I'm comparing the design concepts between using lower order filters with an amount of EQ networks and using higher order filters without or less EQ filters.
The EQ networks mean parallel and series notch circuits, impedance equalization (zobel), and contour circuit.
In brief, I found Thiel speakers utilize lower order filters -- first order -- with a plenty of EQ networks, the brochure states there are about 40% of crossover components being the EQ networks. Compare with my speakers, Braun/ADS, they use second-order filters but without or a bit of EQ networks in some models.
So, I wonder what's the better concept, in theory and practice?
The EQ networks mean parallel and series notch circuits, impedance equalization (zobel), and contour circuit.
In brief, I found Thiel speakers utilize lower order filters -- first order -- with a plenty of EQ networks, the brochure states there are about 40% of crossover components being the EQ networks. Compare with my speakers, Braun/ADS, they use second-order filters but without or a bit of EQ networks in some models.
So, I wonder what's the better concept, in theory and practice?
The mention of the number of components is an additional complication, where does one stand on that. Does it make no difference, some difference? Is it a matter of price or performance?
Another question, are you suggesting high order filters to overwhelm peaks and dips? ..since EQ isn't completely invalidated by filter order and some would use it regardless.
Another question, are you suggesting high order filters to overwhelm peaks and dips? ..since EQ isn't completely invalidated by filter order and some would use it regardless.
Two different concepts means you can't exactly say one is better, because they are different.
Using a higher order filter without EQ may not be the best, but compared to a lower order it can reduce the problem regions a little. Is that to be preferred, I guess it is but why would someone make that decision? Perhaps it could avoid complication by using generic DSP settings instead of measuring? In that case maybe it is a little better.. but not always.
Using a higher order filter without EQ may not be the best, but compared to a lower order it can reduce the problem regions a little. Is that to be preferred, I guess it is but why would someone make that decision? Perhaps it could avoid complication by using generic DSP settings instead of measuring? In that case maybe it is a little better.. but not always.
I sort of understand your question - but it's like "cart before the horse".
Proper cross-over design is TOTALLY dependent on driver characteristics.
So, in actual fact, there is not one fixed BETTER concept.
EG. A lot of woofers only require a first order/6dB per octave LP filter - but the need for driver 'impedance correction'
can be purely up to the upper frequency response of the woofer - and then sometimes a second order is required.
I believe the 'simulation' software you have is probably leading you astray with its myriad of C/O options.
You need a Spectrum Analyzer (software) and Measurement Microphone (hardware) + time & money for experimentation.
Also, read a little bit about "Constant K filters" vs others.
Proper cross-over design is TOTALLY dependent on driver characteristics.
So, in actual fact, there is not one fixed BETTER concept.
EG. A lot of woofers only require a first order/6dB per octave LP filter - but the need for driver 'impedance correction'
can be purely up to the upper frequency response of the woofer - and then sometimes a second order is required.
I believe the 'simulation' software you have is probably leading you astray with its myriad of C/O options.
You need a Spectrum Analyzer (software) and Measurement Microphone (hardware) + time & money for experimentation.
Also, read a little bit about "Constant K filters" vs others.
It depends upon what your design objective is and what sort of trade-offs you're willing to make to achieve it.
It seems, you're looking for a simple answer here. I'm sorry, but it just doesn't work that way in speaker system design. 🙂
Dave.
A 2nd order electrical network will often produce a 3rd or 4th order acoustical result. The Q of the 2nd order filter can be adjusted and tailored to the drivers, and this is actually a form of EQ. In fact, I would say adjusting Q is the most important form of EQ and more widely used than notch filters. But it is not visible or obvious when examining a filter schematic that Q adjustment EQ is present.
I always liked Thiel speakers. 30 years ago, I thought that the reason for the good sound was the 1st order crossover, but later I began to think it was everything else. In order to get a high performance speaker using 1st order acoustical crossovers, he had to pay close attention to diffraction, and driver placement on the baffle. He had to manage impedance carefully. He had to use drivers with exceptionally wide bandwidth, and tweeters with low distortion below 2k. He focused a lot of attention on cabinet structural design.
I think it was all of these OTHER factors that made his speakers so good, not the 1st order network per se...
I think it was all of these OTHER factors that made his speakers so good, not the 1st order network per se...
Nice analysis, I like to think there is no silver bullets, end result is always combination of multiple things.
Reading your list of things considered, all of them would probably improve any loudspeaker regardless of xo slope. Attention to all (important) aspects and balancing them out into a problem free speaker, including the slopes.
Reading your list of things considered, all of them would probably improve any loudspeaker regardless of xo slope. Attention to all (important) aspects and balancing them out into a problem free speaker, including the slopes.
Last edited:
There is no easy answer to that. I would think your driver set and chosen crossover points are going to determine which approach to use. I tend to use more complex networks with imp correction zobels and notch filters when required. As an example I don't see first order crossovers working without driver zobels so I see it more of a what works best scenario.
Rob 🙂
Rob 🙂