down to 1 ohm at 1.5 kHz.
you could try using boxsim and it's crossover optimizer defining your desired impedance to find suitable values.
Edit: by the way, you should have continued in your other thread (maybe you can ask moderators to connect them), it would make your project easier to understand.
Last edited:
Why flatten the impedance? Why use the 12 ohm and 39mF?
The impedance is sometimes flattened inside the crossover to help make the design more straight forward, but here you're using it outside the crossover. It is loading your amp..
I had read on an article about passive crossovers that doing this will reduce load on the amp as there won’t be as many fluctuations in the impedance. These values were just based of what they did to the impedance in the bottom right graph and nothing else, should they be more informed on the drivers specifications??
I haven’t used them in my design but they were what was used in the box that I am going to be using for my design.
I’ll give boxsim a download and have a play around with the values and see what I can make! So you’re saying the low impedance at aroun 1.5k is due to that 270 being too high? I think I’ll remove that zobel network then or maybe move it next to the speaker instead of where it is currently
Threads merged and moved to Multi-Way.
Ignore this for now. Just work on your responses. Make sure the impedance doesn't go too low but focus on the output of the crossover.
The default impedance will be wrong. They don't act like resistive voice coils. Therefore your crossover isn't likely to be right either.
I think you need to learn about component values, i.e. the difference between 'micro-' and 'milli-', some of the components you have are totally inappropriate, and will have very little effect (e.g. 1 uH on a bass driver), or will cost a lot of money, yet be unnecessay (e.g. 22mH on a mid driver). Look at several crossover schematics to get a feel for which order of parts are appropriate.
I’ll make another version and remove the sober network then. The output I should be looking to change is the spl output and not the filter one??