What about the TYPE A network??

[KanedaK]

Hello everybody,

I'm planning to update the networks in my LaScala because I feel they don't sound as clear as they should. What I have now is the original AA networks and they look pretty bad, altough I don't have the required equipment to actually measure them.

I'm also experiencing some channel unbalance, some sounds being stronger on the right speaker than the left one, it's quite subtle but it kinda screws he stereo image, I feel like bad caps in the left speaker might be the cause (?).

I'm using a high power solid state pre-power combo ( DIY Kaneda preamp, vintage Technics SE-A5 power amp, all DC troughout ) wich I'm only using the first watts as confirmed by the power meters on the amp ( I never even reach 1Watt average, except for parties), my living room being quite small 6mx4,5m, and my neighbours quite..huh... touchy. (I'm aware many will read this and think "horn speakers in a medium small space using the first watt of a high power SS amp, that would be the worst mismatch ever made" but I can assure you it sounds quite great, thank you ;o) )

The LS are upgraded with new medium horn lenses (120degrees pattern Electro-Voice SM120A, stock K55V drivers) and new tweeters (Beyma CP25), both were great improvements over the original sound. I had to tame down the tweeters a bit with an L-Pad since they're a bit too hot in my set-up, no matter wich amp i'm using (I've tried many amps during the last 5 years). So far I must say I'm not hearing any negative effect from the L-pads; this might change when the networks are recapped, but I think maybe this is more sensible with low power tube amps having an already low damping factor. That is not the case of the amp I'm using.

I've considered an upmatket network such as the ALK uiversal, but my budget dictates a much cheaper approach for now, so I'm hesitating between recapping my AAs or, even cheaper, converting them to type A networks. Also I'm not using much power and don't feel so concerned about tweeter power handling.

theorically I'm attracted to the 6dB/oct attenuation of the type A, but since both my midrange AND tweeter dispersion pattern are much, much wider than the stock ones, I'm really concerned about driver overlap. I don't wanna end up with bumps and dips in the frequency range between, say, 4,5KHz and 8KHz. IMHO I feel this would be hard to avoid with a first order network (but then, why does it sound better to some people?).

Also, what about changing the x-over frequency from 6Khz to 4,5KHz as advertised by Bob Crites? Would this cause more/less driver overlap?

any thought / experience would be greatly appreciated.

thanks folks!

[Al Klappenberger]

Personally, I think 1st order (6 dB / octave) crossovers stink! They are so sloppy that you literally can't even specify where the crossover frequency is! You get major interference between drivers because of the huge overlap. This sounds nice at low levels because the multiple sources of the same sound tricks you brain into thinking it's "openness" or "ambiance". Turn the level up a bit and the tweeter starts to overload from the lows getting through the single capacitor isolating the tweeter from the woofer. [+o(]

Oh yeah: The bumps and dips in the frequency range you mentioned are VERY often caused by the multiple sources adding together or nulling each other out at random spots all over the room as the two sounds go in and out of phase due to time misalignment.

My advice is to get a cap replacement kit from Bob Crites and fix up your old AA if you don't want to spend the money to truly upgrade.

Al K.

[Guest " "]

you can go the incremental upgrade approach, but odds are the journey won't end until you go with a higher slope xover. The A's or AA's sound great at flea power levels. but my take is that if you are getting into high power levels, you need to get to at least AL-3, or one of the after market higher sloping networks.

[greg928gts]

Hello everybody,

I'm planning to update the networks in my LaScala because I feel they don't sound as clear as they should. What I have now is the original AA networks and they look pretty bad, altough I don't have the required equipment to actually measure them.

I'm also experiencing some channel unbalance, some sounds being stronger on the right speaker than the left one, it's quite subtle but it kinda screws he stereo image, I feel like bad caps in the left speaker might be the cause (?).

I'm using a high power solid state pre-power combo ( DIY Kaneda preamp, vintage Technics SE-A5 power amp, all DC troughout ) wich I'm only using the first watts as confirmed by the power meters on the amp ( I never even reach 1Watt average, except for parties), my living room being quite small 6mx4,5m, and my neighbours quite..huh... touchy. (I'm aware many will read this and think "horn speakers in a medium small space using the first watt of a high power SS amp, that would be the worst mismatch ever made" but I can assure you it sounds quite great, thank you ;o) )

The LS are upgraded with new medium horn lenses (120degrees pattern Electro-Voice SM120A, stock K55V drivers) and new tweeters (Beyma CP25), both were great improvements over the original sound. I had to tame down the tweeters a bit with an L-Pad since they're a bit too hot in my set-up, no matter wich amp i'm using (I've tried many amps during the last 5 years). So far I must say I'm not hearing any negative effect from the L-pads; this might change when the networks are recapped, but I think maybe this is more sensible with low power tube amps having an already low damping factor. That is not the case of the amp I'm using.

I've considered an upmatket network such as the ALK uiversal, but my budget dictates a much cheaper approach for now, so I'm hesitating between recapping my AAs or, even cheaper, converting them to type A networks. Also I'm not using much power and don't feel so concerned about tweeter power handling.

theorically I'm attracted to the 6dB/oct attenuation of the type A, but since both my midrange AND tweeter dispersion pattern are much, much wider than the stock ones, I'm really concerned about driver overlap. I don't wanna end up with bumps and dips in the frequency range between, say, 4,5KHz and 8KHz. IMHO I feel this would be hard to avoid with a first order network (but then, why does it sound better to some people?).

Also, what about changing the x-over frequency from 6Khz to 4,5KHz as advertised by Bob Crites? Would this cause more/less driver overlap?

any thought / experience would be greatly appreciated.

thanks folks!

It wouldn't cost very much to upgrade the caps in the AA networks, that might be a good place to start. Also, don't forget to clean all the connection points of the wiring.

As you say, some people do like the sound of gentle slope filters. I know from my own experience, even at fairly high volume levels (100db at my listening position), I prefer the gentle slopes to the steeper ones on my Behringer DCX2496. With this tool it's very easy to jump from one set of parameters to another to compare. From all I've heard about steep slope crossover networks being better sounding, I was surprised that I preferred the gentler slopes. It's really quite revealing when you get to hear the difference between the two in a A/B test, while listening to music.

I have tried both the 6K and 4.5K crossover points with my Khorns, with Beyma CP25 tweeters, and I really don't notice much of a difference in quality of sound. That is only a 1/3-octave jump in frequency. It's the equivalent of going from 100Hz to 160Hz. Noticeable, but probably isn't going to make a huge difference in sound quality. In terms of overlap of frequencies, with the AA networks, there is nothing to limit the high frequencies coming out of the midhorn, it is designed to let the horn roll off on the high end naturally. So lowering the crossover point of the tweeter will overlap the lower frequencies of the tweeter into the upper midrange more than they already do. If you went with a crossover network that provided the 4.5Khz crossover point both in the tweeter and the midhorn, the overlap would be less. But I have stayed with the 6K crossover point anyway - sounds just as good to me.

Greg

[Al Klappenberger]

As I have said on may occasions, be very careful how you pick your experts.

Al K.

[psg]

Al, please debate the ideas rather than attack the person.

Stuff might have been said in other threads, but people can only really judge a thread on its own. And I didn't see anyhthing from Greg worthy of such a reply...

[Al Klappenberger]

psg,

I BEGAN by stating facts. The "person" simply presented his personal opinion on how something sounds to HIM. The question is the competence of a give "expert", not the "person". Certain people's opinions implicitly carry more weight than others because they are involved in somewhat related endeavors. A person who builds crossover networks, cabinets or houses is no more qualified to say how a stereo "sounds" than a man who operates heavy machinery for a living. The human ear needs a reference to judge the quality of music reproduction. You only get that my listening to live music on a regular basis. Only THAT qualifies a person as a judge of how something sounds! I don't, so I keep my mouth shut about how something "sounds"!

If you want relevant debate, look at this web site: http://demonstrations.wolfram.com/WaveInterference/

The speed of sound in air is 13,560 inch / second so a wavelength at the 6000 Hz crossover is about 2.25 Inches. The path length difference between the squawker and the tweeter represents a different number of wavelengths as you move in either direction from the 6000 Hz crossover. The more gradual the slopes of the filter the wider the spectrum where this mess happens in your room. On the web page, each blue dot represents a single wavelength in distance. Run the animation and you will see just what kind of a MESS driver interference creates in your room. What it "sounds like" to each person is secondary to that fact that it is not correct. It is false ambiance. I compare it to singing in the shower. It sounds good, but it's not "correct"!

A 1st order, 6 dB/octave crossover is the worst possible case for driver overlap spectrum width and a horn loaded speaker is the worst case of time misalignment. To avoid the mess of peaks and dips throughout the room a 1st order network must have perfect time alignment, which is NEVER totally possible, even in the best situation, a coaxial configuration.

Look at the illustration below (from the web site I noted above). It shows about 5 wavelengths between sources. At a 6 Khz crossover, that represents about 11 inches. How does that compare to the path length difference between squawker and tweeter?

Al K.

[PrestonTom]

The OP asked many questions. Let me addrees the first. The slight channel imbalance you are hearing is probably not due to old capacitors. A slight channel imbalance can also be corrected by moving your chair 6 inches to the left.

However, there are a number of reasons to refressh the caps. It does not need to get expensive. If you want to do something fancy, then try "bypassing" the caps with a small film cap. There are a number of threads, so spend some time searching.

My personal opinion is to re-fresh using Solen polyprop caps and bypass them (wire in parallel) with a small Dayton (film, foil whatever). You would be able to get the parts at Parts Express for less than $25. I also agree that it is a great time to clean up the contacts while you are at it.

[Nat Denkin]

Hello everybody,

I'm planning to update the networks in my LaScala because I feel they don't sound as clear as they should. What I have now is the original AA networks and they look pretty bad, altough I don't have the required equipment to actually measure them.

I'm also experiencing some channel unbalance, some sounds being stronger on the right speaker than the left one, it's quite subtle but it kinda screws he stereo image, I feel like bad caps in the left speaker might be the cause (?).

....

theorically I'm attracted to the 6dB/oct attenuation of the type A, but since both my midrange AND tweeter dispersion pattern are much, much wider than the stock ones, I'm really concerned about driver overlap. I don't wanna end up with bumps and dips in the frequency range between, say, 4,5KHz and 8KHz. IMHO I feel this would be hard to avoid with a first order network (but then, why does it sound better to some people?).

....

While there is no free lunch, the ears are rather forgiving. Networks affect both the amplifude and phase of signals with the steeper slopes having a greater affect on both frequency and phase. While some will dispute whether you can hear the phase shifts, consider that you can sense the direction of sound not merely to your left or right, but also in front or behind and above or below. The delay of a signal getting to one ear and then the othe does not fully explain our ability to locate the source of sounds in three dimensions. The explanation of how this works has to do with reflections in the outter ear which cause a frequency dependent phase shift. For best stereo image, you need to consider phase. Keep in mind, you can recreate only the sound as it was mixed. If the engineer merely put the singer on the left feed and the piano on the right, phase is less important. But, if properly mixed, you can image each member of a small group.

A function of your network is to elimate out of band power going to each of your drivers. A 6 dB slope/octave permits lots of signal getting into the woofer and squarker that they do not handle well. A higher slope is generally needed. Higher slopes require more parts and higher precision in selecting (or tuning) them, That adds to the cost, but higher slopes do reduce the overlap between the drivers.

I use the Beyma CP25 and cross them at about 6kHz. They have a good response well below that so you need to consider at least 18dB/octave for them. I am driving the Beyma's directly with a Crown D-45 (triamping) with no problems. During typical music, less than 10% of the power would be going to the tweeters.

Bottom line, recap your AA. I like the ALK universal which I purchased in kit form (you don't save a lot by buying parts yourself!). Others solutions get progressively expensive and your satisfaction may depend more on the music you listen too, room acoustics, and personal preferences.

In one of my setups, I use the ALK universal with the Crites woofer, Atlas mid-range driving a 1" ALK trachorn and Beyma tweeter. In an ongoing project, I am triamping using a Behringer DCX2496 to provide 18dB/oct slopes as suggested by Le Cleac'h to optimize phase response. I am building Jamboree base bins to go with Greg Roberts' Lavera horns. The stereo imaging is quite good, but varies with recordings as sound engineers vary too.

Good luck,

nat

[Hardhead]

On September 6, 2010, greg928gts wrote: "I know from my own experience, even at fairly high volume levels (100db at my listening position), I prefer the gentle slopes to the steeper ones on my Behringer DCX2496."

Yet on August 28, 2010, in a post to a thread he began entitled "I Tested Five Different Tweeters in My Khorns," greg928gts wrote in reference to listening tests he made using the Behringer DCX2496 on five brands of tweeters: "I did mention that I started off with a 12db slope, but changed to 18db because of the amount of midrange overtones making their way through with several of the tweeters. After I did the evaluations, I played around with different slopes with the tweeters, and that's when I realized that the Beyma CP25 had significantly less of those midrange overtones than any of the other tweeters. As such, a 6db slope worked great with them. I also think this may have something to do with the Beyma testing about 3db lower in output compared to some of the others with a pink noise test, yet when I listen to them in my Khorns, they balance out just fine. I think the midrange overtones that are stronger in other tweeters, affect the measured output using a sound level meter."

In light of the above seemingly mutually exclusive statements by greg928gts, I note that, he states on August 28 that during the tweeter tests he heard obviously audible--and therefore obviously objectionable--midrange overtones on four of the tweeters. The objectionalable midrange overtones were reduced by the Beyma CD25 but only because of the reduced output of the CD25. In other words, the 6dB /octave slope worked well only with the CD25 and only because of the lower output of that tweeter and not because the 6dB crossover slope is "more correct." Greg928gts' whole August 28 statement and the results of his entire test of the five tweeters sound to me like an endorsement, not of a 6dB/octave slope or even a 12dB/octave slope or any kind of "gentle" slope crossover that he says nine days later he prefers, but, in fact, an 18dB/octave slope. For the close listening required for his tweeter evaluation, it is significant that greg928gts used the 18dB slope on his DCX2496, which is not considered a gentle slope in anyone's book..

His statement, "I prefer the gentle slopes to the steeper ones on my Behringer DCX2496" doesn't seem to reflect his own previous statements and observations. While he may, or may not, actually use a gentle slope crossover for everyday listening, his own words say that the Beyma CD25 driver is the only reason a gentle slope can work for him and not the steepness--or lack thereof--of the slope itself. Consequently, greg928gts' general statement of his preference for gentle crossover slopes is both inaccurate and misleading without those qualifications.

[radiogram]

Hardhead:

You have a point in that if a tweeter has lower sensitivity then the resulting output will reduce overlap. However the downside to it is that the acoustic output in the highs also are lower and out of balance. Now I do not know clearly if and how Greg had all tweeters tested at the same sensitivity, Greg mentioned that he also tuned by the ear (which I am not sure is a reliable method to match sensitivity at high frequencies, but that's beside the point here).

If I look at this post comparing the plot of K77 vs Beyma (http://forums.klipsch.com/forums/p/40675/372387.aspx), on the surface it may seem that the Beyma actually starts rolling off (i.e on the lower side of the bandwidth not the higer side) even earlier thna the K77 and its extension at the very end may seem moot since it is already at levels about -5dB and further taking into account the -xdB imposed the by crossover and hence should render less overlap than the K77.

However, the average sensitivity of the Beyma in the 5Khz-12Khz range is actually lower than the K77. So, if we match them up then certainly the Beyma will be more extended in the lower nethers in a more audible passband and should consequently result in higher overlap compared qith say K77.

[Al Klappenberger]

Here's a couple random thoughts on all this. Again, I try not to say if something sounds good or bad to me.That's just my opinion and counts no more or no less than anyone eases.

Anyhow, I did some listening tests on several tweeters (Beyma CP25 and the B&C, as I recall) to try to compare dispersion. I have not been able to do it in any repeatable way by instruments, so I tried rotating them around thier axis while listing to white noise. I was more struck by the difference in the sound of the noise between them. The example is when I evaluated Bob's CT125 a few years ago. I noted that the CT125 sounded like "FFFFFFF" while the K77 sounded more like "SSSSSSS". I realized that the difference was that the CT125 simply has a response that extends lower in frequency than the K77! After connecting an ES7500 network to both tweeter to limit the spectrum going to them, the Beyma and B&C sounded identical to my ear! To compare tweeters by ear with all of them in a big rack side-by-side is useless unless you have carefully equalized their sensitivity with L-pads using instruments and limited their frequency ranges with a network that cross over at the lower limit of the tweeter with the highest low cutoff or higher. Doing test like that, even if done correctly are just one person's opinion anyhow! At best in will allow you to choose what tweeter YOU like best with the network you happen to be using at the time.

There is also the question of phase errors. I happen to have 5 (five) different signal generators here that will generate clean sine wave signals. A square wave is a frequency and all it's odd order harmonics (3,5,7 ...). I connect three signal generators up so they were set to 1000, 3000, and 5000 Hz and mixed them together in the correct proportions. I fed the combination to an amp, and oscilloscope and a set of headphones and listened. Since the three generators were not phase-locked to each other as the components of a square wave are, the phases become random and continuously changing. I set the generators for the most stable waveform I could on the scope.At best it looked like a worm squirming slowly around. I found that even by looking at the scope for clues while listening, the sound was a continuous unchanging raspy tone! In the case of a loudspeaker where separate drivers would be reproducing each tone (harmonic), the phase would be random but fixed at some given phase. The only conclusion is that my ears are totally deaf to phase errors between the components of a complex waveform. I just have to assume that my ear / brain combination is typical of all of us. In the case of music, the phase relationships are even more random than with a contrived waveform like the fake square wave I used. Your ear / brain can, however, perceive the phase relationship between a SINGLE sound as heard by TWO ears. This creates a stereo image. A steep slope filter generates faster phase shift with frequency shift (group delay) than a gentle slope one but that does not seem to be important so long as the shift is matched between the left and right speaker. A steep slope filter also has it's maximum group delay right near the crossover where the interaction between drivers is inevitable anyhow. BTW: Group delay and time delay are NOT the same thing and one can NOT be used to compensate for the other in any practical way. I know, I have done lots of group delay equalization in passive filters. .All-pass networks are very expensive at audio and generate lots of loss to a passband!

Al K.

[Johnny's Jill]

His statement, "I prefer the gentle slopes to the steeper ones on my Behringer DCX2496" doesn't seem to reflect his own previous statements and observations. While he may, or may not, actually use a gentle slope crossover for everyday listening, his own words say that the Beyma CD25 driver is the only reason a gentle slope can work for him and not the steepness--or lack thereof--of the slope itself. Consequently, greg928gts' general statement of his preference for gentle crossover slopes is both inaccurate and misleading without those qualifications.

Before we develope our own interpretations of what Greg has written, it would be nice to know what he considers steep slopes. I certainly do not consider 3rd orders steep, nor do I believe ALK does. In fact I believe ALK refers to 2nd and third order networks as gentle slopes.

It is a real shame that ALK and Volti are not business partners, AL's engineering abilities and Greg's skills combined would be an unbeatable combination that would benefit everyone. You guys already fight like sibling, you might as well hop into the ring and sort out your differences.

Craig...out....73

[Al Klappenberger]

" In fact I believe ALK refers to 2nd and third order networks as gentle slopes."

Yes. There really isn't a bright line between gentle and extreme slopes. It's more a matter of what's in the middle! Most specification sheets for drivers want 12 or 18 dB / octave filters. That is what you see on most speakers. Active crossovers offer 24 dB / octave (N=4). To real filter people the entire concept of db / octave isn't even used. It refers to Butterworth response. Most flter people think Chebyshev which is much steeper than the same order Butterworth, especially at higher orders, like N=7, 9 or even 11. I have done as high as N=21. Next up is Elliptic function of orders 3, 5, 7 and 11 or even higher. N=11 is quite common. My ES networks are only roughly N=3 or 4! I simply look at the skirt from the crossover to the point where the attenuation starts to reduce (the arc top) and relate it to a Butterworth to get db / octave. It's only approximate. It turns out to be around 120 dB / octave. They are actually what filter people refer to as "zero placed". That's like elliptic function but not quite. Elliptic function filters have all equal arc tops between "notches" (zeros). A zero placed filter can put the notches anywhere. With only a single "notch" there is only a single acr top so it's had to call it elliptic even though N=3 and N=4 elliptic have only a single notch. Confusing, isn't it?

"It is a real shame that ALK and Volti are not business partners"

It actually started that way. Before the cat-fighting, there were quite a few Trachorns and ALK networks installed in Greg's speaker restorations. This is how Greg got license to build 10 of my Universal networks. I even made a prototype Es300 prototype crossover for his 2-inch horn. A dubious idea! That ended when I had the audacity to actually offer a 2-inch Trachorn! It was his idea to pull the plug. In the future I will be working with Dave Harris of the Fastrac horns instead.

Al K.

[KanedaK]

Wow. Thanks everybody, it's nice to see you guys are as passionate about this as ever!

I guess I'll stay with my AAs for now, just recapping them, maybe adding that bypass cap.

What surprises me is that the Beyma would measure less output than a K77? I've had two different sets of K77, one with the alnico magnets, one with the square magnets, and both sounded less strong than when I installed the Beymas...

[Al Klappenberger]

Your experience with the various tweeters just illustrates the fallacy of listening tests by a single person. You got to start with good instruments and then do the listening. The human ear / brain combination is a very subjective instrument!

Al K.

[greg928gts]

Your experience with the various tweeters just illustrates the fallacy of listening tests by a single person. You got to start with good instruments and then do the listening. The human ear / brain combination is a very subjective instrument!

Al K.

It's not a fallacy, it's a subjective analysis, just like every other part of the audiophile hobby. The instruments are indeed the place to start. Every one of those tweeter manufacturers has rooms full of instruments to design and test their products. They do not need additional testing to provide "a start". They are all good tweeters that "someone" out there is going to like to use. The problem with you Al, is that you fail to recognize that there's any other way to evaluate these components except your own. The fact is, there are many different ways to evaluate sound, and they all have merit. In the end, it's going to come down to what the end-user enjoys for his/her own system. And that IS a subjective evaluation. No amount of re-testing of components by Al K or anyone else is going to change that.

This is why my business model will be successful, because intuitively my customers understand the nature of evaluating sound and music. It's about what we all hear individually. My customers appreciate the efforts I make to implement new ideas into the products that I make and the subjective evaluations of those products that help them decide what might work best for them. For those customers who prefer to look at graphs and charts to determine what they want for their own system, that's fine too. Different strokes.

I appreciate you giving me the opportunity to express this here in retort.

Greg

[Al Klappenberger]

"The problem with you", Greg, is that you have no calibrated instruments, would not know how to use them if you did and therefor do not understand their value. This being the case, all you have left is your ear / brain combination and you try over and over to convince us that what YOU hear is the same as each of us hears also. It isn't true. That is why loudspeaker manufacturers use groups of people to evaluate how a speaker sounds, not just one.

Stick to building speaker cabinets. Nobody, including me, questions your ability in that area.

Al K.

[psg]

If you want relevant debate, look at this web site: http://demonstrations.wolfram.com/WaveInterference/

The speed of sound in air is 13,560 inch / second so a wavelength at the 6000 Hz crossover is about 2.25 Inches. The path length difference between the squawker and the tweeter represents a different number of wavelengths as you move in either direction from the 6000 Hz crossover. The more gradual the slopes of the filter the wider the spectrum where this mess happens in your room. On the web page, each blue dot represents a single wavelength in distance. Run the animation and you will see just what kind of a MESS driver interference creates in your room. What it "sounds like" to each person is secondary to that fact that it is not correct. It is false ambiance. I compare it to singing in the shower. It sounds good, but it's not "correct"!

A 1st order, 6 dB/octave crossover is the worst possible case for driver overlap spectrum width and a horn loaded speaker is the worst case of time misalignment. To avoid the mess of peaks and dips throughout the room a 1st order network must have perfect time alignment, which is NEVER totally possible, even in the best situation, a coaxial configuration.

Look at the illustration below (from the web site I noted above). It shows about 5 wavelengths between sources. At a 6 Khz crossover, that represents about 11 inches. How does that compare to the path length difference between squawker and tweeter?

Al K.

The combfiltering is impressive to look at when the two source signals are of equal strength, but maybe less so when the second source is 6 dB down (the playback lowers the amplitude of one source once, very quickly). Some people say combfiltering is a major issue, but on the other hand I love the sound of my two Klipschorns even when they are fed a mono signal. The comfiltering should sound very bad when you look at a similar plot:

[Al Klappenberger]

psg,

"The combfiltering is impressive to look at when the two source signals
are of equal strength, but maybe less so when the second source is 6 dB
down"

EXACTLY! The next step is to quantify over what portion of the spectrum both sounds are equal enough to interact. This is what I call the "interference window" for the lack of any universally accepted term that I know of. I really have no science to specify what that level is. 6 dB is probably a good number. I have always assumed 10 dB just for comparison between filters, but what is important is that the steeper the filter slope the narrower that "window" becomes. It is the widest with a 1st order network. With a slope of 120 dB / octave, it is very narrow. The consensus of opinion of the people who have invested in networks having sharp slopes like this is that the stereo image comes nicely into focus.

"on the other hand I love the sound of my two Klipschorns even when they are fed a mono signal."

This is no surprise. A gentle slope network in each speaker would generate a pleasing ambiance with a mono program. I suppose extreme slope networks might make mono programs sound a bit dull. I never thought about that. Most people don't listen to much mono programs any more so I suppose it's a minor down-side.

Al K.