Active crossover

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

michal_k

Well-known member
Joined
Nov 12, 2008
Messages
370
Location
France
Hello everybody,

I'm considering a new diy speaker build. After a while I decided on a well known zaph zd5 build. But since I have four channels of more than decent icepower modules, I figured out it would be cool to make it an active speaker. Now, how do I redesign this:
ZD5-crossover.gif

into an active system? I know it's not the simplest crossover ever designed, cause I need few notch filters and a delay element.

I'd be grateful for any suggestions and reading material.

cheers,
Michal
 
I've seen simpler, but passive crossovers are generally a best trade-off compromise. With active circuitry you can make something far better.

Blank sheet of paper time... and start researching different active filter types, etc.

JR

 
Ok, so I checked out some active designs and that's what I don't get so far:

Zaph's crossover does much more than just HP and LP. He employs some notch filters to smoothen drivers impedance and uses a delay network for phase coherence. How to do that in active crossover? Because I still have to, right?
 
Well, there's no such thing as "have to", as long as you like the sound.
Which may, or may not be the case with a well designed crossover.
Unfortunately, there's more to good sound than a flat frequency responce...
:-\
Mr. Linkwitz has everything there is to know about active filters on his website - http://www.linkwitzlab.com/filters.htm
 
Hi,



  most of the digital crossovers available now have very flexible filters and additional processing allowing delay(phase) compensation, and some bands of eq. Even the Berhinger is extremely flexible. I would recommend trying ia digital crossover first, discerning exactly what you want it to do, and the n implement as an analogue cross-over if that is what you want to do. I suspect you will be really happy with a digital crossover. XTA is fantastic, but not cheep.

  You might also want to consider Hypex and Digimoda plate amps, with all the dsp youll ever need built in.


    https://www.hypexshop.com/


  http://www.digmoda.com/

  Kindest regards,


    ANdyP
 
michal_k said:
Ok, so I checked out some active designs and that's what I don't get so far:

Zaph's crossover does much more than just HP and LP. He employs some notch filters to smoothen drivers impedance and uses a delay network for phase coherence. How to do that in active crossover? Because I still have to, right?
The LP is a first-order at ca. 560Hz.
The HP is a slightly damped 2nd-order Butterworth at 4.2kHz; the actual difference between it and a true Butterworth is at lower frequencies (below 2kHz) which does not significantly matters.
The notch filter cuts about 2dB at 950 Hz.
Obviously these filter do underlap. Not really uncommon, to compensate for the rising response of a woofer and the midrange boost many tweeters exhibit.
The all-pass filter (L4, C7, L8, C6) provides ca. 40us of delay at 4 khz, equivalent to 14mm.
 
Hi Andy, thanks for the links, but I think I'd rather go full DIY way, just for the fun of it :)

The LP is a first-order at ca. 560Hz.
The HP is a slightly damped 2nd-order Butterworth at 4.2kHz; the actual difference between it and a true Butterworth is at lower frequencies (below 2kHz) which does not significantly matters.
The notch filter cuts about 2dB at 950 Hz.

I found proper circuits on the website linked by jackies so it will be easy to build. And I'll avoid using inductors which is a good news.

Obviously these filter do underlap. Not really uncommon, to compensate for the rising response of a woofer and the midrange boost many tweeters exhibit.
The all-pass filter (L4, C7, L8, C6) provides ca. 40us of delay at 4 khz, equivalent to 14mm.

So it's 1.5 of period of delay. Again, linkwitzlab has a nice solution of delay element that can be stacked to achieve 540deg phase shift.

Am I thinking right?

 
HI,



  Fair enough, but you still might think about trying a digital crossover FIRST, before you try to build it analogue. You can at least work out exactly and simply what you need your crossover build to DO. It will be much easier than trying to alter each and every parameter in situ.


  Best of Luck, and please keep us updated on your progress.



      ANdyP
 
michal_k said:
Obviously these filter do underlap. Not really uncommon, to compensate for the rising response of a woofer and the midrange boost many tweeters exhibit.
The all-pass filter (L4, C7, L8, C6) provides ca. 40us of delay at 4 khz, equivalent to 14mm.

So it's 1.5 of period of delay.
One period at 4kHz is 250us. I fact the AP filter is tuned to 2.35kHz, which may well be the actual ACOUSTIC crossover frequency. At this point, it gives 70us delay. [/quote]Again, linkwitzlab has a nice solution of delay element that can be stacked to achieve 540deg phase shift. [/quote] Typical active all-pass filters are almost infinitely stackable. If I understand well, the Linkwitz board can be populated with two 1st-order all-pass filters in series, which would be correct.
 
The suggestion to beg,borrow,or steal a DSP based crossover to experiment with is not crazy.  Ideally you could invest in some computer based measurement system and really nail this, but hat has too many opportunities for a novice to mess up.

Passive crossovers by necessity are simple. This is good and bad. Simple low order crossovers have less phase shift associated with them to cause problems with combining between bandpasses, so the signals knit together a little easier. The downside is that slopes are more gradual so keeping bandpasses separate is harder.  Also EQ to correct individual driver response in passive crossovers generally involves throwing away level, since passive eq can only cut and not provide boost.

Just because you can, is no reason to make the crossover more complex than it needs to be. With active electronics you can make a very good "and" simple crossover. Being able to experiment with different crossover approaches by pushing buttons on a DSP, sounds very attractive to me.

Human memory for sounds is very short... Without building up multiple duplicate crossovers, it will be very hard to give them a fair comparison by simply listening between separate builds.

JR

PS: I would resist the temptation to go for a steep 4 pole Linkwitz-Riley, because it looks good on paper, and consider something simpler. Perhaps a derived (one filter subtracted from unity to make the other output), or look at staggered pole, underlapped alignments, where you can keep the crossover order low for modest phase shift with some good frequency dividing.  This is far too complex of a topic to pretend it is this simple, but if you are willing to spend some time, you should learn and have some fun.
 
How about Douglas Selfs latest book "The Design of Active Crossovers" ??. From what I have seen, very
comprehensive.

Frank B.

 
Thanks again for all the answers.

You're right. DSP is a good starting point. I think I'll just use the four outputs of my audiofire and do all the processing on a computer. That way I won't have to invest money in any outboard DSP units. One of the reasons I resisted this solution so much is a bad memory of a university project. I was doing a crossover based on IIR filters with flat phase response(on iirs!). Everything written on motorola 56k in... wait for it... assembly.

Measuring system may be a good idea, this:
http://www.linkwitzlab.com/sys_test.htm#Mic
looks like an easy and cheap thing to build. My plan looks like this now:
A) reading,
B) building boxes & reading,
C) digital crossover & reading,
D) tuning everything to the room,
C) analog version of what i've done before

huh, wish me luck
 
I've found a nice write up on computer dsp systems, pasting here for future reference:
http://www.computeraudiophile.com/content/Getting-started-digital-crossovers-using-Mac

Paper about l-r filters:
http://www.linkwitzlab.com/JAES/jaes_papers76.htm
 
To make accurate measurements of speakers you either need sophisticated measurement gear that can ignore early reflections from nearby surfaces (TDS), build an anechoic chamber, or measure it outdoors. So be careful with those measurements.

4 pole, L-R alignment is good for steep slopes and good driver protection, useful for high power sound reinforcement but IMO fmore than you need for more modest Hifi applications. I lean toward lower order (less poles) crossovers, but I am just repeating myself now. 

I wasn't suggesting that you code your own DSP  ;D...  and I would resist using digital filter types that can't be replicated with simple opamp circuits, unless you are willing to stay in a digital crossover.

JR
 
Really depends on what you'd like to accomplish.

The Revelators do not have huge breakups like Al cones do, but without corrections you'd find them somewhat mid forward. For regular Hi FI some find it musical (whatever it means), but if used for mastering monitors then definitely needs to be addressed. Also the response will greatly depend on the enclosure.

If I were you and was set on the active then I'd consider something like this:

http://www.minidsp.com/

Indeed!--why would I want to make a filter/notch for every freq. when for peanuts I could use something that powerful!!!

It would give you a great ability for response corrections, so you might want to go a bit higher with the crossover point. The XT25 is probably a fine ring radiator tweeter, but again, for Hi Fi. For studio use you would want something with better dispersion. With DSP the 3/4" tweeter might look like a more sensible idea, and depending on the arrangements you make and the power handling needed, you could look at OW1, or a cheaper Seas 22TAFG. On the other hand, it all depends on what is your goal.

The ZD5 is a good design and if you are not experienced with speaker building then I'd suggest just make those as they are. There is nothing crazier than trying to voice a speaker... trust me on that... those things will drive you nuts... unless you are nuts to start with...

Best, M 



.
 
Hi,


  having looked further, minidsp ROOLS . . . .I have a number of poweramps, and a digital crossover plus eq is what I NEED. I dont want to pay oodles, and the solution is right here. Hope it sounds good too . . . .
 
Err..h.  If you want to replace a passive xover with an active one, you are re-designing the speaker.  Good speakers today not only apportion the signal between bass & treble but apply substantial & important EQ to the units.

If I was doing this (as a former pseudo Speaker / DSP / Analogue guru) and assuming the original speaker was good and worth "improving", I would

1  Build the original passive speaker "as-is" as accurately as possible with the correct units.
2  Measure the crossover responses with the units in place & in the correct boxes.  This is not amenable to SPICE as the impedance of the units is complicated and also affected by their mounting.  (There are speaker crossover design progs which allow you to do this but they still need the complex impedance plot of the mounted units)
3  Dream up IIR filters to match this to some accuracy.  eg "Simple Arbitrary IIRs - Lee, AES San Francisco, 2008"

This would give you most of the advantages of an active speaker but not all.

Trying to match analogue active filters instead of an IIR is a serious PITB.  Active filters are much more complicated and inaccurate than passive LCR types to achieve a given complex response.
_____________________

With all this rigmarole, you might as well re-design the speakers from scratch and open yourself to the fun that Marik promises.

For this, you need to make pseudo anechoic measurement.  I strongly recommend http://www.cliowin.com/lite.htm

I've also heard www.audiotester.de is useful but not tried it.  In any case, read the Clio manuals for the best explanation of pseudo-anechoic measurement.

You also need a measurement mike which is where clio scores as it's calibration is one of the few I believe.
 

Latest posts

Back
Top