Autoformerless crossover

[Erik Mandaville]

Very happy with the simple type 'A' network that uses a high pass only on the squawker and tweeter, I had no intention of ever building this. It came by way of an experiment, and, after some work and listening, has resulted in what is shown. What I found is that without question, as far as I'm concerned and think, it's possible to create a very capable network for these speakers without the traditional autoformer. Regardless of the differences concerning attenuation between the two (autoformer and L-pad), I found the fixed, two resistor L-pad more favorable. That's my personal preference, and doesn't make one better than the other, although I have known for years about P. Klipsch's preference in the matter.

This is a straightforward design, but I liked the way the 18dB/octave slope sounded better than the first one I tried in this design, which was the high-pass only first order. How this works out with my single ended amps -- I don't know. I didn't test that. With the Teacs, which is what we have been using for the entire system, it is really satisfactory.

The coils I started with were ballpark values, that I needed to be able to make a network that was independent of the autoformer. Since the difference between the low pass crossover point (400Hz) and upper end (4,000) is one decade apart, I chose to use a narrow band-pass section for the midrange. When I first did this, I didn't have quite the right values to do that -- only an approximation of it. So, using the old 2.5mH coils from our '60s Heresies, I wound the exact values of inductance I needed for the narrow bandpass. The rest, as far as the capacitors needed for the third order in the tweeter, as well as the inductor for the low pass, are textbook. Just as the type 'A' doesn't use impedance (zobel) compensation for either the tweeter or squawker, nor does this network. That I was using very familiar drivers made it easier, in other words. We know of the 9kHz notch for the K55V midrange push terminals, however the 'A' makes no compensation for that. I also left out a corrective notch filter because I've never had a problem with little issue in the response. Checking the specs for the tweeter, I also found that an 18dB slope is suggested for it. That was a coincidence, though. I used one since I'm using a lower crossover point, adding a little extra protection.

The picture shows two boards for each channel. The choke for the low-pass are of course 2.5mH, and measured only very slightly higher than that. The DCR is extremely low. The woofer inductor boards are on the other side of the top sectioin, connected by their own lead to the main crossover board.

I'm glad I started on this unplanned project, because what I have is better than what I had before. That's without testing, which I would like to have done independently.

Erik

[Groomlakearea51]

Two boards per channel? or one board is left and the other is the right? Excuse me if I sound ignorant, but it's very interesting and appears like something worth trying in the future.

[Erik Mandaville]

That's right. The boards are close together, side-by-side, one for each channel. The large inductors for the woofers to the rear are mounted on separate boards, and connect to the tweeter/squawker boards by their own leads. I want to compare them to large AWG air core inductors when I get the wire needed to wind them.

I have been very satisfied with the results. Imaging is particularly good, and that's without our center channel activated. Just two channel stereo. In fact, almost all the music we've been listening to and testing has been without the center, which I always preferred before. I still do for movies, but for music, there isn't even a hint of a hole in the middle and our K-horns are approximately 20 feet apart. I'm not sure if this has to do with the higher order in the tweeter branch, but I am very happy with how things sound. I think the standard type 'A' is still a good network, particularly for lower power amps, but this new one is a significant improvement -- for me.

Thanks for your comments,

Erik

[Marvel]

I would be interested in what you think, using them with your single ended amps, but you may not want to switch them out to try.

Bruce

[Erik Mandaville]

"I would be interested in what you think, using them with your single ended amps, but you may not want to switch them out to try."

Very simple to do if I use the Moondogs. Remember, this is primarily a low order network, and I found it to have slightly lower insertion loss than than the standard type 'A' with an autoformer. The more complex aspect is the tweeter section, which uses a third order, 18dB/octave. Compared to the first try with this network, where I was using a 6dB/octave at 4,000Hz, the sharper cut-off provided by the higher order approach is obviously smoother. It *may* have padded the tweeter down just a little bit, as well.

In any event, the first order narrow band pass on the squawker and single series inductor on the woofer are both not greatly different than the type 'A,' Bruce. Of course there are aspects that aren't similar, given the fact that I'm not using an autoformer at all. The 'A' does not use a band pass for the midrange, which I prefer over just a high pass capacitor.

This network should be easier or at least AS easy to drive with a single-ended amp than the A or AA. I'll try it today and let you know.

Erik

[Groomlakearea51]

Erik,

Reason I'm interested is that one of the pairs of K'horns will be driven at relatively low power levels, and I'd like to "build" a simple crossover set as a project. Perhaps you could post the parts necessary to do this? I'll be using the K-33-B, K-55-V, and probably BEC's CT-125's (although I have K-77's as well).

[Erik Mandaville]

I would be happy to post the complete network as soon as I feel it's ready to do that. The BEC tweeters can handle the lower crossover point I'm using, but I wouldn't use the same frrequency for the K-77s. One of the main things I like about the new tweeters Bob created -- or rather modified for Klipsch Heritage -- is the lower crossover point capability. You could also use the K-77s at 6kHz, and just use a first order network for the tweeter (the cap value needed is a little over 3uf, if I remember right). It's so simple: install the capacitor in series between the input and the driver. For the midrange bandpass, you use an inductor and capacitor in series between the input and squawker, but will need to attenuate it to match the output of the other drivers. The values for those two components in the band pass are determined by the impedance of the driver (16 ohms if not using an autoformer), and the upper (6kHz) and lower (400Hz) crossover points. To match the squawker output, the last in line (after the bandpass inductor and capacitor) is the L-pad. For 3dB, that's made with one resistor in series with the driver, and a second resistor in parallel with it. If you're using an autoformer, I would put that after the bandpass, but you would need to calculate the values for capacitance and inductance based on the reflected impedance imposed by the autoformer. In other words, if the autoformer is showing the amp a 32 ohm load (which is why 13uf is used on the type 'A'), the value for the series inductor after the capacitor needs to similarly take that into account. The autoformer attenuates by way of intentional impedance mismatching, so, depending on what tap you use, you will need to compensate for that by the selected values of capacitance and inductance.

Then just use the usual 2.5mH in series with the woofer. I have two 'antique' Klipsch coils you can have if you'd like them.

Erik

[boom3]

I'm not sure if this is the right point to bring this up, but I'm just back in town and catching up. Why did Klipsch use L-pads in the latest K-horn crossover instead of autoformers?

[Erik Mandaville]

"I'm not sure if this is the right point to bring this up, but I'm just back in town and catching up. Why did Klipsch use L-pads in the latest K-horn crossover instead of autoformers?"

I can't answer that. Some people even use the autoformer in extremely high order networks, so I'm not sure that the sharpness of the cutoff or order of slope was the motivation.

There could be many reasons associated with the decision. In terms of cost, resistors are very inexpensive compared to the autoformer, and can be purchased in huge quantities. The reason I'm using them had nothing to do with cost. It was an experiment, and I like the band-pass-plus-resistor L-pad much more than it's autoformer counterpart. For me, cost is often totally irrelevant, because it *may* have absolutely nothing to do with sonic improvements on a subjective basis. I have 6 unused autoformers, and I prefer simple resistors. The autoformer is heavier, looks kind cool (I think), and I just prefer a simple L-pad. That's all.

I want to stress again, though, that the choice is my own. I'm not saying this is better in a universal sense.

Those using the different Heritage networks have great crossovers! What I made is just a variation, using known drivers (and impedances), crossover frequencies, and attenuation levels. Paul Klipsch preferred not to use resistors, but he also didn't use some of the tweeters, mid-horn types/brands, and other modifications that people have done on La Scalas, Belles, and Klipschorns.

Erik

[Erik Mandaville]

"I would be interested in what you think, using them with your single ended amps, but you may not want to switch them out to try."

Bruce:

I set that up for you. The Moondogs were being used for the first time around with this new network, no autoformer. With that initial try, things were close enough to give me a feel for the difference, but I was mainly trying to get a handle on the difference in attenuation between the autoformer and a two resistor L-pad.

Just as I found in that first case, this new network with the Moondogs is a little louder at the same channel balance setting with the Lexicon than both the autoformer 'A' set up for the usual 6kHz crossover, as well as the 'A' built with an inductor and cap change for the 4,500Hz.

The tweeter with the 18dB/octave rolloff in this last effort is much smoother sounding to me, and the improvement in imaging, as I mentioned earlier, is something that caught my attention. If you like a warmer, more rich sort of sound, what I made might not be the best choice. I'm not saying it does nothing but highlight details at the cost of the music -- not in the least. It's all there, just with a very open sort of clarity, with distinct transients on percussion, strings, piano, and so forth.

If one were to build and find the tweeter to loud, it would be easy enough to install either a fixed or variable L-pad across the driver terminals. For variable pads I've made in the past, those on tweeters were always easier to get the right blend than with midranges.

I also turned off the subwoofer to listen in two channel only. Although there is very decent bass response, I detect better control of the woofer at louder volumes with the Teac digital amps. I was listening at volumes louder than I usually do today, and the only thing about the bass I notice is that start and stop factor of attack and decay and tail-ends of notes; which in the case of the SET amps seem to be maybe a little less short or tight.

Note that only the tweeter is of a higher order, which protects it more at the lower crossover frequency, but to me also sounds 'better' than what I have been used to. If this were an entirely symmetrical network, or one comprised of third orders for also the squawker and woofer, I think the results with a lower powered single-ended amp, because of the potentially greater insertion losses, would be quite a bit different.

Erik

[Marvel]

Erik,

I am currently using DHA2 crossovers in my LS. I love how they work with my Moondogs. I think they have a higher order on the tweeter, but I really can't remember.

Bruce

[Erik Mandaville]

Bruce:

DHA2? Do you have a schematic for it?

Erik

[Erik Mandaville]

Bruce:

I searched around for that network, but the one I found had only a high pass on the tweeter. Is there another version? The Bandpass section is straightforward, and no different from my own except for where I adjusted values for both the interaction between the inductor and capacitor, as well as the lower crossover point. The approach is essentially the same: high and low crossover points set the limits in the pass band, at least as far as the initial calculations are concerned, but because of that interaction, you calculate the values for a narrower corridor passband (narrow bandpass). It's no more complicated than that. The differing element in the DHA2 that I found is associated with the swamping resistor.

I liked the fact that the DHA2 has the cap and inductor in series with one another, with the values of each associated with the same impedance in question -- and also that the autoformer/squawker combination is the termination point of the midrange section. The same is true for a resistor L-pad: It's the last in line AFTER the crossover.

Looks good! If you have a link to a third order tweeter version, that would be interesting to see. It's enjoyable to me to find examples of people experimenting with the many very workable possibilities.

Erik

[Erik Mandaville]

Bruce:

The DHA2 is virtually exactly like the first one I made. I HAD to use different values of 'C' and 'L' in the bandpass, because I had to calculate using the impedance of the squawker, not the reflected impedance in its association with the autoformer. The DHA2 uses a swamping resistor, whereas as the network I designed uses the parallel resistor of the L-pad, combined with the series resistor, to present a 16 ohm impedance.

For all intents and purposes, the following is a DHA2 (where resistors take the place of the autoformer) modified for a 4.5kHz crossover. It is all first order, including the tweeter. I'm going to optimize this one by adding a few more turns on the bandpass inductor, and slightly reducing the value of the capacitor. As good as the third order, 4kHz tweeter sounds, it may turn out that I like the first one I made more.

BTW: if you haven't tried the lower crossover point, (that is if you're planning on a pair of BEC tweeters sometime) I think it would be worth trying. I understand why you're happy with that network.

Erik

[Deang]

You could build this, it looks like it has some potential.

[DrWho]

Interesting network Dean....I wonder if it sounds better with transformers instead of resistors? [^o)]

[Deang]

I don't know. I'm kind of bummed the values for the parts aren't listed. That means a lot of trial and error.

[DrWho]

Values for parts? Heck, just pick the ones that feel right. I've got

some expensive pretty yellow ones I can send you free of charge if you

ever get around to building it. They're non-inductive too. The highs

are a bit rough, but that all clears up when you immerse them in liquid

nitrogen. It looks like we're targeting Channel 5 levels so that's

going to be very important. I've used these in robots too and you

wouldn't believe how lifelike they behaved. I just want you to

experience the same emotional impact with your music. Sometimes I just

don't feel worthy of such exquisite refinement.

[Erik Mandaville]

LOL!

Dean:

A worthwhile and constructive effort there. Thanks. As a schematic, pretty simple, too.

Who:

"Interesting network Dean....I wonder if it sounds better with transformers instead of resistors?"

The older Heritage networks didn't use "transformers." A transformer has a primary and a secondary winding. An autoformer, as it's sometimes referred to, is more of a multi-tapped choke. Thus, to answer your question, my guess is that the literal response would be that, in this case, resistor based attenuation would certainly be more appropriate than a true transformer. You're close though, since one role a transformer plays in this big game has to do with impedance matching, such as the case of an output transformer's matching the comparatively high output impedance of the final stage of a vacuum tube audio amplifier to the much lower impedance of a loudspeaker.

Erik

[DrWho]

An autoformer and transformer are the same thing...

(mutally coupled inductance)