Phase
Hej
I used to beleive the same - intuitively at least. What mede me say that about phase linearisation is the conclusions I drew from Charlie Laubs ACD v2 tutorial (see fig 3). Just affecting the delay of the tweeter compared to the woofer seemed to have quite a substantial effect - admittedly in quite a small frequency band. Maybe my conclusion is wrong and if so, please do correct me! Just for the record, the tutorial is nothing short of brilliant since it allows the reader to play around and get a feel for what's going on and what to do about it. Thanks Charlie!
Hej
I used to beleive the same - intuitively at least. What mede me say that about phase linearisation is the conclusions I drew from Charlie Laubs ACD v2 tutorial (see fig 3). Just affecting the delay of the tweeter compared to the woofer seemed to have quite a substantial effect - admittedly in quite a small frequency band. Maybe my conclusion is wrong and if so, please do correct me! Just for the record, the tutorial is nothing short of brilliant since it allows the reader to play around and get a feel for what's going on and what to do about it. Thanks Charlie!
From a dsp virgin point of view, digital FIR filters are the advantage over analogue. With IIR the advantage is gone.
If i had an external DAC i would consider miniDSP and the like. My dvd player audio output is good however, better than my computer soundcard, so for me analogue is the logical choice. Even wav isn't as clean as the dvd audio. This soundcard is supposed to be ok, better than integrated sound, but to my ears it is not, and bass is rolled off at ~40hz...
So i guess for my situation an external DAC or DSP filters are surplus to requirement.
If i had an external DAC i would consider miniDSP and the like. My dvd player audio output is good however, better than my computer soundcard, so for me analogue is the logical choice. Even wav isn't as clean as the dvd audio. This soundcard is supposed to be ok, better than integrated sound, but to my ears it is not, and bass is rolled off at ~40hz...
So i guess for my situation an external DAC or DSP filters are surplus to requirement.
Yes ofc, delay does affect the response massively but I assumed you had already done that =)
What I meant was using FIR to correct the phase warp by the crossover slopes to get the first phase response to look like the second.
Even if you don't correct the warp as long as the delay is appropriate they will sum more or less the same on the 0 degree axis. The problem is off axis where if you have a steep slope it won't always sum nicely.
Attachments
Last edited:
Hmm, that's really interesting! I would like to read more about phase warp and its mitigation. I will have a look if there is any info on the 'net - let me know if you know any good sources of information.
There is no doubt in my mind over the superiority of dsp crossovers and analogue active. There is also no doubt in my mind about the accuracy and clarity of dsp vs analogue either. It's about accuracy. Crossovers can be set to be exact. Distortion and noise are virtually non existent and delay settings for alignment precise. Of course the analogue "equivalent" is a phase shifter not time alignment? My expererience is with the Rane 23s and Behringer cx3400 and in the past bss 366 omnidrive, "sounding" not unlike an Ultradrive at 1/10 of the price...and currently the now discontinued Alto Stagedrive +.
One caveat, in my current 3 way system, I use a RME ADi-2 dac which feeds a pair of Klark Teknic "Pultec clones" which feeds a tl audio 5013 parametric eq. In fact the KT eq is bypassed meaning they are acting as valve class A buffers into the tl audio.
This then feeds a ChiFi (sorry...superb) class A single ended balanced pre amp...which feeds the crossovers. Power amps again nothing special. Two big Inter M R500's and a 300 on top.
Why so much detail? Because some things make a bigger difference than others! RME dac for the cd transport. Vanishingly low distortion and as good as everyone says it is! For cd playback diminishing returns to buy a "better" one. The Klark Teknik programme eq's. Subtle adjustments can be made and it always sounds natural. As I said in the current system acting as eq bypass class A valve buffers. No complaints. TL Audio parametric...yes. it does have a bit of a character of its own, but it's a great analogue eq all the same and needed in system set up. Depending on the output level of the dac you can just kick the valve section in to colour the sound a tad..
Still with me? Ok so the eq is all analogue and the system set up with calibrated mics/ARTA/Omnimic 2. Which means dsp crossover is doing exactly the same job as the analogue active...no eq, just a delay between mid and treble to align and two crossover points, both LR 24dB/oct...
Repeatedly on a wide range of material the Alto Stagedrive + ("just" 20 bit/48khz) sounds cleaner and more detailed than the Rane 23s which imo, despite being no slouch. sounds thicker and less detailed. Despite achieving virtually identical responses plots the whole image is more defined and detailed through the crossover area at 800Hz with the physical time delay one can set in dsp. What was noticeable with the phase shifter on the Rane was just how hard it was to optimise! Very small changes made little difference to on axis response but the image between the two speakers seemed very sensitive to small variations in the "delay" settings.
So there we have it, using analogue eq and feeding identical signals to either a (discontinued) Alto Stagedrive + and/or a Rane 23s, and using both to perform identical tasks...3 way crossover (125/800Hz) and a delay to the midrange to align my conclusion as you have gathered that a dsp fulfilling just a crossover function with delay sounds cleaner with a very low noise floor with a detailed open sound stage.
Having drawn this conclusion I might consider an upgrade to a "better" spec dsp but given the intrnded application I would expect it to sound little different. Nothing is perfect..probably best I concentrate on being happy with what I have achieved!
One caveat, in my current 3 way system, I use a RME ADi-2 dac which feeds a pair of Klark Teknic "Pultec clones" which feeds a tl audio 5013 parametric eq. In fact the KT eq is bypassed meaning they are acting as valve class A buffers into the tl audio.
This then feeds a ChiFi (sorry...superb) class A single ended balanced pre amp...which feeds the crossovers. Power amps again nothing special. Two big Inter M R500's and a 300 on top.
Why so much detail? Because some things make a bigger difference than others! RME dac for the cd transport. Vanishingly low distortion and as good as everyone says it is! For cd playback diminishing returns to buy a "better" one. The Klark Teknik programme eq's. Subtle adjustments can be made and it always sounds natural. As I said in the current system acting as eq bypass class A valve buffers. No complaints. TL Audio parametric...yes. it does have a bit of a character of its own, but it's a great analogue eq all the same and needed in system set up. Depending on the output level of the dac you can just kick the valve section in to colour the sound a tad..
Still with me? Ok so the eq is all analogue and the system set up with calibrated mics/ARTA/Omnimic 2. Which means dsp crossover is doing exactly the same job as the analogue active...no eq, just a delay between mid and treble to align and two crossover points, both LR 24dB/oct...
Repeatedly on a wide range of material the Alto Stagedrive + ("just" 20 bit/48khz) sounds cleaner and more detailed than the Rane 23s which imo, despite being no slouch. sounds thicker and less detailed. Despite achieving virtually identical responses plots the whole image is more defined and detailed through the crossover area at 800Hz with the physical time delay one can set in dsp. What was noticeable with the phase shifter on the Rane was just how hard it was to optimise! Very small changes made little difference to on axis response but the image between the two speakers seemed very sensitive to small variations in the "delay" settings.
So there we have it, using analogue eq and feeding identical signals to either a (discontinued) Alto Stagedrive + and/or a Rane 23s, and using both to perform identical tasks...3 way crossover (125/800Hz) and a delay to the midrange to align my conclusion as you have gathered that a dsp fulfilling just a crossover function with delay sounds cleaner with a very low noise floor with a detailed open sound stage.
Having drawn this conclusion I might consider an upgrade to a "better" spec dsp but given the intrnded application I would expect it to sound little different. Nothing is perfect..probably best I concentrate on being happy with what I have achieved!
So there's sophisticated discussion of the math of filters focussed on phase issues (which people can't hear when listening to ordinary music sources).
Then folks order the components for building their passive XO. These might have 20% tolerance and so the L and R crossovers can easily end up being rather dissimilar.
No wonder nobody has - to my knowledge - posted a comparison of completed crossover systems performance here.
Not to mention a number of other ways in which the whole theory of crossovers (passive and DSP) is naive. For example, it doesn't take in account directivity indexes of the drivers. Or given the precision of this filter or that filter, that the results might be correct for one ear and wrong for the other and influenced by how the drivers are mounted. Of course with DSP, you can instantly dial-in any changes you want to make the sound better at your chair.
BTW, any measurements I've done with a Behringer DCX2496 are outstanding in all ways.
B.
Then folks order the components for building their passive XO. These might have 20% tolerance and so the L and R crossovers can easily end up being rather dissimilar.
No wonder nobody has - to my knowledge - posted a comparison of completed crossover systems performance here.
Not to mention a number of other ways in which the whole theory of crossovers (passive and DSP) is naive. For example, it doesn't take in account directivity indexes of the drivers. Or given the precision of this filter or that filter, that the results might be correct for one ear and wrong for the other and influenced by how the drivers are mounted. Of course with DSP, you can instantly dial-in any changes you want to make the sound better at your chair.
BTW, any measurements I've done with a Behringer DCX2496 are outstanding in all ways.
B.
Last edited:
Yes, some 15 years ago we supplied a loudspeaker design to a reviewer of standing (Hugh Robjohns) along with the bss 366 omnidrive (£3000) and the Ultradrive 2496.
He was, like us, hard pushed to tell any difference with a marginal subjective preference for the Behringer.
Like much of their 24/96 range they have been available for years. I have 2 x SRC and 3 x Ultracurve, albeit swapped for analogue eq (like the knobs!!) but an SRC is doing sterling work converting format for the TL Audio do-2 digital output board...
The Behringers have never let me down either. Like yourself I suspect, not interested in snobbery, just what works and sounds just fine...and the much whinged about output section isn't too bad either!
Certainly being able to set up crossovers accurately is more important than the name on the front..no difference between Behringer "0's" snd "1's" and other brands.
I would rather have totally accurate set up than subjective guesswork and this "make" sounds "better" than that make? In truth, if blind testing was still popular, most could save a shed load of money....!
He was, like us, hard pushed to tell any difference with a marginal subjective preference for the Behringer.
Like much of their 24/96 range they have been available for years. I have 2 x SRC and 3 x Ultracurve, albeit swapped for analogue eq (like the knobs!!) but an SRC is doing sterling work converting format for the TL Audio do-2 digital output board...
The Behringers have never let me down either. Like yourself I suspect, not interested in snobbery, just what works and sounds just fine...and the much whinged about output section isn't too bad either!
Certainly being able to set up crossovers accurately is more important than the name on the front..no difference between Behringer "0's" snd "1's" and other brands.
I would rather have totally accurate set up than subjective guesswork and this "make" sounds "better" than that make? In truth, if blind testing was still popular, most could save a shed load of money....!
Not sure they do..it depends on the implementation of the filter. IIR filters can replicate the analogue equivalent very closely.
The point is though if accurate set up and exact channel matching is required for optimum loudspeaker performance, then I believe dsp has it. Problem is not so much the filter type but the actual products available to
us? The Rane is considered near the top end but the crossover frequency adjustment is course at best and the levels are set with simple pots. There is no doubt in my mind that the phase shift for "delay" complex to optimise, falls short of the repeatable accuracy of digital time delay.. Perhaps if there was a market for a superior design analogue active many of these criticisms could be dealt with, but it would be expensive and the market seems to think digital has it?
The point is though if accurate set up and exact channel matching is required for optimum loudspeaker performance, then I believe dsp has it. Problem is not so much the filter type but the actual products available to
us? The Rane is considered near the top end but the crossover frequency adjustment is course at best and the levels are set with simple pots. There is no doubt in my mind that the phase shift for "delay" complex to optimise, falls short of the repeatable accuracy of digital time delay.. Perhaps if there was a market for a superior design analogue active many of these criticisms could be dealt with, but it would be expensive and the market seems to think digital has it?
I think this discussion is a bit to black/white. DSP and passive analog both have their place, and even active analog can be usefull (but need to be tailored to the system to work).
I listen to a lot of analog recordings (old vinyl) and i don't want it to go trough a digital stage and processing as that messes up the analog sound that i love. And i know how ot make good analog passive crossovers so i do it a lot.
But for high power systems, systems with little space (like car and mobile systems), pure digital systems or studio monitors, dsp is more relevant, and good dsp can be very high end on sound.
The weak point of dsp is often the AD/DA system, not the processing, and the delay that the processing makes (especially with FIR). Behringer is a perfect example of that, without external dac it's ****, with a good dac it's almost as good as very expensive "high end" systems for a fraction of the price (and good dacs are not that expensive).
I now don't have a dsp system (i had before, but sold it), but one is in the making. But most of my setups will be with passive crossovers or none (as single driver fullrange systems) as i like it more for my goal. And that is what you should keep in mind, my goal may be different than yours, so if you prefer dsp, use it. Both systems have their advantages and disadvantages and you should know them before you decide what fits you best.
I listen to a lot of analog recordings (old vinyl) and i don't want it to go trough a digital stage and processing as that messes up the analog sound that i love. And i know how ot make good analog passive crossovers so i do it a lot.
But for high power systems, systems with little space (like car and mobile systems), pure digital systems or studio monitors, dsp is more relevant, and good dsp can be very high end on sound.
The weak point of dsp is often the AD/DA system, not the processing, and the delay that the processing makes (especially with FIR). Behringer is a perfect example of that, without external dac it's ****, with a good dac it's almost as good as very expensive "high end" systems for a fraction of the price (and good dacs are not that expensive).
I now don't have a dsp system (i had before, but sold it), but one is in the making. But most of my setups will be with passive crossovers or none (as single driver fullrange systems) as i like it more for my goal. And that is what you should keep in mind, my goal may be different than yours, so if you prefer dsp, use it. Both systems have their advantages and disadvantages and you should know them before you decide what fits you best.
Amphion monitors come with passive filters, and they sound surprisingly good. When I heard them at first time, I had been so accustomed to active professional monitors with digital filters, that it was an eye opening experience. They are currently one of the most sought after monitors among professionals. I'm pretty sure that they are more popular than current digitally corrected Genelec ones despite Genelec should offer more accurate FR.
I recently compared my passive Dynaudio small monitors. One unit is the original, and another unit is modified to be used with digital filter. The tests were done in different configurations, sometimes flat FR EQ applied to both, sometimes time correction was applied to digital one. Linear FIR was also tested. I was trying to perform blind test, but after several tests, I started noticing a certain quality that analog filter offers, so it had become almost non blind test because I could tell which is which so instantly. Both are driven by nCore(s).
I don't really remember the reason why I chose this, but at the end the day, I decided to choose the original analog configuration with a little bit of EQ applied, and I'm still using them this way after a few months.
I recently compared my passive Dynaudio small monitors. One unit is the original, and another unit is modified to be used with digital filter. The tests were done in different configurations, sometimes flat FR EQ applied to both, sometimes time correction was applied to digital one. Linear FIR was also tested. I was trying to perform blind test, but after several tests, I started noticing a certain quality that analog filter offers, so it had become almost non blind test because I could tell which is which so instantly. Both are driven by nCore(s).
I don't really remember the reason why I chose this, but at the end the day, I decided to choose the original analog configuration with a little bit of EQ applied, and I'm still using them this way after a few months.
Yes it's an impressive build. I connect it just to measure group delay and finding a golden x-point and transfer the values to my other DSP (8ch)
This function is epic on dcx remote, you slide the x-point on 4 channels while listening:
Attachments
I am soon going to begin a new 3-way build using the Wondom JAB5 amp board. It is a 1 channel analog in - 4 ch analog out with DSP capability (programable with sigma studio), combined with a 4 x 100 W class D amplifiers.
I will be able to compare it to my Hypex Fusion amp setup. Hypex Fusion FA253 for a pair of 3-way speakers is about $1200 USD, while the Wondom JAB5 including power supplies, programing board, and misc cables/connectors is about $320 for a pair of 3 way speakers. I am fully expecting the Hypex to be superior, due to the super-smooth nCore amps and the more sophisticated D/A & A/D processes on the hypex. We shall see...
j.
I will be able to compare it to my Hypex Fusion amp setup. Hypex Fusion FA253 for a pair of 3-way speakers is about $1200 USD, while the Wondom JAB5 including power supplies, programing board, and misc cables/connectors is about $320 for a pair of 3 way speakers. I am fully expecting the Hypex to be superior, due to the super-smooth nCore amps and the more sophisticated D/A & A/D processes on the hypex. We shall see...
j.
I'd be very much interested in the type of amplifier chip the JAB5 design makes use of. Wondom dont want to say, for some reason. Also the diagram in that spec cleverly hides the type of connection between the ADAU1701 and the amplifiers; could be analog, could be digital (chip amps can and do have I2S input...)
Why not start a thread on the JAB5 where you can note your project and progress?