Can I mix orders within a 3-way LR crossover (EG tweeter slope: 24db; mid-range slope(s): 12db; bass slope: 24db)? Or will I mess up the benefits of the LR system? (IE the phase constancy across the freq range).
I know Linn did something like this with at least one of their active crossovers, but not sure how.
Background: I'm using Elliot Audio (ESP) boards, plus the ESP component calculator, and I currently have a 'by-the-book' two way system (also ESP) which sounds stunning.
I know Linn did something like this with at least one of their active crossovers, but not sure how.
Background: I'm using Elliot Audio (ESP) boards, plus the ESP component calculator, and I currently have a 'by-the-book' two way system (also ESP) which sounds stunning.
Last edited:
The electrical slopes alone don't mean much. The combined electrical and driver response should have the slope you are looking for. So you need to measure the acoustic response.
But , yes, I think the slopes at the crossover points should be similar for it to add up as predicted. If slopes are different the phase will also be a problem. But the electrical slopes could be different for LF and HF as long as the acoustic output of the drivers is what you are trying to achieve. Depends on each drivers response and where you cross them over.
But as I said you could get a 24 dB slope with an electrical slope less than what you want because the driver also contributes to it.
But , yes, I think the slopes at the crossover points should be similar for it to add up as predicted. If slopes are different the phase will also be a problem. But the electrical slopes could be different for LF and HF as long as the acoustic output of the drivers is what you are trying to achieve. Depends on each drivers response and where you cross them over.
But as I said you could get a 24 dB slope with an electrical slope less than what you want because the driver also contributes to it.
Last edited:
I'm not sure what you're trying to do, which ESP circuit you're trying to use and what your skill is.
But you seem to think about transient perfect (minimum phase) crossover, specifically the Duelund 3-way.
Of course you can mix the slopes. The LR benefit that you miss is the flat summation at xo frequency, not the phase. But with active xo, the non flat response can be easily equalized.
If you're using passive xo, there's similar way to do it but it seems not what you're looking for.
Thanks all. Since yesterday I found out that for the Linn active crossover, they crossed bass to mid using 12db and mid to high at 24db, which is different to what I suggested earlier. However, the mid has both a 12db and 24db slope, so wouldn't there be a 180 degree mismatch, somewhere?
On the Linkwitz site I found this: Frequently Asked Questions A sentence there suggests why Linn may have chosen 12db for the bass to mid. BTW, I'm using the Kef B110 and B139, the same drivers for which the Linn crossover was designed.
On the Linkwitz site I found this: Frequently Asked Questions A sentence there suggests why Linn may have chosen 12db for the bass to mid. BTW, I'm using the Kef B110 and B139, the same drivers for which the Linn crossover was designed.
Hi,
Here is a design that uses second and fourth order
acoustic L/R slopes : Zaph|Audio - ZDT3.5
Note the driver phase is correct at both x/o points.
rgds, sreten.
Here is a design that uses second and fourth order
acoustic L/R slopes : Zaph|Audio - ZDT3.5
Note the driver phase is correct at both x/o points.
rgds, sreten.
To answer your question, asymmetric slopes don't matter at all.
Don't bother looking at the literature for "the best" crossover type. All of these studies assume that the driver has an infinite passband and a constant phase of zero. Once you throw the driver(s) into the mix, all bets are off.
What is important is to get the output from the drivers that operate together within a crossover region operating in phase and with matched amplitudes (SPLs). What comes out of the driver is the sum of the electrical filters and the driver's own response. Since the driver's own response and the cabinet edge diffraction all influence the amplitude and phase responses these must be included as well. Where the responses are measured is also important. You can get a crossover to sum nicely at one point in space only, because as you move away from that point the relative phase of each driver changes and eventually there will be interference of one kind of another. So where you plan to listen to the speakers relative to their front baffle should also be taken into account.
I usually do some extensive modeling to plan out the loudspeaker and look for issues. Once I have the driver layout and cabinet size/shape figured out, I build the cabinet, install the drivers, and then take lots of measurements. I mean LOTS. On axis, off axis, driver in pairs, nearfield measurements - all of these are important and used in the crossover design in one way or another.
Once you have all the measurements, you can use a comprehensive crossover design tool/package/software to design a crossover that really works as you intend it to. These do not need to be commercial by any means. There is an excellent free passive crossover design tool built in Excel called "Passive Crossover Designer" by Jeff Bagby. I created an active crossover design tool called "Active Crossover Designer" that also uses Excel. Both of these are complicated programs but you can find lots of support for either one in DIY forums (I'm happy to answer questions).
There are some who suggest that you can develop a crossover without any measurements, but I think that this is often a waste of time unless you really know what you are doing.
Don't bother looking at the literature for "the best" crossover type. All of these studies assume that the driver has an infinite passband and a constant phase of zero. Once you throw the driver(s) into the mix, all bets are off.
What is important is to get the output from the drivers that operate together within a crossover region operating in phase and with matched amplitudes (SPLs). What comes out of the driver is the sum of the electrical filters and the driver's own response. Since the driver's own response and the cabinet edge diffraction all influence the amplitude and phase responses these must be included as well. Where the responses are measured is also important. You can get a crossover to sum nicely at one point in space only, because as you move away from that point the relative phase of each driver changes and eventually there will be interference of one kind of another. So where you plan to listen to the speakers relative to their front baffle should also be taken into account.
I usually do some extensive modeling to plan out the loudspeaker and look for issues. Once I have the driver layout and cabinet size/shape figured out, I build the cabinet, install the drivers, and then take lots of measurements. I mean LOTS. On axis, off axis, driver in pairs, nearfield measurements - all of these are important and used in the crossover design in one way or another.
Once you have all the measurements, you can use a comprehensive crossover design tool/package/software to design a crossover that really works as you intend it to. These do not need to be commercial by any means. There is an excellent free passive crossover design tool built in Excel called "Passive Crossover Designer" by Jeff Bagby. I created an active crossover design tool called "Active Crossover Designer" that also uses Excel. Both of these are complicated programs but you can find lots of support for either one in DIY forums (I'm happy to answer questions).
There are some who suggest that you can develop a crossover without any measurements, but I think that this is often a waste of time unless you really know what you are doing.
"basically just very poor for real hifi"?
I have a copule of questions;
What is real hifi?
Why do you make such bold assertions without the slightest bit of backup?
There are many members (me included) who have active systems and would not go back to passive. What makes active so poor, in your estimations?
Abs
Last edited:
A single frequency "cross-over" is TWO Filters.
A High Pass allowing a limited signal to pass to the driver handling the upper frequencies.
A Low Pass allowing a limited signal to pass to the driver handling the lower frequencies.
These two filters can be in series, or in parallel.
The parallel version is easier to analyse.
Most active crossovers use the parallel version. There may be exceptions, but I cannot recall seeing a series connected active crossover filter.
When a 3way is required, one of the TWO Filters is cascaded with TWO more Filters.
When you look at the TWO Filters operating at the same frequency, you usually find they are symmetrical. There are good (technical) reasons for these to not be symmetrical.
A High Pass allowing a limited signal to pass to the driver handling the upper frequencies.
A Low Pass allowing a limited signal to pass to the driver handling the lower frequencies.
These two filters can be in series, or in parallel.
The parallel version is easier to analyse.
Most active crossovers use the parallel version. There may be exceptions, but I cannot recall seeing a series connected active crossover filter.
When a 3way is required, one of the TWO Filters is cascaded with TWO more Filters.
When you look at the TWO Filters operating at the same frequency, you usually find they are symmetrical. There are good (technical) reasons for these to not be symmetrical.
There is absolutely nothing wrong with active crossovers, indeed they are far better than passive ones when a little thought is applied to their design. Well, even with no thought they are better than a passive with no thought.
What sreten is saying is that 99% of active crossovers are designed with little or no regard to the acoustic response of the drivers. In a way he is tarring all active crossovers with the same brush. He is fed up of seeing projects showing 4th-order Linkwitz-Riley electrical filter crossovers being passed off as being just what you need to get great results.
As you pointed out, if there is sufficiently wide response beyond the filters, you can treat the design as a theoretical purely electrical design. The problem is that there are not really that many drivers you can do that with. So the result is that you need to deviate from textbook filter responses and crossover points. You are likely to need a different slope/order filter on the bass/mid low-pass to the treble high-pass. You will probably need to tune the filters Q to arrive at your desired acoustic Q. There may also be a need for notch filtering or baffle step correction as well.
You MUST do acoustic measurements to design any crossover and an active crossover does not have any less need to do so. Yes, it can have steeper slopes which can reduce the problem, but it still doesn't mean you can use a textbook filter.
But I think most people in this thread are saying the same.
Last edited:
That's why an active is subjectively better than passive. Just like tube amps, you can design one with eyes closed and it will still sound good.
How good we are at active and passive will determine our success to implement each of them. Users with no skill, they can build ESP's active with any drivers and ready to go. They cannot do equally well with passive (e.g. ready made crossover).
Many people who can design both active and passive, go back to passive after some time with active. One important exception is Linkwitz. I think it is because his electronic skill is so superior and he can see the good commercial aspect in the active.
My problem so far with active is the quality of the electronics. Many people try to go beyond LM3886 (Linkwitz seems okay with this chip amp). They try hundreds of Watts iddle power in amps, pure class-A, high slew rate, all-tubes and iron, etc. These active components of active crossover is really a PITA in this situation.
That is true. If somebody is going to slap together a crossover, active is the lesser of two evils.
I think there are real pitfalls when a DIYer attempts to replace a well designed passive crossover in a well designed commercial speaker, with a textbook active. People doing this then making a comparison of active vs passive, well the contest is invalid.
But for a beginner wanting to build a new speaker, despite the failings of a textbook active, it gets them going. I've done it myself as I'm sure many on here have. It just means there's room for improvement. And as long as the person understands that later on he's going to have to measure their speaker and make changes, that's fine. Sometimes we have to build something and get a little reward, to help keep us going along the long road
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.