So the only way to design a loudspeaker crossover correctly is to take very accurate measurements of each of the driver's frequency and impedance curves with their corresponding acoustic and electrical phase response, in their final enclosures and after they are fully burnt-in. Also waterfall plots must be made to see if there are any parts of the frequency spectrum that take a little too long to decay. And add in a few off-axis frequency measurements to check baffle edge diffraction problems and some distortion measurements for spotting any other nasties. And don't forget to measure the output from the port either - especially with a two-way system, there can be quite some midrange energy radiating from the bass reflex port!

After thorough interpretation of all this data, the relevant sections are fead into accurate and versatile simulation software and several crossover typologies and options are modelled. This is where over 30 years of experience comes in handy to know what has potential and what is a complete waste of time. This modelling results in about 5 or 6 different crossover schematics that all seem to produce a more or less flat summed frequency plot, an amplifier friendly impedance curve and are kind enough to the driver / cabinet combination used. They may vary in steepness of the slopes, crossover-point, have lots of additional correction correction-networks or be relatively simple, etc, etc. Each of these 5 or 6 different crossovers are then built in real-life with standard quality components and listened to exstensively. This part of the crossover design is quite interesting as it lets you hear how the various driver / crossover combinations work (or don't work) together. Some combinations "harmonise" better than others but they all have their own sonic signatures. After this lengthly period of evaluation you are left with about half the amount of crossover schematics that still seem to sound okay. Then it is time to measure again. These measurements will show any potential "problem area's" and together with their sonic characters help narrow down the options. Just to keep things simple: say you are left with two filter possibilties that both sound quite good. With the recent measurements in the back of your head, further modelling of these crossovers and even more listening sessions, you try to "tune" the two options until you have the feeling that you can "squeeze" them no further. Let them both compete with each other, like a sort of speaker "shoot-out". When crossover A seems to be the winner soundwise and concerning measurements, then "tune" crossover B even more until it seems to be on top. Keep this process going as long as possible combining constant measuring with listening to all types of music, from string quartet to big-band, from electronic dance to large orchestrial and choir works (see my Music Page for inspiration). Don't make the mistake of only using these so called audiophile recordings of jazz-trio's you often hear at hifi-shows. They always sound good, even on a crappy system. Give them something healthy to eat like a nice helping of Johann Sebastian Bach's Weihnachtsoratorium. This part of the design has nothing to do with personal taste nor is it part of any "voicing" stage. All that should be on your mind during this stage are things like balance, neutrality, sound-stage and coherence.

After a while you should be left with a crossover schematic that measures very well and sounds very good. This one will serve as the basis for the final design stage. Now it's time for fine-tweaking, system matching and a little bit of personal taste. The crossover is now built with high-grade crossover components on the more critical positions. This can get a bit tricky because you have to take into account the signature of the loudspeaker drivers, etc and the signature of the crossover components. They all have to blend together well to form a synergy that takes the system to a much higher level. There is no point in just buying "the best" components money can buy, the chance that you will get a perfect "match" that way is rather remote. What I usually do first with a two-way system, is to start with the midwoofers low-pass inductor (L1 in the schematic further down this page) because the choise here is limited: air-core's of various wire diameters or AWG's, round wire, flat wire, hexagonal wire, litze wire, copper foil with polypropylene or copper foil with impregnated paper - that's it! When you have decided on your favourite inductor then it's time for the tweeters main high-pass capacitor (C7+C9 in the schematic). In this case I mixed two flavours roughly in a 1:7 ratio to obtain the taste I like. From experience I know these two go together well like salt & pepper, peanut butter and jelly, fish & chips, gin & tonic, etc, etc. Another sonically critical component is the resistor that is directly in the tweeters signal path (R3 in the schematic). I put may favourite resistor here: the Mundorf M-Resist Supreme, very affordable, reliable, spatious, detailled and smooth. This substituting process is repeated for all the components throughout the crossover, obviously some positions are more critical than others. For example, to obtain a maximum in coherency, I chose the same brand and type of inductor (just a smaller gauge) for the high-pass inductor parallel to the tweeter. But the inductor in the correction network also parallel to the tweeter, is a normal small gauge baked wire air-core (L3 in the schematic).

The final icing on the cake are the bypasses. These are components with a very small capacitance compared to the main value. You don't see any difference to the standard measurements when they are in place, but they do make small audible enhancements in things like smoothness, micro-detail and imaging. These parts are so cheap that you can use them in all positons and blend multiple types till you obtain your favourite blend.

For the first version, I built the woofer low-pass section and the tweeter high-pass section on two seperate boards made from 9mm Baltic Birch plywood so that they are a bit easier to mount inside the cabinets. Also it makes the photo's easier to explain. The crossover boards have holes drilled at the necessary posistions and the components are held into place by strong, high quality Hellermann Tyton tie-raps. All connections are first crimped together to ensure good mechanical contact. Then the connections are soldered with RoHS compliant lead-free solder with 3% silver content. The crossovers supplied with the kit are built the same way except they come on one board as shown at the bottom of this page.

The photo's on the left show the woofer low-pass section with the large Wax Coil dominating the other corrective elements. Basically it is a 2nd-order filter with a Zobel network to flatten the woofers impedance due to inductive rise caused by the voice-coil. A small RC-network parallel to the Wax-Coil tames the woofers peaking in the octave from 2500 to 5000Hz. The Wax Coil's copper foil leads are insulated with shrink tubing to avoid short circuit should they accidentily come into contact with other conductive parts. Note the little Vishay MKP1839 bypass capacitors across all the capacitors in the woofers parallel circuits. The two photo's on the right show the tweeter 2nd-order high-pass section. The tweeter series capacitor is an example of hifi "Haute Cuisine" and consists of a blend of Intertechnik Tri-Reference and Mundorf Supreme Silver-Oil, garnished with some Vishay MKP1837 and MKP1839 bypass capacitors with a 100K carbonfilm resistor and Styroflex capacitor for topping. Parallel to the tweeter we see an LCR-network that tames some excess energy in the lower treble. The capacitor in this network is also bypassed by a Vishay MKP1839. The green resistors in the parallel sections are Jantzen Audio SuperRes Non-inductive.

The schematic shows the main filtering components in red and the fine-tuning bypass elements in green. The tweeter high-pass capacitor consists of the parallel connection of C7 + C8 + C9 + C12 + C13 + R6. The carbonfilm resistor's capacitance of a few hundred picofarads works the same way as a bypass capacitor. Both drivers are connected in positive phase. The crossovers are only supplied in matched pairs, fully assembled and tested as part of the L' Orfeo kit as shown on the photo below - the cd is not included, it is just there for scale reference!

No part of this website may be reproduced in any form without written or e-mail consent from the author. The designs are free to use for private DIY-purposes only, commercial use is strictly prohibited. Copyright Humble Homemade Hifi © 2000 - 2012