Aren't 4th order LR crossovers supposed to pass squarewaves?

[Bricolo]

Hi,

Here's my problem: I'm working on an active crossover project, and have just tested the LR4 alignment.
I thought that LR4 were supposed to pass squarewaves correctly (the summed output should be a squarewave), but mine doesn't

What's wrong?

Here's the (simplified) topology.
I think it's a LR4 alignment, since Q of one stage is 1/(3-Gain) and gain=1.58. Q=0.707 (butterworth)
2 cascaded 2nd order butterworth form a LR4

http://www.diyaudio.com/forums/attachment.php?s=&postid=518858&stamp=1101235143


input signal (red), the low pass output (green) and high pass output (blue)
http://www.diyaudio.com/forums/attachment.php?s=&postid=518852&stamp=1101234956



And here's the summed output

http://www.diyaudio.com/forums/attachment.php?s=&postid=518862&stamp=1101235302


Thanks for help

 

[SY]

I thought that LR4 were supposed to pass squarewaves correctly (the summed output should be a squarewave)

I don't think that's true.

 

[pxr5]

you output will look nothing like a square wave.

 

[pxr5]

from memory i think its only summed 1st order which will. google search and some typical results will be found

 

[pxr5]

also this is 1st order INCLUDING speaker driver transfer function, generally the higher the order the less like a square wave

 

[Henrik Juhl]

Not square.

Bricolo,

Neither do I think the throughput should be square

Have you seen this article about 'transient perfect' filters?

http://www.geocities.com/kreskovs/CrossoverdocN.html

See also his main page on: http://www.geocities.com/kreskovs/John1.html

Some foods for thought and very inspiring.

 

[pxr5]

I'm am misconceived

 

[Davey]

What's wrong?

Nothing. Your simulation results look correct to me. LR4 crossovers, as well as many other, won't sum and reproduce a square wave identical to the input. This is normal.

Cheers,

Davey.

 

[Bricolo]

OK, thank you guys. I was mistaken...

 

[hifi]

bessel are suppose to pass square waves pretty well since they feature a flat groupdelay in the passband so not to delay any of the squares harmonics.

 

[Bricolo]

Bessels have a (nearly) constant group delay

but only the low pass

 

[Bill Fitzpatrick]

What is the crossover freq?

 

[ScottG]

A 1st order highpass with a 4th order LR lowpass will, (or multiples of 1st orders ending with a 4th order LR lowpass).

 

[Bricolo]

What is the crossover freq?

in the exemple, 1.6kHz

A 1st order highpass with a 4th order LR lowpass will, (or multiples of 1st orders ending with a 4th order LR lowpass).

With the same cutoff frequency? (at -3dB for 1st order and -6dB for the LR4)

 

[Bill Fitzpatrick]

Let's see what it looks like with a 100Hz square wave. You're feeding 1KHz now.

 

[Bricolo]

Re: Not square.

Bricolo,

Neither do I think the throughput should be square

Have you seen this article about 'transient perfect' filters?

http://www.geocities.com/kreskovs/CrossoverdocN.html

See also his main page on: http://www.geocities.com/kreskovs/John1.html

Some foods for thought and very inspiring.

Are thos the only transient perfect crossovers?

 

[Bill Fitzpatrick]

I'm almost 100% sure that when I was prototyping fourth order filters many years ago I was getting a square wave out of the summer.

I think the reason you are not is because the input square wave (1KHz) is too close to the crossover frequency(1.6KHz). Try 100Hz.

 

[Bricolo]

yes, it looks a little like a squarewave

 

[Pan]

I'm almost 100% sure that when I was prototyping fourth order filters many years ago I was getting a square wave out of the summer.

I think the reason you are not is because the input square wave (1KHz) is too close to the crossover frequency(1.6KHz). Try 100Hz.

Well isn´t the point to get a decent time behaviour around the Fc..?

Well above or below a low or highpass the roll off has less and less influence.

/Peter

 

[Davey]

I think some further reading is required here.

Here's a link to some information from the man that invented the Linkwitz-Riley 4th-order crossover:

http://www.linkwitzlab.com/frontiers.htm#F

The phase distortion from a 4th-order LR crossover is real. The group delay is not constant and these will not pass a square-wave undistorted. This is inherent in their design.
The big question is.....is this phase distortion audible?

Cheers,

Davey.