TO get the full potential of an active system!!! thoriated said:
You will find there are many users(*) who are very happy with this unit, most them use it as a permanent set-up, I am one of them.
Go to Yahoo, there is DCX2496 user forum there.
FYI some passive x-over components alone for a single pair of speakers costs much more than the DCX. eg. $600 vs $250.
One big problem in passive systems are tolerances in driver manufacturing. A pair of tweeters or midranges will hardly exhibit exactly the same resonant frequency. Sometimes mismatches may be quite large (20% or more, particularly in low cost drivers). Furthermore, Fs will drift with aging but not by the same amount in both channels, probably resulting in increased mismatch.
These tolerances and drifts have little impact on the raw driver frequency and phase responses, thus they can do very little to hurt the performance of an active system, but they result in impedance plots that are quite dissimilar between units (the bumps move!) Since the frequency and phase response of passive crossovers are strongly dependent on driver impedance, any mismatches result in poor imaging.
The performance of an active system is always more repeatable and stable with time.
These tolerances and drifts have little impact on the raw driver frequency and phase responses, thus they can do very little to hurt the performance of an active system, but they result in impedance plots that are quite dissimilar between units (the bumps move!) Since the frequency and phase response of passive crossovers are strongly dependent on driver impedance, any mismatches result in poor imaging.
The performance of an active system is always more repeatable and stable with time.
Like I said before, if you are designing the crossovers correctly, that will all be accounted for. When designing a passive crossover you don't just pull some random readings from a single driver and say, ok this will do. Especially for a Diyer, you are optimizing for the drivers you have. however any decent speaker company is going to be using stringent QC methods in testing the drivers to ensure that whatever tolerance issues they have are small enough to have a minimal effect on the frequency response. And again, as I said before, if a tweeter has a 20% tolerance problem, its going to effect the active system too.
You can't look at Passive crossovers at their worst and active crossovers at their best, thats not a fair comparison.
You can't look at Passive crossovers at their worst and active crossovers at their best, thats not a fair comparison.
I'm glad you're happy with it. The electronics in my audio system, however, are much more transparent than this unit, and introducing it would seriously degrade several important areas of audio reproduction such as detail, low level linearity and imaging. Deterioration of this magnitude is nothing that I have to be concerned about with well chosen passive crossover elements.
For home audio I design my own speakers and amplifiers, and I'm not concerned about the unit-to-unit variability of cheap drivers, and I can arrange my passive xovers to have less than 0.5 db of loss worst case and tailor them to my speakers as much as I want to. I don't need an active xover to biamp, either. The nature of my speakers allows for easy time alignment from close to 100 hz on up. Yes, I could design or use active xovers, but there is just going to be more coloration (and probably outright deterioration) added by their relatively complex circuitry (and A/D/A conversions and dgital domain word truncations, if applicable), than from a dozen or fewer passive components of good quality.
Btw, for a living, I design electronics that include 24bit/96 ks/s codecs and DSPs for high quality audio/video applications.
The DEQ isn't actually all that transparent, IMO, either. One of the engineers from Cabasse has talked about this with their own digital speaker management adventures, and basically, it still remains a system of tradeoffs. You are always making tradeoffs with whatever path you choose. In the case of a digital speaker management controller, you must convert the signal from analogue to digital first in many, but not all, cases. That first conversion has the potential to degrade sound with digital errors. It adds a point where noise can be added to the signal, interference picked up, distortion added, etc. Then you have to process the signal, and while generally able to happen in a fairly clean manner, you still have the unfortunate problem of throwing away bits, which causes you to lose transparency, dynamic range, and again, add distortion. At this stage of the game, it doesn't matter if you feed the system of digital signal or not. Then you have to convert the signal to analogue using a D/A converter. At this point again you have the chance for further errors in the conversion process, added noise, added distortion, etc etc. In the case of an analogue low level crossover, again, you run the risk of adding noise and distortion within the complex circuit, and much in the same way that a complicated passive crossover must have a lot of parts in the signal path, so must a very flexible active analogue filter. Then the biggest drawback here is that its very difficult to implement the level of flexibility that DSP can have, and so they are never as capable, and often more expensive.
Then you have passive crossovers. Passive crossovers, because they are acting on a high level signal, will not add any noise, and distortion is negligable. However, to properly design them requires a strong working knowledge of filter theory, good modeling software, measurement capability, and careful planning. As I said before, I fully believe this is true of any crossover, but a lot of people feel they can implement with better results something like a Dcx2496 than with a passive crossover (I don't agree with this). Passive crossovers have the disadvantage of needing to handle a lot of power, and thus the parts are larger and more expensive than the parts used in an active crossover. Suddenly esr of the parts starts to matter quite a bit, and so implementing textbook electric 4th order or higher crossovers isn't very sensible. For instance a 6th order low pass would have 3 inductors in the signal path of pretty considerable size, and if it was being used down below 300hz, they would not only be very large, but have to dissipate a lot of power. Often this means more creativity is needed to implement such steep slopes. In home audio this is all less of an issue since the parts would rarely be dissipating more than a handful of watts, and even during peak transients, we are still talking less than a few 100 watts. However in pro audio environments, its not shocking that active quickly became the preferred choice. From a reliability stand point alone it was much better. I remember seeing video's from 60's concerts such as woodstock where the passive crossovers are mounted externally, and during setup you often saw a stage hand soldering in a new part to replace a bad one. You can imagine that got old rather quick.
However, from a capabilities standpoint, anything the DCX or even DEQ can do can be accomplished with passive crossover networks as well. Delay, phase shift (which is delay), any crossover slope, eq, etc. Neither of those units can do transient perfect, and as I have said before, I stand by that even units capable of mathematically transient perfect can not achieve electrical or acoustical transient perfect transfer functions (DSP could give an electrically transient perfect slope, but the speakers own electrical phase shift would disrupt that).
Then you have passive crossovers. Passive crossovers, because they are acting on a high level signal, will not add any noise, and distortion is negligable. However, to properly design them requires a strong working knowledge of filter theory, good modeling software, measurement capability, and careful planning. As I said before, I fully believe this is true of any crossover, but a lot of people feel they can implement with better results something like a Dcx2496 than with a passive crossover (I don't agree with this). Passive crossovers have the disadvantage of needing to handle a lot of power, and thus the parts are larger and more expensive than the parts used in an active crossover. Suddenly esr of the parts starts to matter quite a bit, and so implementing textbook electric 4th order or higher crossovers isn't very sensible. For instance a 6th order low pass would have 3 inductors in the signal path of pretty considerable size, and if it was being used down below 300hz, they would not only be very large, but have to dissipate a lot of power. Often this means more creativity is needed to implement such steep slopes. In home audio this is all less of an issue since the parts would rarely be dissipating more than a handful of watts, and even during peak transients, we are still talking less than a few 100 watts. However in pro audio environments, its not shocking that active quickly became the preferred choice. From a reliability stand point alone it was much better. I remember seeing video's from 60's concerts such as woodstock where the passive crossovers are mounted externally, and during setup you often saw a stage hand soldering in a new part to replace a bad one. You can imagine that got old rather quick.
However, from a capabilities standpoint, anything the DCX or even DEQ can do can be accomplished with passive crossover networks as well. Delay, phase shift (which is delay), any crossover slope, eq, etc. Neither of those units can do transient perfect, and as I have said before, I stand by that even units capable of mathematically transient perfect can not achieve electrical or acoustical transient perfect transfer functions (DSP could give an electrically transient perfect slope, but the speakers own electrical phase shift would disrupt that).
I understand what you are saying SY, but it can be done. I use a Ladder Delay network for delay, and I took the idea from Zaph. I know its not his, but I had no experience with them until reading his ZD5 article. However they do work fine, and it is a way of doing delay in a passive crossover.
I do think there is a HUGE difference between a properly implemented digital crossover and what a lot of people end up having to do. It's a mixture of design issues, as I've talked about, and how the end user implements it. Any time I have seen a DCX2496 used in a system demo, the volume has been set so high that the bit dumping issue wouldn't arise. I probably shouldn't get into naming names here, but at a certain electronics show, a certain vendor was showing off their new speaker with digital active crossover (which I think was a behrenger). Sounded great, everyone liked it, but as the show went on, I went back and asked if I could turn the volume down to a more reasonable level. As soon as I did that the tell tail distortions, loss of transparency, etc of the bit dumping issue arose. The manufacturer quickly ran up and turned it back up, and told me that it was only optimized for that higher volume range. Ha, yeah, because these units aren't designed to be used that way, its not about how he optimized it.
having used and fine tuned the Behrenger DCX2496 and DSP2496, it has been my experience that they can not be optimized to work over a wide range without throwing away bits, unless you use an analogue volume control after the crossover, but before the amps. Having talked with various designers of these units and digital circuits in general, it has been my impression, from these so called experts, that this is a limitation of digital crossovers and speaker management systems right now, and while some are much better implemented than others, its still an issue. I recall Cabasse saying that had fixed the problem with the La Sphere's digital crossover, yet I've read numerous reviews of the system which seemed to feel it had a sense of grain and hash at very low levels. It also didn't measure all that well in stereophile as I recall, and there were some missing measurements, I think, which I suspect would have further shown some of the shortcomings in such a design.
I do think there is a HUGE difference between a properly implemented digital crossover and what a lot of people end up having to do. It's a mixture of design issues, as I've talked about, and how the end user implements it. Any time I have seen a DCX2496 used in a system demo, the volume has been set so high that the bit dumping issue wouldn't arise. I probably shouldn't get into naming names here, but at a certain electronics show, a certain vendor was showing off their new speaker with digital active crossover (which I think was a behrenger). Sounded great, everyone liked it, but as the show went on, I went back and asked if I could turn the volume down to a more reasonable level. As soon as I did that the tell tail distortions, loss of transparency, etc of the bit dumping issue arose. The manufacturer quickly ran up and turned it back up, and told me that it was only optimized for that higher volume range. Ha, yeah, because these units aren't designed to be used that way, its not about how he optimized it.
having used and fine tuned the Behrenger DCX2496 and DSP2496, it has been my experience that they can not be optimized to work over a wide range without throwing away bits, unless you use an analogue volume control after the crossover, but before the amps. Having talked with various designers of these units and digital circuits in general, it has been my impression, from these so called experts, that this is a limitation of digital crossovers and speaker management systems right now, and while some are much better implemented than others, its still an issue. I recall Cabasse saying that had fixed the problem with the La Sphere's digital crossover, yet I've read numerous reviews of the system which seemed to feel it had a sense of grain and hash at very low levels. It also didn't measure all that well in stereophile as I recall, and there were some missing measurements, I think, which I suspect would have further shown some of the shortcomings in such a design.
pjpoes said:
I'm an active guy thru and thru, but yeah in regards to the dcx this is very true indeed.
I currently run a deqx, prior to that I had the same system setup with the dcx, I was dumb then (well, read dumber than I am now) and so there were a few problems with the setup. However once I'd learned a bit more, this problem mentioned here made sense. I was always playing it loud. That's OK in and of itself, as I don't mind loud (an awful lot do mind) and I'm in the lucky position geographically that I can play loud (no neighbours) but still, it was definitely a limitation of the unit. At the time I didn't really realise what was happening, but when I learnt about digital vol control and throwing away bits etc (I used to use the dig vol control on the source unit) I realised a lot of the reason I played it loudly was that it sounded better loud.
I know we are talking 'budget' active solutions here, and the deqx certainly does not qualify as budget, and i know that everyone who describes their system here is 'wowed' and 'impressed' by their diy system, but since I have had the deqx in my system, and learnt and gotten it fully 'dialled in', I simply have not ever heard another system that even comes close. OK I don't haunt hifi shows etc (we don't have many here anyway!!), but I have heard quite a few $80-100 000 systems in various places that don't even come close.
Of course in this case we are not really talking digital active crossovers, but rather dsp. And again, unless you really know what you are doing you can muck it up as well, BUT given time and that you have done 'no harm', you can eventually get it right too.
It is very hard for us to describe what we hear (in a way that get's thru to s/one else that is) but what I get from my system that I have never heard anywhere esle to anything even remotely approaching mine is the ambience and information contained in the recording ,it is all around you.
And to be honest, this is quite recent really (since the system was finally tweaked with the correct measurements and calibrations), 8 months or so ago.
I too used to join in with the chorus of 'poor recordings' and so on.
Well, let me tell you, the quality of the recording (in the vast majority of cases, sure there are some truly awful ones out there) is the least of our worries.
I could be completely wrong in the why of it, but my running hypothesis is that the results I get are due to the correct phase of all the drivers, both individually and in concert, thruout the entire fr.
Again this is a function of dsp I guess, but the most convincing demonstration of that I have ever heard was with a single driver system, so that meant there were no x-overs at all obviously. Anyway long story short, it meant that all that was done was 'driver correction'. Well you would have to hear it to understand, totally different system.
Same phenomenon, when it was corrected (and back to my working hypothesis) and so all the phase info was presented correctly, the soundstage just opened up and became ambient.
As I say, I could have the reason wrong, and I understand we are really talking a different kettle of fish, but due to those reason I simply cannot believe a passive crossover could even begin to approach it, as the simple introduction of it would immediately muck up the phase (let alone that the individual drivers could not be corrected anyway).
Heh heh, it also helps my belief that I could not design a passive crossover to save my life!! Sure, something this complicated takes some work, but truly my mind boggles when I read about x-over design. And at a guess, some of the davanced x-over design techniques being discussed here would hardly, if ever, be implemented commercially (all the more reason to diy) and are of a very high level of sophistication. And I'd say techniques only understood by a very small minority of the dity community.
So I'd say a lot of caveats when saying a 'well done and properly done' passive network can be compared to an active setup.
But don't forget the level above them all (even tho clearly not within the domain of the original post) and that of dsp.
Zaph ZD5 Again, this is what gave me the idea to learn about them. I don't know much more than, they work, and how to design and implement them. They don't appear to have any effect on the amplitude response.
Wait, 180 degree phase shift? Flip the wires on your speaker, 180 degree phase shift, no amplitude change? Do you mean 180 degrees from the other driver, then use the ladder delay I linked to above. It's a lot of parts, so yeah, its bulky, but hey, it can be done, and its not unheard of. I have used them in my Focal towers as well as in the full range drivers of my Tripole speakers. The tripoles are very small speakers, so the crossovers are as compact as I could make them, using very small (20 gauge inductors) parts, I was able to fit them on a 2" by 2" pc card.
Personal opinion here, but I think PCI sound cards make great crossovers with the right software. I think there is some wonderful software available if you look around.
Wait, 180 degree phase shift? Flip the wires on your speaker, 180 degree phase shift, no amplitude change? Do you mean 180 degrees from the other driver, then use the ladder delay I linked to above. It's a lot of parts, so yeah, its bulky, but hey, it can be done, and its not unheard of. I have used them in my Focal towers as well as in the full range drivers of my Tripole speakers. The tripoles are very small speakers, so the crossovers are as compact as I could make them, using very small (20 gauge inductors) parts, I was able to fit them on a 2" by 2" pc card.
Personal opinion here, but I think PCI sound cards make great crossovers with the right software. I think there is some wonderful software available if you look around.
SY said:
There's BruteFIR for linux, but it is only an engine for running filters. To generate crossover and EQ filters you need some serious software and measurements. You could do the filters free with GNU Octave (MATLAB clone), but it requires some DSP expertise. Then there's Acourate by Dr. Ulrich Brüggemann. It looks like a very nice package, but it is a bit pricey for the average DIYer.
Soundeasy is the best put together software with the most advanced interface and complete abilities, in my opinion. They have recommendation for sound cards on their website, but you know how it goes, buy a decent 2in 8out or better pro audio interface card. I need to keep looking, I have a free demo software I got that I've used as well that I like quite a bit. I will post that when I find it as well.
Dont you hate when you forget the name of something, can't find it on your computer (even though you just used it a few days ago), and can't find the website. Well I finally found it Loudspeaker Frequency allocator
I'd expect things would go like this...
1)
You will have to take your speakers apart by removing the drivers or the terminal cup on the back. You will need to disconnect the current crossover. And put an extra set of terminals on the back. When you reconnect your drivers the screw holes will be just a little looser than when you started.
2) You get the $200 Behringer crossover. Hook everything up and quickly determine the crossover frequency is somewhere from 1800 to 3khz. You can probably do this all by ear without any fancy equipment. Just get a pink noise test tone CD.
3) After 10 hours of intense listening, you are confused why it sounds slightly different than before. In some ways you think it sounds better, but mostly you conclude that the cheap crossover has made it actually sound worse.
4) You decide it must be the cheap crossover. So you go out and buy a Behringer DCX2496 ($300) as an upgrade. It sounds better, but still not better than the original crossover.
5) You go out and buy 2 Benchmarkmedia DAC1's and a computer and some expensive measurement microphones. Now it actually DOES sound better than the original crossover. Of course, your source signal is better and that's the real reason why it sounds better, not because its active.
6) You go back to designing passive crossovers.
Active is all I use and the above story is almost autobiographical in nature. Consider it me poking fun at myself.
I see the active crossover as a design tool for creating new speakers. You can quickly test dozens of crossovers by pushing a few buttons. Once you have figured out what sound you like its time to design a passive crossover and set things in stone.
If your goal is to make your current speakers sound dramatically better look elsewhere. I had a commercial speaker from 10 years ago I disassembled for the same passive to active conversation project. I thought I could make it better.
What DID make it better was upgrading the tweeter to a better model.
1)
You will have to take your speakers apart by removing the drivers or the terminal cup on the back. You will need to disconnect the current crossover. And put an extra set of terminals on the back. When you reconnect your drivers the screw holes will be just a little looser than when you started.
2) You get the $200 Behringer crossover. Hook everything up and quickly determine the crossover frequency is somewhere from 1800 to 3khz. You can probably do this all by ear without any fancy equipment. Just get a pink noise test tone CD.
3) After 10 hours of intense listening, you are confused why it sounds slightly different than before. In some ways you think it sounds better, but mostly you conclude that the cheap crossover has made it actually sound worse.
4) You decide it must be the cheap crossover. So you go out and buy a Behringer DCX2496 ($300) as an upgrade. It sounds better, but still not better than the original crossover.
5) You go out and buy 2 Benchmarkmedia DAC1's and a computer and some expensive measurement microphones. Now it actually DOES sound better than the original crossover. Of course, your source signal is better and that's the real reason why it sounds better, not because its active.
6) You go back to designing passive crossovers.
Active is all I use and the above story is almost autobiographical in nature. Consider it me poking fun at myself.
I see the active crossover as a design tool for creating new speakers. You can quickly test dozens of crossovers by pushing a few buttons. Once you have figured out what sound you like its time to design a passive crossover and set things in stone.
If your goal is to make your current speakers sound dramatically better look elsewhere. I had a commercial speaker from 10 years ago I disassembled for the same passive to active conversation project. I thought I could make it better.
What DID make it better was upgrading the tweeter to a better model.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.