speakers with a "smooth" impedance curve

If I can clarify something here-

If the speaker was voiced to work with transistors, the smoother impedance curve is going to be a real boon if you are trying to use tubes.

If the speaker is designed for tubes, the impedance curve likely will not matter too much. So both Duke and Trelja are correct as this is an issue that is very use-specific.

for more information see:
http://www.atma-sphere.com/papers/paradigm_paper2.html

Daedalus, what you describe is what I do in my speakers, instead of using an outboard impedance equalization filter. And I would estimate that designing for a smooth impedance curve as well as a smooth frequency response curve roughly quadruples the workload it takes to get the crossover right. But then as you said smooth is smooth, and amps like that.

And by the way, your crossovers are particularly well done; I can tell that attention was paid to the power response, not just the on-axis response.

Thanks for laying things out so well, Duke. I will fire off an e-mail in the near term. Anyway, we're way past due on sharing an e-mail or three.

Lou, I think your statement about being able to design a smooth impedance into the loudspeaker is a very good point, and fits into what I've been espousing perfectly. No crossover parts needed to accomplish it, simply good, well-thought out design. That makes the speaker a friendly partner for the type of low-moderately powered transformer coupled tube amplifier I happen to prefer.

I will say that from afar, your loudspeaker designs inherently make that easier than most on the market these days. I think ala your using more drivers. So, you are taking advantage of handing off to the next driver up the ladder before the rising impedance due to voice coil inductance rears its head.

As most loudspeakers today use something like a 6.5" midrange/midwoofer handing off to a 1" dome tweeter somewhere in the 2000 - 2500 Hz region, that impedance rise can become a problem.

The cookbook solution is a Zobel network, which flattens the rise, but I feel they do far more harm than good in the ways I've previously mentioned. Again, I accept I might not be in the majority here, as a buddy of mine and I have the EXACT same loudspeakers, which were an outgrowth of the pinnacle of Bud Fried's lifework (Valhalla System, though with what we've been able to move forward with in terms of crossover design), even to the point of independently arriving at more or less the same Zeta value of somewhere in the 1.0 - 1.1 (Bud's commercial designs, as well as his personal pair which I own come in around 0.6 - 0.7, personally, I find that value a bit too forward) range. The only difference between our speakers is the fact that he feels the Zobel makes the speaker, and I feel it breaks it. As I like to say, that's why they make vanilla AND chocolate.

This impedance rise is what I blame for so many of today's systems sounding overly forward and bright, as there is simply too much energy in the presence region. Especially, when driven by a tube amplifier, which is putting its power better into these higher impedances, and manifests itself in terms of the hard, glassy presentation that people tear their hair out trying to ameliorate with room treatments, cabling, and trying to put even more tube equipment (preamplifier, CD player, tube buffer, etc.)into their system.

I find a lot of the more creative and forward thinking designers these days are shifting the crossover point upward, in the effort to stay far away from the resonance frequency of the tweeter, as they feel that is the cause of the brightness. I would suggest the better way to go is do what is required to move downward to avoid the impedance rise of the midrange/midwoofer driver altogether.

Trelja,

I tend to agree with your point regarding excess energy in the presence region - sometimes it seems to me that it's almost ubiquitous in high end loudspeaker designs. But... I've never particularly associated this with models that cross to the tweeter at any particular frequency. It's always seemed more universal.

I'm not familiar with some of the technical issues at work here and therefore don't follow your point re: impedence rise vs. crossover frequency. Is this rise associated with the crossover network itself or do you refer to some other phenomenon?

Thanks in advance.

Marty