[rblnr]

Posted this on the VMPS forum as I'm an owner of said speakers, but got no response:

Can someone explain (or point me toward some info.) on what what the differences are between first, second, third and fourth order crossovers? I know the slopes get steeper, but that's about it. And what is the benefit of first (or quasi-second) order xovers in the VMPS line?

I ask this question primarily because through all the speakers I've owned or auditioned over the years, I almost invariably prefer ones that wind up having first order crossovers -- they sound(ed) more organic and alive. There are obviously many design factors at work here, but am curious why this might be. Some great designs, notably the Waveforms (4th order) I owned prior the my 40s/LRC spread were great in certain areas, but missed somehow.

[TheChairGuy]

rblnr,

I've moved your question to 'The Lab' Circle as the technically minded folks would seem to be able to best answer your question.

TheChairGuy
Facilitator Audio Central.

[rblnr]

Anywhere I might get a reply would be cool.

[TheChairGuy]

I wish I could help, my man, but the q's are outta' my league and realm of expertise....

AudioGeeks, please help this poor sonofagun.....pretty please

[JohnR]

Hi, the order of the crossover determines in essence how steep the cutoff between the drivers is.  A first order crossover has a 6dB/octave rolloff, a second order has 12dB/octave, and so on.

6dB is half the signal (voltage wise), so suppose you had a first-order crossover at 3 kHz. At half the frequency, ie 1.5 kHz, the tweeter is receiving only half the signal that it receives above 3 kHz.

For a second-order crossover at 3 KHz, at 1.5 kHz, the tweeter will be receiving a quarter of the signal.

The woofer signal is similar but in the other direction. At and on either side of the crossover point, both drivers are effectively producing signal. In the first order case, this area of overlap is quite wide. With a second order crossover, it's narrower, and with 3rd and 4th, narrower still.

But there is another issue than just the amount of signal. The higher the order of the crossover, the greater the amount of phase shift that the signal goes through. The total phase shift is 90 degrees times the order. It so happens that with a first-order crossover (and assuming perfect coincident drivers...) the *sum* of the signal from the two drivers (in the area of overlap) has no phase shift. All higher orders (except for some very special and complex configurations) do have phase shift in the summed signal.

What does the phase shift do? Well, in effect, it changes the way the waveform appears in time. There is lots of disagreement about how and whether this is audible, and I won't take any sides. When it comes down to it, loudspeaker design is a whole load of compromises, and it's pretty hard to separate what it due to what.

For example, higher-order crossovers generally have more capacitors in series with the tweeter. Is it the phase shift that causes a perceived difference, or the extra capacitors?

HTH

[randytsuch]

For example, higher-order crossovers generally have more capacitors in series with the tweeter. Is it the phase shift that causes a perceived difference, or the extra capacitors?
...

From what I know, the higher order the crossover, the more parts it takes to implement.  So, a 2nd order crossever should be simplier, and have less parts than a 3rd order.  To me, fewer parts is usually a good thing.

OTOH, a shallow slope (2nd order) means the tweater has to go lower, and the woofer has to go higher, so you are asking your drivers to do more.  That's a big tradeoff.

Randy

[Arctos]

Bravo JohnR!!,
You have just made many things clear that have been puzzling me for quite some time.  Thanks for the great post!!

RDS

[Roy Johnson]

JohnR covered most everything.
I would add, from my experience, the advantages of a first-order crossover are in clarity, dynamics, timbre/texture, and imaging- the whole gamut. Along with less amplifier interaction.

There are 1st-order filters out there which are very complicated, from the designer's choice of drivers and cabinet design. Anytime one adds more parts, low-level signals are lost, nuances of pitch and timbre are obscured, and dynamic contrasts are reduced. It sounds laid back and hazy.

A simple 1st-order circuit is very difficult to design, because the drivers cannot require any "correction", nor can the cabinet be a factor in the resulting tone balance.

If you are ever going to judge a speaker quickly, put on an artist whose notes span a given crossover point. You'll hear everything change/collapse/pinch/bite/become grainy/sharp in that tone range. Don't let anyone tell you "it's the recording". To know that for yourself, listen to it on a decent pair of headphones- which have very little phase shift as a rule. You'll hear nothing wrong with "that" recording.

I know of no crossover circuit, except the first-order ones, that is time-coherent. Meaning that the woofer and tweeter are not only "in-phase" on every tone shared, but also that those two drivers start and stop together.

Higher-order filters (2nd and up) inject a time delay between the drivers, a delay that is different at every frequency. It is for this reason alone that people's opinions of those speakers change depending on the music and who is doing the listening. The time delay means that depth of image is foreshortened in that tone range, that a tweeter's image is "separate" from the woofer, that some notes are piercing, others are smeared, and that the higher-range instruments do not occupy the same acoustic space as what is heard from the woofer/mid.

Because the different orders have different time delays, the audible effects are different. If the crossover point to the tweeter is 2-4kHz, you hear a 2nd-order adds a graininess, like the older AR speakers from the late 70's/early 80's. A third-order circuit is very soft in its imaging in that tone range, and dynamically not quite right. A 4th-order crossover, implemented "by the book", will sound pinched in that tone range, and quite forward. To reduce those side-effects, many 4th-order speakers have their crossover points between woofer and tweeter separated- the tweeter crossing over a little higher than "textbook" and the mid a little lower. A dip in the frequency response is thus induced at that frequency, when measured with test tones or pink noise. Most times, this technique also "tones down" the tweeter's LF-reflection off the surrounding cabinet. That is, if you think that a sonic splash off the front is ok for the music.

To do its timing thing correctly, a first-order crossover circuit requires that the woofer, mid, tweeter lie at the same distance (acoustically) from your ear. If they do not, then that crossover, with those drivers, in that cabinet, is still not a minimum phase shift design- sort of close, but no cigar.

I know of no co-axial speaker that is time coherent, nor of any single-driver cone speaker.
Full-range single panel speakers are/can be minimum phase- and IMO why they are enjoyable, even at the expense of sitting dead on center because of that large panel vs. the short treble wavelengths.

Hope this helps!

Best,
Roy Johnson
Green Mountain Audio

[Ravi]

Roy, I have a ton of respect for you, BUT, I must disagree a little bit with you  

I have heard speakers using all types of XOs, and I can't say that there is a general rule for what order XO sounds best.

For e.g. some standout speakers and the XO topologies they use:
1. Linkwitz Orion (4th order LR) regarded by many as the best of the best
2. BESL System 5 (4th order LR) very smooth and refined
3. Joseph Audio Pearl (Infinite Slope), breathtaking, just like live

I built a pair of speakers for my sister that use a 2nd order (acoutic) XO, and they sound fantastic.  The woofer is surface mounted to provide better time alignment.  I hear no grain whatsoever as you mention.  The design used Dennis Murphy XO.  Actually, I cannot tell the XO point by ear, thats how well they blend.

Now I've heard speakers using 1st order, like VMPS RM40, GR Diluceo, etc., and as good as they are, I can't say they are my favourites.

Anyways, that was just my experience.

[rblnr]

John/Roy much thanks for the info. as I try to expand my basic understanding and jibe it w/my experience.

I would add, from my experience, the advantages of a first-order crossover are in clarity, dynamics, timbre/texture, and imaging- the whole gamut

This has been my experience.  The first pair of speakers that really grabbed me were Phase Research (defunct) Little D's -- time-phased and first order.  Grew up on Chicago blues bars, these sounded alive.

Various speakers in between, then post-college w/a few dollars did a massive listening search, unimpressed till I  heard Epos ES14's.  Knew immediately that these were for me, they got it right somehow.  Turns out they were first order, virtually no crossover, but not time-phased. The drivers more or less roll off naturally into each other  Epos ES25's after that, similar, but more bass.

Bought Waveform Mach Solos, with matching center and surround pods; did this on reputation.  Incredibility well-built, accurate, the designer is a breath of fresh air in a hype filled industry.  They sound good, but something's missing.  Texture maybe, they lack the organic flow and palpability the Epos' had.  Took me a couple of years to admit what I knew pretty quickly.  These are fourth order, btw  and it is this experience that really got me wondering about XO's.  

Now have the VMPS RM40's w/a LRC center.  Feel like I got the music back.  Would like to hear some Green Mountain speakers sometime.

Muchas gracias all.  Am up for any other info.  Think I ought to stop messing around and get to work now.

[Arctos]

I have a question I would like to throw into the mix:  If time-aligned means that the average distance from the listeners ear to the woofer and the tweeter are the same (someone correct me on this if I'm wrong) then as the listener changes elevation (sits down or stands up) this would bring the drivers into or out of alignment.  Is this a true statement?

Also, when we talk about distance from the listeners ear, do you mean from the voice coil, the center of the driver, or some other measurement?


Thanks!

[_scotty_]

Arctros, you have struck at the heart of the designs' limitations.
It only is time aligned on one axis and will only reproduce a squarewave
under anechoic conditions. Time alignment requires the physical alignment
of the acoustic center of the drivers or its'electrical equivalent. These designs also have a ragged off axis response as a result of the first order
crossover slope and the lobing that results. This can lead to response anomolies as a by-product of the way it interacts with the room acoustics.
The quasi-second order series crossover offers the same phase coherent
design potential with the advantage of a steeper slope. A combined electrical and acoustic roll-off rate of 12dB/oct can be achieved with the attendent reduction in lobing and off axis response errors if the right drivers are chosen. The problem lies in choosing drivers that will work
with the crossover. Most of the cone materials popular at the present time
have resonance problems at the top of their passband which makes them unsuitable for this crossover. They require the crossover to correct for their inherent nonlinear  response characteristics. This is not possible
without departing from the quasi-second order circuit topology.
  That being said I still prefer the sound of a quasi-second order design when properly done over all the other parallel crossover designs
I have heard, particularly higher order designs implemented passively.

[Roy Johnson]

I forgot to add this in my earlier post:

Arctros is correct about the time delay/time advance if you stand up- this is true for any speaker. Fortunately, the amount of phase shift induced to a first-order design is still far less than exists in speakers with higher-order crossovers.

Scotty- may I correct some common misconceptions?

You say, "The quasi-second order series crossover offers the same phase coherent design potential with the advantage of a steeper slope [compared to a parallel 1st-order circuit]."

Yes, same phase relationships, but no- not any steeper slope. In fact, quasi-2nd order filters have slightly less rolloff for a half-octave each side of the crossover point, and finally, ten times! lower/higher than the crossover point, provide only one more dB attenuation. There is no rate-of-rolloff advantage, no matter what drivers are used. The slope-comparison between all the crossover filter-orders are graphed by the B&O engineers in the AES Audio Anthology papers from Old Colony Sound Labs.

"A combined electrical and acoustic roll-off rate of 12dB/oct can be achieved with the attendent reduction in lobing and off axis response errors if the right drivers are chosen."
Yes, but then you have the phase shift of a second-order filter, without question.

"The problem lies in choosing drivers that will work with the crossover.... This is not possible without departing from the quasi-second order circuit topology."
Finding the right drivers is a big problem, but they do exist, such that the quasi-2nd order circuit is not necessary.

"That being said I still prefer the sound of a quasi-second order design when properly done over all the other parallel crossover designs I have heard, particularly higher order designs implemented passively."
I'm sure I'd agree with you, as there are very few properly-implemented, simple, parallel 1st-order crossover speakers out there, for reasons of poor driver choices and enclosure design.

Note that, to the signal passing thu it, a series 1st-order filter (the quasi-2nd order filter) does offer any electrical or other physics-based advantages over a parallel 1st-order circuit, where each driver has its own circuit. It has the disadvantage of the signal passing thru the extra series components.

Yet I know that with certain drivers, the series circuit does work better in every way, but that has to do with this circuit's interaction with those drivers' inherent damping (Qes, Qts) via the higher series impedance `tween each driver and the amplifier. It is certainly not because of "better coupling to the amplifier"?? as Bud Fried said, quoted by Marty deWulf. Mr. Fried did not say why that statement was true.

Best,
Roy Johnson
Green Mountain Audio

[_scotty_]

Roy,from AES journal June 1971,Vol 19,number 6, J. Robert Ashley and
Allen L. Kaminsky,Active and Passive Filters as Loudspeaker Crossover Networks, page 499 Quasi-Second-Order Passive Crossover Networks.  Paraphrasing, Kaminsky derived a steeper filter by relaxing the constant impedance constraint,Greek lower case z equals a=1 and new constraint equals z=a=0.5 relaxed, and the filter yields 12dB/oct on either side of the center frequency and 6dB/oct there after. Quoting directly,{The low frequency channel cuts off at 12dB per octave at Wo and 6dB per octave above 2Wo;the high frequency channel cuts off at 12dB per octave at Wo and 6dB per octave below 0.5Wo.Hence the appellation of quasi-second-order seems appropriate for this filter. From page 292,Top left of page, JAES May 1977,Vol 25,No.25 .{Curves for other values of a are shown in Ashley and Kaminsky[7]. The crossover network will have constant input impedance at a=1. Quoting from Ashley and Kaminsky lower on same page,499,{The class of filters exemplified in this section {such filters are possible for any order }have desirable characteristics that do not stop at the lack of phase error.One of these is that the quasi-second-order filter has half the components of a conventional second order filter. The other is that component tolerances are not as important as in conventional networks,since a slight error in the crossover point affects both channels in a complimentary manner.}
 
Another advantage I have noticed is better transparency due to the crossover parts being in a shunt position across the driver rather than
in series with it. I also have to echo I.M. Fried's observation of greater  dynamics with this filter topology. See link  http://users.tpg.com.au/users/gradds/series_cross-overs7.htm
My apologies for not foot noting my earlier remarks.
                               Regards, Scotty

[Harmon]

Alright guys, lets name as many 1'st order and quasi 2'nd order crossover loudspeakers that in your opinion are well implemented. I believe that the Meadowlark Nighthawk and Heron2 are good 1'st order loudspeakers.

[tinear99]

My Green Mountain Europas are first order (Hi Roy!) And one of the best speakers I have heard, especialy at their $880 list price. The speaker is about as clean and honest as you can get, has a very unique sound to it. What you put in is what you get. Roy does know he stuff. I encourage anyone that wants to learn a few things about crossover & speaker design give him a call. Very refreshing to talk with a manufacturer with no BS and no hype of his product. I think he applies his knowledge and degree in physics to his designs. I will have a review shortly as I wait a few months to write so the "newness of purchase" doesn't  influence my thoughts.

Other first orders that I know of: Vandersteen and I think Merlin. My old Maggie 10.1s had one cap & inductor in the network also.

[JohnR]

Folks, this is the Lab so let's keep the discussion technical. Thanks

[Harmon]

Folks, this is the Lab so let's keep the discussion technical. Thanks

I respect your opinion on keeping the discussion technical but it would not hurt to push the technicalities into the real world and discuss the implementation of crossovers to commercial loudspeakers. It may help to guide people into the right loudspeaker.

[_scotty_]

Harmon, right now and for the forseeable future there will be no One Right
Speaker. People will always choose that which sounds best to them.
As long as we are trying to recreate reality with a mechanical transducer
we are probably doomed to a crude approximation of a shadow of reality,
despite hyperbole to the contrary by manufacturers. As far as first order or quasi-second-order designs and their merits are concerned you virtually have to go listen to a loudspeaker you are interested in. An opinion expressed here or a consensus reached about which is best
is no substitute for first hand experience with a product.                          

Who has done a good job of implementing these types of designs would be good topic for discussion in Audio Central.

[JohnR]

I know of no crossover circuit, except the first-order ones, that is time-coherent. Meaning that the woofer and tweeter are not only "in-phase" on every tone shared, but also that those two drivers start and stop together.

Um, this last bit is incorrect. At the crossover frequency the two drivers are 90 degress out of phase. It's only the summation of the signals from the two drivers that has no phase shift.