I am in the process of a project creating a 4 channel chipamp with active crossover to drive my B&W 685s. So first step was to strip down the speaker and see what it is made of. The bass-mid driver seems reassuringly heavy. The tweeter has a sort of tube on the back - I don't know what that is all about but seems to be well-regarded. The existing crossover is surprising. One capacitor in series with the tweeter and one inductor in series with the bass-mid. Both high quality components so no arguments there, and I know there is a school of thought that says the simpler the better, but a first order filter on both drivers doesn't seems promising in terms of phase/coherence. So it will be interesting to hear what these being driving by a 4th order linkwitz-riley sound like (and if I can tell any difference!). Pictures of existing crossover and components just for fun. I am reusing the boards and replacincg the inductor with a short and the capacitor with back to back 2200muF electrolytics (as DC protection for the the tweeter).
Those multilayer air-cores B&W use get to be surprisingly lossy by the crossover frequency due to proximity effect. The crossovers inside my CM1s looked almost the same as yours.
Too bad you took it apart. A good thing to do would have been to measure the freq responses from the various drivers as-is, so you could reproduce those in the DSP.
Jan
Jan
There are some measurements of the whole speaker here : https://www.soundstagenetwork.com/i...&catid=77:loudspeaker-measurements&Itemid=153
The cross over design is closely linked to the cabinet design. IIRC the B&W is a ported enclosure so you're getting a different roll off response to a sealed enclosure (and vice versa) so the crossover design is impacted by this.
Speaker design is a seriously complex branch of engineering. Designing good amps is easy. Designing good speakers is not.
A good book on these subjects is 'The Loudspeaker Cookbook' by Vance Dickason where the cross over order wrt cabinet design is discussed in some depth.
🙂
Speaker design is a seriously complex branch of engineering. Designing good amps is easy. Designing good speakers is not.
A good book on these subjects is 'The Loudspeaker Cookbook' by Vance Dickason where the cross over order wrt cabinet design is discussed in some depth.
🙂
I think he just wants to convert it to multichannel and DSP xover. That's a piece of cake if you have the original responses.
That's what I do with the ESLs. Record the response at the output of the step-ups and recreate it in the DSP. Done.
jan
no, its the best place to start, with the raw response of the drivers.
Cool project m10!
I have done similar action on other 2 ways. Be sure to implement a baffle step correction as well, it may be needed if running a regular LR joint. Also run a 12 db test, 24 can be somewhat hard i have noticed. You may also benifit from a ~0.10 ms delay on the tweets.
I have done similar action on other 2 ways. Be sure to implement a baffle step correction as well, it may be needed if running a regular LR joint. Also run a 12 db test, 24 can be somewhat hard i have noticed. You may also benifit from a ~0.10 ms delay on the tweets.
Sometimes yes, but in the case that no drivers or acoustic conditions will be changed then it is best to avoid acoustic measurements unless you are instead redesigning the crossover.
Well easily reinstated - a total of 4 solder connections! But I have access to another pair. So I am going for a straightforward active conversion and will compare the two versions. It always strikes me that speakers and rooms are such a compromise its more of matter of which I prefer the sound of than as to which measures flatter.
Read his post. That's exactly he wants.
"So it will be interesting to hear what these being driving by a 4th order linkwitz-riley sound like"
A proper crossover is when that 4th order Linkwitz-Riley response is acoustical, so he needs the raw acoustical driver responses to begin with.
Yes - some of you are just way, way ahead of my modest technical aspirations! I like messing with electronics. Years ago I built myself a set of successful speakers (well everyone that heard them was 'impressed') which I implemented using active crossovers and multi-amping. So I am just amusing myself again. I do appreciate that DSP is now 'the way to go' at least technically. I also know I could measure impulse responses and software implement crossovers in response to those etc. But that is way beyond where I desire to go. However .... the next iteration on my list is to use a Zoudio amp/DSP board to do the same sort of thing over again, only digitally and to see how all that sounds. Then maybe toy with adding a subwoofer. All good clean fun!
Besides anything else I measure them as close to 1ohm dc resistance (about 0.75 in fact) so given the nominal 6ohm driver that is quite lossy. If nothing else I will gain some efficiency here!
Let's consolidate then. If you use LR filters then you won't have LR at the speakers because they change it. What will you do? ... at the very least you will want a way to confirm the polarity you choose is acceptable. You can do this by listening but it would be more informative if you measure.
Thanks for all your input. Points noted @AllenB ! Here is an update and my design approach.
I have found some apparently good measurements of the drivers and complete speakers here http://www.audioexcite.com/?page_id=6164. The existing passive crossover is making use of the woofer's drooping response above 4k - so it is a first order crossover above 4k which 'maybe' sums to 2nd order with the rolloff. But the FR of the whole speaker shows this is only partly successful as it seems to show a pretty pronounced hump around that frequency. I still think this is a pretty low-budget approach by B&W despite all their technical clout.
Inspecting those driver response curves I am looking for where they are jointly the 'flattest' over a reasonable range. My eyeballing of the data suggests somewhere around 2.8Khz looks the most promising area to invoke a high order (4th) LR electronic crossover. So that is what I am going to try. Looking at those graphs I cannot see any reason why reverse phase is empirically suggested.
I will update with progress. If I can source the necessary gear at a reasonable price I may end up doing an A -B comparison of my active implementation vs the existing passive.
BTW I realise this is all very 'belt and braces' stuff. I got the speakers very cheap and its just a bit of fun. Its DIY after all!
I have found some apparently good measurements of the drivers and complete speakers here http://www.audioexcite.com/?page_id=6164. The existing passive crossover is making use of the woofer's drooping response above 4k - so it is a first order crossover above 4k which 'maybe' sums to 2nd order with the rolloff. But the FR of the whole speaker shows this is only partly successful as it seems to show a pretty pronounced hump around that frequency. I still think this is a pretty low-budget approach by B&W despite all their technical clout.
Inspecting those driver response curves I am looking for where they are jointly the 'flattest' over a reasonable range. My eyeballing of the data suggests somewhere around 2.8Khz looks the most promising area to invoke a high order (4th) LR electronic crossover. So that is what I am going to try. Looking at those graphs I cannot see any reason why reverse phase is empirically suggested.
I will update with progress. If I can source the necessary gear at a reasonable price I may end up doing an A -B comparison of my active implementation vs the existing passive.
BTW I realise this is all very 'belt and braces' stuff. I got the speakers very cheap and its just a bit of fun. Its DIY after all!
the author of the above site goes on to suggest crossover improvements and sites "improved/better phase tracking"
however, the graphs presented do not appear to concur.
however, the graphs presented do not appear to concur.
If you decide to take the DSP route you can always use the methode of first EQ-ing the drivers flat (don't go overboard a tweeter is not a midwoofer 😉 ), and after that applying a textbook LR filter.
If that's the route you take make sure you have some good measurements of the drivers in the enclosure without any filters attached. Also study the distortion profiles to pick a suitable X over point.
If the DSP allows you can also go fancy with FIR to get a linear phase crossover.
I found this to be an exellent write-up of the procedure: https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf
If that's the route you take make sure you have some good measurements of the drivers in the enclosure without any filters attached. Also study the distortion profiles to pick a suitable X over point.
If the DSP allows you can also go fancy with FIR to get a linear phase crossover.
I found this to be an exellent write-up of the procedure: https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf