building a crossover

[NAD3020]

first of all i'd like to apologize for my atrocious english , but i'll try to be clear.

i'm building a crossover for a 3-way loudspeaker. all the spearkers are from RCF, maybe most of you don't know what RCF is but those speakers sound wonderfully and the source is a NAD C375BEE.
i was wondering if with such a configuration the difference between a 12db crossover and a 24db one could be heard.
i saved money for a long time for this, so i think i'm building a 24 db crossover to avoid the problems with phase rotation and all that stuff, but i know that inductors and capacitors must have low tolerances.

are good MKP 5% (jantzen, mundorf) or MKT 5% caps enough for a 24db crossover?
should the tolerance be lower, like, don't know, 3%?

thanks

 

[annunaki]

first of all i'd like to apologize for my atrocious english , but i'll try to be clear.

i'm building a crossover for a 3-way loudspeaker. all the spearkers are from RCF, maybe most of you don't know what RCF is but those speakers sound wonderfully and the source is a NAD C375BEE.
i was wondering if with such a configuration the difference between a 12db crossover and a 24db one could be heard.
i saved money for a long time for this, so i think i'm building a 24 db crossover to avoid the problems with phase rotation and all that stuff, but i know that inductors and capacitors must have low tolerances.

are good MKP 5% (jantzen, mundorf) or MKT 5% caps enough for a 24db crossover?
should the tolerance be lower, like, don't know, 3%?

thanks

No offense is intended here, but to be honest if you are not sure that you can hear a difference between the two, much less what the effects of doing such designs will be, I am not so sure you should be building a crossover from scratch. I would highly suggest that you look into a kit for your speakers or something to help you out.

A well done 2-way crossover is no small task and a 3 way adds even more complexity.

 

[NAD3020]

No offense is intended here, but to be honest if you are not sure that you can hear a difference between the two, much less what the effects of doing such designs will be, I am not so sure you should be building a crossover from scratch. I would highly suggest that you look into a kit for your speakers or something to help you out.

A well done 2-way crossover is no small task and a 3 way adds even more complexity.

No offense taken.

I don't know the difference between the two because i've never heard a 24db crossover with my own ears, but i've built several crossover myself.
I know that you have to attenuate certain speakers, to take care of frequency response, bandpass gain, crosstalking between inductors and so on.

For example, i know i could add a 1,0uF MKP cap to a 33,0uF electrolytic cap in a 12db crossover without changing too much the cut frequency.
I don't know if a variation of +- 1,0uF is gonna change something in a 24db one.

That's why i'm asking about tolerance.

But, if you really think i'm a noob, tell me where i can buy a good 24db crossover.

 

[highfigh]

first of all i'd like to apologize for my atrocious english , but i'll try to be clear.

i'm building a crossover for a 3-way loudspeaker. all the spearkers are from RCF, maybe most of you don't know what RCF is but those speakers sound wonderfully and the source is a NAD C375BEE.
i was wondering if with such a configuration the difference between a 12db crossover and a 24db one could be heard.
i saved money for a long time for this, so i think i'm building a 24 db crossover to avoid the problems with phase rotation and all that stuff, but i know that inductors and capacitors must have low tolerances.

are good MKP 5% (jantzen, mundorf) or MKT 5% caps enough for a 24db crossover?
should the tolerance be lower, like, don't know, 3%?

thanks

24dB gives plenty of phase shift and insertion loss- the drivers will determine the need for specific crossover type. You can't make them do what they aren't able to do. The first thing that's needed is to find out what the drivers are doing with/to the frequency response and then design the crossovers.

I didn't see atrocious English. I read more of that from people who use English as their first and only language than from those who use it as their second, third or fourth language.

 

[annunaki]

Highfigh is correct here. The drivers typically determine the need for a 4th order filter.

I rarely see these used in a passive design. Typically the crossover is coupled with the natural response of the driver which will yield results that act as a 4th order filter in practice.

 

[TLS Guy]

first of all i'd like to apologize for my atrocious english , but i'll try to be clear.

i'm building a crossover for a 3-way loudspeaker. all the spearkers are from RCF, maybe most of you don't know what RCF is but those speakers sound wonderfully and the source is a NAD C375BEE.
i was wondering if with such a configuration the difference between a 12db crossover and a 24db one could be heard.
i saved money for a long time for this, so i think i'm building a 24 db crossover to avoid the problems with phase rotation and all that stuff, but i know that inductors and capacitors must have low tolerances.

are good MKP 5% (jantzen, mundorf) or MKT 5% caps enough for a 24db crossover?
should the tolerance be lower, like, don't know, 3%?

thanks

RCF are high powered pro drivers for PA and band and not domestic use. They are an Italian company.

You can not pull the order and crossover points out of the back of your neck.

What drivers are you using? They have decent data sheets so it should be possible to model the crossovers. You must model a crossover first.

Those drivers are very high powered. You will not be able to use their potential with passive crossovers, they will go up in smoke.

To get the proper performance from those drivers you need active electronic crossovers with a separate amp to each driver. So for a pair of three ways you would need six amp channels.

 

[NAD3020]

I didn't see atrocious English. I read more of that from people who use English as their first and only language than from those who use it as their second, third or fourth language.

24dB gives plenty of phase shift and insertion loss

I rarely see these used in a passive design. Typically the crossover is coupled with the natural response of the driver which will yield results that act as a 4th order filter in practice.

Insertion loss shouldn't be a problem.
But, I chose a 24db crossover to avoid phase shift.
"Plenty of phase shift", what do you mean? shifting phase leads to quality loss?

And to annunaki, the midrange has a resonance frequency of 300Hz and the woofer's frequency response starts to weaken much higher, so what should i do?

by the way, is it true that a 12db crossover with the all drivers connected with reverse polarity has pratically no phase shift?

RCF are high powered pro drivers for PA and band and not domestic use. They are an Italian company.

You can not pull the order and crossover points out of the back of your neck.

What drivers are you using? They have decent data sheets so it should be possible to model the crossovers. You must model a crossover first.

Those drivers are very high powered. You will not be able to use their potential with passive crossovers, they will go up in smoke.

To get the proper performance from those drivers you need active electronic crossovers with a separate amp to each driver. So for a pair of three ways you would need six amp channels.

Thanks, as soon as i'll be back from university i'll post those data sheets.

 

[TLS Guy]

Insertion loss shouldn't be a problem.
But, I chose a 24db crossover to avoid phase shift.
"Plenty of phase shift", what do you mean? shifting phase leads to quality loss?

And to annunaki, the midrange has a resonance frequency of 300Hz and the woofer's frequency response starts to weaken much higher, so what should i do?

by the way, is it true that a 12db crossover with the all drivers connected with reverse polarity has pratically no phase shift?



Thanks, as soon as i'll be back from university i'll post those data sheets.

The phase shift of a fourth order crossover is 360 degrees at crossover. The phase shift of a second order crossover is 180 degrees at crossover.

However there are phase shifts within the drivers also, so design is not simple.

The crossover slopes and crossover points are dictated by the Thiel/Small parameters of the drivers and above all the acoustic response of the drivers. It is not an arbitrary process at all. You must understand that the crossover largely determines the sound and quality of the speaker. You can have the best drivers ever and if the crossover is not right on, it will be a dreadful speaker.

One of my big concerns is what you intend to use these speakers for. They are very high powered drivers. You can not make a very high powered speaker with passive components. The reason is insertion loss. Half your amp power is lost in the crossover more than that for a fourth order three way. So that lost power generates heat in the crossover and will burn up the crossover. You will not be able to use the power those speakers are capable of, with passive crossovers.

The types of drivers I see on the product list are best used with electronic active crossovers, that are inserted before the power amplifiers, with the speakers being connected directly to the amplifiers.

 

[NAD3020]

Woofer: lf15n301
Midrange: mr10n301
Tweeter: nd3020-t3

I'll use them in my house and i'm not buying them because of the power they can handle, but because of the sound quality, i just love them.
I've already built a loudspeaker with cheaper RCF components and a 12db crossover and they used to sound wonderfully with my NAD3020 (now broken, that's why i changed amp).

No one wants to tell me the tolerance a cap for a 24db crossover should have?

The phase shift of a fourth order crossover is 360 degrees at crossover.

360° phase shift = 0° phase shift = no phase shift?

 

[TLS Guy]

Woofer: lf15n301
Midrange: mr10n301
Tweeter: nd3020-t3

I'll use them in my house and i'm not buying them because of the power they can handle, but because of the sound quality, i just love them.
I've already built a loudspeaker with cheaper RCF components and a 12db crossover and they used to sound wonderfully with my NAD3020 (now broken, that's why i changed amp).

No one wants to tell me the tolerance a cap for a 24db crossover should have?



360° phase shift = 0° phase shift = no phase shift?

The tolerance of the caps is the least of your worries. A 360 degree shift is phase is a shift and puts the mid a cycle ahead of the woofer at crossover and the tweeter a whole cycle ahead of the mid at crossover.

However you still have no clue how to design a crossover as the order of the drivers and acoustic slopes are part of the total equation, and as I pointed out, many other factors as well. For instance the inductance and DC resistances of of the voice coils affect all calculations. You must model a crossover before building it.

 

[highfigh]

Woofer: lf15n301
Midrange: mr10n301
Tweeter: nd3020-t3

I'll use them in my house and i'm not buying them because of the power they can handle, but because of the sound quality, i just love them.
I've already built a loudspeaker with cheaper RCF components and a 12db crossover and they used to sound wonderfully with my NAD3020 (now broken, that's why i changed amp).

No one wants to tell me the tolerance a cap for a 24db crossover should have?



360° phase shift = 0° phase shift = no phase shift?

The tolerance should be whatever you want to pay for. If you want accuracy and remove variables, go with tighter tolerances but you need to realize that the driver's tolerances under load will probably negate the tighter tolerance of the cap(s) and coil(s). Also, if you drive the speakers hard, the voice coils will become warm/hot and that leads to increased resistance, altering the crossover points. In extreme cases, it leads to thermal compression.

360° can't equal 0°. The relative position is the same but in absolute terms, it's not.

In music, this would be considered "playing behind the beat".

 

[NAD3020]

The tolerance of the caps is the least of your worries. A 360 degree shift is phase is a shift and puts the mid a cycle ahead of the woofer at crossover and the tweeter a whole cycle ahead of the mid at crossover.

How do i solve this then?

However you still have no clue how to design a crossover as the order of the drivers and acoustic slopes are part of the total equation, and as I pointed out, many other factors as well. For instance the inductance and DC resistances of of the voice coils affect all calculations. You must model a crossover before building it.

Then tell me what i should do.
Please, teach me everything you guys think i should learn, i'm a noob, i know that, i'm asking for some guidance.

 

[TLS Guy]

How do i solve this then?



Then tell me what i should do.
Please, teach me everything you guys think i should learn, i'm a noob, i know that, i'm asking for some guidance.

There are problems in designing speakers that are insoluble and you are working you way to understanding just one of them.

As a noob you are not going to be able to design a three way crossover, without a lot of research, reading and experiment. There is more than we can teach you on a forum. If you want to go ahead you will have to trust me to design one for you.

However a model is just that, and there will be real world differences. It is really tough to design a crossover without the speakers in your hands to modify the crossover as required at the end of construction. This is particularly true of three ways, which are orders of magnitude harder than two ways. I never advise the novice to start with a three way crossover.

 

[Swerd]

I'm coming into this discussion late, but I have to say that I agree with TLS Guy and all the others who are telling you that your plans to "throw" a crossover at your drivers will not result in success.

To get an idea of what is involved in crossover design, read this link. If, after reading it, you find that it is above your understanding of crossovers, good. Consider that your first step in learning.

Your question about capacitor tolerances deserves an answer despite the fact that, as TLS Guy said, it is the least of your problems.

In DIY speaker building, a simple rule-of-thumb says that capacitor values ±10% will work in a crossover. So, if a design calls for 7 µF, and you can only find capacitors of 6.8 or 7.5 µF, either one will work. I would choose the 6.8 µF cap and not loose any sleep over it. Spending extra money on capacitors with tighter manufacturing tolerances than about ±5% is a waste.

Of course, all this assumes that you have a properly designed crossover.

 

[GranteedEV]

How do i solve this then?



Then tell me what i should do.
Please, teach me everything you guys think i should learn, i'm a noob, i know that, i'm asking for some guidance.

Have you read vance dickason's book? While it may not quite be for total beginners it's full of lots of very useful information.

 

[NAD3020]

Have you read vance dickason's book? While it may not quite be for total beginners it's full of lots of very useful information.

Thanks man...

There are problems in designing speakers that are insoluble and you are working you way to understanding just one of them.

As a noob you are not going to be able to design a three way crossover, without a lot of research, reading and experiment. There is more than we can teach you on a forum. If you want to go ahead you will have to trust me to design one for you.

Would you really design one?
Man that would be great...

I still don't know if i'm getting Ciare speakers instead, they're a little bit cheaper than RCF ones but at least i would have the money for a NAD C515BEE.

In DIY speaker building, a simple rule-of-thumb says that capacitor values ±10% will work in a crossover. So, if a design calls for 7 µF, and you can only find capacitors of 6.8 or 7.5 µF, either one will work. I would choose the 6.8 µF cap and not loose any sleep over it. Spending extra money on capacitors with tighter manufacturing tolerances than about ±5% is a waste.

If i need theorically, after i did the math, a 7,05µF cap, i will put in parallel four 1,5µF cap and one 1µF cap.
The tolerance i chose was ±0,1µF. But if you're saying there's no difference between 6,8 and 7, alright, thanks for the advice.

Thanks for the links.

 

[Swerd]

If i need theorically, after i did the math, a 7,05µF cap, i will put in parallel four 1,5µF cap and one 1µF cap.

I'm not sure you understood the first point in that article I linked. It said:

1.) Textbook formulas will not produce a satisfactory crossover because……Textbook formulas assume a resistive load, like 8 ohms, when in reality a speaker’s impedance is very reactive. As a result the only way to design a crossover that works as intended is to measure the real impedance of the speaker and use this complex impedance in the design stage.​

By "textbook formulas" he means that math formulas used to calculate crossovers all assume a single value for speaker impedance, like 8 ohms, when impedance varies with the frequency. That's what "reactive" means. The textbook math formulas ask for a single value for impedance, when in reality you need to know the impedance data for the full range of frequencies that you intend to use for the driver.

The tolerance i chose was ±0,1µF.

The manufacturer of the capacitors chooses the tolerance, not you. A tolerance of 0.1 µF is unrealistic. And if it were available, it would be excessively expensive.

 

[TLS Guy]

Thanks man...



Would you really design one?
Man that would be great...

I still don't know if i'm getting Ciare speakers instead, they're a little bit cheaper than RCF ones but at least i would have the money for a NAD C515BEE.



If i need theorically, after i did the math, a 7,05µF cap, i will put in parallel four 1,5µF cap and one 1µF cap.
The tolerance i chose was ±0,1µF. But if you're saying there's no difference between 6,8 and 7, alright, thanks for the advice.

Thanks for the links.

You can't design a crossover round theoretical values. If you use CIARE drivers I can't design a crossover for you, as they don't provide sufficient data. If you use those drivers you would have to do your own measurements before I could start. You can't design a crossover either without accurate measurements.

 

[NAD3020]

You can't design a crossover round theoretical values. If you use CIARE drivers I can't design a crossover for you, as they don't provide sufficient data. If you use those drivers you would have to do your own measurements before I could start. You can't design a crossover either without accurate measurements.

Which data do you need? On another forum i know a guy that works for Ciare and i'm pretty sure he knows what we need...

I'm not sure you understood the first point in that article I linked.

The only problem here is my english...

It said:

1.) Textbook formulas will not produce a satisfactory crossover because……Textbook formulas assume a resistive load, like 8 ohms, when in reality a speaker’s impedance is very reactive. As a result the only way to design a crossover that works as intended is to measure the real impedance of the speaker and use this complex impedance in the design stage.​

By "textbook formulas" he means that math formulas used to calculate crossovers all assume a single value for speaker impedance, like 8 ohms, when impedance varies with the frequency. That's what "reactive" means. The textbook math formulas ask for a single value for impedance, when in reality you need to know the impedance data for the full range of frequencies that you intend to use for the driver.

Knew this.

The manufacturer of the capacitors chooses the tolerance, not you. A tolerance of 0.1 µF is unrealistic. And if it were available, it would be excessively expensive.

But i choose the tolerance when i do the math to calculate which caps i have to put in parallel when i'm projecting the crossover.
If i did the math and the result was 67,1µF i would've put a couple of 33µF caps and one 1µF cap, where probably for you a couple of 33µF caps would've been enough.

nevermind anyway.

 

[TLS Guy]

To design a crossover a full set of the Thiel/Small parameters are required for all drivers. Also the acoustic data is required for each driver. This needs to be an accurate frequency response graph, at least on axis and preferable off axis as well. The entire bandwidth of the driver must be shown so that any peak/cone break up modes that may have to notched can be dealt with.

I need to point out that we need to model the crossover before you purchase any drivers. It is not possible to make a passive crossover for a lot of driver combinations. Appropriate driver selection is key to getting a good overall design.

As a further remark, I note you seem to be setting your sites on pro drivers. In general pro drivers make for very poor domestic speakers. They are intended for active designs in large spaces.

Since this is a domestic application you are much better off selecting good domestic drivers from companies such as Vifa, Scanspeak, Peerless and SEAS, and a number of others.

As you probably gather I have serious reservations about how you are going about this project. I have to tell you I'm seeing pretty much a full house of beginner mistakes.