How much lower than 8uF what increments?
Will less power to the tweeter result in less volume?
Thank you
Two things that the woofer crossover affects...where it cuts out (it's rolloff), and the shape/steepness etc. They should match the tweeter or one or both will require adjustment to match.
Other examples: You might change your brand of tweeter and if you get the response the same as the old one you could leave the woofer crossover as it is...and you can (generally speaking) change the shape or cut off point for both (with the original woofer and tweeter) and still be OK matching them at the crossover.
Even more reasons why circuits might look different: There are a number of different ways to produce a circuit that will do exactly the same thing. You'll also sometimes find that some parts of a complicated crossover are actually making very small changes to what the circuit is doing, and it would sound reasonable even without them.
First of all, lemme just say that this is an amazing article. I am so gratified pouring into the knowledge of these great audiophiles writing here.
Thanks, you've helped me immensely in my quest for a better understanding of speaker technology.
I just wonder how on earth you came up with the number 12.6 for the tweeter's crossover?
I'm not doubting ya, it's probably some mathematical constant based on the physical properties of the sound, but i am none the less curious as to how or why.
Thanks, you've helped me immensely in my quest for a better understanding of speaker technology.
I just wonder how on earth you came up with the number 12.6 for the tweeter's crossover?
I'm not doubting ya, it's probably some mathematical constant based on the physical properties of the sound, but i am none the less curious as to how or why.
I registered just to say thank you AllenB - thanks for a fantastic article !
I've had cobbled together speakers all my life, and I'm finally going to bite the bullet, buy some brand new cones and try to make something of my own that I can be proud of.
The crossover has always put me off in the past, seeming like a hybrid of advanced physics, extreme mathematics and voodoo (possibly benefiting from the sacrifice of several small animals).
Your article doesn't make it look easy, but at least now it does look 'do-able'.
Thank you for your time mate
I've had cobbled together speakers all my life, and I'm finally going to bite the bullet, buy some brand new cones and try to make something of my own that I can be proud of.
The crossover has always put me off in the past, seeming like a hybrid of advanced physics, extreme mathematics and voodoo (possibly benefiting from the sacrifice of several small animals).
Your article doesn't make it look easy, but at least now it does look 'do-able'.
Thank you for your time mate
I tried estimate substitutions on a graphing calculator and found out it can make the system do some weird things if the overlaps start wrestling. Things like low impedance dips , crazy zig zags in the spl levels around the crossover. Adjusted both sides of the xover and it smoothed the spl plot very well.
I hope this is read as helpful, not criticism. The thread is excellent.
I hope this is read as helpful, not criticism. The thread is excellent.
This is good, and how it ideally happens. The drivers are mounted close to each other using a similar slope and cut-off frequency and their responses complement each other, and their phase response remains parallel.
Often one driver is closer to the listening position than the other. There will be a change in phase related to this. If one slope is changed this can trade the response level of the driver against the response level of the two in combination (phase), trying to find the best response.
Ideally the crossover will cause no peaking in the impedance, or maybe a slight smooth dip through the region.
I've been reading this post several times, finding it very informing and inspiring.
Finally I’ve decided to use the concept to work out a new crossover solution for my Opera Tebaldi speakers. They’re a beauty to look at, but the sound quality is fatiguing. As far as I’ve found out the problem seems to be the two sets of mid-woofers that play too loud, and have far too high crossover value (I’ve seen reviews saying the value is 7500Hz.)
My first shot was to replace the original crossover with a new solution using better quality parts and cables, but keeping the same parts value.
That has not been very successful.
So I’ll give it another go with new crossovers, using the calculations given by Allan.
I have a few questions that I hope someone can answer:
1) The speaker has two mid-woofers, and two woofers. Would you use the same crossover values for both elements of the same kind? If so, would two elements in parallel make impact on the calculations?
2) There are some references to notch filters in the thread. Is there a straight forward explanation for how to make use of these filters?
Ole Petter
Finally I’ve decided to use the concept to work out a new crossover solution for my Opera Tebaldi speakers. They’re a beauty to look at, but the sound quality is fatiguing. As far as I’ve found out the problem seems to be the two sets of mid-woofers that play too loud, and have far too high crossover value (I’ve seen reviews saying the value is 7500Hz.)
My first shot was to replace the original crossover with a new solution using better quality parts and cables, but keeping the same parts value.
That has not been very successful.
So I’ll give it another go with new crossovers, using the calculations given by Allan.
I have a few questions that I hope someone can answer:
1) The speaker has two mid-woofers, and two woofers. Would you use the same crossover values for both elements of the same kind? If so, would two elements in parallel make impact on the calculations?
2) There are some references to notch filters in the thread. Is there a straight forward explanation for how to make use of these filters?
Ole Petter
With a speaker like this, I'd probably assume that cables and component brands would do little to improve it. I also think that since using this method of crossover design isn't going to take advantage of the measurements taken by the manufacturer, it would be best to keep as much of the original crossover as you reasonably can.
It's another thing if you identify a reason maybe to lower the crossover point. My concern with that would be that the current tweeter will pushed much harder. While I probably wouldn't choose 7k5Hz for this crossover point, the manufacturer did. As a guess I'd think that crossing the tweeters closer to 3Khz might have an effect on them, not to mention that 3kHz is a region where you really need to get it right.
Have you considered tweaking the crossover as it is?
It's another thing if you identify a reason maybe to lower the crossover point. My concern with that would be that the current tweeter will pushed much harder. While I probably wouldn't choose 7k5Hz for this crossover point, the manufacturer did. As a guess I'd think that crossing the tweeters closer to 3Khz might have an effect on them, not to mention that 3kHz is a region where you really need to get it right.
Have you considered tweaking the crossover as it is?
Thanks for a quick response.
I have considered tweaking the existing filters. Actually I have made my own with new parts, but I find the construction complex to change.
I have included a picture of the crossover design. The manufacturer claims the crossover for the mid-woofer is 2 KHz, but a test review I have come across point out that it is in fact closer to 7,5kHz.http://www.opera-lautsprecher.de/downloads/tests/Opera_Tebaldi_AudioReview_2005.pdf
My idea was to go for a refresh start with new filters that I can tweak from a point where I know what I’m doing, unless you see obvious changes I can do to the existing design?
Ole Petter
Thanks for the link this does help but I'm sorry, I can't read italian. I notice that there is a filter at 7k5 but I am guessing this is not the upper crossover point. This looks as if it might be lower.
Have you been through the process of finding the best room location for these and aiming them (not always directly at yourself)?
The first tweak I would then look at would be the small series resistor in the mid crossover (change it from 1 to maybe 2 or 3 ohms), and decide whether this is making an improvement. If it does, the right amount to change might even be less than this.
Have you been through the process of finding the best room location for these and aiming them (not always directly at yourself)?
The first tweak I would then look at would be the small series resistor in the mid crossover (change it from 1 to maybe 2 or 3 ohms), and decide whether this is making an improvement. If it does, the right amount to change might even be less than this.
Thanks AllenB,
I'll give that a try. Still think I'll make a new filter for the bandpass. I have all the parts I need, and with three terminals it’s easy to put it externally and play around with some alternatives. I also have a set of Scan-Speak 15W/8530K00. They may turn out to be a bit more pleasant than Seas E0015 W15CY-001 that are in use now. You may argue that it may not be a wise thing to do. But to me it’s a way of learning by trial and error. I’ll make sure to keep the original untouched, so I can return the speaker in its original version.
Ole Petter
I'll give that a try. Still think I'll make a new filter for the bandpass. I have all the parts I need, and with three terminals it’s easy to put it externally and play around with some alternatives. I also have a set of Scan-Speak 15W/8530K00. They may turn out to be a bit more pleasant than Seas E0015 W15CY-001 that are in use now. You may argue that it may not be a wise thing to do. But to me it’s a way of learning by trial and error. I’ll make sure to keep the original untouched, so I can return the speaker in its original version.
Ole Petter
Like others have said, this is great write-up for a beginner. Of course, having questions answered usually leads to more questions being raised. I'll try to organize my questions by number so its easier pick them out. I'm just wanting to make sure that what I've learned/concluded so far is correct or if I'm totally off base. For reference my background is in mechanical engineering but I've studied some electronics - mostly DC L,R,C series/parallel/combination and digital technology but I am venturing into the AC realm since I want to build speakers
1. My main question is about the voltage/current phase shift that occurs when using inductors and capacitors. Is this a concern when designing a crossover? Is there a way of calculating the power factor of a crossover circuit? From what I've read, 45* and 90* phase shifts are the worse case scenarios where 2x or 4x the power will have to be dissipated by the amplifier.
2. The different orders of filters - mostly looking for verification here. 1st order is 6dB/octave, 2nd order is 12dB/octave, 3rd order...4th order... on down the line doubling dB/octave each time. Its my understanding that the higher order filters are used to cross over rapidly from one driver to the next and would be used when there is not a lot of overlap between the two drivers' frequency response. 1st and 2nd order filters provide a smoother, more gradual transition from one driver to the next and can be used when there is a lot of overlap in frequency response. What exactly makes a filter 1st, 2nd, 3rd order, etc. both in terms of components and layout?
3. Sensitivity - Assuming two speakers are of equal impedance, will a speaker with a higher sensitivity rating sound louder when driven with the same amount of power? That's how I interpret the rating of dB per 1W or 2.83V but I have read some articles that have confused me about this.
Thanks in advance for taking the time to help me out!
1. My main question is about the voltage/current phase shift that occurs when using inductors and capacitors. Is this a concern when designing a crossover? Is there a way of calculating the power factor of a crossover circuit? From what I've read, 45* and 90* phase shifts are the worse case scenarios where 2x or 4x the power will have to be dissipated by the amplifier.
2. The different orders of filters - mostly looking for verification here. 1st order is 6dB/octave, 2nd order is 12dB/octave, 3rd order...4th order... on down the line doubling dB/octave each time. Its my understanding that the higher order filters are used to cross over rapidly from one driver to the next and would be used when there is not a lot of overlap between the two drivers' frequency response. 1st and 2nd order filters provide a smoother, more gradual transition from one driver to the next and can be used when there is a lot of overlap in frequency response. What exactly makes a filter 1st, 2nd, 3rd order, etc. both in terms of components and layout?
3. Sensitivity - Assuming two speakers are of equal impedance, will a speaker with a higher sensitivity rating sound louder when driven with the same amount of power? That's how I interpret the rating of dB per 1W or 2.83V but I have read some articles that have confused me about this.
Thanks in advance for taking the time to help me out!
Not particularly. Of course you need to take the crossover phase issues into account so that the response from a given pair of drivers will be reasonably in phase, e.g. in the direction of the listening position.
The total response has a phase response that corresponds to the individual drivers and filters (including the unseen filters that make up the response of the drivers themselves). This phase response is usually not of much concern but in extreme cases it can be enough to affect what is heard.
Where the drivers don't sum completely (which along with the crossover includes their physical configuration), some of the sound will be lost by way of cancellations. Lobing normally occurs where the drivers become further/closer looking from above/below with relation to each other. This isn't necessarily a problem and is a common occurence.
Notwithstanding this, the power to the circuit can be assessed by looking at the complex impedance into the crossover (not including excess sources of current sinking, obviously). Normally the impedance isn't too far from consistent and resistive, but you need to take into account a few different aspects of the drivers and their enclosures, for example, to know whether such a flat impedance is actually desirable in the bigger picture.
Last edited:
Yes, where third is 18 and fourth is 24.
It is a good starting point to use no less but no more than necessary. Designing a crossover begins with the physical aspects of the drivers/cabinets. If you have designed yourself into a situation where the drivers need to be crossed near their limits, any issues they have at the fringes will of course be asking for a steeper crossover. This is fine to a point, an extremely fast crossover or a driver/cabinet outside its limits can be a bad thing.
The standard topology starts with a single series component making a first order filter. A further component in parallel with the load makes it second. Each new component can go in series/parallel etc.
Some refer to the filter based simply on this. The impedance of the driver (the load for the filters) will affect their response. This can appear to create a total (e.g.) 1.5 order filter but in reality this would be more of a coincidence where it did.
Furthermore, the driver's response is subject to mechanical filtration and the total combined response of the driver/cabinet/electrical filtration including driver load, makes something that might be referred to as the acoustical response.
Different designations for different situations of course. The acoustical response is probably relevant more often.
Yes.
The efficiency of the motor is one aspect of this. Efficiency is normally traded around bandwidth. Acoustical conditions, the cabinet etc also have an effect, e.g. efficiency may change for a given cabinet above a certain frequency based on the radiation angle.