Assuming one has the amplification available, going the DIY route with digital crossovers does seem very advantageous.
I have questions in doing so.
1) Would one still put in analog Baffle Compensation Circuits to aid in response? Does it make sense to do so?
2) How does this affect tweeter responses and blending on the lower end of its range?
3) Would it still be necessary to use analog shaping filters with the tweeters?
4) Or, is all of it unecessary as many of the compensation circuits are solutions to issues associated with analog crossover circuits?
Dear Annunaki,
First of all truth in advertising, I have no hands on experience using digital filters, but I'm getting increasingly tempted to get my feet wet.
Let me tell you what little I know.
This Australian firm have blazed the trail. I have watched their progress with interest.
http://www.deqx.com/DEQX-Linear-Phase-Digital-Crossovers.html
Their gear is very expensive, as is so often the case with those in first.
Behringer have entered the field with units at a tempting price point to play.
Meridian have had a high end system using digital crossovers and powered speakers for some time. It is expensive, but highly regarded. I have a feeling history will regard it as seminal product.
Now the advantage of digital crossovers is steep slopes with no phase, time or group delay issues due to the crossover. Phase shifts due to the inductive nature of moving coil drivers will still be with us, as will problems of time because of driver placement and the positions of their true acoustic centers.
Now as to your first question would diffraction compensation be necessary? I would say yes if small bass/mid drivers are used on a narrow baffle. How you do this in the digital domain I don't know, but I'm sure there is a way.
Now as far as blending is concerned that is the big attraction. Usually in analog filters either active or passive one is working with low order filters. The result is driver overlap.
The problem is then to blend the natural roll off of the driver slopes with the crossover slopes to sum to a flat frequency response. No easy task as driver slopes frequently change order, and then the crossover slopes have to be the inverse.
The next problem to solve is keeping the drivers out of their trouble zones. Because of the preoccupation with rigid driver cones, woofers hit the wall hard and generally have very unpleasant break up modes often in the 4KHz range. These frequently have to dealt often with notch filters and all the attendant phase problems. Ted Jordan debunked the issue of rigid cones back in the 60s, with elegant experiment and advanced math. The Jordan Watts modular loudspeaker was the result. The problem is very few have realized the significance of his work.
Now with the tweeter, the pass band issue is keeping the tweeter well away from its fundamental resonance. The same applies to a midrange, plus the same issues as a woofer. I personally like to have a tweeter 24db down at resonance.
The other issue with the large driver overlap using low order filters, is comb filtering. This is where the response is made irregular by adjacent drivers canceling at some frequencies and reinforcing at others.
Now to answer your question. If I were to use a digital filter I would use steep slopes and operate the drivers well in their pass bands. That would solve the issues I relate above.
As far as shaping the tweeter response, I would choose one that is very smooth above were it starts its lower end roll off and obviate the problem.
Now all of this it seems to me would make three and four way designs much less problematic. Good 10 inch woofers could be driven further up their operating range, and this would make this diffraction loss issue less acute, as well as keeping the mid range units well in their pass bands.
If the whole scheme is as as good as it seems, the issue is going to boil down to selecting the optimal crossover frequency for the selected drivers, driver placement and level matching. I just hope it is not to good to be true. However if drivers a selected with very flat pass bands I don't see why it should not solve a lot of thorny issues.
Now the selection of amps is important for the tweeter I feel. The tweeter will not like turn on thumps or any DC off set what ever. So I feel a an amp with a cap in the output would be mandatory. A good use for a nice old quad 303. Also I think there should be circuits to disconnect the tweeter and probably a lot of mid ranges at turn on and for a few seconds after.
I know most of what I have written will be old news to you, as I can tell you have wrestled plenty with these issues. However I think this is an important issue, and likely to bring about big changes in audio, that will ultimately involve all members of these forums. And I feel the educational aspect of these forums is paramount.
I hope I have answered your questions, even if in a roundabout way.