To me putting active crossovers into an inexpensive home bookshelf speaker is analogous to sticking the sequential transmission from an F1 car into a Ford Fiesta. Sure in racing situations the F1 sequential will be better than the factory stock transmission in the Ford. But in the Ford, the engine is still under powered and the car is still designed as a domestic economy car. The F1 would not only be a poor match in the Ford, working quite poorly, but you could never take advantage of its virtues in such a car. Putting an active crossover designed for pro audio use into a bookshelf like this will not only produce worse results than you had before, but the benefits it would have, would never be appreciated with the speaker in the first place.
Actually, to me changing inexpensive home bookshelf speakers to active is where the biggest gains can be made. The standard crossover is likely as simple as it can be and not that well optimised. With a little time and knowledge a good active crossover could be developed for it. Yes, using really good drivers would be a lot better in the end, but I think it's an ideal play project. As long as you don't fall into the trap of textbook crossovers that mr sreten so rightfully despises.
Yeah but this guy is wanting to use a Behringer active crossover, which can do nothing but give text book crossover slopes. I mean, look at those graphs I showed, and with some cheap speakers, they response can be even more ragged and peaky. A text book slope can't deal with that, BS isn't the only problem to contend with, and so you end up needing to make a custom active crossover or using a very expensive unit. The only way I could see achieving good results would involve running the output of one crossover into the input of a second, and so on 3-4 times in order to give the varied slopes needed to deal with all the problems.
The DCX2496 allows independent low-pass and high-pass filters on each output channel. Allowed filter types are BUT6, BUT12, BES12, LR12, BUT24, LR24, BES24, BUT48, LR48. It also allows to apply a phase shifter to each channel in order to compensate for the phase shift due to the natural roll-off of the neighbour driver. Finally, up to 9 parametric EQs per channel may be applied (40 maximum or so) to match group delay peaks and dips between drivers and get optimum summing.
In other words, it allows to tailor filter response to speaker requirements starting from ideal filters.
In other words, it allows to tailor filter response to speaker requirements starting from ideal filters.
I recently purchased a Behringer CX3400 with RCA/XLR cables and am very happy with it. There's a number of "buts" though to the active approach. My principal reason for going this way was that I found that I was spending an inordinate amount on passive components while trying out different designs AND I already had the necessary preamp and power amps on hand. I also avoided buying a plate amp for subwoofers this way.
In retrospect I should've bought a DCX2496 instead even though its more than twice the price. As mentioned in previous posts, with the CX3400 (and CX2310) your only filter choice is a 4th order Linkwitz Riley -- you can't implement a BSC or notch filter without additional passive components. You can't tame a rising response with a 1st order filter and so on.
On the plus side (again as already mentioned) you don't need to account for inductance or resistance of the drivers with active crossovers. You can handle some differences in driver sensitivity (up to 6dB with the CX3400) w/o extra components and have great flexibility in selecting the crossover points.
Getting back to the original question though, I would not recommend going active for a relatively inexpensive two-way -- its not cost effective. You'd be better off buying some measuring equipment and checking to see whether the original crossover can be improved (and if yes doing it passively).
In retrospect I should've bought a DCX2496 instead even though its more than twice the price. As mentioned in previous posts, with the CX3400 (and CX2310) your only filter choice is a 4th order Linkwitz Riley -- you can't implement a BSC or notch filter without additional passive components. You can't tame a rising response with a 1st order filter and so on.
On the plus side (again as already mentioned) you don't need to account for inductance or resistance of the drivers with active crossovers. You can handle some differences in driver sensitivity (up to 6dB with the CX3400) w/o extra components and have great flexibility in selecting the crossover points.
Getting back to the original question though, I would not recommend going active for a relatively inexpensive two-way -- its not cost effective. You'd be better off buying some measuring equipment and checking to see whether the original crossover can be improved (and if yes doing it passively).
no I disagree that it can tailor the slopes and stand by what I've said. Again, I direct you to my graphs to see what I'm talking about. You can not make a transfer function with a slope that changes with the Behrenger. All it can do is make standard textbook slopes of varying types. Parametric eq could be used to fake this sum, but would not be ideal, as a crossover causes a db step loss per octave, where as a parametric eq is more of a notch filter with variable notch width, depth, and frequency. The least expensive crossover I know that can do this is from DBX and is roughly 3500 dollars. The only other option I know of is the computer software based solutions.
Here this is simulation graphs from an active crossover programing software. This software is used to upload the custom slopes into high end active filters like those from Dolby. I start with the raw response of the drivers, then I use completely symmetric slopes of slopes and frequency, then I optimize based on what most active crossovers can do, then finally I fully optimize using features not available on most crossovers. These are features that can be achieved with passive slopes very easily though, with no more than 4 parts per driver. In fact, only 3 parts in series total (two in the woofer and one in the tweeter).
raw response
symmetric slopes, symmetric frequency, 48db per octave at 3000hz, no delay, no eq.
optimized with delay, phase, some parametric eq, non-symetric slopes, frequency, and Q. Notice there is still a rise in the response that has not been accounted for.
This begins to get into territory that normal crossovers can not do. This used variation in the Q of multiple parallel crossovers (Changing the specific slope of the 48db per octave crossover), but still using all parallel crossovers at the same frequency. As you can see, things look a lot better, some might even argue they look good enough that further refinement with the eq would fix the rest. However, there is still a rise in the response that isn't being accounted for, and its roughly 3db's from 1-2K.
This response graphs shows all of the optimizations I have been talking about, including different slopes, frequencies, and q's for all the 4 paralleled crossovers. This allows the response shaping that can not be achieved with any of the active crossovers I have used or seen under 3500 dollars. Here I have removed that 3db rise.
For reference, these graphs were created from FRD files I created from actual measurements of a Focal TC90TDx tweeter and Scan-Speak 7" revelator midbass (18W...).
raw response
symmetric slopes, symmetric frequency, 48db per octave at 3000hz, no delay, no eq.
optimized with delay, phase, some parametric eq, non-symetric slopes, frequency, and Q. Notice there is still a rise in the response that has not been accounted for.
This begins to get into territory that normal crossovers can not do. This used variation in the Q of multiple parallel crossovers (Changing the specific slope of the 48db per octave crossover), but still using all parallel crossovers at the same frequency. As you can see, things look a lot better, some might even argue they look good enough that further refinement with the eq would fix the rest. However, there is still a rise in the response that isn't being accounted for, and its roughly 3db's from 1-2K.
This response graphs shows all of the optimizations I have been talking about, including different slopes, frequencies, and q's for all the 4 paralleled crossovers. This allows the response shaping that can not be achieved with any of the active crossovers I have used or seen under 3500 dollars. Here I have removed that 3db rise.
For reference, these graphs were created from FRD files I created from actual measurements of a Focal TC90TDx tweeter and Scan-Speak 7" revelator midbass (18W...).
ok I take back some of what I said. The shelving I was doing through paralleled crossovers can be achieved with the shelving filter. I forgot it had this ability. That could also be used for baffle step compensation if need be.
However the amount of slope variations I made would still not be possible with a DCX2496. However it is possible with the Brownsound active crossover. I keep looking at that and thinking, ah what I could do with that, then I price it all out, think it through, and back off. Ah some day, when spending upwards of 1000 dollars on a diy digital crossover doesn't sting so much, I will buy one of those. What I would really like to do is build one with 8 inputs (4 cards) and between 16 and 21 outputs. However thats roughly what, around 2000 dollars plus.
However the amount of slope variations I made would still not be possible with a DCX2496. However it is possible with the Brownsound active crossover. I keep looking at that and thinking, ah what I could do with that, then I price it all out, think it through, and back off. Ah some day, when spending upwards of 1000 dollars on a diy digital crossover doesn't sting so much, I will buy one of those. What I would really like to do is build one with 8 inputs (4 cards) and between 16 and 21 outputs. However thats roughly what, around 2000 dollars plus.
HiFiNutNut said:
Our digital crossover software called XOverWizard does exactly that
Try it out on your own PC, if you got some individual driver measurements of a loudspeaker. You can downloaded it here: XOverWizard
I was sloppy when I wrote:
What pjpoes provided was an example of the potentially great flexibility of active crossovers. I don't have the DCX2496 or the simulation software (?) pjpoes has in order to try other scenarios but I wonder if a better starting point would be a lower xover point. Note I'm not familiar with the performance of either driver beyond the published T/S params, impedence and freq resp graphs.
I did not mean to imply that the DCX2496 could solve all problems that passive components can. I did say the principal attraction of active crossovers was the high cost of continually buying passive components.
What pjpoes provided was an example of the potentially great flexibility of active crossovers. I don't have the DCX2496 or the simulation software (?) pjpoes has in order to try other scenarios but I wonder if a better starting point would be a lower xover point. Note I'm not familiar with the performance of either driver beyond the published T/S params, impedence and freq resp graphs.
I have a DC2496 unit and I use it to experiment with many speakers configurations, this unit is very flexible and also I don't have to buy lots of passive components. They can be very expensive and you cannot resell them after usage, DCX has a very good resale value if you decide not to use it after a while.
After I narrow down to one configuration that I like I will try passive x-over and compare it against active.
cheers.
I keep reading people saying that passive crossovers are much more expensive over time. I don't think that is the case at all.
First, you can use mechanical connectors in your passive crossovers (you don't have to solder), so all the parts are removable and re-usable, should that be your desire.
Second, if you are building a new speaker, what will you do with the old one? With a passive design, you can sell it as a fully functioning system to anyone with a single amp. With an active, you can't sell a fully functioning, plug and play system, unless your buyer has the same active setup you do. If you sell the active crossover with the speaker, then you have to replace it with something new for your new active speaker, at considerably higher cost that a new passive crossover.
You can't move the old active speaker to another room, unless you have a second set of active gear. With a passive deisgn, the speaker can be used elsewhere in your home with only a single amp needed.
The only way I see active as cheaper is if you throw away your old speaker when you do something new and what cost did you have in your time and materials in building the old speaker.
First, you can use mechanical connectors in your passive crossovers (you don't have to solder), so all the parts are removable and re-usable, should that be your desire.
Second, if you are building a new speaker, what will you do with the old one? With a passive design, you can sell it as a fully functioning system to anyone with a single amp. With an active, you can't sell a fully functioning, plug and play system, unless your buyer has the same active setup you do. If you sell the active crossover with the speaker, then you have to replace it with something new for your new active speaker, at considerably higher cost that a new passive crossover.
You can't move the old active speaker to another room, unless you have a second set of active gear. With a passive deisgn, the speaker can be used elsewhere in your home with only a single amp needed.
The only way I see active as cheaper is if you throw away your old speaker when you do something new and what cost did you have in your time and materials in building the old speaker.
If you, like me, continually try out different enclosures and driver combinations passive is much more expensive. For example, the latest project I'm trying out is Martin King's U- and H-Frame designs (http://www.quarter-wave.com/OBs/U_and_H_Frames.pdf) using 2nd order Linkwitz-Riley xovers at 150Hz and 125Hz. For stereo pairs, the price of the large inductors and capacitors required exceeds the cost of the CX3400 I bought.
When I'm done playing with the enclosures I'll reuse the Alpha 15As in another project - ie. speaker building is a hobby. I don't sell the finished projects or normally intend to sell the drivers. If I intended to do that I'd use an active crossover to experiment and voice the speaker and then convert it to passive.
Since the passive crossover components are tailored to the drivers and the enclosure they frequently don't "fit" the next project. In the case of the U- and H-Frame experiments the 2nd-order xovers are used to flatten the rising response of the woofer in these specific enclosures.
When I'm done playing with the enclosures I'll reuse the Alpha 15As in another project - ie. speaker building is a hobby. I don't sell the finished projects or normally intend to sell the drivers. If I intended to do that I'd use an active crossover to experiment and voice the speaker and then convert it to passive.
Since the passive crossover components are tailored to the drivers and the enclosure they frequently don't "fit" the next project. In the case of the U- and H-Frame experiments the 2nd-order xovers are used to flatten the rising response of the woofer in these specific enclosures.
Holdent,
I still don't buy it. First, you are choosing the situation where the passive components are the most expensive and active most palatable. Of course it is a more feasible comparison if you are talking a low pass at 100-150hz and only one amp. Still, you could use Jantzen P-cores inductors at about $20 each, depending on your drvers impedance, and non-polarized electorlytic caps at about $3ea. for each 200uF. You could easily have a pair of passive crossovers for under $50. You can most certainly reuse all your passive components if you run them parallel or in series to make larger or smaller values. You can always unwind inductors to make them smaller, or just wind your own at an even lower cost. The CX3400 is about $130 and it is about as basic an active crossover as you can get and is certainly not able to reproduce, on it's own, what a typical passive crossover design does, as others have demonstrated in posts above.
You are also not including the cost of the amps for each channel, especially significant when we are talking about active for a multi-way speaker. Add a $300 amp in your scenario and I could have built about 9 pairs of passive crossovers for your scenario for the same price, and that is if I only use the passive parts once.
OK, so you don't want to sell your old speakers to support your DIY habit. Many other do. You also don't want to give your old speakers to friends, family or reuse them yourself in another room? You prefer to continue to reuse the same old drivers for however long it takes to justify your active investment? Most DIYers I know want to try all the newest drivers, not reuse all the old ones. With passive you have all that flexibility. Not so with active, not at least for the supposed lower cost. Also, as I've mentioned, the passive parts can indeed be reused over and over again. I do it all the time.
I still don't buy it. First, you are choosing the situation where the passive components are the most expensive and active most palatable. Of course it is a more feasible comparison if you are talking a low pass at 100-150hz and only one amp. Still, you could use Jantzen P-cores inductors at about $20 each, depending on your drvers impedance, and non-polarized electorlytic caps at about $3ea. for each 200uF. You could easily have a pair of passive crossovers for under $50. You can most certainly reuse all your passive components if you run them parallel or in series to make larger or smaller values. You can always unwind inductors to make them smaller, or just wind your own at an even lower cost. The CX3400 is about $130 and it is about as basic an active crossover as you can get and is certainly not able to reproduce, on it's own, what a typical passive crossover design does, as others have demonstrated in posts above.
You are also not including the cost of the amps for each channel, especially significant when we are talking about active for a multi-way speaker. Add a $300 amp in your scenario and I could have built about 9 pairs of passive crossovers for your scenario for the same price, and that is if I only use the passive parts once.
OK, so you don't want to sell your old speakers to support your DIY habit. Many other do. You also don't want to give your old speakers to friends, family or reuse them yourself in another room? You prefer to continue to reuse the same old drivers for however long it takes to justify your active investment? Most DIYers I know want to try all the newest drivers, not reuse all the old ones. With passive you have all that flexibility. Not so with active, not at least for the supposed lower cost. Also, as I've mentioned, the passive parts can indeed be reused over and over again. I do it all the time.
I'm not "choosing" a situation to illustrate my point -- this is my situation and the reason I went active. As well in a previous point I also noted that I already had the required electronics (other than the active crossover) so no additional cost there. I also noted earlier that active couldn't do everything and that for the purpose of the thread originator passive would be the most cost effective. [Incidently, I didn't pay $300 per amp, I got 3 NAD 2150 amps for $276 on eBay (US - not counting shipping) less than 6 mos. ago).]
Wow - where did this come from?! I made no statement justifying an expenditure by saving on driver purchases!
I have hand wound inductors and its not as basic as you imply -- I've found online calculators only get you into the ball park and you still have to test them (more expense) to ensure they are the value you want. After working out the cost of a meter against the savings of DIYing its better buying the inductors off the shelf unless you need a lot.
This is starting to beat a dead horse. I already noted
and for the originator
Different horses for different courses.
pjpoes said:
Actually you can fake all the required changes with the phase shifters and parametric EQs, at least until you run out of them. For example, a parametric EQ placed at the cut-off frequency equals a change in Q, and with two parametrics (or shelving) you can simulate different cutoff frequencies for the parallel sections comprising an high order filter. Shelving parametrics also allow to simulate frequency-dependent slopes.
Then again, it's all about poles and zeroes.
Now consider that all this is real-time user-adjustable.
The DCX2496 is very versatile, but it's true that it would be even more if it allowed to adjust independent Qs and cut-off frequencies for the bi-quads comprising higher-order crossovers. This wouldn't require more processing power, only a more complex user-interface (to be understood by even less people, considering that it's already complex for anybody not understanding crossovers well).
I would love it if somebody made a modified version of the firmware giving user access to the Qs and the cutoff frequencies of the internal biquads.
EDIT: Remember that the first goal is acoustical phase matching on-axis between every pair of speaker drivers, and only then, the second goal is flat response (that is easily achieved through global EQ).
I have some experience getting some performance out of old and naughty PA systems with DCX2496. Once phase is matched, it's a pleasure to equalize, you can feel the slightest change in EQ sliders, while systems that are not summing properly are a pain and seem to ignore your moves at times. Some people becomes puzzled at the results of old PA + DCX2496 + careful adjustment.
Actually, if someone could make software like that offered by ground sound would be fine. One in which you can not only take real measurements and alter the parameters in real time, but also import the frd files as a starting point, as it makes the modeling so much easier.
I still feel, from my experience and modeling, that you are generally better off, given the home audio environment, using passive crossovers in any design under say 5 grand or so. If you include amps, which you really need to, you probably would need to increase that number some. As I recall the dcx2496 only has one high and one low shelving filter per output, right? That means that you could use one for the peak, but then not have one for the baffle step. I tried to use a parametric eq to adjust the peeking in the response of that driver, but was unable to reduce it as smoothly or easily as a simply parallel resistor and appropriate crossover values.
Lower crossover points would not solve the peaking problem as it would exist regardless of how low you go. The driver peaks, with no crossover at all, at around 6-8db's at the top of its response, and this needs to be addressed no matter what. If you don't address it then the off axis response will be compromised. Actually, I didn't show this to help prove my point better, but the simplest solution to the problem is to use much shallower LR slopes for the bass with some parametric eq and shelving. It's still not ideal, but good enough for most.
I think for many of the people who lack the measurement and simulation software abilities that I have, or better even, probably end up with a response that looks like the first graph I showed, or maybe the second one at best.
I also want to reiterate that the graphs I showed were not showing what the Behrenger can do, it was showing what a top end digital crossover with appropriate programming software can do. The Groundsound unit could do this, but you are talking roughly 1000 dollars total when all is said and done. The closest things in the pro audio world that can do this, that I know of, are the top end DBX and Dolby units with appropriate external control software and measuring gear. The Behrenger can not do what I did, in order to make my point I actually used the Behrenger software, put in certain values in the behrenger, then transfered the transfer functions into the other software to simulate the results. The best I could do was the first 3 graphs with the Behrenger. Then using some "trickery" with the behrenger (Running the output of one crossover into the input of another externally is the only way to equal this), and then also adding a parallel resistor to the woofer, I got the second to last results. The final results involved things that the Behrenger can not do at all.
I still feel, from my experience and modeling, that you are generally better off, given the home audio environment, using passive crossovers in any design under say 5 grand or so. If you include amps, which you really need to, you probably would need to increase that number some. As I recall the dcx2496 only has one high and one low shelving filter per output, right? That means that you could use one for the peak, but then not have one for the baffle step. I tried to use a parametric eq to adjust the peeking in the response of that driver, but was unable to reduce it as smoothly or easily as a simply parallel resistor and appropriate crossover values.
Lower crossover points would not solve the peaking problem as it would exist regardless of how low you go. The driver peaks, with no crossover at all, at around 6-8db's at the top of its response, and this needs to be addressed no matter what. If you don't address it then the off axis response will be compromised. Actually, I didn't show this to help prove my point better, but the simplest solution to the problem is to use much shallower LR slopes for the bass with some parametric eq and shelving. It's still not ideal, but good enough for most.
I think for many of the people who lack the measurement and simulation software abilities that I have, or better even, probably end up with a response that looks like the first graph I showed, or maybe the second one at best.
I also want to reiterate that the graphs I showed were not showing what the Behrenger can do, it was showing what a top end digital crossover with appropriate programming software can do. The Groundsound unit could do this, but you are talking roughly 1000 dollars total when all is said and done. The closest things in the pro audio world that can do this, that I know of, are the top end DBX and Dolby units with appropriate external control software and measuring gear. The Behrenger can not do what I did, in order to make my point I actually used the Behrenger software, put in certain values in the behrenger, then transfered the transfer functions into the other software to simulate the results. The best I could do was the first 3 graphs with the Behrenger. Then using some "trickery" with the behrenger (Running the output of one crossover into the input of another externally is the only way to equal this), and then also adding a parallel resistor to the woofer, I got the second to last results. The final results involved things that the Behrenger can not do at all.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.