High order crossovers

Higher order loudspeaker crossovers ABSOLUTELY and EMPHATICALLY rob dynamics!!!

Insert a single capacitor in series with a driver, and the amount of sonic degradation will more than disappoint you. Multiply that by the complexity of a crossover, and the impact manifests itself accordingly.

Although crossover complexity normally correlates with slope, it does not absolutely.

Additional components making up things like Zobel networks, notch filters, L Pads, etc. must be considered, and why many (like me) believe they create more / larger problems than they solve. Think about not just crossover slope, but overall parts count, as well. Each part acts like a speed bump.

Thiels employ first order crossovers, yet with all of the added circuitry, represent some of the most complex crossovers, with the highest parts counts around. THAT’S why they’re so difficult to drive, demanding of the partnering amplifier, and simply will not come alive unless fed correctly.

Having owned and been around the entire JSE / Infinite Slope lineup, their obvious and fatal flaw is the very dynamics of the loudspeakers. Though they possessed many qualities such as notable timbre and ease of listening, they simply lack the ability to stop and start / react, ability to startle, and overall excitement IOW...dynamics.


Perhaps the biggest reason the Frieds I was a part of had such fantastic dynamics and the overall ability to excite in a most beguiling manner was Bud sorting all of this out so long ago. Frieds used simple first order crossovers, with a minimum of components. Most who remember know he employed the series crossovers he learned from Dynaudio (even if they themselves don’t), which not only present a more natural and coherent presentation, and far easier load on the partnering amplifier, but the ability to implement a more gentle or steep (as desired) slope IN A FIRST ORDER(!) crossover; that’s impossible in the parallel networks used in 99.9% of the market. Bud’s raison d’etre was the resistive, not reactive, loudspeaker, all in the attempt to recreate his life changing experience of hearing Stokowski in Philadelphia as a child. In fact, Ralph Karsten told me he developed his Atma Sphere amplifier when he owned a pair of Frieds (or IMFs?) back in the day, which even today present an astoundingly wonder match with his amplifiers

@audiokinesis, "Sorry but I must disagree that there is a correlation between parts count and how difficult a speaker is to drive."

I don’t see myself at RMAF this year, but share your warm sentiments, Duke. It’s been too long since we’ve gotten together, and would love to see you and your wife again. Hope you’re both doing well!

I’m thankful this topic has opened up an interesting discussion. However, disagreement obviously does not mean disproven. Could you shed a bit more insight into how you landed on your assertion, and potentially provide some specifics, please? Obviously, from my time at Fried and your designs, we’ve arrived at different places.

@prof, "My Thiel 2.7s are in fact notable for their dynamic sense of liveliness. It’s one of their most salient features, and having auditioned a great many other speakers recently, they remain among the most lively and dynamic I’ve heard. And I’m driving them with 140W/side (CJ tubes) so it’s not like they require some powerhouse amps to come alive."

Two things, Rich. First, if you don’t consider 140 wpc of tube power A LOT, we are about apart as two people can get. I’ve had a lot of tube amplifiers in my system, and once they rise above 60 wpc coming from at least 4 output tubes per channel, I’ve most definitely reached the serious power and slam territory. I should clarify that with several of the loudspeakers I used which implement first order crossovers, I produce the sort of dynamics that has people just about jump out of their skin with 10 - 35 wpc tube amplifiers.

Secondly, although I have the highest level of respect for Jim Thiel and his designs, few have ever been as famously unsympathetic towards amplifier designers. Along those lines, I have nothing untoward that position, and in fact, respect his conviction. Your Thiels can most assuredly produce dynamics, but require amplification that can also just about serve welding needs to do so. A simplified crossover would change that, without question. Again, I do not question Jim Thiel. His design choices produced exceptional loudspeakers with may notable qualities, and the success of his venture testifies to that. Just that ease of drive was never one of them.

Thank you for providing the reviews for the JA loudspeakers. Note the comments from Atkinson come from partnering the speaker with MBL and Pass Labs amplifiers, again something different than what I consider the demands of a first order crossover.

Finally, as an actual example to illustrate my point of the impact the slope crossover imparts on the loudspeaker’s friendliness to an amplifier, I will bring up the PSB Gold i and the Vandersteen 2. Both speakers come close in their drive complement and stated sensitivity, and differ in the choices implemented by their respective and very talented, very successful desigers. Richard Vandersteen staked his claim on first order crossovers, Paul Barton employs steeper slopes. I do not imply it’s an apples to apples first order vs higher order crossover comparison, but I definitely think we have two end products that we can put in a drag race for our discussion here. The Vandersteen 2 attained as much popularity as any loudspeaker in the past generation, partly because they present a friendly load to an incredibly wide swath of amplifiers and so implicitly have a huge potential customer base. The Gold i, while also selling into the thousands present a surprisingly tough load. When I sold them would only come alive with the brute force Carver Lightstar amplifiers, though would come close to knocking walls down with them. That’s dynamics, but again, with huge demands. Not even the big Adcom monoblocks could wake them from what I considered slumber. As the HEA market pivoted towards tube amplifiers over the past 20 years, the ubiquity of the Gold i collapsed.

Again, crossover slope absolutely rob dynamics. Substantially so

@audiokinesis, "@trelja, can you tell me what the actual mechanisms are by which crossover parts count and/or conventional crossover topologies constrain dynamics? There may be effects I’m unaware of that are worth taking into account, even if they aren’t necessarily the primary cause of compression. You may be correct that a steep crossover slope constrains dynamics, but can you tell me why? The "why’s" interest me a great deal."

Thank you, Duke. The "why’s" interest me a great deal, as well! That’s where the fun and advancement come from. I believe the issue under consideration is actually quite a simple thing. To make the point let’s go back to Electronics 101.

Resistors represent a fundamental component in the toolbox. Resistors reduce voltage, reduce gain, reduce dynamics. We use that to our advantage. When a mother, wife, roommate, dorm captain, guy sitting next to us on the bus, police officer, etc. tell us to turn the sound down, we make things quieter by adding more resistance to the respective audio signal, and (hopefully) restore the peace. We reduced the volume / dynamics from an objectionable level to something others can accept. That’s obviously what the volume control on a preamplifier does. We also use resistance to reduce the voltage through the succeeding stages of our power supplies to suit the needs and limitations of the downstream components. Likewise, as tweeters traditionally play louder than woofers, we use resistance to lower the tweeter’s voltage which translates to volume and dynamics to some level closer to the woofer in order to balance the sound between the two drivers out, and produce an overall better result. Resistors, no matter how small cannot and do not add voltage or volume or dynamics, and they cannot and do not leave the voltage or volume or dynamics unchanged. Resistors, no matter how small, reduce the voltage, reduce volume, reduce dynamics. Add resistors, no matter how small, to loudspeaker crossovers, and we reduce volume / dynamics of that loudspeaker.

Inductors are another fundamental electronic component, and typically a part of the crossovers discussed in this thread. They block part of the musical signal from getting to a driver, intentionally so, as we cross over from one driver to another. Every piece of wire, regardless of its elemental composition or length, has resistance. Increase the length of wire and we increase its resistance. Longer wires have measurable and meaningful levels of resistance. Inductors are made from a long (can be 50’, 100’, or even 300’) piece of wire, coiled upon itself. In fact, one specification of an inductor will be its resistance. In other words, on some level since an inductor consists of a long piece of wire, an inductor is a resistor. Again, resistors reduce dynamics. Use an inductor, and you’ve increased resistance and reduced dynamics. Moving from the typical 16 or 18AWG to an 8 or 12AWG often brings the comment of the obvious increase in dynamics and slam. Why? Because its resistance has decreased. Not to the zero not having the inductor at all would represent, but it shows adding components adds resistance which reduces volume and dynamics, and vice versa.

Capacitors are another fundamental electronic component, and like inductors, usually a part of crossovers discussed in this thread, and also block part of the musical signal from getting to a driver, intentionally so, as we cross over from one driver to another. Capacitors also have resistance, commonly noted as ESR (equivalent series resistance). No need to repeat the same statements as inductors, other than to say add a capacitor and we’ve added resistance which reduces volume and dynamics.

The number and position of inductors and capacitors in the crossover leg in the typical parallel network determines its slope. Second order crossovers have an additional inductor and capacitor from a first order crossover. Third order crossovers have another inductor and capacitor from a second order crossover, and so on. Like for like, fourth order crossovers have more inductors and capacitors than third order crossovers, which have more inductors and capacitors than second order crossovers, which have more inductors and capacitors than first order crossovers. Beyond that, there are compensation networks, made up of additional resistors, inductors, and capacitors a loudspeaker designer may add to the basic layout to suit their design goals. These will also to varying degrees, add resistance to the overall crossover. Additional components work in an additive, in terms of resistance, manner. As previously stated multiple times, adding resistance reduces volume and reduces dynamics.

Again, crossover slope absolutely rob dynamics, substantially so

@audiokinesis "So it is not obvious to me that series resistance reduces the voltage ratios that a driver sees, and therefore I do not see how it would reduce the dynamic contrast. The volume level being reduced is not the same thing as the dynamic contrast being reduced."

@erik_squires "Resistors are the most linear parts in the entire audio chain, except wire."

While I’m trying to keep my posts as brief as possible, I’m glad the topic of resistors came up. I also used to think like the points raised. As I built speakers, I spent an incredible amount of time on crossovers, listening and auditioning, tweaking values and configurations, then doing it all over again. Moving up in complexity did impact dynamics, to the point of losing the ideal match between amplifier and loudspeaker because of it. I also noticed the damage resistors did to the sound, and it left me perplexed. Even the best resistors I found sounded surprisingly and disappointingly bad (though not nearly as bad as capacitors) in comparison to nothing. Later, I began building amplifiers, and noticed how even more profound the effect of resistors really are.

Along the way, I was fortunate to work as a material scientist / R & D chemist during the time when electronic materials needed reformulation because of the environmental legislation (CE, ROHS, etc) requiring the elimination of materials such as mercury, lead, cadmium, etc. Among our many products, my company provided the resistor materials to CTS, Dale, General Motors, Ohmite, TRW, Vishay, and others that manufacture the resistors such as get used in HEA, although our audio world is like a bucket of water in the ocean. It proved the most exciting time in the industry since the 1960s when the original formulas were developed. I was incredibly lucky to be at the right place at the right time to do such work in going back and essentially getting to start from a clean sheet of paper, and to learn how such technologies impacted our hobby.

Knowing what’s actually inside quickly proves thinking those resistors coming anywhere close to the copper and silver wire we employ in our cabling and internal wire as simply and fundamentally wrong. It opened my eyes to why resistors sound so poor, with some sounding much worse than others.

One of the fundamental characteristics of resistors is temperature coefficient of resistance (TCR), which you can find in most resistor product / sales literature. TCR quantifies how much the value of the resistor changes in relation to temperature, down to -55C and up to 125C. As most of the world (especially, the military) considers it imperative a resistor not change value in relation to temperature, a neutral (0 PPM - parts per million) TCR is the goal of most companies producing resistors. And in fact, most modern (unlike the old carbon composition) resistors do exceedingly well with that, carrying a TCR of +/- 50PPM or even +/- 25PPM. Not getting too technical, except to say that means the resistance in relation to temperature remains quite stable. On the other hand, we know the resistance of metals typically behave quite the opposite of that, as resistance rises (substantially, in fact) with temperature, which we refer to as positive TCR. For example, the resistance increase of copper is 0.393% for every increase in degree Celcius, and vice versa. As I said, it’s substantial. That’s exactly what Duke described in terms of thermal compression of a driver, which definitely trumps crossover design and implementation - when it comes to dynamics and everything else. At some point, the resistance of the voice coil due to its temperature rise precludes it from behaving as it did previous to that in handling the signal passed to it.

How do we produce resistors with the impressively neutral TCR specifications I mentioned? The formulations, logically, include materials that have substantially negative TCR profiles to balance / shift things from positive toward the desired 0 TCR. What sort of materials possess negative TCRs? Semiconductors (materials that only conduct electricity under certain conditions) or non-conductors / insulators such as ceramics (bismuth trioxide most famously, but also compounds like titanium dioxide, barium titanate, boron triphosphate, calcium oxide, and on and on and on) and infinitely (literally) varied glass formulations. Although these materials may not conduct electricity themselves, they can exert a strong enough influence on the resistor as a whole to bring the TCR down to the desired level. Glass provides the fusing material for the resistor to attach (literally melts and fuses) to the substrate as it goes through the especially violent sintering process in a high (1100F - 1600F) temperature furnace, as well as the means of forming the conductive (quite far away from what "wire" looks like) matrix of the metallic conductive components, as well as often (and, hopefully) benefiting the TCR situation. In fact, as the old timers used to say, "It ALL comes down to the glass." The main driver for formulating new products was the elimination from lead and cadmium from the glass. Lead oxide, which as in other fields such as leaded crystal and decorative glass, yields very low temperature melting, wonderful performing glass, and provided the extremely high-performing resistors that came out of the 60s. The presumption of most around the world was the new "green" materials would never come close to the ones they were replacing due to the removal of lead and cadmium. In fact, most of our competitors green products looked downright scary. Through a two-pronged approach, but mostly from the grace of God, I was fortunate to actually end up with resistors that were superior to the old products, to the astonishment and disbelief of my boss who developed our products in the 60s. Our testing showed them to be the finest, highest performing low resistivity (0.01 Ohm/square - 10 Ohm/square) resistor material in the world.

Wirewound resistors fare no better, as instead of using the sort of wire materials we might expect in HEA, they consist of alloys of base metals such as nickel and chromium, again with negative TCR compounds as part of their proprietary formulations to produce the desired neutral TCR specification. I mostly avoid the religious wars on cable, fuses, etc., but my experience makes me to allow for the possibility of yielding positive results by employing much improved conductive materials. And instead of a 0.5" or 1" wire in a fuse, consider the very, very many feet that makes up a wirewound resistor.

As one can imagine, reducing the TCR of such heavily positive materials down to 0 or 25, 50, 100, or even 500PPM, typically requires substantial amounts of these negative TCR materials. To anyone who has chased audio cable with increasingly pure levels of copper or silver, say from 99.9% to 99.9999% through the reduction of compounds (what we call impurities) like oxygen in the quest for improved sonic performance, this admittedly must come as more than a shock. Conductive content as part of the overall (conductors, sintering aids, glass, additives, dopants) end product could even drop as low as 5 or 10%, though it’s typically far higher in order to produce a reliable conductive matrix. Products get sold as a series, normally each product representing a decade of resistance 10X higher or lower than the next in the lineup. More conductive products contain more of the conductive material. Less conductive (more resistive) products contain less, up to a point, that is. At some point, a lack of conductive material no longer supports a conductive matrix, and one moves up to a different series, featuring a higher resistance element as its conductor.

It may sound less pristine and "right" than we imagined, yet while still leaving out a lot of the details, that’s how resistors get made. Look at electron micrographs of the end result, and some will leave you to wonder how any HEA component using them could ever allow the purity and beauty of the music to pass through without looking like a car wreck.

Does anyone still think resistors are close to "wire"? And is it now even just a little easier to think the pinball trip through these resistors could possibly negatively impact dynamics, or by multiplying their number or complexity of the circuits that employ them could result in same?