OK, perhaps a simple xo will do, without any thinking of acoustic response , slope symmetry, phase match etc. - just to get sound out of it. Then don't be surprised if it sounds worse than the original passive speaker. I was speaking of hifi construction with dsp, taking the best out of it, if one would go the difficult route anyway.
I don't agree with you sir. I think it is hard to do a passive crossover but by going active, it becomes easy. Thats one of the advantages. You can simply tune the sound by ear by changing the slopes and xover points using the digital crossover. How much time would it take to do that with passive parts?
Whats so hard about selecting a crossover point and a crossover slope? Its easy when you do it digitally.
Repeat a few are overthinking this.
The crossover shown is nothing more than a 12dB/oct lowpass feeding the woofer and a 18dB/oct feeding the tweeter.
Make same crossover frequency and slope actives, feed 50W/15W amps with those signals and drive the speaker components.
It will *start* sounding GOOD and , of course, can be tweaked to death.
Don't forget that KEF Engineers *already* did all the work, so following them should not be a bad idea.
Why think otherwise?
And some seem to forget this is **DIY Audio** Internet Forum 😉
What's perfectly justifiable in a Major Company Design Team, agree with that, may be a too heavy workload for a regular Forum Member.
The crossover shown is nothing more than a 12dB/oct lowpass feeding the woofer and a 18dB/oct feeding the tweeter.
Make same crossover frequency and slope actives, feed 50W/15W amps with those signals and drive the speaker components.
It will *start* sounding GOOD and , of course, can be tweaked to death.
Don't forget that KEF Engineers *already* did all the work, so following them should not be a bad idea.
Why think otherwise?
And some seem to forget this is **DIY Audio** Internet Forum 😉
What's perfectly justifiable in a Major Company Design Team, agree with that, may be a too heavy workload for a regular Forum Member.
Might be worth doing a frequency sweep and measuring the voltage on the drivers... Would not be hard to do this, even manually, to confirm just what the xover is doing... Make it easier to transfer it to your DSP.
Can you post the crossover schematic? Three components could be a lot of things. First order, second order series crossover, mixed slopes...
What people are saying is it will likely take you a lot of time to learn to measure properly, interpret what you are measuring and then incorporate that knowledge into a crossover design. Not impossible, and if you are doing it for the learning experience, great. Just don't expect perfection (or even to get close to the original sound) right away.
Just FYI, Sallen-Key is not a filter function but an electronic topology that can be used to reproduce different filter transfer functions that include Butterworth, Bessel, Linkwitz-Reilly or something in between "named" filters. Most of the active filter boards you will find are going to be Sallen Key (usually unity gain variants), although some will be state variable or multiple feedback variants. Each topology has its benefits and drawbacks.
Oops, forgot the XO was in the first post. That looks like electrical first order (with impedance correction) low pass and an 3rd order high pass. Likely the low pass combines with the driver's own response to make a 2nd or 3rd order acoustic response. The 3rd order high pass likely accomplishes some or all of the phase matching that has been mentioned. Again fairly easy to approximate actively. Follow the electrical filter orders. The low pass could be an R-C filter feeding a buffer. High pass make a second order unity gain Sallen-Key followed by a C-R high pass into a buffer. The R-C across the woofer may also provide some EQ. See http://www.linkwitzlab.com/ for various EQ modes. Have fun, but expect a learning curve. PM me with your email and I'll send you a spreadsheet.
Looking at it again, the low pass is more 2nd order with the R in the RC changing the Q. Use a variable gain Sallen Key filter. The filter component values are easy to calculate and you can trim the Q to flatten out whatever bump KEF was working on.
What people are saying is it will likely take you a lot of time to learn to measure properly, interpret what you are measuring and then incorporate that knowledge into a crossover design. Not impossible, and if you are doing it for the learning experience, great. Just don't expect perfection (or even to get close to the original sound) right away.
Just FYI, Sallen-Key is not a filter function but an electronic topology that can be used to reproduce different filter transfer functions that include Butterworth, Bessel, Linkwitz-Reilly or something in between "named" filters. Most of the active filter boards you will find are going to be Sallen Key (usually unity gain variants), although some will be state variable or multiple feedback variants. Each topology has its benefits and drawbacks.
Oops, forgot the XO was in the first post. That looks like electrical first order (with impedance correction) low pass and an 3rd order high pass. Likely the low pass combines with the driver's own response to make a 2nd or 3rd order acoustic response. The 3rd order high pass likely accomplishes some or all of the phase matching that has been mentioned. Again fairly easy to approximate actively. Follow the electrical filter orders. The low pass could be an R-C filter feeding a buffer. High pass make a second order unity gain Sallen-Key followed by a C-R high pass into a buffer. The R-C across the woofer may also provide some EQ. See http://www.linkwitzlab.com/ for various EQ modes. Have fun, but expect a learning curve. PM me with your email and I'll send you a spreadsheet.
Looking at it again, the low pass is more 2nd order with the R in the RC changing the Q. Use a variable gain Sallen Key filter. The filter component values are easy to calculate and you can trim the Q to flatten out whatever bump KEF was working on.
Last edited:
^^^^ That's the point.
All concerns expressed here are valid, I'm not dissing them **at all** , but at the same time I think the OP should start with something, as of *now*.
And my (educated) guess is that even first results will be good, simply by mimicking what KEF (who are not fools) did.
And from there on he can experiment.
The point is not getting blocked before even plugging the soldering iron by a desire to get First Prize on the first hit.
This is fun place to be, and a great learning tool to boot 🙂
All concerns expressed here are valid, I'm not dissing them **at all** , but at the same time I think the OP should start with something, as of *now*.
And my (educated) guess is that even first results will be good, simply by mimicking what KEF (who are not fools) did.
And from there on he can experiment.
The point is not getting blocked before even plugging the soldering iron by a desire to get First Prize on the first hit.
This is fun place to be, and a great learning tool to boot 🙂
Yes indeed. This will yield a "baseline" set of measurements that can duplicated (initially) and allow further fine-tuning after that.
If a person wants to start from scratch that's valid too, but at least measuring/saving the electrical filter responses that KEF created is terrific piece of information to fall back on if needed.
Cheers,
Dave.
I have been looking around for the Behringer CX2310 instead of a fixed active crossover, seems nice to be able to tune frecuency and gain of each low/high-pass. Any experience with it?
The first problem I have with the CX2310 is it is fixed in symmetrical 4th order LR electrical filters. (this is the generic crossover route that you were warned against) You see by the schematic you made that KEF determined that the drivers are best served with asymmetric filter responses. Other issues include balanced in out (are your amps and sources balanced?) and lack of phase adjustment or EQ.
Better in Behringer would be the DCX2496, while more expensive allows all the functions you need. Another popular digital crossover is miniDSP. I haven't used it, but as mentioned earlier, Linkwitz says it sounds good enough for him, and it is the least expensive of the ready to go DSP XO options. I use Thuneau's frequency allocator and an old M-Audio Firewire 410 (Firewire 610 is current) for prototyping. Any ASIO sound card works with Allocator, or you can use ASIO4all to make your current multichannel card compatible. There's a free trial available.
If you want to try the CX2310 route, hope your tweeter is flat for an octave or so lower than the original crossover frequency and use a lower crossover frequency. This may work for lower listening levels, but the tweeter may not like lower frequencies at higher SPLs.
Industry standard seems to be about 0.5 mm xmax on tweeters, pretty much limiting them to crossing near 2KHz.
Better in Behringer would be the DCX2496, while more expensive allows all the functions you need. Another popular digital crossover is miniDSP. I haven't used it, but as mentioned earlier, Linkwitz says it sounds good enough for him, and it is the least expensive of the ready to go DSP XO options. I use Thuneau's frequency allocator and an old M-Audio Firewire 410 (Firewire 610 is current) for prototyping. Any ASIO sound card works with Allocator, or you can use ASIO4all to make your current multichannel card compatible. There's a free trial available.
If you want to try the CX2310 route, hope your tweeter is flat for an octave or so lower than the original crossover frequency and use a lower crossover frequency. This may work for lower listening levels, but the tweeter may not like lower frequencies at higher SPLs.
Industry standard seems to be about 0.5 mm xmax on tweeters, pretty much limiting them to crossing near 2KHz.
No, you don't really need to.
A transfer function measurement across the existing passive crossover represents the exact amplitude response you should use to create a line-level equivalent.
This assumes the testing amplifier (and the amplifier used to drive the new, active speakers) used when creating that measurement has low output impedance and is functioning as a voltage source.
Cheers,
Dave.
Hi,
In reality its probably a serious waste of time and money unless
you get serious, tri-amp and convert to a properly optimised
2.5 way speaker. YMMV but TMM 2 ways, here 3/4" tweeter
with twin 6.5" bass units is an inherently compromised setup.
FWIW it smacks of a speaker shoehorned into an AV range,
and not much about it appeals to me, especially as a possible
candidate for very a simplistic conversion to active.
Describing it as an 8 ohm speaker is appalling nonsense,
they are clearly 3 to 4 ohm nominal impedance speakers.
rgds, sreten.
In reality its probably a serious waste of time and money unless
you get serious, tri-amp and convert to a properly optimised
2.5 way speaker. YMMV but TMM 2 ways, here 3/4" tweeter
with twin 6.5" bass units is an inherently compromised setup.
FWIW it smacks of a speaker shoehorned into an AV range,
and not much about it appeals to me, especially as a possible
candidate for very a simplistic conversion to active.
Describing it as an 8 ohm speaker is appalling nonsense,
they are clearly 3 to 4 ohm nominal impedance speakers.
rgds, sreten.
Last edited:
I am thinking that the speakers, while working with one low-pass and one high-pass signal, there are two woofers in parallel.
Does it make any difference if I use 3 amplifiers, one for each cone instead of the two woofers sharing one amplifier, and feeding the same low-pass signal to each of the two low-pass amplifiers?
Does it make any difference if I use 3 amplifiers, one for each cone instead of the two woofers sharing one amplifier, and feeding the same low-pass signal to each of the two low-pass amplifiers?
I know impedance will be half if paralleling the two woofers, but I have to add that it should be no problem because I am using one BPA board for each cone (3x LM3886 in parallel). So if one LM3886 is able to drive a 4ohm load, 3 in parallel should drive ""approximately"" 1,4 ohms.
You can use three amps with the same input signal, the speaker won't care.
If you are looking to improve the performance of the speakers, as consider making it a 2.5 way as suggested recently. Roll off the two woofers that are furthest from the tweeter at baffle step frequency. You can take the initial drive from the "main" woofer high pass and then add another filter to get the proper response.
The issue with your logic in post 34 is that you are talking about a bridged amplifier. The effective load seen by a bridged amp is half of the speaker impedance. So, with 2 chips in parallel on each side, you get the current capability of a single chip with twice the voltage swing. For your BPA, the minimum load should be 2.8 ohms. You don't want to operate LM3886 anywhere near its limits on a regular basis. The effect of the SPIKE protection circuit is VERY audible. Seriously consider using separate LM3886 amps per driver, perhaps paralleled chips.
If you are looking to improve the performance of the speakers, as consider making it a 2.5 way as suggested recently. Roll off the two woofers that are furthest from the tweeter at baffle step frequency. You can take the initial drive from the "main" woofer high pass and then add another filter to get the proper response.
The issue with your logic in post 34 is that you are talking about a bridged amplifier. The effective load seen by a bridged amp is half of the speaker impedance. So, with 2 chips in parallel on each side, you get the current capability of a single chip with twice the voltage swing. For your BPA, the minimum load should be 2.8 ohms. You don't want to operate LM3886 anywhere near its limits on a regular basis. The effect of the SPIKE protection circuit is VERY audible. Seriously consider using separate LM3886 amps per driver, perhaps paralleled chips.
I forgot to say that while being a BPA, I wired it to switches and inputs so to be able to use the 4 boards as separated amplifiers. That is, 2 bridged boards could drive 2/3 of the impedance of what a single LM3886 (there are 3 IC per board) can drive, near 2,6 Ohms. A single board can drive (theorically) 1,3 Ohms.
That said, I have 4 amplifiers that can drive 1,3 Ohms each and have the voltage swing of a single LM3886. That's why I am asking if driving each woofer with a dedicated board worth it or am I good by driving those two woofers in parallel with a single board. If it is advisable to drive them separately, I would use another separate 20W amplifier just for the tweeters.
That said, I have 4 amplifiers that can drive 1,3 Ohms each and have the voltage swing of a single LM3886. That's why I am asking if driving each woofer with a dedicated board worth it or am I good by driving those two woofers in parallel with a single board. If it is advisable to drive them separately, I would use another separate 20W amplifier just for the tweeters.
Are you saying your BPA-300 is bridging two boards with three parallel chips? Are all boards non inverting?
If that's the case, use a board for each woofer, with an extra filter for the bottom woofer to make it a 2.5 way. Don't go nuts with a 20W tweeter amp, just get a standard LM3886 board and use one chip per tweeter. That simplifies your power supply requirements and allows plenty of reserve power for the tweeter. You won't blow the tweeter if you implement your crossover properly.
If that's the case, use a board for each woofer, with an extra filter for the bottom woofer to make it a 2.5 way. Don't go nuts with a 20W tweeter amp, just get a standard LM3886 board and use one chip per tweeter. That simplifies your power supply requirements and allows plenty of reserve power for the tweeter. You won't blow the tweeter if you implement your crossover properly.
This is exactly what I built, I even used the same boards from this forum. But I added the ability to use it as an stereo amplifier bridging boards, or as a 4 channel amplifier using each board for a channel.
DIY BPA300 6x LM3886 300W audio Amplifier
Looking at the schematic I recall that boards were inverting, right?
I finally bought a Behringer CX2310, I am waiting for some cabling to start playing with it. Now that I think about it, being inverting boards, I will have a problem of phase correlation if I use a non inverting amplifier for the tweeters. This should be no problem as the Behringer allows me to invert the HP or LP if I want to:
http://www.behringer.com/assets/CX2310_P0132_Front_XXL.png
DIY BPA300 6x LM3886 300W audio Amplifier
Looking at the schematic I recall that boards were inverting, right?
I finally bought a Behringer CX2310, I am waiting for some cabling to start playing with it. Now that I think about it, being inverting boards, I will have a problem of phase correlation if I use a non inverting amplifier for the tweeters. This should be no problem as the Behringer allows me to invert the HP or LP if I want to:
http://www.behringer.com/assets/CX2310_P0132_Front_XXL.png
There's a real easy solution for using a non inverting amp for the tweeters, reverse the connections at one end of the speaker cable. Just plug the positive cable from the speaker to the negative on the amp and vice versa.
If you want, build the amp with the terminals reversed internally, so the connections are "normal".
Of course, the inverting amp may have already done this for you. Are the connections to the positive post to the chip or to the power supply ground? If ground, just use a non inverting amp on the tweeter and you'll be fine.
If you want, build the amp with the terminals reversed internally, so the connections are "normal".
Of course, the inverting amp may have already done this for you. Are the connections to the positive post to the chip or to the power supply ground? If ground, just use a non inverting amp on the tweeter and you'll be fine.
Last edited:
Yeah, I didn't think about it, it is as easy as that. Reverse connecting the speaker cables or pressing the invert signal button on the Behringer crossover.
- Status
- Not open for further replies.