[Tyson]

The Journey
Recently I completed a set of DIY speakers, the Ella's, which you can see here:

http://www.audiocircle.com/index.php?topic=73155.0

Before that, I was running VMPS RM40's.  With a passive crossover for a while, but actively for the last year I owned them.  Moving to active was a big improvement in dynamics, bass, and clarity. 

When I built the Ella's, I simply re-used my active setup.  Once I got them assembled, I ran them with the same crossover points and slopes that the designer built for them.

They sounded terrible.

Taking Control
It's clear to me now, that they were designed for a big room, being an MTM-WW 2.5 way design.  The baffle step compensation wasn't as-much needed in my room, so the lower midrange and upper bass was overwhelming.  The woofers were spaced too far apart, and lobing errors and non-smooth tweeter transition was a result. 

I stuck with the MTM design for a while because I wanted to like it, it was an elegant solution to a lot of the problems inherent to speakers.  At least on paper.  In the real world, I never could get them to sound right.

Going 3 way
So, I ditched the MTM design and went with a MT-WWW 3-way design.  I was finally able to get the mids and tweeters fully integrated.  But now there was another problem - lower mids were weak, and mid-bass was incredibly boomy. 

You see, the previous 2.5 way design had baffle step compensation built in, and the low woofers used a shallow crossover slope, so they were playing way up into the mids, and reinforcing those freqencies.  When I went to a true 3 way, the lower woofers no longer did that.  So everything above 360hz (my bass-mid crossover point) was incredibly weak up to about 900hz.  Everything below 360hz was boomy as hell and all-over-the-place due to room interactions.  The biggest spikes was right at the 360hz crossover point, and between 170 and 180hz (again, a room mode). 

Fixing the Bass
To fix it, I notched out the 180hz mode with a -8db eq setting.  I tried the same at 360, but it was hard because I had the crossover there and I didn't want to move it.  So, rather than use EQ, I simply moved my midrange crossover point up to about 380, and my bass crossover point down to 320.  This "gap" actually allowed the drivers to stop re-inforcing each other (and the room mode) and gave a perfectly flat response with no EQ needed. 

Mids and highs
Finally, the other thing you should know - every driver has frequency resopnse anomolies.  Add to that the fact that putting it in a box screws it up even worse, and you have a lot of issues, particularly in the midrange.  For example, the Peerless Nomex driver I use for mids is supposed to be relatively flat up till 3khz.  But in reality it's response drops like a stone at 1.5khz.  So your tweeter better be able to cross at 1.5khz, or you'll have a saddle in your FR in the upper mids/low treble.  MOST 6.5" drivers will have a serious drop in this area, due to their size, especially if you measure even a little bit off axis. 

Plus, it has a major peak just before it drops off.  And a major drop in response below 900hz due to baffle step issues.  How to fix?  Well, I tried damping down the peak with EQ but it made everything sound muffled.  So I figured I'll try it from the other side - deal with baffle step first, then come back to the peak.  So, I dialed in a 12db/octave boost for the mids, starting at 750hz.  This brought the 750hz-360hz suckout back in line. 

Looking at the response graph, it was getting pretty smooth at this point.  Only major thing left was that now there seemed to be a deep "saddle" dip between 800hz and 1khz.  So, I dialed up 5db of EQ boost at 900hz with a Q of 5 (Q defines how wide or narrow your boost goes - 1 is very wide, 10 is very narrow).  Result?  Almost perfectly flat response from 35hz to 16khz in room!  Trust me, that is very, very rare.

The Lesson
Why have I told you this story in such detail?  So you can see what goes into making a decent sounding speaker.  There is no "magic" in crossovers, they are just a means of getting a well tuned speaker.  A speaker with a passive crossover has undoubtedly gone through a measuring and incremental improvement process similar to the one I describe above.  The big difference is this - with a passive speaker, you have no control over how your speaker interacts with your room.  Having EQ ability and a means of fine tuning crossover settings gives you a much better chance of getting outstanding sound in your room.

Plus, if your speaker uses a 6 inch midrange, you are going to have serious issues between 1khz and 2khz, particularly off axis.  Most tweeters are not robust enough to cross over below 2khz, unless using a steep filter.  And steep passive filters are very expensive, and usually sound bad.  Active crossovers allow you to use a steep filter without that same degradation, and it keeps your signal phase-coherent to boot. 

Exceptions - If you have a 5 inch midrange, that pushes you almost to 2khz for a good crossover point, and a 4 incher is just over 2khz.  Also, if you have a waveguide on your tweeter, it allows it to have more power response down low, and thus a lower crossover point without stress.

Why go active?  Control.  With the right tools (such as a DCX2496, modded of course, an inexpensive mic, and an inexpensive USB mic-preamp), you have the ability to tune your speakers to your room to a far greater extent than "moving them around" would ever allow.  Plus you get better dynamics, better bass, better detail, and a more coherent sound than almost any passive system is capable of.

One other bonus - you can swap drivers very easily, assuming they will fit in your cutouts.  I already swapped a Peerless HDS tweeter for an SB Acoustics Ring Radiator, and the new tweeter is much more to my liking.  I think I might try a Scan-Speak Revelator woofer, or maybe even an Accuton woofer next for midrange duties.  You can't exactly do that with a passively based system. 

You might say, "Oh, but the cost of having 3 amps!".  I'd say that you don't have to spend as much on each amp in a 3 way system.  You can get a low powered amp for the tweeters, a moderate powered amp for the mids, and an inexpensive higher powered amp for the woofers.  And because the direct-couple with the woofers, they don't have to be insanely overbuilt like amps driving passive speakers do.  That means they can be cheaper.

Maybe active isn't for everyone, but maybe think about it the next time you are thinking about "upgrading" your current speakers or current amps.

[Tyson]

And, so we don't lose sight of the fact that it's all about the music - I sit here (not long after getting the above settings finally "dialed in") listening to the Britten Violin Concerto, a piece that never appealed to me much before.  But, I'm hearing interactions, melodies, passion, things I'd never heard before.  I'm enjoying it immensely.  It was a lot of effort to get to this point.

It was worth it.

[JimJ]

I learned the ways of the active setup in car installs - it'd be easier to win the lottery than find a pair of passives that are ideally set up for a vehicle's interior.

My dream setup at home involves going active

[Tyson]

The average audio room is not much better, particularly for bass.

[macrojack]

Watch yourself, Tyson. You could get booted right out of here if you continue to address the cause of the problem. Much of the activity here is fueled by band aid sales. Tweaks and mods and padding and cables and isolators and insulators and so on and so forth. In some circles science is trusted less than witchcraft.

But I may as well confess ... I agree with you completely.

And another benefit of active crossovers is they make your speakers more efficient and therefore able to operate with less power' Passive crossovers are a notorious power sponge.

[drphoto]

Tyson, do you think the 'killer app' for an audio company would be a high performance but affordable digital crossover/eq unit? In other words, an 'audiophile' version of the Behringer unit. Seems like this could open up a world of opportunities, especially to use pro-sound drivers. It seems like some smart programmer could figure out a way to use feedback from a mic in the room to auto set the parameters to get a flat (or close) eq.

Even if this processor was fairly pricey, seems like you would not need (as) expensive amps and speakers. Of course, sources still need to be of high quality.

It seems like more and more people are beginning to realize that in order to achieve realistic playback, the key is dynamics. I agree w/ you that active amping is a good solution to this issue. 

BTW: what is the basis of the mods done by MikeG and TomS to the Berhinger? Is it the usual upgrades to power supply, replacing jacks or what?

[Tyson]

drphoto - when I was at RMAF, the Emerald Physics people were developing exactly this type of technology.  If/when they come out with it, it should be a killer app. 

The mgalusha mods to my unit were extensive.  I lost the post that describes everything that was done, but maybe Mike will be by and see this and can fill it in.

[lonewolfny42]

Try these links....

http://www.audiocircle.com/index.php?topic=47777.msg427921;topicseen#msg427921

http://www.audiocircle.com/index.php?topic=50790.msg455491;topicseen#msg455491

http://www.audiocircle.com/index.php?topic=57499.msg509831;topicseen#msg509831

http://www.audiocircle.com/index.php?topic=57836.msg511055;topicseen#msg511055

http://www.audiocircle.com/index.php?topic=59508.msg530400;topicseen#msg530400

And one that was for sale...now sold...

http://www.audiocircle.com/index.php?topic=60908.msg544468;topicseen#msg544468

[JimJ]

Tyson, do you think the 'killer app' for an audio company would be a high performance but affordable digital crossover/eq unit?

Already out there. Check out the DBX DriveRack (not the PA version).

Rane makes a nice one too, but it's pricier.

Not sure how "audiophile" approved they are...but does it matter

[mgalusha]

The DCX I built for Tyson is not like any of the other I did. It does not have the Linear Audio analog board though in other respects is very similar.

For the analog output's Tyson's uses 1/2 of a National LM4562 per channel and no coupling caps. It's configured as an impedance balanced output. The nice thing about this is the original output board can be used though with some rework since the original circuitry is not used. I also changed the levels to be more suited for consumer use to try and maximize the available dynamic range.

The analog input is on a separate daughter board using a another LM4562 as a buffer feeding some Burr Brown diff amps to generate the differential signal needed for the AD converters. There is another bit on the board to provide the offset voltage for the A/D converters.

His does have the Oettle input receiver/clock/SRC board and the 4396 DAC chips.

Mike

The mgalusha mods to my unit were extensive.  I lost the post that describes everything that was done, but maybe Mike will be by and see this and can fill it in.

[macrojack]

Already out there. Check out the DBX DriveRack (not the PA version).

I have the Drive Rack PA and mostly like it although I have two quibbles. One has to do with a strict need to shut down and fully discharge both amps before I shut off the DRPA. Failure to observe this caution results in a rifle shot through the speakers.
The other has to do with a faint rushing sound that emits from my horns whenever the gear is on even if the preamp is muted. The dbx seems to be the cause. It is faint, as I said, but it is quite evident at night when the rest of the house is silent.

So naturally I think about upgrading this unit for something that does the same thing better. Why is the Drive Rack PA not recommended? I just read that it has been replaced by the PA+ model that has soft tart so that would take care of one problem. What would I gain by moving up in the Drive Rack line? Their higher end models are very pricey.

[Coytee]

For biamping (or triamping by using two units) I rarely hear people talk about the Electrovoice Dx38.  I've had one for several years now and recently bought 3 more off Ebay.  They needed a new display.  While I was talking to an authorized service center (who deals with various brands) I'll just say that he came down pretty negatively on some of these other brands (dbx & Behringer).  He was talking from a service perspective I guess....  I don't know. 

Regardless....  one might also look at the Dx38 if you have biamp needs (2 in, 4 out all XLR) and if you have triamp needs you could still use the same unit but you would need two of them using one for each channel.

Interestingly, doing that would also give you a 4th output where you could run a line to a sub...?

[Kevin Haskins]

I agree with your conclusion, active speakers are better but disagree with many of your premises.   

Active is nice because it allows for more design freedom, it allows you as the designer to choose the appropriate amp rather than letting consumers pick one through trial and error and it allows you to have more control over the final product.   It also gives you more tools to work with low frequency issues that are impossible or very expensive to deal with in a passive design.    It gives you more effective power, makes matching transducers easier and it is easier to implement complex crossovers.   An opamp, few resistors and caps are typically a $1 item per opamp side, a passive filter is a large inductor ($10), large cap ($5) and 10W sand cast resistor.    If you are looking at the cost of the network, active is actually cheaper.   

Speaker design is more complex than putting drivers in a box, and picking a crossover though.   The measurement and design process is too complex for such simple solutions.    Active or passive, the design of a loudspeaker requires multiple measurements, on multiple axis and it requires a knowledge of the acoustical properties of the transducers.    It is best left to those who have the equipment and experience/knowledge if you hope to get good final results.   The idea that you can mount up your favorite driver, place a measurement mic in your room and let the DSP/computer figure out the proper crossover is a pipe-dream that ignores several basic principles of good design.

In terms of the dispersion properties of a 6.5" driver, it does narrow sooner relative to a smaller driver.   With a traditional dome, with no directivity control, you get a hole in the off-axis measurements that has been problematic to deal with.   You can, with directivity control match up a 1" device to a 6.5" device without a problem and alleviate the need for doing a 3-way.   

Two way....  20deg. off-axis.



45 deg off-axis



70 deg. off-axis




There are other examples where some of the better design teams are doing the same at Harmon, Paradigm etc...

[Tyson]

Actually, I think your post re-inforces exactly what I was (trying) to say in my first post.  Looking at your graphs, the crossovers are at 1.5khz, which is pretty much where it has to be to avoid beaming with a 6.5 inch driver.  This results in a smooth, even, transition between drivers, particularly off axis, which IMO results in better, more consistent sound.  Paradigm in particular has long been an exponent of the importance of this.

My contention is that it's easier (and sounds better) to cross a tweeter this low using a steep active filter than it is to use a passive one.  Most manufacturers recommend not crossing 1 inch tweeters over below 2khz, particularly if you are going to use a shallow 1st or 2nd order slope. 

For me, the other key point is baffle step - in passive systems, the compensation has been built, or "tuned" to some hypothetical room.  With an active system, you can take direct control and tune it yourself. 

But you are right, building a speaker from scratch (even an active speaker) is not an easy task.  Good, accurate measurements with the drivers in the box, and knowing how to interpret and use those measurements, is critically important.  But, the nice thing about an active system is you don't have to worry about passive parts like caps and coils causing impedance changes in the drivers and thus altering their FR.  That simplifies things considerably.

[drphoto]

Kevin was probably referring to my simplistic proposal to simply let a 'smart' crossover figure out settings for itself.

[Tyson]

Ah, he's probably right about that.  It might get you part of the way, if you are starting from scratch, but I do think you need to know something about how the drivers work and speaker design.

[Kevin Haskins]

Actually, I think your post re-inforces exactly what I was (trying) to say in my first post.  Looking at your graphs, the crossovers are at 1.5khz, which is pretty much where it has to be to avoid beaming with a 6.5 inch driver.  This results in a smooth, even, transition between drivers, particularly off axis, which IMO results in better, more consistent sound.  Paradigm in particular has long been an exponent of the importance of this.

My contention is that it's easier (and sounds better) to cross a tweeter this low using a steep active filter than it is to use a passive one.  Most manufacturers recommend not crossing 1 inch tweeters over below 2khz, particularly if you are going to use a shallow 1st or 2nd order slope. 

For me, the other key point is baffle step - in passive systems, the compensation has been built, or "tuned" to some hypothetical room.  With an active system, you can take direct control and tune it yourself. 

But you are right, building a speaker from scratch (even an active speaker) is not an easy task.  Good, accurate measurements with the drivers in the box, and knowing how to interpret and use those measurements, is critically important.  But, the nice thing about an active system is you don't have to worry about passive parts like caps and coils causing impedance changes in the drivers and thus altering their FR.  That simplifies things considerably.

I'm in full agreement.   

The above is actual measurements, but a simulated crossover.   The design is active and it is much easier than doing it passive.   I also have some EQ for the bottom-end, and a subsonic filter to prevent overexcursion of the midwoofer which is impossible to do with the passive design.   You could also do a limiting filter or other tricks but I like the KISS principle. 

[Kevin Haskins]

Kevin was probably referring to my simplistic proposal to simply let a 'smart' crossover figure out settings for itself.

Yea... I'm not a big proponent of them simply because it takes some knowledge of the devices, and acoustics to get good results.   They conceivably could be a good tool, but in-room measurements, even with infinite signal processing are less than desirable to do good design work with.    You have to look at the devices on multiple axis of measurement and it helps to understand what wiggles are related to what.   Some of those wiggles in the response are things we should leave alone, others you have to deal with via changing cabinet design (a resonance in the cabinet) and things like baffle diffraction, which show only on-axis should be left alone.    The computer doesn't know all of this.   At some point, you could design a program and system sufficiently complex to be able to deal with all of those things, but it will never be cost effective because the market for such a widget would be too small.   It is cheaper to hire a human for the few times you actually need to design something.   

What is currently on the market is woefully insufficient for the average guy to get good results with. 

[Housteau]

Tyson, do you think the 'killer app' for an audio company would be a high performance but affordable digital crossover/eq unit? In other words, an 'audiophile' version of the Behringer unit. 

BTW: what is the basis of the mods done by MikeG and TomS to the Berhinger?

I have to agree that once I went back to actively triamping with my new speaker system a whole new world opened up for the better.  I am fortunate to have one of Mike Galusha's modded Behringers.  The work done was extensive and inclusive of the Linear Audio analog board.  That basically turns the DCX into a system preamp where both the input and output levels for the channels can be adjusted in the analog domain.  I don't think enough has been said about the preamp section itself, which I find to be quite transparent, especially using the analog in jacks.  I just run my line level sources directly into it and I find it better than going straight digital in.

Even the unmodded Behringer DCX sounds pretty amaizing and transperent when allowing it to do both the A/D and D/A functions by connecting line level sources to it.  The downside of the unmodded units is the need to attenuate the output levels going into your amps so the input levels can be increased to max out the resolution of the digital conversions. 

A friend of mine has used one off and on experimenting with fairly resolving speakers, such as Martin logans, Von Schweikert etc.,  and we cannot tell it is in the loop except for the positive attributes and potential it brings with it.  I guess it could just be that all the positives so outweigh the negatives that this little non-audiophile unit can really do the job, even though it doesn't cost a fortune.  At its price it is basically a disposable piece that can readily be replaced.  Build quality is not great on stock units and that is probably why the cost is so low.

[Tyson]

The other cool thing is that if I ever feel like dropping in a different midrange driver or a different tweeter, I can do that without issues, as long as they fit the cutouts I have in my boxes.  And as Kevin indicated, you can easily put in a subsonic filter to prevent over excursion of your drivers, which really opens up the ability to use the drivers in a sealed cabinet, which to my ears always gives cleaner and tighter bass.  On my speakers in particular, since I did the wiring of the drivers in such a way that I can go from my current M-T-WWW config back to an M-T-M-WW config very easily.  There's a LOT of flexibility that the DCX offers you, both from a driver, and from an amp standpoint.  Toss in the ability to dial in almost any EQ curve and any frequency (and multiple frequencies), you've got an insanely flexible solution to a host of (normally ignored) problems.