Zobel networks

[servicetech]

I was doing a bit of research on Zobel networks and have concluded that a low pass section of a crossover shouldn't work in theory w/o one. A typical woofer has an impedance 2-4 times at the crossover frequency. A 8ohm woofer may have an impedance of 32ohms a 3Khz rendering the series inductor designed for a 8ohm circuit basically useless. Yet many premade crossovers say nothing about this, how can a premade crossover effectively provide roll off for the low pass section? Are Zobel networks critical in 12db/octave Xovers or does the "shunt" capacitor take care of the roll off?

[ThomasW]

Zobels aren't used on woofers, doing so simply isn't practical.

One doesn't see zobels on generic crossovers since a zobel is specific to the driver it's designed for.

The slope of the crossover has nothing to do with whether or not a zobel is used.

[JonMarsh]

There are two types of zobels employed for speakers-

1) LCR zobels to deal with the fundamental resonance of the driver- necessary if a driver is being crossed near that resonance whether it's a midrange, tweeter, or woofer. An example would be a three way system with midwoofer resonance near a low crossover point, such as some Wilson Audio systems use. They can be big and expensive to implement.

2) Inductive rise RC zobel, to deal with inductive impedance rise- this appears to be what you're talking about. The degree of inductive rise varies with motor design in the driver. Faraday rings help a lot, and crossing lower helps, too. With strong inductive rise, it can be much harder to hit the exact target acoustic transfer function without a zobel, and the zobel can also be trimmed to help with amplitude shaping and BSC.

[mrogowski]

Impedance compensation is usually applied at the beginning of designing the crossover, since it is easier to work with a constant impedance value than an ever-changing one. When the crossover is built to satisfaction, the compensation network is usually removed by making minor adjustments to other components in the network that will yield overall impedance that it satisfactory to the design.

Best,
Mark

[Mark Seaton]

There are two types of zobels employed for speakers-

1) LCR zobels to deal with the fundamental resonance of the driver- necessary if a driver is being crossed near that resonance whether it's a midrange, tweeter, or woofer. An example would be a three way system with midwoofer resonance near a low crossover point, such as some Wilson Audio systems use. They can be big and expensive to implement.

2) Inductive rise RC zobel, to deal with inductive impedance rise- this appears to be what you're talking about. The degree of inductive rise varies with motor design in the driver. Faraday rings help a lot, and crossing lower helps, too. With strong inductive rise, it can be much harder to hit the exact target acoustic transfer function without a zobel, and the zobel can also be trimmed to help with amplitude shaping and BSC.

Jon of course covered things nicely here. In my own course of learning about this back in college, I used it as the topic of a few lab projects and did learn a few interesting things. For clarity, and based on the history I've seen, I usually only refer to the inductive compensation using a parallel R-C as "a Zobel network." When talking about compensating for the resonant related impedance peaks with a conjugate network, I either refer to it directly as full impedance compensation or as a conjugate network.

Regardless the terminology, both have their uses. A crossover definitely does not require a zobel or impedance compensation to function well. You will probably need some form of either if you are trying to get textbook filters to respond as you calculated them to assuming fixed, resistive loads.
IMO, the textbook filters are great for setting targets, but you will be better served to have a real impedance and magnitude measurement with a program that simulates the crossover.

The first speaker I built in college was a basic, Seas, 2 way bookshelf. The crossover, which served as the topic of a few lab experiments, included a zobel on the woofer, and a full impedance compensation network on the tweeter. The tweeter + conjugate network measured as a 5.2 Ohm load +/- 0.2 Ohms, or so the HP analyzer said...

Fast forward to the "audiophile party system" that I whipped up for an old college friend Brendan (posted on my own forum)... The main speaker used an Eminence Beta12CX + Selenium compression driver. While I will probably tweak the final response some later, the crossover topology won't likely change. The full crossover using this compression driver and 12" woofer was done with 5 components total. h:

I have been working more and more with full range/main LCR designs, and in the case commercial products, it is a trade of money invested in the crossover vs. the drivers. A crossover needs to be as complex as required for the intended design, but more so doesn't necessarily make it better. If we step back from the idea that a the individual drivers have to have a seemingly ideal response sans XO, many more options open up, with some having some surprising performance benefits. In the end it is the end result, where the whole can be superior to a simple sum of the parts.

The point? Focus on making sure the results deliver what is needed, and don't get too caught up in what others think is the "right way" to get there.

Cheers,

[Dennis H]

Yet many premade crossovers say nothing about this

And that's one of the reasons why premade or homemade 'textbook' crossovers seldom work well. As Mark said, you may or may not need Zobels for a particular driver but what you do need is to measure the drivers' SPL and impedance curves and plug them into crossover design software rather than relying on textbook formulas. There's free software available so there's no reason not to do that.

[servicetech]

Where can I get the free software?

[ThomasW]

http://www.htguide.com/forum/showthread.php4?t=26544