Best three way cross over?

Big question! It sounds like it might be a perfect time to consider actively bi-amping your speakers. Using the active x-over between the bass and mids, would eliminate that part of you x-over causing problems. (By the way, you didn't mention your speaker model.)
The best electronic x-over (in many peoples opinion) is the Krell KBX. Krell will make a custom board for almost any speaker for $200. There are KBX's for sale on this site and on ebay, right now.
Also, I recently changed from passive to active x-over using the KBX.........the single biggest improvement that I have ever made to my system.
Richard

This site http://www.sound.au.com/ has a very lengthy article on the merits of biamping/triamping loudspeakers using active crossovers. It also has a project on designing a very nice phase coherent 4th order linkwitz riley crossover. Marchandelec.com also has many electronic crossover kits. Once you opt of electronic equalization you'll need to adjust the gains on your individual amplifiers to smooth out the frequency response since your three drivers probably don't have the same efficiencies/sensitivities, and the passive network was designed to attenuate this. There was another thread here just a few days ago asking about electronic crossovers too--it might have some info to help. The electronic if done right will probably increase the performance. You won't lose power in the passive network-in many cases the equivalent of halving the amplifiers power, and intermodulation distortion can be reduced. Passive crossovers in three-way systems can become unpredictable at higher output levels because the temperature of the voice coil changes and therefore so does its electrical characteristics. It would help to state what speakers you have--PSB or Avalon Eidolons? Another thing is that the quality of parts in your passive network could possibly be better. With seperate amps you'll make full use of there effective damping to a given driver. But do contact the manufacturer of your speakers to find out just what they've done with the crossover--sometimes the crossover can be used to smooth the frequency response of the system even in a single drivers operating range--not just the crossover region, which may complicate things. I've been toying with biamping myself, but in the case of two way designs I'm still uncertain if its worth it. However, with a three way it many cases it is.

There may be a simple solution to your problem. Some speaker drivers improve dramaticaly with the implementation of a zobel network across the driver . Both my friends Dunlavys and Pipedream speakers were improved with zobels placed across their drivers. The Dunlavy's Bass driver in particular sounded "muddy" before a zobel was used (we didn't realize it was "muddy" until AFTER we did the zobel, boy were we shocked at how much it improved!). It is extremely unfortunate that many speaker manufacturers don't implement (or understand) how beneficial these circuits can be. In many cases a Zobel makes perfect electrical sense in improving the drivers control by the amplifer, and can do wonders for even the most highly regarded "recommended" speakers. It amazes me that so many audiophiles think that these speakers' designs cannot be improved upon. You may also see contradictory responses to my zobel recomendation that state things like: "but zobels have issues such as their power robbing ability". Unfortunately, this is not the case, and just misinformed audiogon speaker tweakers, who don't really understand what a zobel does, or how these circuits are almost always a "win-win" scenario when they are implementated correctly.

Ehider, in response to the zobels, are they ever used when a speaker is being driven with an electric crossover? And more specifically what do they do to improve performance? I've only seen them used with high capacitance speaker cables like Goertz to keep unstable amplifiers from oscillating (If memory serves). I'm just curious/asking since I don't now all that much about them. Its just that the disappointing aspect of passive networks is that the only way to smooth the frequency response of two drivers of different sensitivities (not uncommon) is to knock the more efficient driver down to that of the least efficient driver (unless you're sticking a horn on one the less efficient ones, not common). So technically using electronic equalization will allow the driver to perform at its full sensitivity and make the most effective use of the amplifiers power. A not uncommon padding of 3 db on a driver is the equvalent of needing an amplifier twice the power to reach the sound pressure level (compared to an active design all things being equal of sorts). I'm just dwelling on it now; I get a little teary-eyed sometimes when I see a 95db efficient driver built into a system that only carries a 90db system sensitivities--5 db just thrown away :( No, watts aren't that expensive (although they can be if you want them too) but.....if you have your heart set on some little 10watt class A amp 5 db's alot.

Thanks for the responses. I have Tyler Acoustic Speakers with (2) 8" Seas Excel Drivers in each of the Bass Cabinets, and a Seas 6.5" Mid and Millenium Tweater in the Monitors. The speakers are fabulous and I love them to death. But at the same time noticed an increase in degredation in sound going from a 2 Way to a 3 Way crossover (I previously had just a pair of Tyler Monitors) and have realized that in going for a three way design there is a whole nother set of high pass filters my midrange goes through to get to the mid driver. i.e. more mud!
I have now completely removed the internal crossovers and made external crossovers which will allow me to experiment more easily. Just doing this has helped as in the process I also directly connected the internal wire to the speaker leads, etc. I also now have tri wired from the amp to the crossover...and then on to the speakers. This has had a huge benefit. I have contacted Tye at Tyler Acoustics and he has been wonderful and very helpful (he has even sent me som new capacitors for the mid high pass filters). I'll let you know what we come up with further in a future post.
I see now that I simply had no idea how much ALL passive crossover components degrade the signal. It is amazing. Just bypass the crossover - briefly for experimental purposes - and it is amazing the immediate increase in clarity. Its astounding. Truth be told, I have always been very pleased with the clarity of my speakers, and I have compared them favorably to tons of top notch speakers....the closest I think I have found is actually the Avalon Eidelon you mention Ezmeralda11. I was told they have a new type of 3 way crossover.....anyone know anything about the Eidelon crossover.....whatever it is it sounds very good! The only drawback to that speaker is it cannot be bi or tri wired directly, it has a single terminal for wiring. Anyway, hearing the speakers direct (not through the filters) has led me to see why you are all mentioning active (powered) crossovers.
I was hoping not to have to bi or tri-amp, but it seems this is the only option besides the zobel network Ehider mentions. BTY I found this acticle on it: http://www.trueaudio.com/st_zobel.htm Ezmeralda11
maybe you will understand it better than I, and can explain what it does from this site to us all in layman terms! I also looked into the Krell KBXs and for a three way system (and to not use passive components at all) I would need to pairs of them as they are just two way crossovers. Gets kind of pricy, along with the purchase of another set of amps to use the active/electronic crossovers. Does a good less expensive active exist? Is there such a thing as an active crossover the would work AFTER the amp stage? Is there any option besides passive components using a single amp?

Hi Mp: certainly you've thought about, or have even actually tried, to upgrade that offending crossover cap. with a different component of equal value capacitance but of better quality? There are a number of brands available (Solen, Hoveland, MIT multicap, Musicap, etc.) so it seems that some experimentation in that regard might be the first & easiest approach to cleaning up your midrange.