Shunt only parallel network

[Hezz]

Has anyone ever tried to make a shunt only parallel crossover with maybe a four way system.

I used to have an Audiophile friend named John who worked at a local shop that I bought equipment from. He had a speaker designing friend named Terry who had worked at Wilson audio. I believe this is the same Terry that is now building the Escalente design speakers. Which have gotten the best review I have heard from a speaker.

Anyway, John used to say that Terry's best sounding set of speakers used a 3db crossover network. I wonder if this was accomplished by only using passive parts in shunt and keeping all parts out of the primary signal path of each driver.

Could this be done in say a four way with very smooth wide ranging drivers with low Fs and high power handling.

Say you had a 1 inch dome tweeter and 2 inch dome midrange, a six inch mid/bass and a woofer. You have the three upper drivers mounted very close together in a top cabinet and the woofer in a lower cabinet mounted in close proximity to the other drivers.

[kelticwizard]

Originally posted by Hezz
Has anyone ever tried to make a shunt only parallel crossover with maybe a four way system.

I think you probably could. For a long time, I have been wondering why the audio industry abandoned the old idea from the fifties and sixties of providing woofers with a smooth rolloff in the midrange-with no components at all.

People on this forum have mentioned seening older woofers with some putty like substance on the voice coil near the spider, which serves to smooth out the woofer's midrange response and provide a natural downward slope. AR used this approach with much success for years. There are others.

To my knowldege, this method introduced no phase shifts like a series inductor would. Yet it is rarely seen now, even though the audio world seems to be more concerned about phase shifts versus crossover area width than ever.

If one wanted to get a higher slope rolloff, you could put in a shunt capacitor. Yet you rarely see such a woofer or natural crossover today.

Below is the frequency response for an AR 17 woofer with no components. It ws tested in a smaller than optimum box, so forget the hump in the bass range. Instead, look at the natural rolloff with no additional components. The only irregularity occurs many decibels down below the midpoint, and will be inaudible.

[kelticwizard]

As for a midrange or even tweeter to mate it with, I often wonder why people don't try a mid or tweeter with an extremely low Qts, which will have a natural rolloff well above it's resonance. i thinkit's a 6 dB rolloff, and a parallel component could take it up to 12 dB with no series components.

[sreten]

Hmmm.......

Shunt only parallel crossover = one very dead amplifier.

Shunt only series crossover is a possibility.

/sreten.

[Hezz]

I guess it would be difficult in a parallel topology to keep the combined impeadances high enough to be safe. But going to a series network kind of nullifies the benefit of having nothing in the signal path for the drivers.

But the question remains. Do capacitors do more damage to the signal quality than the inductance and back EMF of the other drivers. I know that there are good sounding series crossover designs and some swear by them.

[kelticwizard]

Originally posted by sreten
Hmmm.......

Shunt only parallel crossover = one very dead amplifier.

/sreten.

That would certainly be true if the parallel element is a single component-either choke or cap.

I wonder, however, if instead of a single component the parallel element is a network. Perhaps, when combined with the impedance characteristic of the driver, the total speaker impedance can be kept in a safe range.

Yes, I am aware that a series LC network has a frequency where R = 0 ohms, so a simple LC network would be out as the parallel crossover element. But perhaps a more complex network?

[sreten]

Originally posted by Hezz
But going to a series network kind of nullifies the benefit of having nothing in the signal path for the drivers.

Hmm..... theorectically not true, by definition with only shunt components
the sum of voltages across the drivers always sum to the input voltage.

Much touted but largely illusory benefit though....

/sreten.

[sreten]

Originally posted by kelticwizard
But perhaps a more complex network?

Not really. for a parallel network with shunt only components all
you can affect is the input impedance, not the drivers responses.

/sreten.

[Hezz]

Not really. for a parallel network with shunt only components all
you can affect is the input impedance, not the drivers responses.

If this is the case, in theory it could be done. But the shunt components would not be used to shape speaker response only input impeadence. The drivers themselves would have to have the proper frequency, sensitivity and roll off characteristics.

[sreten]

Originally posted by Hezz


If this is the case, in theory it could be done.

What could be done that is remotely sensible ?

/sreten.