I'm currently building a pair of transmission line speakers for my final year project, and they're going to be active.
I've dealt with most of the speaker design and amp modules etc...
But I'm getting held up on building the crossover.
I've sorted out rectification, and a dual rail supply, so I've got +12v, -12v and 0v.
I've got the Douglas Self book, but most of the designs are Sallen-Key.
My question being - if I built a 4th order Linkwitz Riley slope in a basic Sallen-Key design, with a buffer... Will that be at all suitable for use in my active speakers? It seems a little... basic.
I know that I haven't considered time delay yet, but just as a basic crossover, will this provide any sort of realistic sound quality? (Let's presume for now that it's using NE5534 opamps, although I'm open to suggestions)
Just to give an idea of what sort of "basic" I mean, something similar to this (pulled from google images)
http://sound.westhost.com/p09_fig1a.gif
I have actually built a circuit similar to that in Multisim and calculated my value, and the frequency responses are fine (although the voltages seem bizarre), but I know that doesn't relate to how it would sound.
Thanks in advance for any help, and sorry about the long post!
I've dealt with most of the speaker design and amp modules etc...
But I'm getting held up on building the crossover.
I've sorted out rectification, and a dual rail supply, so I've got +12v, -12v and 0v.
I've got the Douglas Self book, but most of the designs are Sallen-Key.
My question being - if I built a 4th order Linkwitz Riley slope in a basic Sallen-Key design, with a buffer... Will that be at all suitable for use in my active speakers? It seems a little... basic.
I know that I haven't considered time delay yet, but just as a basic crossover, will this provide any sort of realistic sound quality? (Let's presume for now that it's using NE5534 opamps, although I'm open to suggestions)
Just to give an idea of what sort of "basic" I mean, something similar to this (pulled from google images)
http://sound.westhost.com/p09_fig1a.gif
I have actually built a circuit similar to that in Multisim and calculated my value, and the frequency responses are fine (although the voltages seem bizarre), but I know that doesn't relate to how it would sound.
Thanks in advance for any help, and sorry about the long post!
penguinpaul,
That is exactly the road I am going down for my own design. The L/R 4 alignment is analogous to the same thing in a passive crossover of the same design and is one of the most popular designs while staying away from 2nd and 3rd order filters that you would have read about in the Self book on electronic crossovers. I would have no problem using this type of filter for many different speaker design applications.
That is exactly the road I am going down for my own design. The L/R 4 alignment is analogous to the same thing in a passive crossover of the same design and is one of the most popular designs while staying away from 2nd and 3rd order filters that you would have read about in the Self book on electronic crossovers. I would have no problem using this type of filter for many different speaker design applications.
Penguinpaul, I fully concur with Kindhornman. Just one small addition.
What is very important is that the phase behaviour of the drivers arround their crossover frequencies match. Linkwitz Riley does this very well, electrically. But, when designing your crossover, you also will have to look at the phase shifts that are caused by the natural roll off of drivers. Both the filter and the driver are minimum phase, which means that there is 90 degree phase shift per filter order. Drivers in a sealed enclosure typically roll off 12 dB/octave, which is a second order filter slope.You either stay well away from the natural roll off frequency of your drivers, or you have to incorporate them into your design. For example, if your midrange has a natural roll off below 400 Hz, you can cross over at this point, but you will only need a 2nd order electrical filter to achieve the desired slope.
I hope you can do measurements, because othwerwise it will be difficult to get it right.
What is very important is that the phase behaviour of the drivers arround their crossover frequencies match. Linkwitz Riley does this very well, electrically. But, when designing your crossover, you also will have to look at the phase shifts that are caused by the natural roll off of drivers. Both the filter and the driver are minimum phase, which means that there is 90 degree phase shift per filter order. Drivers in a sealed enclosure typically roll off 12 dB/octave, which is a second order filter slope.You either stay well away from the natural roll off frequency of your drivers, or you have to incorporate them into your design. For example, if your midrange has a natural roll off below 400 Hz, you can cross over at this point, but you will only need a 2nd order electrical filter to achieve the desired slope.
I hope you can do measurements, because othwerwise it will be difficult to get it right.
Thanks for the replies guys.
It's a 2 way design - my driver (theoretically) starts to roll off at 3.6kHz. My tweeter (again theoretically) rolls off at 2kHz.
I am planning on somewhere around 3kHz for my crossover point.
I've previously done an assignment on the phase changes of filters, especially butterworth, so I understand that - however, Douglas Self suggests that these phase responses are not really audible. Does this sound accurate, or is it subjective?
I will try and sketch out what will happen above 3.6kHz just so I can make sure nothing too bad happens.
The circuit schematic I linked to - is that really adequate? Will that simple circuit really be usable for audio?
It's a 2 way design - my driver (theoretically) starts to roll off at 3.6kHz. My tweeter (again theoretically) rolls off at 2kHz.
I am planning on somewhere around 3kHz for my crossover point.
I've previously done an assignment on the phase changes of filters, especially butterworth, so I understand that - however, Douglas Self suggests that these phase responses are not really audible. Does this sound accurate, or is it subjective?
I will try and sketch out what will happen above 3.6kHz just so I can make sure nothing too bad happens.
The circuit schematic I linked to - is that really adequate? Will that simple circuit really be usable for audio?
Penguin,
I do not agree that you have to have the acoustical function end with a fourth order slope, if the slope does have some rolling from the device itself then it only adds to the stop band and is not in my opinion going to be an audible effect. Many designers have gone with much higher slopes such as 48db per octave without a problem and I see this chasing of the acoustical slope requirement having to limit to the acoustical slope response only as being a true red herring in the overall course of things.
I do not agree that you have to have the acoustical function end with a fourth order slope, if the slope does have some rolling from the device itself then it only adds to the stop band and is not in my opinion going to be an audible effect. Many designers have gone with much higher slopes such as 48db per octave without a problem and I see this chasing of the acoustical slope requirement having to limit to the acoustical slope response only as being a true red herring in the overall course of things.
If anything, looking at that overall slope would be just to show that I understand what's happening - just for the sake of extra detail in my coursework.
- Mostly because I'm not specifically after a 4th order slope - I'm after something that will get the job done, so if it starts to roll of at more than 24dB per octave after 3.6kHz, this isn't a problem.
- Mostly because I'm not specifically after a 4th order slope - I'm after something that will get the job done, so if it starts to roll of at more than 24dB per octave after 3.6kHz, this isn't a problem.
Phase shift by itself is not readily audible, so you don't have to worry too much about that.
The other issue is the phase shifts the drivers exhibit around crossover. If these don't align properly, the result will be a strange lobing pattern and no bloody way to get the FR flat at more than one location.
The other issue is the phase shifts the drivers exhibit around crossover. If these don't align properly, the result will be a strange lobing pattern and no bloody way to get the FR flat at more than one location.
Vacuphile,
I bring it up over and over and it seems to go over most peoples heads that a conjugate LRC network on the speaker terminals and not a simple RC Zobel network does much to solve not only the impedance rise of the device and the tendency to turn to a reactive shift is controlled with the conjugate network and also will flatten the phase shift greatly. This is should be the first part of any network design before analysis of any other section of the crossover. Once this situation is taken care of the rest of the crossover becomes much less of a critical problem.
I bring it up over and over and it seems to go over most peoples heads that a conjugate LRC network on the speaker terminals and not a simple RC Zobel network does much to solve not only the impedance rise of the device and the tendency to turn to a reactive shift is controlled with the conjugate network and also will flatten the phase shift greatly. This is should be the first part of any network design before analysis of any other section of the crossover. Once this situation is taken care of the rest of the crossover becomes much less of a critical problem.
- Status
- Not open for further replies.