[JDUBS]

I haven't seen anyone say this.

With the DEQX, its about as easy as it gets....which is pretty easy.  Takes a little time but its not a difficult process.

-Jim

[Tyson]

These should get your started off on the right foot:

http://www.hifizine.com/2010/12/prototyping-4-way-open-baffle-speaker-with-the-minidsp-2x4/

http://www.hifizine.com/2011/03/refining-a-4-way-open-baffle-speaker-minidsp-2x4/

[PRELUDE]

Any suggested starting points?  My random internet googling, etc., only tipped me off to the idea that to realize any significant improvement from bi-amping, I would need to do this, but there is very little I was able to find on how...

This isn't something I am likely to rush into, but wouldn't mind doing at some point.

I do want to clarify one thing, though.  Both amps are identical (Class D audio 254) and are vertical (i.e., one amp board for the left speaker, the other for the right). I was under the impression that this was the "safest" way to do bi-amping in terms of matching gain, phase, etc.

http://sound.westhost.com/bi-amp.htm

[wgscott]

Thanks.  That is actually one of the things I had found and read, and if you don't know what to  look for, it is easy to miss that there are implicit complications, eg:

1.  The resistor(s) in the low-frequency crossover are there in part to compensate for different efficiencies in bass vs. midrange/tweeter, and for every 3dB difference in efficiency, you are dissipating half of the input power as heat.  Eliminating this eliminates the wasted power/heat, but now requires adjustment of the relative gain for the bass amp vs the treble/midrange amp to compensate for that resistor's elimination.  At least that is what I now gather.

2.  The passive crossover shifts the (relative) phase 180°

[YoungDave]

How does (did) this setup compensate for differences in amplifier gain?

A "dumb" active crossover permits level adjustment, typically by variable resistor, and permits selection of crossover point by either variable resistor or selected resistor value.  So, implemented with a simple active crossover, one could replace the passive system with an active system with very similar parameters.  I did use 24 dB/octave LR crossovers on my VMPS system, in place of the lower-order passives, but I kept the same crossover frequencies.  This made the transition simple until I was ready to step up a notch (or ten) into a DEQX.

I made a box with 6 crossover networks and a good power supply, so that it would accommodate the 3-way speakers and a subwoofer.  I crossed at 55, 225, and 5900 Hz, if I recall correctly, and used 10-turn linear pots with turn counters for the level adjustment.  The linear pots give me a lot of precision compared to audio taper, and 10 turns of it at that.  Level adjustment makes a BIG difference, and it is a lot easier to discern that when adjusting in real-time from the rack instead of back-and-forth from behind a speaker.  Not to mention, speaker pots do not give 10-turns of adjustment - the whole range encompasses about 2/3 of a turn, which is pretty coarse.

Bottom line, I had an easy-to-implement active crossover with the same crossover points as the original speakers, readily adjustable levels, and a very low learning curve.  Cost me maybe a couple hundred bucks but I did build the thing myself, so add a few hours' time value.  I still keep this crossover as a spare in case one of my DEQX's breaks.

The DEQX is about an order of magnitude more powerful and at least an order of magnitude harder to implement.  It is a big-time leap, and I certainly endorse it, but a modest approach as described above is easy and still offers a large improvement over passive.

I have pics & schematics around somewhere if anyone is interested.

And yes, to get back to the original question, I certainly concur that one should remove the passive components from the circuit, for reasons explained in another post.

[Pez]

ALL crossovers shift phase. The least offensive are 1st order active. The most offensive are steep slope 48 dB or higher passive. Steep slope actives do nasty things to phase as well. Yet people love to use them in active for reasons I don't understand.

[wgscott]

Yet people love to use them in active for reasons I don't understand.

To minimize intermodular distortion?

[wgscott]

I have pics & schematics around somewhere if anyone is interested.

Yes, please.  And if you can recall where to get those pots...

[PRELUDE]

Here is a new active crossover for all needs. www.accuphase.com

[Pez]

To minimize intermodular distortion?

Usually speakers measure flatter with steeper slopes, but I don't like the way steep slopes sound. In the end that's what matters most IMO, not measurements, but to each her (or his) own.

[JohnR]

Usually speakers measure flatter with steeper slopes, but I don't like the way steep slopes sound. In the end that's what matters most IMO, not measurements, but to each her (or his) own.

I would like to counter the implication that a preference for steeper slopes is because of the "measurements." I've done both and in my system/setup, prefer the sound of the steeper slopes.

[Pez]

Interesting, I have heard many speakers I like that have steeper slopes, but I can never seem to get my setup to sound as good with them.

[YoungDave]

I have pics & schematics around somewhere if anyone is interested.
Yes, please.  And if you can recall where to get those pots...

There is a pic of my unit & a schematic in my gallery at http://www.audiocircle.com/image.php?id=8635. 

I bought the pots and turn counters from Mouser.com.  It has been several years, and I do not have the part numbers, but their online catalog is pretty easy to search.  I had the front panel made for me, to my specs, by front-panel-design.com.

I really liked Dr White and his approach to this business, and regret that he closed up shop.  His kits are no longer available.  I really do not want to post any of his copyrighted material.  However, instead of etching your own boards and trying to build a White Noise Audio crossover like mine from scratch, you can get complete units, kits, or just boards from several sources, including Marchand http://www.marchandelec.com/xm9.html, AmpsLab http://www.ampslab.com/crossover.htm, and Audio-Kits http://www.audio-kits.com/catalog/item/4071180/3884480.htm.

I have not totalled up the cost, but a few hundred $ for the crossover is probably not going to be the main factor - the cost of amplifiers will probably be considerably more.

My main point here, trying to not get to deep into minutiae, is that there are simple, i.e., non-dsp, means to get into active crossovers, including DIY, and the results can be superb.

By all means drop me a PM if you want any more DIY details and I'll try to help.

[JohnR]

I'm sure I am being too simple-minded, but if the active crossover has the job of sending everything above 150 Hz to the higher-frequency midrange and treble, and everything at or below 150 Hz to the bass, what would the passive crossover do to each of these signals, since they have already been split?  In other words, if there is nothing left for them to process, why wouldn't they simply pass the signal, relatively unscathed?

I don't think this question has been answered yet. A crossover is not a "brick wall" filter - it has a gradual transition, which occurs on the "pass" side as well as the "stop" side of the nominal crossover freqency. For example, here is the frequency response of a first-order filter:



Here is a steep filter - you can see the transition is much sharper but it's still there:



If you leave the passive crossover in, your active crossover can only "add" to this kind of response i.e. make the slopes steeper. If your crossover frequency was not the same as in the passive crossover, then you would also have to boost one output near the crossover frequency to try and counter its effect. That would be problematic. With the same crossover frequency, I think it probably is doable but you would need to be aware that the response you get from each driver includes the response caused by the passive crossover. If you are OK with using steeper slopes (and I see nothing wrong with that) and are using a digital crossover (with choice of lots of different slopes), you might be able to get it to work well. However, it would still most likely be "better" to remove the passive crossover. At the least, it might be worth a try just to get started and see how it goes, use it as a way to learn about how to set up an active crossover, before deciding to remove the wiring from the drivers. At that point you might also decide to go full 3/4 way. You may decide that rather than destroy the resale value of the speaker you would rather go active with something DIY. That is all assuming that you can measure as mentioned by others.

[JohnR]

I wondered about the idea of going from (DAC-generated) analog input, back to digital, and then back to analog again, in terms of degrading the sound quality.

"Everything" degrades sound quality It's a matter of making tradeoffs / optimizing within constraints. Paul Spencer has documented some listening tests / comparisons of various digital crossovers (the only ones I'm aware of) -

http://redspade-audio.blogspot.com/2010/12/active-crossover-listening-tests.html

With regard to the miniDSP specifically (can't comment on others as I don't use them), there is a digital I/O card available for it.

[JohnR]

Interesting, I have heard many speakers I like that have steeper slopes, but I can never seem to get my setup to sound as good with them.

To speculate about that for a minute, there are a couple of factors that could be considered. One is the dispersion pattern, the other is different types of drivers. With a shallow slope, you have a much more gradual transition between the drivers; with steep slopes any "discontinuities" on either side of the crossover point will be more pronounced. I'd be inclined to consider the first a more likely candidate for investigation, in my speaker I "engineered" the baffle and chose the crossover points to provide the best match in dispersion patterns, so steep slopes didn't introduce a sudden change in dispersion pattern. On the latter, some claim that they can hear the differences between different "types" of drivers around a crossover point, which I've not personally experienced but if it is possible to hear then in your case the transition from the woofers to the planar mids may work against the steeper slope.

It's not that I didn't like the first order crossover, and it's impulse response was measurably cleaner, but the steeper one sounded clearer and more dynamic to me. I didn't spend any time trying to figure out why. It's quite possible that with different drivers (perhaps just the tweeter) I would have had a different preference.

[macrojack]

John R makes a good point. Why compromise the value of your B&Ws only to wring slightly more performance from mass-produced mediocrity?

If you want to find out what this is about, you can follow the DIY suggestions here. High performance drivers are your starting point. Suitable cabinets are next. These must be designed for your specific driver choice. Horns can be a good idea. Gedlee kits might speed you along. If, like me, you don't make cabinets or care to arm yourself with test equipment, you can find a DIYer who wants to move on to the next project and needs to sell a previous triumph. You will still need to make it fit your space.

With an active system, you are not dealing with hit or miss. If it isn't right at first, you can still make it so. And it needn't be intimidating at all. It sounds right if you like it.

I actually hired a professional sound guy to set mine up. He did it by ear and I'm very pleased.

[neobop]

All of the analysis of a passive crossover is a gross simplification, There are many other considerations. It's much more degrading than Pez has stated.

The 1st order Butterworth filter is considered minimum phase and constant voltage. Crossover sums flat at -3dB for each driver. This is at the area of summation. There's a 90 degree phase difference between the 2 sections over the rest of the response. This causes a 15 degree tilt in vertical polar response with drivers separated by a distance of 1 wavelength at X-over frequency. That would be -15 degrees for same polarity of the tweeter relative to woofer. The group delay is 0 for both sections at .16u sec. 

Minimum phase requires exact driver alignment. With 1/2" of misalignment the X-over is no longer minimum phase and group delay. Frequency resp has a dip of 2.5 dB in high pass (w/misalignment). Besides being sensitive to driver alignment the 1st order is likely to make the tweeter resonate within the stopband. Often tweeters with a low frequency resonance 2 octaves below X-over point, will resonate. I use this passive crossover and tweeter resonance is a problem. This requires tweets with an extraordinarily low res freq. For every other type X-over you trade off one aspect for another. A steeper slope will minimize problems of driver distortion within stopband and resonance, and be less sensitive to misalignment. However phase and group delay are all over the place depending on the type. There's fourth order Butterworth, Linkwitz-Riley, Bessel, Legendre, Gaussien and Linear-Phase (similar to Bessel).

You really need a computer model to combine 2 different slopes with the 2 drivers. In a 3-way it gets much more complicated. Passive crossover components seriously degrade the sound. Large value coils for 150Hz low pass are usually horrible iron core or huge air core inductors. Giant caps at that freq will also seriously degrade. I would suggest considering leaving the passive in place, disconnected or leaving it alone, but not using both.
neo

[wgscott]

John R makes a good point. Why compromise the value of your B&Ws only to wring slightly more performance from mass-produced mediocrity?

The $2K mass-produced mediocrity is the most expensive audio purchase I've ever made...

[wgscott]

There's fourth order Butterworth, Linkwitz-Riley, Bessel, Legendre, Gaussien and Linear-Phase (similar to Bessel).

I actually know what the last four of those mean, but I am still pretty much intimidated out of doing this.