[Roger A. Modjeski]

First: The problem and the bad news

Below is a picture of a crossover that the maker is obviously proud of. I must admit it is impressive and congratulate them on a beautiful execution of a bad idea. What is the bad idea? Speaker level crossovers are the weakest link in the high-end audio system. Wouldn't the money spent to make such an expensive crossover be better spent on a second amplifier and electronic or passive line level crossover??? These are the questions so few audiophiles are willing to ask.

http://live.audiogon.com/i/mn2005/h/1113690388.jpg

Here is the insurmountable problem. At the speaker level all the components of the crossover have to pass amperes of current. At 100 watts a 4 ohm speaker and it's crossover experience peak currents of 10 amperes or more if the impedance is lower at that frequency. No matter how good the capacitors are, they have to handle these currents and let me assure you they don't like it a bit. Capacitors store charge, they are not designed to pass current, it isn't in their nature.

The inductors add hundreds of feet of 16, 18 or 20 gauge magnet wire. Look at how many are in the picture below. Seems odd to ignore this common, inexpensive wire, but it's inside where it is not seen. Given the aesthetic attention given to speaker cables it becomes obvious that most audiophiles are only concerned with they can see.

Lets not forget the resistors that are there to convert audio signals into heat. Besides that, 5-100 watt power resistors generally have internal brass wire crimps to connect the nasty resistor wire, which is a poor conductor by definition, to the nasty leads. These crimps can oxidize over time making matters worse.

It is my opinion that any speaker maker who is offering a high end speaker offer it without a crossover at a discount that you can apply toward another amplifier to do it right. When hi-fi started, speakers were cheap in comparison to amplifiers. Now it's the other way around. There are many systems with inexpensive amplifiers that are a fraction of the cost of the speakers they drive. In some systems the cables coming in and out of the power amp cost more than the amp itself. This makes no sense at all. 

From their appearance alone I can see that the components and construction costs of these speaker level crossovers is indeed more than a second quality amplifier.

Now the good news

The simple alternative is both inexpensive and elegant: Crossovers are just filters with slopes of 6, 12, or 18 dB per octave. They are implemented electronically with passive components alone or combined with tubes, transistors or ICs. I have been making 6dB/octave crossovers for knowledgeable listeners who have already caught on to the improvements and flexibility they provide. A two way unit should have a level control to turn down the more sensitive amplifier. A three way would have 2 level controls. They can also be made with one level control for each amplifier for total flexibility. One can easily find the setting that gives the mix of bass, treble and midrange (in a 3 way system) that the speaker manufacturer intended. In addition, the listener can adjust that blend to his liking. A little more bass, a little more treble at the turn of a knob.

Here are the technical advantages:

1. The audio currents in line level crossovers are in micro amperes, one million times smaller than the currents in a speaker level crossover.

2. There are no inductors. No matter how large the wire used to wind them, they add resistance right where you don't want it, in series with the woofer. There goes your damping (woofer control).

3. 1% sub-watt resistors are made with metal films that are much better than resistance wire.

4. Capacitors are of small values in the area of .01 microfarads v.s. microfarads to 100s of microfarads in speakers. The QUAD ESL has a 200 microfarad electrolytic cap that all the audio has to pass through along with series power resistors, ceramic capacitors and miles of wire in the chokes that drive the concentric rings. None of the mods available address the circuits that drive the rings as they operate at 6,000 volts. 

5. It is easy to move the crossover point around and control the output levels so that any speaker and amplifier can be used.

6. No power is wasted and amplifiers can be of smaller size which generally sound better. While the woofer might require 100 watts from a solid state amp with high damping, the tweeter can be driven by a little 5 watt single ended tube amp.

7. Most speaker crossovers have over a dozen components (count them in the picture). A 6dB/octave 2 way crossover requires one capacitor and one inductor. For 3 way add one more inductor and one more capacitor. For 12 db slope add one more component per driver. For 18 dB add one more. Given that relationship, infinite slope would require an infinitie number of components.  The center crossover in the picture above has 15 components that I can see. There are 4 big inductors whose wire, unwound, would allow you play your speakers a few houses down the block while connected to your amplifier back at your house.

Before you plunk down the money on your next speaker purchase ask the maker if he would be willing to build the speaker without the crossover and give you an appropriate discount. By the amount of that discount you will quickly know how he values his crossover.

I recently fell in love with the little FE 103 speaker that we now offer. I didn't expect it to sound so good. It's just a simple single driver in a box. I have to attribute the detail and coherency I hear to the quality of the Fostex driver and the absence of a crossover. Obviously this little speaker is not going to play high levels or deep bass. Most of the single driver speakers I have heard sound very thin, lacking in the lower midrange. To solve this problem I devised a simple passive 9 dB equalizer that can go inside our amps or between your preamp and your power amp. I was astounded to hear things I had never heard before out of a cone speaker. It changed my opinion about what cones can do.

I have been asked if this EQ could be put between the power amp and the speaker. If it were, all the aforementioned component problems would be encountered and the power amp would have to be turned-up 9 dB which corresponds to a 83% power loss i.e. 8 amplifier watts for every watt delivered to the speaker.

[bummrush]

seems like quite a waste of resources wire etc,So what are the reasons for this type of crossover deal?Chrishma Audio isnt a fly by night deal.

[zybar]

Roger,

What are your thoughts on digital crossovers?

George

[Vapor Audio]

I'd love to post my thoughts on this, but it could be taken as argumentative ... and this is your circle.  If you'd like a discussion, consider posting in the lab section.

[Roger A. Modjeski]

Not only is it a waste of resources it's not the best way to get good sound. It is the standard way to go because because it's easy for the manufacturer, the dealer and the user. It is far from the best solution. I would hope that listeners will start to ask the pertinent questions. There is no argument, other than laziness, that i can imagine to support speaker level crossovers as the best way to go in any high-end system.

You have to understand that for years the dealers determined what the manufacturers were allowed to do in certain circumstances. No dealer would want to be bothered with explaining the benefits of multi- amplified systems. Even though the dealers are less in the picture these days the users have not required the manufacturers to get outside the box.

For those of you who know the movie "Network" the news-anchor, played by Albert Finney, says he has run out of bullshit, he just doesn't have any more. Unfortunately the beforementioned purveyors of speakers have not run out. There is a wonderful scene where he tells his viewers to go to the window, open it and shout "I'm mad as hell and I'm not going to take it anymore". If you will let me play that role: I want you to call up your dealer and the maker of your speaker and ask him. "Why are you still making speakers with built-in crossovers. How can you claim you are making an ultimate product when you ignore the weakest link?

As to digital crossovers: I know some people who have liked them at first. They are economical and not a bad tool for the experimenter to find his slopes and crossover points but in the long run their owners take the information and buy something analog.  Keep in mind that a digital crossover has to take in an analog signal, convert it to digital, run it through a lot of microprocessor math and convert it back to analog. That's a lot to do and they are filled with parts. Why torture your signal with all that when a capacitor and resistor can make a 6dB/octave crossover. If steeper slopes are needed that is not too much harder. Often preamps have enough drive to do all that passively. If the input impedance of your amplifier is more than 100 times the output impedance of your preamp (and is usually is) I can make a passive 6 or 12 dB/octave crossover. If not, a few tubes, transistors or good op-amps can provide the necessary buffering. When it comes to sub-woofers I find 24dB/octave is necessary to keep male vocals out of the sub if the crossover point is 100 Hz. If the crossover point is 50 Hz or less 12 dB slope is OK.

The Beveridge RM-3 Universal crossover I designed in 1979 does all that with very simple circuits. All the high pass is done with unity gain, class A complementary buffers that employ only two transistors in as unity gain buffers.

I will take TurboFC3S's suggestion and post this in the lab section.

[sbrtoy]

Curious as to why no one has offered an easy to implement product like the RM-3 more recently with modern circuit design and components?  Marchand makes analog units, but other than that I am not aware of any. 

[6BQ5]

You have to understand that for years the dealers determined what the manufacturers were allowed to do in certain circumstances. No dealer would want to be bothered with explaining the benefits of multi- amplified systems. Even though the dealers are less in the picture these days the users have not required the manufacturers to get outside the box.

I would not agree necessarily.

Back in the late '70s and early '80s, the speakers I most coveted were Linn's DMS Isobarik.




Linn explicitly recommended tri-amplification of this speaker, and every dealer had them pretty much setup this way in showrooms (otherwise, they would sound plain and anemic). Linn sold plug-in crossover cards for their amps to accomodate. Naim had the same philosophy with their top-end speakers and sold outboard crossovers too. The built-in DMS crossover was a real bear, it was so complex, you needed extremely powerful amps because the crossovers would sap the life out normal amps ... the impedance curve was all over the map because of 6 drivers per speaker. Very easy to hear the differences when the were passively and actively networked.

The Linn-Naim flat-earth approach gained a lot of converts however. Blew me away the first time I heard the DMS. I eventually got the little sister, the Sara 9, which was also a bear to amplify too.

So ... Linn and Naim for sure convinced their dealers somehow, some way. Now going to the other extreme with something easy to drive ...

I like what some people on the Klipsch discussion forum's have done. Paul W. Klipsch, just before he died designed his ultimate, swan-song speaker which was called the Jubilee.

 

Unfortunately, the accountants convinced him not to put it into production because they were sure they would never make money on it. Now the story goes, that there is an annual pilgrimage by Klpsch-heads to Hope, Arkansas where some people heard the prototypes and were blown away as to how much better the Jubilee sounded compared to the Klipschorn.

The parts and design are well known by regulars on the board through insider information. So quite a few have done the DIY route in building their own Jubilee. It simply is on another level, has a sensitivity of over 106 db/w/m ... thus can easily run off low powered amps. Also ... it's a 2-way, such that some people have gone the bi-amplification route with external crossover. It does not sound like a typical horn, and distortion is super low. Some people with very exotic electronics gear have opted for the Jubilee. Bonus too ... these do not have to go into corner's like the Klipschorn, thus can be optimally placed in a room.

Here is an example of someone's DIY project. And if you scroll to the bottom, you will also see some scary crossovers too.
http://www.jubilation.ws/

If I had time, room, and the money ... without hesitation, that's the route I would go.  (Bi-amped of course).

P.S. That's a newer style Tractrix horn you see on top, not the typical exponential horn.

[studley]

"The QUAD ESL has a 200 microfarad electrolytic cap that all the audio has to pass through along with series power resistors, ceramic capacitors and miles of wire in the chokes that drive the concentric rings. None of the mods available address the circuits that drive the rings as they operate at 6,000 volts."

I have the Quad 989s.  I assume they have something similar.  If so, can anything be done about this i.e. can i remove this circuitry and use a line level crossover like you describe  with multiple amps? (I've been told that Quad ESLs don't have normal crossovers like cone driver speakers do - ??)

[pbrstreetgang]

Im sorry I dont see the apeal in those Klipsch. Alls it is is a free standing horn and a bass bin. Why wouldnt an Avantguarde, Oris, Edgar, or a million other of the same recipe be so transcendent? Maybe just eliminate the box and go CS2

[MarkR7]

Why have any crossovers at all?  They just impede the purity and coherence of my music!  Probably explains why I gravitate towards speakers that don't use crossovers, especially on the all critical midrange.  I don't see myself selling my Andra 1s any time soon, and I also like the Reference 3A Grand Veenas and Diapason Adamante IIs and IIIs.  I am sure there are others.  Once you go crossoverless, you won't go back!

[konut]

While I agree with what most of Mr. Modjeski writes, there are some quibbles that I have with some of his assertions. I would submit that the room, or acoustic environment, is usually the weakest link in a high end music "system". This is a minor point and more of a degree of standpoint than absolute difference. After all, even the finest speaker, of any kind, will not sound its best in a poor acoustic space.
     While speaker level crossovers have all the drawbacks outlined above, they offer 1 advantage that cannot be denied, convenience. I realise that his is not a argument for them, I point this out simply to explain their popularity even among those who are seeking an absolute zenith in sound quality. With all the other choices in audiophiledom, SS vs tubes, vinyl vs CDs vs server, single driver vs 2, 3, or 4 way, power conditioning vs none, etc, line level crossovers get short shrift when it comes to decisions about sound quality. Its a level of complication very few are willing to take on.
   As a user of all the different combinations of crossovers, or absence of crossovers, I can make some personal observations. The first electronic xover I ever used was a JBL 2 channel active with plug-in modules for different frequencies. This was in 1974 in a PA system consisting of Altec Lansing Voice of the Theater bass bins, with JBL drivers, crossed at 800hz to Bose 800 PA cabs. There was simply no comparison to the stock passive Altec xover. I'm sure most of this was due to the fact that there must have been frequency shaping components inside the passive Altec xover to compensate for the compression horn in the stock design. The other advantage of the active JBL xover were due to the many advantages cited by Mr. Modjeski.
    My next encounter with a line level xover was with the optional passive line level xover offered with the Magnapan Tympany 1D. This was a small box with a left and right input and a lift high, left low, and right high, right low output crossed over at 800hz to the respective panels on the 1D. There was a 3db insertion loss with this unit and even with a good quality preamp, and the notorious low sensitivity of the Magnaplanars, the speakers required over 2kw of total power to get the playback levels to realistic levels.
    The next experience I had with an active xover was with a cheap Behringer CX 2300 2 way stereo, 3 way mono unit. I used this with a few different speakers. Although it never failed to clean up the crossover region and make the speakers more dynamic, it always imposed its own sonic signature on the sound.
   To appreciate the benefits of a crossover-less design, I've used an Omega Aperiodic 8. This speaker uses a Visaton B200 in an aperiodic loaded cabinet. While the articulation of this speaker is superb, the rising response of the driver required the use of a frequency shaping network to take the bite out of the 7khz region. This lowered the ultimate efficiency of the speaker and imposed its own sonic signature(phase anomalies?). Perhaps the network executed at the line level would have had less of an effect on the sound, but not being an electronics engineer, I could not get anyone who was qualified, interested enough to design the comparable line level network. The other significant drawback of this, or any single driver design, is the narrowing dispersion characteristic as frequency rises. I subsequently added a sub woofer utilizing a Marchand XM46 line level high pass filter to cross over at 70hz, 24db per octave. This filter is virtually transparent and has only a 1db insertion loss. The sub has its own 5 band parametric to low pass and equalise.
     I have gone back to a 2 way design(ahem), a SP Tech Timepiece Mini. I'm still using the Marchand and sub with superb results. While the Mini has the traditional passive xover, it's of such high quality, and individually hand tuned to the tweeter, that the benefits of the horn design(dispersion), and more linear behavior, far outweigh any advantages of the crossover-less Omega.
    Back in 2005 I had an exchange of opinion with David Ellis of Ellis loudspeakers on this very subject.
http://www.audiocircle.com/index.php?topic=16410.0

    Over time, I have come to these conclusions:
1. Passive line level crossovers offer the potential for the most transparent reproduction.
2. There are few, if any, drivers suitable for use with simple, symmetrical, line level passive xovers. Shaping networks are required to linearalise the   response, or out of passband resonances, thus lowering overall efficiency(insertion loss) negating other advantages.
3. Only the most expensive active electronic x overs(implementation) are transparent enough for accurate reproduction.
4. Modern, properly designed, and executed, passive speaker level crossovers come very close to line level xovers.
5. The market is so small for the line level xover(passive or active) that only the most devoted, and renown, designer will ever get them to be accepted by the general audiophile market.       
   

Mr. Modjeski, I salute you. You are fighting the good fight. Best of luck with your endeavours.