Active Crossover for the Green DIYer

[snmhanson]

I have built exactly one pair of speakers plus a center channel, all based on existing plans here at htguide. I am hopefully going to soon start on my second paid in the near future, once again based on existing designs. However, I would eventually like to start building my own designs. However, in all honesty I don't think I'll be able to find enough time to learn the nuisances of passive crossover design. I have read a bit about active crossovers and it seems that it could be a way that I can create my own designs without worrying too much about ending up with a pile of worthless caps and resistors. I am hoping someone can answer these questions to help clarify a few things for me though.

First, is active crossover speaker design generally easier than passive crossover design or is it just different?

Second, would someone with basically no crossover design knowledge or experience be able to figure out the basics of designing an active system in a reasonable amount of time? I understand there is a ton to learn but I am just talking about the basics here so that I could end up with speakers that I could stand to listen to.

Is there any danger when working with an active system of damaging the parts or, even worse, causing a fire or other damage?

What exactly do you need for an active system besides what would normally be required? Obviously the active crossover (whether digital or software based), an amp channel for each driver and some decent testing equipment. Anything else? As far as the amp, would a multi-channel Emotiva amp work well, or would it be too much power?

Finally, if I created a successful active design is there a relatively easy way to convert it to a passive design or would it require starting from the beginning for the passive design?

Thanks for any information. On the surface, active crossovers seem to be the ticket for people who aren't engineers or electricians to get into speaker design. If you don't have it right, as verified by your testing equipment and your ears, you just make some adjustments and try again. Or, maybe it's not that simple...

Thanks,

Matt

P.S. I also understand that cabinet design is part of the process and is a learning curve in itself. I would utilize winISD initially for that and be open to other options as well. If I decide to go down the active road that is.

[BobEllis]

Welcome to the world of speaker design. It's a path to obsession.

Active crossover design has all the issues that you have with passive design, except dealing with driver impedance. So in a way it is slightly easier. They are not the panacea that some make them out to be, but I find analog active crossovers easier to design and tweak than passive. Don't minimize the importance of measurement technique, either.

Crossover design has a learning curve whether passive or active. "Textbook" filters usually don't get you there, most drivers require a bit of EQ and then there are baffle step compensation and phase alignment. A good reference for active crossover design is http://www.linkwitzlab.com/index.html.

If you aren't up to learning to design analog filters, there are several dsp options such as the miniDSP which allows you to fairly easily program the transfer function you desire. I use frequency allocator from www.thuneau.com for prototyping but prefer to go analog for the finished project.

Many of us active crossover devotees use a large (30+uf) cap in series to protect the tweeter in case of mistakes in turn on sequence. You don't want a DC thump taking out your tweeter.

An Emotiva amp will be fine. You'll attenuate the tweeter level so that you don't overdrive it, but you will have plenty of power for transients. I run 150W amps for both woofer and tweeter without issue.

You'll know the transfer function of your active XO, but implementing it passively can be difficult because the driver impedance interacts with the crossover components. The other way around is much easier.

The DSP/PC based active options are the easiest to tweak, but there is still quite a bit to learn to get it right. It's a fun journey. Enjoy.

[Amphiprion]

I was going to make a post, but Bob pretty much covered it.

[Saurav]

FWIW, I'm using an Emotiva for my active speakers. I'll say that working with active digital XO's is easier than passive analog XOs. Not having to deal with the driver impedance is a big component of it. Also, being able to simply dial in delay on each driver makes phase matching a lot easier. Don't get me wrong, it doesn't mean you can just drop one in and make it work. But in my experience, working with something like the DCX2496 was a learning curve but doable, while passive XO design is a black art which I've pretty much given up on ever mastering.

You will need measurement capabilities, and ideally, simulation / optimization software. Figure a few hundred bucks for just software and hardware tools by the end of it, if you want to really get into it. But the entry level cost is a lot lower, there's good free software that will do a lot of what you need, just not as conveniently.

There's also the advantage of being able to re-use everything on your next design. No more looking at your stash of inductors and wishing you had one that was just a slightly different value. Although I suspect once you've built enough passive XOs, maybe that ceases to be a problem

[---k---]

I thought this post was going to be about saving electricity. ops:

[BobEllis]

Guess my higher powered A75s don't qualify. I needed that Ryan, thanks.

[snmhanson]

Thanks for all the posts. Am I now going to be known as the socially conscious DIYer? I guess I should have said novice DIYer but I thought green would better describe my complete lack of crossover understanding. To set the record straight, I'm no tree hugger (not that there's anything wrong with that). I AM, however, a complete novice in DIY speaker building. Semantics...

Initially, Bob's post was half English, half Greek to me. However, I did a bit of research and think I have a bit of a grasp into what he is saying. The other posts helped as well.

The miniDSP looks like a promising option for me. Not too expensive and fairly straight forward to program. If I understand it correctly, you program the curve and parameters on a computer and upload it to the DSP, right? So the downside would be that you can't do on the fly tweaking. Or, can you have a computer hooked up and make adjustments as you are using the DSP? Where do you get the software for a miniDSP setup? Does it typically come with the unit or do you go out and find you own compatible software? Also, the miniDSP has 2 ins and 4 outs so I would need two for stereo 3-way. Are there any comparatively priced 3-way stereo DSPs out there or would I be stuck getting two miniDSPs?

Stepping up (or is it sideways) to a hardware based unit such as DCX2496 appears that it would give me more tweakability but also be a bit tougher to get it right. Still, it would be really easy to experiment on the fly and see how making certain adjustments effect the sound. Overall, though, the hardware based DSPs are similar to the software based DSPs other than the interface, right? (DSP and active crossover are the same thing, right?)

As far as the simulation software eluded to in Saurav's post, what would be an example of that and what is the advantage over just hooking everything up and tweaking directly with the active crossover? Does it have tools to help you optimize the rig so that you don't spend hours tweaking? Does it allow you to simulate driver combos before you go spend the money on them? I know there's a purpose, it's just not totally clear to me yet.

I can see that speaker building can be an addiction. I just took my first hit and I am coming back for more. At least it's a healthy, or at least not unhealthy (if I take the mdf dust out of the equation) addiction. Plus, I'll learn a lot in the process and be able to impress my friends and relatives.

Thanks for the help,

Matt

[Saurav]

I've just very recently started looking into miniDSP. Read through their FAQ and forum, some of your questions are addressed there. In their opinion "DSP is just DSP" is incorrect, and they list some advantages of their approach over the DCX. Yes, you'd need 2 boards for a stereo 3-way setup. They sell software "plugins" for $10 each that enable various kinds of functionality. It looks like the plugins have a fairly easy to control UI, but I've only seen screenshots. The DCX will also have issues with gain levels in a home system (it's designed for pro audio), so you'll need some extra steps to deal with that. I use an Emotiva LMC-1 as a 6-channel volume control downstream of the DCX.

Personally, I'm not sure you're stepping up from miniDSP to DCX2496. I've considered replacing my DCX with miniDSP, but I'm still researching.

Regarding optimizers - you can download a demo of LspCAD and play with it. Basically, your workflow is something like:

* Measure the drivers in your final speaker / baffle, import those measurements into LspCAD

* Tell LspCAD "I want a 250Hz LR2 and 1800Hz LR4 slopes on my midrange, and flat in between", ditto for tweeter and woofer

* 'Design' a crossover in LspCAD where you drag and drop logical / functional blocks, so you build up a circuit with an HP, an LP, throw in some notch filters roughly around where you see driver resonance peaks, some shelving filters to handle baffle step

* Then run the optimizer, and it tries all combinations of all values, trying to work with the filter / EQ blocks you specified, and your driver's response, to generate the final acoustic response which matches your target. FWIW, you can use LspCAD to optimize passive XOs too, in that case you'll design a circuit with resistors, caps and coils. This is quite fuzzy. For example, sometimes it'll move it away from the resonance spike I'm trying to address, so I have to lock that down, i.e. tell the optimizer not to change those values. Stuff like that

* Once the optimizer is done, take those values, input them into the DCX, take another measurement to see if it looks right

* Repeat

It'll probably be clearer if you download the demo and play with it. But like I said, there is a non-trivial learning curve.

Take a look at this thread, it covers some of what you'd need to do. penngray was going through the measurement software, and I was helping him with the crossover calculations.

[snmhanson]

Thanks for the posts Surav, answered several of my questions but brought about twice that many. I'll download the demo lspCAD software and play with it a bit as well as taking a look at your suggested reading. I am leaning towards the miniDSP as it seems to be the "simplest" of everything I've seen so far and a great way to get my feet wet. I also need to get some measuring equipment going and am thinking about the Dayton mic and TrueRTA. I better keep myself in check though - I am also in the midst of studying for the CFP exam. As tough as it may be to put it before speaker building, that has to take precedence. Still, life can't be all work and no play.

Matt

[Saurav]

HOLM Impulse
ARTA (demo version is free)
Room EQ Wizard

Add the Dayton mic, and an appropriate preamp / soundcard, and you have a very competent measurement setup. Look around, there are lots of threads on what hardware / software you need to get a measurement setup going.

There's really no point in investing in anything else until you have a decent measurement setup and learn how to use it correctly. IMO.

[cjd]

ARTA is a tremendous piece of software, and the free version does everything but let you save data - enough to let you learn a LOT about measuring before dropping any cash on the software (though you need hardware already...) - and it's (relatively speaking) affordable, too.

[Saurav]

You can export individual FR measurements with the free version of ARTA. It's not convenient, but it's doable.

[Thooms]

I went the DIY route - I built a pair of Rod Elliott's 4-way LR4 boards - I figured they'd be ample for any future projects. A lot cheaper than the pre-made options too - at the expense of ease of setup (you're swapping resistors rather than doing it on a computer - not a big deal IMO).

Depends on how handy you are with a soldering iron I guess, but they're a straight forward and enjoyable build.

The active approach is very appealing - plus a DIY active crossover can be built for the price of a complex passive network.

Another benefit is for prototyping - especially for those of us who aren't lucky enough to have a gazillion Ls and Cs knocking around to try stuff out with.

[Saurav]

For me, the main issue with boards like that is that you can pretty much only implement textbook crossover filters. I could be wrong, but does it allow any EQ? For example, combine a shallow notch filter with a textbook filter to get a complex transfer function? If you pick your drivers really well, it's possible that all you'll need are standard filters, but look at any of the completed projects here, and you'll see that the transfer functions are usually more complex than that.

I've seen boards that have filter and EQ sections, so that would take care of my concern. Maybe this one has that as well, and I just didn't read it carefully when I'd looked into it.

[Thooms]

This is very true - one of the reasons I still haven't got around to building any speakers yet is that I want 'easy' drivers to work with

There is some provision for a kind-of EQ on this board, but it's nothing more than broad-band response shaping - no notches or anything like that. It's not a substitute for expensive testing, but as a speaker-building noob it should allow different things to be tried reasonably easily.

Notches/all-pass filters are going to be a stripboard job with this one

A small EQ could probably be assembled reasonably quickly too.

[Paul Spencer]

Matt,

There's nothing to stop you coming up with your own completely original designs. A good passive design is great for simplicity, but if building different designs becomes a hobby then active makes a lot of sense. MiniDSP costs a similar amount to a passive crossover, but you can use it in any number of designs - no need to solder each one.

Do you want to get your feet wet? Grab a MiniDSP and do an active version of what you have now.

You'll need a basic measurement setup:

http://redspade-audio.blogspot.com/2010/08/basic-measurement-setup.html

Quickstart measurement tute:

http://redspade-audio.blogspot.com/2010/12/how-to-measure-your-crossover-with-holm.html

Guide on designing your own box:

http://redspade-audio.blogspot.com/2010/04/how-to-us-winisd-pro.html

You don't need as much power with active, but headroom doesn't hurt either. As a very rough guide, a tweeter only needs 15% as much power. So if you had 100w driving a fullrange speaker, about 15w is now fine. In a HT with an active sub, the mid to about 80/120 might need about 40%. So just about any amp has enough power. If you have a beefy Emotiva and find you need more channels, you can just add some low cost modest amps for the tweeters.

I was involved in some comparisons with DCX and MiniDSP. We found that a modified DCX could not beat the MiniDSP. Who expects the $100 options to match the $1500 one?

You can read about it here:

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

If I were you, I'd start with MiniDSP. It's a low cost way to get your feet wet. It's slightly less convenient that you need a PC to change settings, but you really do need to hook up your PC for measurements with DCX anyway.

If you want to do things like use digital in, MiniDSP is more user friendly. DCX can do it, but you need AES otherwise you get serious clipping issues. I had to attenuate my Bluray player to about 3% digitally to get it to work.

Another thing you can do. You could actually buy a nice box, then box together MiniDSP + power amps, have shorter cable runs and make it neater. In a 2 channel setup you could add a diy preamp and some power amp modules and set up an active system with source plus one box.

[Deward Hastings]

What everybody said <g> except I think active is *significantly* easier than passive.

Re. miniDSP . . . my first take was "it's too easy" . . . I kept trying to find ways to make it more difficult (and couldn't). Having been up the learning curve for the DCX may have helped, but not all that much. Even the bi-quads are fairly simple. The "programming" software comes with it . . . $10 each for the various available modules (which are themselves pretty flexible). You buy one with it, and more later if you need.

For simulation software there's always the neglected "right hand side" of the "Passive Crossover Designer" spreadsheet . . . which pretty much covers the basics of active designs (a lot of people use PCD . . . they apparently never try the right side, or they wouldn't build passive crossovers).

If there are any "design tricks" to making your intro easy I'd say start by adjusting your drivers to flat an octave or two past crossover (ignoring breakup peaks and nulls), then notch those peaks (and, where appropriate, notch, don''t boost, the nulls), and then apply the crossover. That way you can easily change and compare the crossover frequency and slope, and because you're already working with "idealized" drivers phase takes care of itself (after you dial in the acoustic center delay adjustment).

For amplifiers use whatever you want. In your final design you might want to use different amps for different drivers, whatever best suits. At some point in the design phase you decide your target maximum SPL . . . that and the sensitivities of the drivers determines the power needed for each. One of the many advantages of active design is that impedance and sensitivity don't matter . . . it greatly increases your range of driver choices. Using lower power amps and tweeter protection is not a bad idea when experimenting . . . "mistakes were made" is the story of my life . . .

Anybody want a box of "coils and caps" ? ? ? (actually just kidding . . . I'm going through the coils and 10 mfd "Madisound specials" pretty fast using them as noise filters in power supplies . . .).