DIY active crossover

Hi Of course there is a bunch of projects on the net , but I never tried to build any of . I tried some old single-end design but there were a lot of dynamics issues and collorations. So, my question is do you know some accurate and "high end " design? Maybe some discrete design? :

There is an old school 1979 Brian G. Wachner design that sounded damn good
and would be easy to DIY in some form. (BGW Model 20).  I'll try and dig up
a schmo.

GARY

I am casing this one up right now. It uses opamps.
It's for my living room HiFi system.
There is also a 24dB 2 or 3-way crossover on Rod Elliot's site (basically just another filter stage added after the first, with some resistor/cap doubling.)
http://sound.westhost.com/project09.htm

Here is an edited copy of my post over there......

Project 81 -- 12 dB active 2-way Linkwitz-Riley alignment and phase coherent
http://sound.westhost.com/project81.htm

My intentions are to use this 12db crossover to properly blend (crossover) my front channel B&W speakers, and an HSU subwoofer......which makes it a 2.1 channel system.

Theory in short....
The P81 is used on the output of my preamp and from there, is sent to 2 amps...highpass is sent to front channel amp, and  lowpass to the subwoofer amp....crossed-over @ 80 Hz.
What this does, is "relieve" my front channels from having to produce much below 80 Hz, an area that mini-monitors often struggle at (flabby, poorly-defined bass). and gives that responsibility to the sub. Plus, the crossover point will be well tuned, rather than relying on my sub's adjustable lowpass filter only (there's no highpass filter on my sub or preamp).
The sub's one filter is defeatable, and the P81 alone, defines the crossover point for satellites and sub.

I have built the P81, installed it temporarily in my system, and seems to be doing it's job....
(click on pic)
(unshielded ?? noise ??? >> NO << ....how about that ? :shock: )

associated gear...
http://www.hsuresearch.com/products/vtf-2-mk3.html
http://www.bowers-wilkins.com/display.aspx?infid=2301&sc=hf

Click to expand...

Also, LINKWITZ LAB has much info, theory, boards, and schematics for building your own.
http://www.linkwitzlab.com/filters.htm

=FB=

I'm going to do one in the future. I've got an ATC 10" bass, ATC dome mid and a Morel tweeter waiting, and a speaker cabinet is mostly done.

What my plan is, is to first take long measurements of the system with a digital cross-over. After I've settled on the cross-over frequencies & other things, I'm going to duplicate them in analog. Based on what I've researched regarding the cross-overs ATC use, I should be getting away with 24db/oct crossover frequencies and some phase alignment in mid-tweeter xover frequency. Baffle step is going to be taken care of, too of course (I've got to check my notes, I think I took this into account when designing the cabinet) and there is going to be a slight bass boost in the low end since it's a closed cabinet. It's going to be a lot of work but I know I'm going to succeed with it. 

http://www.linkwitzlab.com/filters.htm

Here's a bunch of good stuff, too.

Thanks guys, I'm familiar with http://www.linkwitzlab.com/filters.htm and http://sound.westhost.com/project09.htm but never heard about old school 1979 Brian G. Wachner design. That will be great to see the schemo  .

What my plan is, is to first take long measurements of the system with a digital cross-over. After I've settled on the cross-over frequencies & other things, I'm going to duplicate them in analog.

Click to expand...

I did exactly the same thing , I'm using upgraded DCX2496 + Hypex amps and system measures flat, sounds OK but I still miss some details. That's why I decided to try analog active. I'm still not sure which design to try  :

I just ordered the DCX2496, and from reading from diyaudio.com and various other sources there will be some problems with the unit, mainly with the CS8420 sample rate converter and the NJM4580 opamps used everywhere. Also the PSU might need upgrading. Anyway, I don't think analog active can bring any life to any well working digital system. I mean if you are mixing in a box and go the digital route to the DSP, use some good quality DAC's with reasonable slobe anti-aliasing filters then how can you go wrong?  If you are missing details then there are propably some time-dependent issues like some kind of reverberation in the room. Flat on-axis response is not a measurement of transparent system anyway, it may be the off-axis response has some notches which prevent some room interaction. Also with MTM systems the vertical lobing may be a problem. In general the room acoustics in studios are problem as you should have some reverberation left to mimic home listening environment. Not to say that an all analog active system would be bad or worse than the half digital one, at least if you don't have to tweak the system anymore. Building such system kind of freezes your (speaker system) development process and let's you concentrate on other issues (like listening music ) which is always a good thing. However, at some point I would like to have a system that doesn't need tweaking or changing of components. I think at diyaudio site there was a guy who actually had built a system (think it was Geddes' Summa spekers) and he had lost all interest in building speakers  because he had a system he was happy with. I think it is the way to go.

I just ordered the DCX2496, and from reading from diyaudio.com and various other sources there will be some problems with the unit, mainly with the CS8420 sample rate converter and the NJM4580 opamps used everywhere. Also the PSU might need upgrading.

Click to expand...

Yes I did all available upgrades and DCX sounds much better than stock (BTW stock is terrible sound wise). I removed "upgraded analog PSU" and returned to stock one because it simply sounds OK, just need to be L/C filtered to rid of some noise.

I mean if you are mixing in a box and go the digital route to the DSP, use some good quality DAC's with reasonable slobe anti-aliasing filters then how can you go wrong?  If you are missing details then there are probably some time-dependent issues like some kind of reverberation in the room. Flat on-axis response is not a measurement of transparent system anyway, it may be the off-axis response has some notches which prevent some room interaction

Click to expand...

Unfortunately I use DCX analog in because I have large analog console. Acoustics is done pretty well, and I claim that some details are missing because I compared system with similar but passive crossed speakers in same room. So, it measures flat, I don't have any EQ problems but i can't hear all details i hear on my passive monitors. At the end, I tried to do all passively but mine drivers are not ideal for passive crossover  That's why I would like to try some analog active and to hear the difference. If that doesn't sounds "better" I can go back to DCX...

Moby said:

Hi Of course there is a bunch of projects on the net , but I never tried to build any of . I tried some old single-end design but there were a lot of dynamics issues and collorations. So, my question is do you know some accurate and "high end " design? Maybe some discrete design? :

Click to expand...

Designing an accurate "active" crossover begs the question "accurate" to meet what target response? Stand-alone active crossovers generally aim for steep symmetrical band pass filters that don't interact too severely in the crossover regions between adjacent outputs.  Passive crossovers, are a huge tradeoff in trying to steer the right frequency ranges to the right drivers with minimal losses, good driver protection, and again good combining in transition regions.

Designing a dedicated active crossover for a known loudspeaker, means you can design a filter set from scratch that much better matches what the specific drivers want to receive. Ideally you want to look at the strengths and weaknesses of each driver. The performance of the drivers will often lead you in a one direction, or another.  Say you are crossing over between two drivers where one is still pretty flat, but the other is already in trouble (phase shift and roll-off). For this situation you want a steeper slope on the marginal driver to keep it out of trouble and rely upon the stronger driver to pick up the slack. This suggests an asymmetrical filter topology like the "derived" or subtractive filter topology. One conventional multiple pole HPF or LPF, is subtracted from unity to create the adjacent driver output. This derived output will not have as steep of a slope as the convention filter, but the nice thing about this approach is because the signal is derived , it will recombine to unity (most crossovers filters don't). The stronger driver can tolerate the more gradual slope, and it uses less parts with no need to precision match the two filters.

In general for loudspeaker crossovers, less phase shift is better, so the best sounding in that regard is 1 pole. But obviously the one pole filter has the slowest roll-off. Again here is where you need to look at the drivers and see how quickly you need to roll off between drivers to protect and serve. Most passive crossovers are lucky to generate 2 pole crossover responses.

It would be instructive to reverse engineer the passive crossover to understand the approach the original speaker design engineer chose.

Sorry about the non-answer answer, but IMO the question is not which high performance circuitry to use, but what filter curves to target.

Good luck.

JR

Sorry about the non-answer answer, but IMO the question is not which high performance circuitry to use, but what filter curves to target.

Good luck.

Click to expand...

No problems, I like your answers anyway  . Well, I wanted to ask about circut because I already have idea about curves and crossing points generated and tested with DCX.

Moby said:

Acoustics is done pretty well, and I claim that some details are missing because I compared system with similar but passive crossed speakers in same room. So, it measures flat, I don't have any EQ problems but i can't hear all details i hear on my passive monitors. At the end, I tried to do all passively but mine drivers are not ideal for passive crossover  That's why I would like to try some analog active and to hear the difference. If that doesn't sounds "better" I can go back to DCX...

Click to expand...

Are you sure you don't compare apples to oranges with the passive vs. active monitors? Simple change of location (of the speakers, I mean the other are soffit mounted and the other free-standing) will change the room response a lot, and if your passive monitors were not flat it may be the reason why they did show some detail which the flat EQ'ed active didn't. Which kind of monitors the passive are actually? Not the Tannoys?

How about the DCX setup, have you properly time-aligned the drivers? This is something which is more difficult to do with analog active setup, you can use "HP allpass" or "LP allpass" or both in series like Linkwitz does to introduce some delay. Unfortunately you can't simulate this with DCX because it's missing the allpass filters as well as Linkwitz transfrom function (for closed woofers, though it can also be used with tweeters).  Have you tried the system with only single midwoofer connected, I think it would help as well as raising the mid to lf crossing frequency a bit (Genelec uses 400 Hz and 3.5 kHz with the 1035B for an example, it is also a TMM configuration speaker which helps). Also using a tweeter with a waveguide would help bringing the detail in (there is the quite new Seas with built-in DXT waveguide which might match into your system, http://www.seas.no/index.php?option=com_content&task=view&id=184&Itemid=179 ).  Anyway, nearfield monitors always bring more detail than main monitors simply because the sound is not coloured with the room reflections (because of the close promiximity of course, though with nearfields the reflections from the mixing desk may be a problem), I think it's unrealistic to expect both to have the same level of detail. Also if you need more bass extension you might also combination of nearfields and the bass cabinets of the mains (the nearfields obviously should be time-aligned to the woofers).

mhelin , lot of questions and every one is resonable  Yes I agree with evry single word you said and I tested almost everything. At first I must say that I'm really happy with my mains, I would  like to try to improove things a bit. I don't have a "huge problem". Just a few details  : the nearfields I like is my new ones  ;D

Regarding tweeter time align that's the only thing keeps me with DCX. I simply like delay function.  I tried aligning tweeter physically 28mm but it sounds a bit "nasal" . I also tried waveguide , that's bit smoother but it has some horn diffraction I really hate 

Moby said:

the nearfields I like is my new ones  ;D

Click to expand...

ZD5's, look very pretty. I think there was a thread on building them at diyaudio.

ZD5's, look very pretty. I think there was a thread on building them at diyaudio.

Click to expand...

Well I started from ZD5 but ended with something else  Tonny Gee was kind to help me design completely new crossover.
I also forget to mention that I experimented with DCX on this speaker too. Some transients are missing again 

My new crossover is not directly related to the application being discussed here, however the circuits are similar.
It "sounds" better than  using the built-in LP filter in the sub's amp.
I put my sub/satellite 12dB crossover in it's case. with a relay muting circuit for no power-on/off "thump".
Being 12dB, the sub (low pass) is inverted. It is extremely quiet.
At some point, I may try (build) another, with a 24 dB slope.
Now I will tune the room by speaker placement, etc.

Pics....

http://i5.photobucket.com/albums/y177/Midiot/DIY/DSCN3584.jpg
http://i5.photobucket.com/albums/y177/Midiot/DIY/DSCN3576.jpg
http://i5.photobucket.com/albums/y177/Midiot/DIY/DSCN3582.jpg
http://i5.photobucket.com/albums/y177/Midiot/DIY/DSCN3583.jpg
http://i5.photobucket.com/albums/y177/Midiot/DIY/DSCN3586.jpg

=FB=

Here is another one on the net...."Moamps" / Pass

http://www.passdiy.com/pdf/articles/phasecrx.pdf
http://www.passdiy.com/gallery/hi-lo-xover-p1.htm
http://www.diyaudio.com/forums/showthread.php?postid=319300



=FB=

Moby said:

ZD5's, look very pretty. I think there was a thread on building them at diyaudio.

Click to expand...

Well I started from ZD5 but ended with something else  Tonny Gee was kind to help me design completely new crossover.
I also forget to mention that I experimented with DCX on this speaker too. Some transients are missing again 

Click to expand...

Reagarding the XO you tried with the nearfields using DCX was it similar topology (3rd order I guess if this is the one you implemented: http://i52.photobucket.com/albums/g8/mobidik/ZD5_nearfield_alternative.jpg )? Well, passive driver has to deal with real impedance of driver so you better measure the FR & phase of that nearfield crossover first (electrically directly from drivers, not acoustic response) and then reproduce that in DCX.

Also you could try this one (3rd order Bessel xo) with the mains:
http://www.rane.com/note147.html

Many main monitors (in big studios) use horns and compression drivers for better transient response. It may be that even though the response looks good the soft dome tweeter can't do it with real stuff. I think you could try the metal dome SS 980000 (http://www.tymphany.com/d2904_980000) instead the SS 970000 you use (I assume D2905/D2904 have similar motor and need minimal changes in crossover). Also shallow waveguides like the ones Genelec uses should create no problems with diffractions (they are not real horns).

Reagarding the XO you tried with the nearfields using DCX was it similar topology (3rd order I guess if this is the one you implemented: http://i52.photobucket.com/albums/g8/mobidik/ZD5_nearfield_alternative.jpg )? Well, passive driver has to deal with real impedance of driver so you better measure the FR & phase of that nearfield crossover first (electrically directly from drivers, not acoustic response) and then reproduce that in DCX.

Click to expand...

Yes, that's the crossover  I tried to simulate that in DCX as I mentioned but I tried to copy the electrical curve from simulator 
Also, that was just a quick test just to confirm my previous conclusions.

Also you could try this one (3rd order Bessel xo) with the mains:
http://www.rane.com/note147.html

Click to expand...

No, Thanks, will try 

Many main monitors (in big studios) use horns and compression drivers for better transient response. It may be that even though the response looks good the soft dome tweeter can't do it with real stuff. I think you could try the metal dome SS 980000 (http://www.tymphany.com/d2904_980000) instead the SS 970000 you use (I assume D2905/D2904 have similar motor and need minimal changes in crossover). Also shallow waveguides like the ones Genelec uses should create no problems with diffractions (they are not real horns).

Click to expand...

Sorry, but why I need wavegudie since I aligned tweeter digitally? Or you are talking about in case I switch to analog crossover? And yes, tweeter choice can be the problem. To be honest I don't have great results on Genelec's (I'm talking about big ones, small are toys  ), but that's very subjective since some people have the results. I liked the PMC's and they use soft domes http://www.pmc-speakers.com/imgResize.php?img=images/products/147cea987a0098_imgResizeCAPC89SZ.jpg so I tried to follow that . But it's possible that 970000 is too smooth for my ears. What do you think about ceramics dome tweeters? I never heard them but I listened the diamonds and I liked  ;D

Freq Band said:

Here is another one on the net...."Moamps" / Pass

http://www.passdiy.com/pdf/articles/phasecrx.pdf
http://www.passdiy.com/gallery/hi-lo-xover-p1.htm
http://www.diyaudio.com/forums/showthread.php?postid=319300



=FB=

Click to expand...

I was thinking about MOX crossover, but that discrete opamp looks kinda funky. Not sure.... Did you tried that?

Moby said:

I was thinking about MOX crossover, but that discrete opamp looks kinda funky. Not sure.... Did you tried that?

Click to expand...

No, but I'm thinking it might be something to try. http://www.delta-audio.com/Active%20filter%20two.htm


Although not as flexible as the MOX , Borbely Audio sells kits (boards + parts)
http://www.borbelyaudio.com/eb1102215b.asp


=FB=