[audiojerry]

I guess that's why I've never personally heard a good active system: I'm too young!  

[nathanm]

Whaddaya mean Jerry, I thought you liked the Mackies?  Thems be fully active!  

There's a difference between the fun of tweaking and the fun of buying new stuff and trying it out.  Like a 3-way DIY active system would require the end user to do all the leg work.  Find the right gain for each driver, the right volume balance, polarity, crossover point, slope...you would have a wealth of different sounds at your fingertips - but it might also drive you mad.  Whereas with a store-bought active system you assume someone else already jumped through all those hoops to get the speaker to sound correct.  

When I actively biamped my Norh 7.0s there was a big difference between that and using the passive XO board.  The best XO frequency seemed to be what the passive was supposed to be - 2K but wow could you really get a range by sweeping that frequency; you move that dial and your ear is picking up on this weird swishy phasey sound and you just don't know what is "right" - it was kind of maddening.  You kind of lose your bearings as to what the music should actually sound like.  A slope adjustment would've been cool, but my crossover didn't have that.  Basically I said screw it and went back to the passive board.  

I'd hate to see what would happen if I had that digital XO of Behringer's.  I would probably have a nervous breakdown.  But if you truly enjoy tweaking and not just buying new things to try, going active will DEFINITELY get you into tweak heaven.  The crossover controls alone will provide hours of fun, but then throw in different amps...cripes you'll be lucky if you ever listen to a song for fun again!  Heh!

[audiojerry]

Whaddaya mean Jerry, I thought you liked the Mackies? Thems be fully active!

You're right, Nathan, sorry. In the context of price considerations, the Mackies offer a nice, compact solution to good sound. I did like the Mackies. What I should have said was, I never heard an active system that I'd want for myself.

[TheChairGuy]

I, like most of you, have heard some great speakers at every price point.  I'm not sure if I ever heard 'active' speakers before my little experiment.

My memory doesn't allow for A/B comparo all the time, but I got real live-like music coming from those cheapo Radio Shack LX-8's.  For what it cost to set it all up, it is really outstanding.  It sounds like music...what more is there to be said?  

The Norh MultiAmps simply don't exhibit the problems encountered in passive role....they are passably good in active and were not satisfying driving the Radio Shacks (okay, abysmal passive crossover likely here from factory), Vandersteen 1C, Camber 5.0Ti, and Quad 11L's...the last sets of speakers that have been used in my home .

I still got some tinkering to do, but I am simply more pleased with the sound than I thought I'd be for the modest cost.  I'm sold on the idea of active crossovers...at least for DIY 2 ways.   It is a more cost effective way to achieve higher fidelity.

[WerTicus]

how is tweeking a crossover any different in active than in passive? its not really is it?!  actually its easier because of the elimination of impedance problems.

[DVV]

how is tweeking a crossover any different in active than in passive? its not really is it?!  actually its easier because of the elimination of impedance problems.

It's easier because you don't have to mess around with high power components, and in particular, because you don't use wire-wound inductors. Instead, you have electronically simulated inductors.

However, while easier, it's just as full of pitfalls, depending on the actual construction of the electronic XO. Remember, this is not a matter of swapping one capacitor and/or resitor, you typically have to change several of them at once for proper operation. If, for example, you want the change any crossover point, you have to change the components for both the high pass and low pass section, so that they continue to coincide.

Cheers,
DVV

[DVV]

...
When I actively biamped my Norh 7.0s there was a big difference between that and using the passive XO board. The best XO frequency seemed to be what the passive was supposed to be - 2K but wow could you really get a range by sweeping that frequency; you move that dial and your ear is picking up on this weird swishy phasey sound and you just don't know what is "right" - it was kind of maddening. You kind of lose your bearings as to what the music should actually sound like. A slope adjustment would've been cool, but my crossover didn't have that. Basically I said screw it and went back to the passive board.

That is the problem of buying a universal crossover board; what you describe is what I have also heard once with such a setup.

In my view, the actual crossover points should not be adjustable, but this of course assumes you have bought a ready made system, where as you say, somebody has already jumped through most of the loops. Obviously, the XO slopes would be fixed as well, but this could still afford the designer to use different slopes in any combination he chooses to use (e.g. 6 dB/ocy for bass to mid XO and 12 dB/oct for the same spot for the mid from bass, etc). In my experience, this is where many of the potential problems with otherwise superior electronic XOs arise, and is best avoided by fixing points and slopes.

However, I would seriously consider a full parametric equalizer for each driver, because active or passive, the key problem of reducing room effects is still there all the same. That plus gain for each driver.

Obviously, this would jack the price up by a fair share (as I wouldn't be using run-off-the-mill pots, but would go for some quality products from people like Bournes, Alps, Noble, TDK, etc). Yet I feel this is the best solution overall, of course, with an electronic bypass control.

Cheers,
DVV

[WerTicus]

im going for a fixed point 24db / oct electronic crossover... im just trying to decided where i want those points fixed before i start rather than have to change it later

obviously there is a good chance ill be changing it later anyways but ill try to get it right first time!

[DVV]

im going for a fixed point 24db / oct electronic crossover... im just trying to decided where i want those points fixed before i start rather than have to change it later

obviously there is a good chance ill be changing it later anyways but ill try to get it right first time!

24 dB/oct? Are you sure you really need them that steep? Remember, even in electronics XOs, higher orders still carry phase penalties.

Just asking ...

Cheers,
DVV

[WerTicus]

well its actually technically twin 12 db /oct filters so it adds up to 24db oct

....

though i suspect i need this phase issue thingy explained somewhat!

[DVV]

well its actually technically twin 12 db /oct filters so it adds up to 24db oct

Er, um, ... it doesn't work out quite that way. If your say bass driver is cut by 12 dB/oct on top, and your say mid is cut by 12 dB/oct down, what you end up with is a 12 dB/oct, or second order filter.

You have 24 dB/oct only if your filters are in a straight series, one after the other.

In my view, 24 dB/oct is excessive. While it will of course filter very well, it will also tend to introduce quite a bit of phase shift and could well inhibit dynamics (sort of slowing down fast passages).

though i suspect i need this phase issue thingy explained somewhat!

Well, in very loose terms, each and every filter ever made by anybody will, by default, introduce some phase shift. As you know, phase shift is changing the time relationships of the output signal as compared to the input signal. Essentially, capacitors tend to do that rather well simply because their impedance is non-linear (VERY non-linear), but they could also be introduced by poorly designed electronics (essentially, electronics with a small open loop response, which need much feedback to linearize them).

Subjectively, we tend to hear phase shift as a blurring of the sound, as its going out of focus, and the first to die is low level spatial information, what usually gives us what we call "air". If really overdone, we will tend to say something is wrong, the sound is unreal, even unnatural, but that bad is rare these days.

The greater the filter cut rate, the more capacitors you need to use, and consequently, the greater the opportunity for more phase shift. Also, the charge and discharge time of more capacitors tends to lengthen, which then tends to slow down the delivery and mess up the sound dynamics.

Please understand, this is all in very loose terms, I'm trying to describe what is essentially very subjective - however, also quite measureable.

Which brings us back to filters. Obviously, one is faced with a dilemma: use steeper slope filters for better driver coupling (less driver extension outside its designated range) with phase and dynamics problems, or use less steep filters with better dynamics and less phase shift, but also more driver range overlapping.

In my view, 12 dB/oct represents about the best overall compromise between the two, but I must add I have heard some really good electronic XOs which used 6 dB/oct for driver top cutoff and 18 dB/oct for mid lower filtering - in other words, a combination of two different slopes.

This is where the art part of the deal comes in. The designer must have an excellent practical knowledge of the drivers he is using to be able to make the appropriate choice, and using 3X drive also helps because the critical 800-3,000 Hz range is usually covered by a dedicated driver. My choice were drivers from Son Audax, but there are other drivers from other manufacturers which are just as good, and some probably better. The reason why I opted for Son Audax is because I happen to like their arguably softer sound than most others, but another reason was because of their uncommonly linear impedance modulus, which makes them easy to drive and allows me to concentrate on other aspects of the amps more than on sheer brute power. Lastly, I have yet to hear bass drivers made as well as Son Audax makes them (this one at least, 10" long throw, plastic over paper, 7 lb magnet, die cast frame, real rubber surround).

It seems to me this is the key part of designing a good active system - you need to use just as good drivers as you would in a passive design (i.e. no cost-cutting because you have direct amp drive with no passive XO), you need to know the drivers you work with and you need to select them very carefully. The electronics part is then fairly straightforward, but not cheap if you go for the max.

For example, you need to decide how to supply your amps with power - use one single supply line for voltage and current gain stages, or use separate (slightly higher for voltage gain stages to compensate for natural voltage drops) lines, regulated or not, and, most important, will you regulate the output stages as well? This will lock them into an almost ideal operating state, but it practically doubles your outlay for the power amps. And it could well open the door to pure class A amplification, done by the book.

Remember, your mid and tweeter drivers will be very happy with just 30-40 watts, since they are by default efficient and there is no power-robbing passive XO in between.

Decisions, decisions, ...

Cheers,
DVV

[WerTicus]

oh okay dvv i understand how this phase shifting can occur now.

the filter does have two 12db in series.... though there would be no reason why i couldnt skip over the last filter for the various stages and make it a 12 db oct device.

or possibly (i have yet to actually look into it) make it into a 4 way crossover

here is an image of the plans for one channel of the kit active xo im in the process of working on.  Would this design constitute junk in your book?



as you can see i could remove the last filter op am in each stage for 12 db if i wanted to... also note that the cap values are for the default xo points and can be changed for whatever feq i want.

You can see the mid range driver gets a high pass and low pass put onto it so its effectivly getting twice as much phase shifting as the bass and tweeter! ugghhhhh oh well the kit only cost me $50 but if you think it might be workable with 12db then i can still use it i suppose!

[DVV]

oh okay dvv i understand how this phase shifting can occur now.

the filter does have two 12db in series.... though there would be no reason why i couldnt skip over the last filter for the various stages and make it a 12 db oct device.

Well, I didn't delve too deep into the schematic, but I don't think you can simply drop out a part of the circuit like that, some modifications would probably be needed to keep it in balance.

or possibly (i have yet to actually look into it) make it into a 4 way crossover

Technically, you could, no problem, but I would not advise it. With good speaker selection, a 3 way will prove to be the best overall compromise between conflicting requirements.

Where I would spend a lot of time if I were you is in finding a mid range driver capable of LINEAR reproduction in the 800-3,000 Hz range, and if at all possible, going up to 5 kHz. This would eliminate any crossover in the most critical range of our hearing, where our ears are naturally most sensitive and where say 70% of all the action is.

I'd look over the offer from people like Son Audax, Vifa, ScanSpeak, Peerless, Morel, Focal, etc. I would bypass Dynaudio because of their bass drivers - to me, they sound far too plasticky, but as ever, this is very subjective.

here is an image of the plans for one channel of the kit active xo im in the process of working on.  Would this design constitute junk in your book?
...

Well, I wouldn't be caught dead using TL0xx series of general purpose op amps in any audio patj, save perhaps a DC servo (essentially, an idiot circuit dealing with whatever below 1 Hz, but even so, I'd go for temperature stable LF411).

My choice of op amps is described in a text posted on http://www.zero-distortion.com , where you will also find the data sheets for all relevant chips.

Since you are dealing with relatively high voltages (0.5-1 V), noise is not too crticial, so you don't need any ultra low noise op amps, which also tend to cost quite a bit. My choice would be (and is) AD826AN, which in my view has just the right set of compromises, but above all, which sounds right.

Another problem you should consider is how to get the signal from the crossover to the power amps. Basically, you have three choices:

1. The amps sit next to the XO, and you have long cables to the speakers (which means the amps reside outside the speakers),

2. The amps are located in or on the speakers, which implies long runs of balanced cables to the speakers, and

3. The XO sits very near the speakers, or is in them, in which case you need just one cable run from the preamp to the XO.

In view of the fact that this will be probably something like 2-3 meters (7-11 ft), I would strongly advise use of balanced lines - NOT provided for on your schematic. Balanced is far more immune to airborne induced noise than standard RCA lines, and it seems rational to me that if you lay out the bread for an all active system, it deserves the best you can give it.

How will that actually look in your system?

Cheers,
DVV

[JohnR]

Is that the Jaycar kit?

[WerTicus]

yeah thats they jaycar kit....

... its obviously not using anything but the cheapest parts in it - i was going to modify it for better parts anyways but since i dont want 24db oct i might not even try.

though i dont see why i couldnt just remove those last filter stages dvv.. i rekon it would work

i still have not decided on the orientation though i know i will be using that sliced paper cone scan speak 2.5inch midrange driver which is flat from 300hz to 6 khz or so
or maybe that 4.5" inch kevlar one for mid (13M/8636)... its pretty much the same specs really but with higher power handling

[sfdoddsy]

I'm not sure that you should dismiss 24dB slopes so blithely. A 24dB Linkwitz Riley is phase accurate, and 12dB slopes have their own phase problems. If you are a phase freak you might as well stick to 1st order slopes.

However these have their own compromises. Your drivers will need to well bahaved over a much wider range, and you may have problems with over driving them if you play at all loud. I'd rather have a few minor phase problems than the sound of driver being overloaded.

I use 24dB L/R crossovers in my avtive speakers, as do many others. The reason you seldom see them in passive speakers is that they are difficult and expensive do passively, especially for bass.

Cheers

Steve

[sfdoddsy]

I'm not sure that you should dismiss 24dB slopes so blithely. A 24dB Linkwitz Riley is phase accurate, and 12dB slopes have their own phase problems. If you are a phase freak you might as well stick to 1st order slopes.

However these have their own compromises. Your drivers will need to well bahaved over a much wider range, and you may have problems with over driving them if you play at all loud. I'd rather have a few minor phase problems than the sound of driver being overloaded.

I use 24dB L/R crossovers in my avtive speakers, as do many others. The reason you seldom see them in passive speakers is that they are difficult and expensive do passively, especially for bass.

Cheers

Steve

[WerTicus]

ahhconflicting information!

why does it always have to be a trade off

i want perfectly coherant and power handling!    and 99db an octave!

im actually looking into a digital crossover like the behringer ULTRADRIVE PRO DCX2496

http://www.behringer.com/02_products/prodindex.cfm?id=DCX2496&lang=eng

[DVV]

I'm not sure that you should dismiss 24dB slopes so blithely. A 24dB Linkwitz Riley is phase accurate, and 12dB slopes have their own phase problems. If you are a phase freak you might as well stick to 1st order slopes.

To be able to even discuss this subtopic, we need to agree on what point becomes the true limit, i.e. at what cut (- xx dB) the output of any driver becomes immaterial. Filters are conventionally declared at their nominal -3dB points, but it's fairly obvious they will not drop dead after that point, but will go on working for at least a bit more. Well, if you happen to be using a Smith-Papoulos filter, with its -370 dB/oct you really could say the only point of interest is the -3dB point, but these are generally used in RF communications.

I would propose that we view the -8 dB point as the last of interest, since that's the point loudspeaker manufacturers like to use when quoting frequency responses. Since this is not a voltage, but a power figure, -3 dB is in fact half of nominal and -8 dB is really 15.87% of its nominal output.

However these have their own compromises. Your drivers will need to well bahaved over a much wider range, and you may have problems with over driving them if you play at all loud. I'd rather have a few minor phase problems than the sound of driver being overloaded.

All filters ever made by anybody are a set of compromises. Each type has its pros and cons. For example, a Chebyshev filter has a wonderful cut rate, but it produces a small rising "hump" just before it goes into full action, a fact exploited by Technics (to name but one) to extend their tuner frequency response up to 17 kHz at just -0.5 dB, but keep it down 39 dB at 19 kHz (pilot tone). Obviously, for that particular purpose, this type worked best.

In crossovers, I would always think Butterworth or Linkwitz-Riley, because these two have shown themselves to be the best overall sets of compromises.

As for slopes, I do believe 18 dB/oct, or third order, has shown itself to be the best overall set of compromises between phase linearity and cut rate. So, at 18 dB/oct, if my driver is cut at -3 dB at say 800 Hz, doubling the frequency adds the filter cut, so by 1.6 kHz, it is -21 dB down. This is exactly 126 TIMES (!) down - so what overload are we talking about? What overhang?

As for overloading the drivers, if this should happen, it will be the result of:

1. overdriven drivers (do this in any configuration and bad things happen);

2. Poorly selected power ratings of the drivers used in comparison with the amps installed;

3. Poorly selected power ratings of the amps used, which could lead to catastrophe if any amp should clip.

Lastly, regarding phase shifts from crossovers. The cure for this is usually simple, assuming there are no fundamental design flaws - use high quality parts, and use fast op amps. Fast in this context is of course relative, but you could find that some op amps, despite their modest specs, do surprise (e.g. OP27) in terms of sound quality. However, I prefer to use AD826AN for its sound quality and speed, in my view an uncanny combination brought into one.

Cheers,
DVV

[sfdoddsy]

There's always conflicting information in audio. That's what makes it interesting.

 

I understood you were being recommended 12dB slopes, hence the question mark with the drivers. 18dB would cause feewer problems, and 24dB even less.

Of course, you should match your crossover to the drivers rather than the other way around. I use a digital crossover which allows me a fair amount of flexibility in the chice of crossover points andslopes (6dB to 48d).

I ended up with L/R 24 because they measured and sounded the best, as well as being those recommended by the man who inspired the speaker.


Cheers

Steve