Typically you define the system parameters (2-way? 3-way or more?), the cabinet (size, dimension), the enclosure type (sealed? bass reflex? transmission line?) and you select the drivers for the job. Next the crossover type will emerge on its own.
For instance a significant proportion of 2-way systems are Linkwitz-Riley 4th order (total electrical and acoustical).
For instance a significant proportion of 2-way systems are Linkwitz-Riley 4th order (total electrical and acoustical).
Best sonic is achieved when no crossover part is in series with the driver. But then we need some filters to reduce distortion and breakups and help flatten the frequency response. And that's the art. Simple but not too simple.
I had a woofer with rigid cone. No one had come up with a filter that sufficiently tame the driver. But with luck I came up with no parts in series with the woofer. But the parallel filters (notches) are expensive because it should be a conjugate and applied at middle frequency band.
I had a woofer with rigid cone. No one had come up with a filter that sufficiently tame the driver. But with luck I came up with no parts in series with the woofer. But the parallel filters (notches) are expensive because it should be a conjugate and applied at middle frequency band.
If you are referring to series crossovers (as opposed to parallel crossovers) then no, in the passband a driver is still in series with a crossover part. It is not in series with any crossover parts only in the stop band.
If it is not a series crossover then a notch filter in parallel without any crossover parts in series will only sink the impedance to close to zero at certain frequencies making it impossibly hard to be driven by the amplifier.
Unless you are talking about some exotic topology I am unaware of - but then only exotic situations call for exotic crossovers (e.g. elliptic filters?) and they are not applicable to every design.
If it is not a series crossover then a notch filter in parallel without any crossover parts in series will only sink the impedance to close to zero at certain frequencies making it impossibly hard to be driven by the amplifier.
Unless you are talking about some exotic topology I am unaware of - but then only exotic situations call for exotic crossovers (e.g. elliptic filters?) and they are not applicable to every design.
I'm not aware of any topology. Just several notch filters to help smooth-out the natural roll-off of the driver. Furthermore, it works because the filters work in synergy for both the woofer and the tweeter.
There is rarely a text-book crossover filter for a good speaker.
Perhaps you could be a little more clear on what you mean by not having anything in series with the driver by showing a sketch of the filter. Normally when you have a driver in series with amp out with no other passive parts in series with it, then the driver receives whatever voltage amplifier sends and there is no change in the FR, despite the parallel wired elements.
Best sonic comes from a speaker without crossover. But this is possible only with a few select drivers probably designed for that purpose. We know there are many such monitor like the MM DeCappo where the woofer has no filter and the tweeter only has one capacitor. This is the kind of simplicity that I always SEEK with a given driver combo, before going with the steepest filter possibility.
Often, the response is not sufficiently smooth, but we can add notch filters to smooth out the response (and later use the natural roll-off to cross with the tweeter) to an acceptable level.
Crossing the tweeter to match the woofer natural roll-off is rarely successful. That's why I said I was lucky because the woofer and the tweeter was just match for this trick. The filter for the woofer also work for the tweeter. The frequency location of the roll-off was also perfect to have both drivers in phase.
I think there is a 'famous' speaker that use more or less the same trick. 'Midnight Sensation' or something, but from the same designer. I will check later if I have time.
In its simplest form a crossover is intended to combine with the natural response of drivers to achieve a target. The target is both a combination of order and phase characteristics. Think of phase as the drivers movement signature. Different crossover targets will have phase characteristics that align or offset drivers movement so one driver or other will lead, follow or be in sync.thereis no ideal as each provides pros and cons. It depends on which pros are more important to you.
@johnego Thank you very much for your detailed clarification. It makes perfect sense now! I digged and found this quote:
Of all the early 3A designs, probably the most popular was the MM, often noted for its clarity and its bass, which went surprisingly deep for a stand-mounted minimonitor. It was also noteworthy for having no crossover applied to its midrange-woofer -- the driver’s natural upper-frequency rolloff was used to integrate its output with that of the 1” tweeter, which had a simple high-pass filter to roll off its low end.
As you mentioned, the drivers will have to be specifically designed for this.
For me personally, I am not shy to deploy steep slope filters to combat cone break-up and distortion. All drivers have those problems and if we are not addressing them electrically we would be fixing them mechanically (or a combination of both). Virtually all high end speakers today relies on the former because it is more effective and easier.
Also, using the natural roll-off of the woofer does not really eliminate crossovers - it eliminates only the electrical part of it. There is still the acoustic crossover and we are subject to the same lobing issues as everybody else. In fact the problem will now be worse because the shallower slope means the problem will affect more frequencies and we will not be able to push the crossover frequency lower to reduce the lobing pattern.
Of all the early 3A designs, probably the most popular was the MM, often noted for its clarity and its bass, which went surprisingly deep for a stand-mounted minimonitor. It was also noteworthy for having no crossover applied to its midrange-woofer -- the driver’s natural upper-frequency rolloff was used to integrate its output with that of the 1” tweeter, which had a simple high-pass filter to roll off its low end.
As you mentioned, the drivers will have to be specifically designed for this.
For me personally, I am not shy to deploy steep slope filters to combat cone break-up and distortion. All drivers have those problems and if we are not addressing them electrically we would be fixing them mechanically (or a combination of both). Virtually all high end speakers today relies on the former because it is more effective and easier.
Also, using the natural roll-off of the woofer does not really eliminate crossovers - it eliminates only the electrical part of it. There is still the acoustic crossover and we are subject to the same lobing issues as everybody else. In fact the problem will now be worse because the shallower slope means the problem will affect more frequencies and we will not be able to push the crossover frequency lower to reduce the lobing pattern.
Last edited:
Yes. And why I seek simple passive filter is because I have long experience to understand what is good for me in music listening. Either in amplifier design or speaker design, the importance of speaker drive and control (as well as 'timing') is high from my point of view.
My tolerance for (any kinds of) 'distortion' is very low, even for second order harmonic distortion that many tube lovers love. But we must know where the distortion is, what the distortion sounds like, in order to know how we can avoid what we need to avoid.
I kind of give up with steep slope, when it is passive. Amplifier output design is already hard, especially with low transconductance transistor like lateral fets. The passive crossover elements will make things worse.
About shallow slope: When a notch is applied at the right point along a rolling-off slope, the slope will become steeper (e.g. Bainter notch, elliptical slope). This is how we can adjust the natural roll-off with notch filters. At the same time the notch filter will also 'flatten' the breakup usually found along woofer's roll-off frequency. And in general, woofer's natural roll-off is much steeper than tweeter natural roll-off.
About more frequencies affected by shallow slope: There is no problem with shallow slope as it is normally seen in first order filters. It is even preferred as long as the sum with tweeter response is okay.
About lobing at high crossover point (fx): Lower fx is rarely good with low diameter dome tweeters. But some small woofers have good dispersion at around its natural roll-off frequency.
Yes, you are right about the parallel notch filter increasing the slope of the roll-off, but there are still 2 remaining problems: 1, the notch filter lowers the impedance too much and makes the amp harder to drive, and 2, if a parallel notch filter can cut the woofer to any significant degree it will cut the tweeter section as well.
Whereas if you allow just one inductor in series with the woofer, just one, that same notch filter can easily attenuate cone break-up by 20dB or more without cutting the tweeter, and the impedance curve remains well-behaved without dipping too low.
I will reiterate that avoiding an electrical crossover is not completely avoiding all crossovers, because you still have the same acoustic lobing to deal with.
There is nothing interesting about getting rid of that series inductor. It will have some insertion loss and you will have to counter that with more amplifier power, that's all. It will still use far less amplifier power than a parallel notch filter. The voice coil itself is an inductor, and its DC resistance is an order of magnitude higher than the series inductor in the crossover. If the series inductor is evil it is no more so than the voice coil itself.
The impedance issue will be a major problem. A pure voltage source is impossible to attenuate with a parallel notch filter because by definition the voltage across the woofer is invariant with respect to the load impedance. A parallel notch filter only works if there is output impedance. If I hazard a guess you may have a tube amp with a tube output stage. A solid state amp with an output impedance of 0.01 ohm will be hard to attenuate with a parallel notch filter without practically shorting the circuit.
Most low-budget 2-way speakers are made with the single capacitor high pass, no low pass topology, where the entire crossover is comprised of only a single part. In fact I think many car stereo 2-way speakers are exactly built like this. You never see a metal cone in this application. A paper cone dampens the cone break-up making it not offensive.
I can't possibly imagine subjecting an aluminum cone woofer to this treatment. I bet your woofers are made of paper.
I think there are high end speakers doing the same, like the Reference 3A MM you found. Of course a lot more care would be required to design a woofer which could use no electrical crossover at all and still sound great - not even any parallel notch filters.
Whereas if you allow just one inductor in series with the woofer, just one, that same notch filter can easily attenuate cone break-up by 20dB or more without cutting the tweeter, and the impedance curve remains well-behaved without dipping too low.
I will reiterate that avoiding an electrical crossover is not completely avoiding all crossovers, because you still have the same acoustic lobing to deal with.
There is nothing interesting about getting rid of that series inductor. It will have some insertion loss and you will have to counter that with more amplifier power, that's all. It will still use far less amplifier power than a parallel notch filter. The voice coil itself is an inductor, and its DC resistance is an order of magnitude higher than the series inductor in the crossover. If the series inductor is evil it is no more so than the voice coil itself.
The impedance issue will be a major problem. A pure voltage source is impossible to attenuate with a parallel notch filter because by definition the voltage across the woofer is invariant with respect to the load impedance. A parallel notch filter only works if there is output impedance. If I hazard a guess you may have a tube amp with a tube output stage. A solid state amp with an output impedance of 0.01 ohm will be hard to attenuate with a parallel notch filter without practically shorting the circuit.
Most low-budget 2-way speakers are made with the single capacitor high pass, no low pass topology, where the entire crossover is comprised of only a single part. In fact I think many car stereo 2-way speakers are exactly built like this. You never see a metal cone in this application. A paper cone dampens the cone break-up making it not offensive.
I can't possibly imagine subjecting an aluminum cone woofer to this treatment. I bet your woofers are made of paper.
I think there are high end speakers doing the same, like the Reference 3A MM you found. Of course a lot more care would be required to design a woofer which could use no electrical crossover at all and still sound great - not even any parallel notch filters.
Last edited:
If you look at typical combined speaker impedance curve, you will see that low impedance only happen in certain frequency (e.g. a 'saddle'). There is no such issue if the notch frequency is high enough, i.e. impedance related with tweeter.
The nature of a notch filter is it work with a narrow band frequency.
Yes, and I have mentioned that the same filter work both for woofer and tweeter. For example, the resonance frequency of the tweeter is exactly at the exact position where the woofer has the critical breakup.
With notch filters we are not dealing with 20dB. A peak is usually less than 4dB, especially in good woofers.
It is not for avoiding lobing, distortion and the other crossover design issues.
Have you tried active crossovers? Many people prefer active crossovers than passive. IF you experience the same, then may be you will understand why avoiding those inductors can be beneficial.
No, I don't use tube amps. My speaker is a working design. There is no problem with impedance and subjective sound.
Yes, for low budget speakers you need minimum filter components. And such simple filter will only possible if the breakup is not severe, i.e. when the driver is using paper cones.
I avoid the series elements, not to save money. The filter elements cost me more than $200 I think (It's a conjugate notch filters).
I understand your confusion. It is not aluminium, nor paper. I have mentioned that many others had use steep and lower frequency filters but the 'ringing' was still there (subjective distortion severity has nothing to do with the size of the peak). I solved the issue with the opposite approach. That should answer your question how we choose crossover frequency when designing our speakers.
Did you measure the impedance of your design? What was your minimum impedance?
Could you share with us your take on lobing?
Many people have reported drastic improvements by removing the crossovers of their speakers and replacing them with active crossover/bi-amplification. This is generally true if the speakers in questions have sonic flaws to begin with. They certainly will reap significant benefits from steeper crossover slopes hiding those sonic flaws. These same modders will not be able to improve upon truly great speakers with properly designed passive crossovers.
For me, my design process was to play with a 4th-order LR active crossover first, and then work on the passive until the passive beats the active soundly. The key here is I 'played' with the active but I 'worked' on the passive. Of course if I equally 'worked' on the active with EQ, phase alignment and all the trimmings it was going to sound just as nice (there is nothing that a passive crossover can do that an active crossover can't). However I must be fairly confident that no casual modder can take my design and achieve any significant gain by replacing my passive crossover.
If an inductor robs me of 0.5dB of power I just crank up the amp by the same amount. That's all.
Last edited:
I forgot but no amps had been harmed in the process.
I did my homework regarding lobing just because I could. Honestly I think lobing is overrated. There are more serious tonality issue that can be caused other than by lobing. If you listen to live concert by sitting on the first row, second row, third row, each has different sound but you never think that only one position has the right tonality.
I have different way to look at it. But to be on the same page, we must be sure that we are aiming at the best sounding thing in the world, where even very small things matter and no components and amplifications are perfect.
For me its like comparing apple with orange. Each has its own issues. Especially when the active is analog. We deal with current noise, voltage noise, etc. that we dont deal with in passive design.
We are all climbing up the ladder to perfection - a lofty but unattainable goal. But as we move up the food chain, our new speakers are more revealing and more flaws in the recordings become obvious. Recording quality has nothing to do with open reel tape vs vinyl vs DSD vs SACD or what you will, but everything to do with the skill of the recording, mixing and mastering engineers, the acoustics of the venue, microphones and their placement, and even the positioning of the performers.
There are many recordings I used to like but due to equipment improvement they no longer sound good to me. The repertoire narrows until the flaws of the speakers themselves become a limiting factor again. Then, hunt for the next speaker, and the cycle repeats itself with the search for even higher caliber recordings.
My entire design (drivers+crossover parts, excluding cabinet) is less expensive than your notch filter alone. Assembling a cohort of star players does not automatically give rise to an invincible team. Players of more humble origins who are put in the right positions and work together seamlessly are more likely to win the game.
There are many recordings I used to like but due to equipment improvement they no longer sound good to me. The repertoire narrows until the flaws of the speakers themselves become a limiting factor again. Then, hunt for the next speaker, and the cycle repeats itself with the search for even higher caliber recordings.
My entire design (drivers+crossover parts, excluding cabinet) is less expensive than your notch filter alone. Assembling a cohort of star players does not automatically give rise to an invincible team. Players of more humble origins who are put in the right positions and work together seamlessly are more likely to win the game.
You make an unfair assumption.
A system with cheap components, in the hand of a good designer will sound better than a system with expensive components in the hand of poor designer.
But if you assumed that the designers are equally good, the expensive one will beat the cheap one, all the time.
When you drive an amp below its rated minimum load impedance, it does not always result in damage. Some amplifiers, however, will become unstable, oscillate on its own, ring, or otherwise output things that are not in the original signal.
A good designer, among other things, looks at the impedance curve of the whole system to make sure that it is stable for the amplifier to drive at any frequency.
This article explains some of the crossover basics.
Myths & Facts about Loudspeaker Crossovers: Identifying Legitimately High Fidelity Designs | Audioholics
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.