Linkwitz-Riley is NOT Butterworth
Circlotron is right -- to get a maximally flat 4th order Butterworth filter, you would have to cascade different sections.
However, if you use a low pass 4th order Butterworth filter for an LF driver, and a high pass 4th order Butterworth filter for an HF driver, you don't get good performance when the signals are added up again acoustically, because the phase responses of the signals mess everything up near the crossover region.
The idea of the Linkwitz-Riley filter transfer characteristic, [two identical cascaded Butterworth 2nd order sections], is to improve the result of the acoustic addition of signals, at the cost of slower initial rolloff.
I don't remember if the result is perfect summing (I think it might be) but I do remember that it is a better result after summing than using 4th order Butterworth transfer characteristics for both HF anf LF.
Circlotron is right -- to get a maximally flat 4th order Butterworth filter, you would have to cascade different sections.
However, if you use a low pass 4th order Butterworth filter for an LF driver, and a high pass 4th order Butterworth filter for an HF driver, you don't get good performance when the signals are added up again acoustically, because the phase responses of the signals mess everything up near the crossover region.
The idea of the Linkwitz-Riley filter transfer characteristic, [two identical cascaded Butterworth 2nd order sections], is to improve the result of the acoustic addition of signals, at the cost of slower initial rolloff.
I don't remember if the result is perfect summing (I think it might be) but I do remember that it is a better result after summing than using 4th order Butterworth transfer characteristics for both HF anf LF.
Hi sonnya and Circlotron
If the two 2nd order sections are of the butterworth type (i.e. with a Q of 0.707) and of same cutoff frequency then it would of course make sense: This is a 4th order Linkwitz-Riley filter.
ALL L-R crossovers are 6 dB down at the crossover frequency, independant of order (2nd, 4th, 6th .....).
Regards
Charles
If the two 2nd order sections are of the butterworth type (i.e. with a Q of 0.707) and of same cutoff frequency then it would of course make sense: This is a 4th order Linkwitz-Riley filter.
ALL L-R crossovers are 6 dB down at the crossover frequency, independant of order (2nd, 4th, 6th .....).
Regards
Charles
Feedback
Emitter followers have (nearly) 100% negative feedback.
[ sorry, I just had to get that out. One of my pet peeves is reading audiophile gripes about feedback. ]
Emitter followers have (nearly) 100% negative feedback.
[ sorry, I just had to get that out. One of my pet peeves is reading audiophile gripes about feedback. ]
Yeahhh!!! okay know this starts over again...
Of course it has feedback... But compare it with opamp's with global feedback.
What i mean when talking about feedback is when this is done with a circuit with large gain and then you compensate it down to the desired gain with the feedback pair like when you look at a ZEN amplifier or amplifiers with global feedback compensating for two or more gainstages!
And of course this site is about Linkwitz Riley filter... I think thats what this thread is about!
Butterworth filter requires some calculation and the ability to add gain into the circuit. Thats why you need opamp's or circuit with gain to do the filter, but the filterpath is the same circuit as for Linkwitz Riley filter.
I hope this clears some things up a bit!
Sonny
Of course it has feedback... But compare it with opamp's with global feedback.
What i mean when talking about feedback is when this is done with a circuit with large gain and then you compensate it down to the desired gain with the feedback pair like when you look at a ZEN amplifier or amplifiers with global feedback compensating for two or more gainstages!
And of course this site is about Linkwitz Riley filter... I think thats what this thread is about!
Butterworth filter requires some calculation and the ability to add gain into the circuit. Thats why you need opamp's or circuit with gain to do the filter, but the filterpath is the same circuit as for Linkwitz Riley filter.
I hope this clears some things up a bit!
Sonny
Nelson Pass said:
Also, don't forget that it's quite hard to make a 4th order filter real good if you include component tolerances.
mirlo wrote
It is in fact the case that both output signals sum to unity independant of frequency, since the filter output signals are always in phase, also throughout the crossover region (here it would be correct to say that they are out of phase by 360 degrees !!!)
But this would assume perfect drivers without any rolloff (which in turn wouldn't need any crossover at all).
If the driver's transfer function can be taken into account, then the L-R crossover will have the advantage of a flat amplitude response, paired with good IMD and dispersion characteristics. But I don't think this will be an easy task to achieve.
But it is still by no way transient perfect.
But even ordinary active textbook 4th order L-R crossovers seem to allow decent results, I can't imagine why they are so common with P-A. systems and active studio monitors otherwise.
Regards
Charles
It is in fact the case that both output signals sum to unity independant of frequency, since the filter output signals are always in phase, also throughout the crossover region (here it would be correct to say that they are out of phase by 360 degrees !!!)
But this would assume perfect drivers without any rolloff (which in turn wouldn't need any crossover at all).
If the driver's transfer function can be taken into account, then the L-R crossover will have the advantage of a flat amplitude response, paired with good IMD and dispersion characteristics. But I don't think this will be an easy task to achieve.
But it is still by no way transient perfect.
But even ordinary active textbook 4th order L-R crossovers seem to allow decent results, I can't imagine why they are so common with P-A. systems and active studio monitors otherwise.
Regards
Charles
For those of us who do not like op-amps
Use a JFET follower. I bet even sonnya could design one without the use of a simulator [joke].
Jocko
Use a JFET follower. I bet even sonnya could design one without the use of a simulator [joke].
Jocko
Today I maked some test with high-pass section .
The speaker was PAudio BM-D750 2" compresion driver and a big horn. Amplifier used was QSC USA1300 .
Channel 1 conect to a 24dB/oct pasive bandpass filter and compresion driver.
Channel 2 inpout from a 24dB/oct active bandpass filter.
( schematics is in first thread ).
Freq. band for this test was : 3000Hz - 30Khz. injected by PC signal generator .
Test 1 : SPL 1W/1m (with Behringer Test Mic and PC analyzer ).
Pasive CH. 104 dB
Active CH 109 dB
Test 2 : SPL 100W/1m
Pasive CH. 122,5 dB
Active CH 129 dB
Test 3 : Sound quality :
Pasive CH : Poor ( Something like compresion ) , not so liniar and add some coloration.
Active CH : ( Adjusted at same SPL with pasive Channel) Very clear and linear sound. Some coloration but very small at mids-high (3.25Khz - 5,7Khz) . Very sharp at 8 - 12khz.
My first conclusion based by this short test :
Active channel is much better than passive . It sound louder and clear. But biggest diference was at small volume .
Regards !
The speaker was PAudio BM-D750 2" compresion driver and a big horn. Amplifier used was QSC USA1300 .
Channel 1 conect to a 24dB/oct pasive bandpass filter and compresion driver.
Channel 2 inpout from a 24dB/oct active bandpass filter.
( schematics is in first thread ).
Freq. band for this test was : 3000Hz - 30Khz. injected by PC signal generator .
Test 1 : SPL 1W/1m (with Behringer Test Mic and PC analyzer ).
Pasive CH. 104 dB
Active CH 109 dB
Test 2 : SPL 100W/1m
Pasive CH. 122,5 dB
Active CH 129 dB
Test 3 : Sound quality :
Pasive CH : Poor ( Something like compresion ) , not so liniar and add some coloration.
Active CH : ( Adjusted at same SPL with pasive Channel) Very clear and linear sound. Some coloration but very small at mids-high (3.25Khz - 5,7Khz) . Very sharp at 8 - 12khz.
My first conclusion based by this short test :
Active channel is much better than passive . It sound louder and clear. But biggest diference was at small volume .
Regards !
Re: For those of us who do not like op-amps
No no .. I can't!!!
I admit that i am connected to my PC and i will not do anything without consulting it (WIFE???)
I do not even remember how to calculate on them anymore
Back to reality! Not a bad idear Jocko with the JFET instead... Same stray capacitance as mosfet when using 2SK170 but a lot lower noise, but higher output impedance.... in a buffer design...
This can be a really low noise filter when using them!!!
Sonny
Jocko Homo said:
No no .. I can't!!!
I admit that i am connected to my PC and i will not do anything without consulting it (WIFE???) I do not even remember how to calculate on them anymore
Back to reality! Not a bad idear Jocko with the JFET instead... Same stray capacitance as mosfet when using 2SK170 but a lot lower noise, but higher output impedance.... in a buffer design...
This can be a really low noise filter when using them!!!
Sonny
Do we have a concensus here as to the definition of phase
coherent? I have always assumed that it meant that you
could sum the outputs of high and low pass and get a
square wave back. Any thoughts?
coherent? I have always assumed that it meant that you
could sum the outputs of high and low pass and get a
square wave back. Any thoughts?
- Status
- Not open for further replies.