Is this active x-over design worth while?

[klitgt]

Hi

I have seen lots of threads about active crossovers here. They all use opamps, I want to try descrete components if possible. I found the attached diagram in an old HIFI magazine and I want to have opinions on this.

it is a 12dB/octave, why go for more if the speakers can handle it.

The Cf and Rf components have to be calculated according to the formula Fc=1/2*pi*square root 2*R*C.

Any comments??

[EC8010]

Line level crossover filters are invariably based on the Sallen & Key topology that you've shown. You only get nice simple equations if the amplifier within the loop has lots of gain. the circuit you've shown should have quite a bit of gain, so it's rather like an op-amp in that respect - it just doesn't have the DC stability conferred by a differential pair input stage.

A very popular configuration is the 24dB/oct Linkwitz-Riley crossover, which is made of a cascade of two 12dB/oct Butterworth filters. The reason it's so popular is that it's insensitive to the fact that the two drivers are not coincident.

Finally, all Sallen & Key filters have increased distortion within the pass band compared to that same amplifier configured for the same gain but without filtering. In other words, you want a nice low-distortion amplifier before you wrap Sallen & Key around it.

[klitgt]

Originally posted by EC8010
Line level crossover filters are invariably based on the Sallen & Key topology that you've shown. You only get nice simple equations if the amplifier within the loop has lots of gain. the circuit you've shown should have quite a bit of gain, so it's rather like an op-amp in that respect - it just doesn't have the DC stability conferred by a differential pair input stage.
***
The DC offset is blocked by the caps at the output of both high pass and low pass sections
***
A very popular configuration is the 24dB/oct Linkwitz-Riley crossover, which is made of a cascade of two 12dB/oct Butterworth filters. The reason it's so popular is that it's insensitive to the fact that the two drivers are not coincident.
***
What does "not coincident" mean?
***
Finally, all Sallen & Key filters have increased distortion within the pass band compared to that same amplifier configured for the same gain but without filtering. In other words, you want a nice low-distortion amplifier before you wrap Sallen & Key around it.
***
??? I do not understand this statement???
***

Anyway I am happy that someone finally responded to my thread

[forr]

"it is a 12dB/octave, why go for more if the speakers can handle it."

Below the crossover frequency down to the resonance, the excursion of a hi-pass filtered loudspeaker is constant for constant input to the filter, despite the fact that electric power applied is falling.
This is a very good reason to go more than 12 dB/o.

~~~~~~ Forr

§§§

[EC8010]

Not coincident means the drivers don't occupy the same point in space.

Sallen & Key wrote the original paper on how to make filters with CR networks and amplifiers, rather than LC networks. The circuit you showed is a Sallen & Key filter. Unfortunately, it increases the distortion produced by the amplifier compared to that amplifier without Sallen & Key around it.

[Ricky]

As a professional designer of loudspeakers, I would say that at least 18dB/ Octave is preferrable. The problem that you can have with 12dB is the steepness of the slope ( or lack of it). Let's take, for example, a ribbon tweeter. These perform very well in the case of the Visaton MHT12, from around 2500 to 3KHz but, to function correctly, they need quite a steep filter slope say at least 18dB/ Octave. The 24dB Linkweitz Riley crossover is good because it gives you a nice steep slope BUT, you get the equivalent of 1st order phase response.

If it were me, I'd go for the above option.

Best of luck.

Ricky.

[phase_accurate]

The 24dB Linkweitz Riley crossover is good because it gives you a nice steep slope BUT, you get the equivalent of 1st order phase response.

Dream on !!!

Regards

Charles

[OliverD]

My personal experience is that lower order crossovers, no matter whether active or passive, analog or digital, sound better. Which is a reason to select drivers that don't need steep slopes.

Anyway, probably a matter of taste...

klitgt, did you have a look at Nelson Pass' active crossover or the (similar) MOX? They do use opamps, but discrete ones...

[darkfenriz]

Any comments??

The discrete opamps in the attachment are so primitive, that thay simply won't work correctly I think. Especially this 'open collector' operation makes little sense.
A question:
is in-built frequency compensation of opamp really neede for low-pass filter? Or does the filter compensate opamp itself?
regards

[phase_accurate]

The frequency compensation is there to achieve a stable NFB-LOOP.
In the case of a Sallen-Key filter where the op-AMP is used as a buffer full feedback (or almost full in case of gain) is applied and therefore the OP-AMP must be stable for the desired gain independant of being used for low- or high- pass.

Regards

Charles