How are active crossovers so simple/clean?

[Ken L]

I'll use the Behringer unit.

The Behringer or the driverack pa, would probably be your best bet.

On another PC, I have the upgrades that Thorsten Loesch recommends for the Behringer unit - would of course require some soldering - however, should be what the unit needs to be top notch.

I can't say enough good things about the flexiblity of the unit. When listening, You can tell the difference between 2 or 3 hz crossover point -10 hz can easily be heard - when I set the delay by ear - (the sub drivers are 19 inches ahead of the mid\FR) the subs slowly seemed to move up to the center of the horns - presumably because you are hearing more overtones from the horn -

I am EQ'ing 1.5 DB in one place and 2DB in another -

If you had told me that being able to fine tune and tweak these things would have meant so much I wouldn't have believed you. It's like it just snapped the sound stage into a field with greater depth.

The combination of EQ, best placement and choice of crossover type and order - all combine with the Azurahorns and subs for very, very nice sound. And my drivers are FE206E's for 180 bucks a pair.

I believe that we are turning a corner with the digital crossovers that will change how we look at things in the future. Whether or not that turns out to be the case, it has certainly changed how I look at things

Regards

Ken L

PS if anybody is interested in those upgrades to the Behringer e-mail me and I will send them in a few days - I'm temporarily running on another box to debug it before returning it to it's owner - MS service pack 4 for Win2Kpro was causing some instability _frown_

 

[SY]

tcpip, my saga continued:

4. Both, but more passives than actives- the laws of statistics. I certainly did see a lot of active designs in Speaker Builder, I'd read a lot of Linkwitz's stuff, but for the basic layout of this particular set of speakers, my inspiration was more from Ken Kantor (NHT 3.3, for example) and Roy Allison's passive designs. With one big exception, I don't recommend going from a passive to active in already-optimized designs. The big exception is lower frequency crossover points, where the advantages of biamp/active are overwhelming.

5. My xovers were pretty specific to the details of my design, and without consulting my notebook, I'll have to be approximate: sub to woofer at about 80 Hz, woofer to mid at about 300, mid to tweet at about 3K. There's nothing magic about these numbers- they were chosen because of specific considerations of the drivers used and my desired dispersion. Looking at the mid/tweet transition, for example, I wanted the point high enough to not endanger the tweeter at moderately high SPL but low enough so I didn't get the common horn/trough characteristic in the horizontal polar patterns. Low is also good for avoiding woofer breakup modes.

6. I did a combination of in-room and outdoor measurements. The in-room ones included nearfield and a LOT of off axis measurements. Some with the room, some with the room gated out. The power of time-gating is one of the major reasons to use FFT/MLS.

7. I'm fairly clueless on using SPICE and similar sim software, so do as I say, not as I did. Use it. What I actually did was misuse my speaker CAD software to simulate the effects of various filter functions until I hit my target. Translating that into component values is just some algebra. My build quality was... spartan. Radio Shack perfboard. It works fine.

8. This was about my 20th speaker design. It was somewhat easier than the first few

9. I'd do the whole thing with Behringer or similar units if I were to do it again.

10. I've got about zero experience with 10 band units, so can't really answer that.

11. It sounded really, really good. Great definition and balance. If cost were no object, I would have spend more time on cabinetry, using more exotic materials, more modular construction, and had PC boards designed for a discrete XO circuit, probably using tubes and FETs.

And now I'm on to line sources, a whole new Universe to explore!

 

[sfdoddsy]

I use the dbx Driverack 260 digital crossover to essentially duplicate the crossover and EQ that SL uses for the Orions, although my speakers use different drivers so I had to modify everything, or at least measure everything to confirm the settings.

There's no way I could have done it without a digital crossover/EQ, I am just too clumsy and stupid. Another huge advantage is that you can do room correction for the bass, as well as tweaking the sound to your own and your room's preferences.

The big question, of course, is whether one of these will sound as good as a custom analog unit such as the ones SL designs, with the same drivers and cabinets. Given SL's knowledge versus mine, I would suspect the answer is no, but given how good my speakers sound, it would be close.

The area SL feels the digital units lag is in phase and delay relationships, that analog allows him to teak things more accurately. There is a section on this on his website but the math sails over my head.

The Behringer does allow you to play with delays, and the Driveraack allows you to adjust phase for each driver in 1 degree increments, so you do have some control.

I can't detect any sonic nasties from inserting the Driverack. I also plan on getting a DCX 2496 for my centers and surrounds, which use analog active crossovers and EQ at the moment and are a royal pain in the *** to adjust.

So my humble opinion is if you are going active, go digital.

Cheers

Steve

 

[wintermute]

Thanks. I know exactly what you mean. And I can assure you I'm a total newbie. I've just been doing reading up, that's all. The first speakers I ever made were a mono boombox using a TDA2020 amp and a single dual-cone FR driver in a ready-made particle-board box which used to vibrate and rattle merrily. This was in 1985

But perhaps we Indians are a bit more careful before spending for parts, because our disposable incomes are much smaller than those in developed countries, if converted to a common currency?

But the learning is fun. Specially if we can tap the wisdom of those who are ahead of us, like you.

Tarun

Well I guess we are almost identical total Newbes My current speakers are ones that have been modified over the years from some that were given to me in 1985 (chipboard) and the only thing thats original are the enclosures (minus the front baffle)..... I had been playing around with speakers in open backed boxes for a few years before that but nothing serious. Funnily I've only just joined this forum too, and have been doing lots of reading, and running into the same sorts of doubts, but in differing areas.

I guess in some ways we are not newbees, but in the theory side of things we are! It's a really steep learning curve and just when you thing you're getting on top of things some new mountain shows up . I think the important thing is to stop at some point and do something, Otherwise you will never know whether the sims are actually right! You may have a winner but you won't know until you try it.

Most of my anxiety is about whether or not I'm going to be throwing away money, hey if it doesn't work and I've spent the equivalent of about US$700 on drivers, I'm not going to be real happy! I can buy a nice 20mm Nikon lens for my SLR for that amount of money!! If the money wasn't an issue I would just go ahead, and if it didn't work rack it up to experience and try again,

I guess you need to reach a point where you are confident that it isn't going to be a total botch up, and you should at least get good value for money even if it doesn't fully meet your expectations. I guess you can always take SY's example and sell them to some of your friends and start on a new project!

Regards,

Tony.

Regards,

Tony.

 

[tcpip]

Re: ^ the reason why

i went to full range single driver horns.

Aha... Why do you think I bought those Jordan drivers? I'd been reading about single-driver speakers for a bit before that at melhuish.org and elsewhere.

I figured that anyone who has the space to keep horns and the skills to fabricate them would be in a different category compared to lowlies like me. So I tried box designs with the JX92S. And then I discovered that its highs roll off somewhat earlier than promised (because they failed to reproduce even tape hiss as well as the dome tweeters on my Wharfedales), and their bass expectedly rolls off early (I made a bookshelf-sized ported boxes). So I began to explore ways to add a tweeter for the very high treble (single cap XO, to come into effect above 6KHz, maybe above 10KHz). And I began to explore good clean bass options. Sigh... Here I am. Exploring crossovers. And plucking hair out. One by one.

I haven't heard horns, but in non-horn designs, I've not read, seen or heard anything which convinces me that a full-range single-driver speaker will satisfy my eclectic tastes in music (combination of Al Di Meola, von Karajan, Diana Krall, Dire Straits, Indian classical vocal, Wishbone Ash, Led Zepp, Beatles, Louis Armstrong, Hindi film music, Clapton, Cohen, GDead, and lots in between). However, I believe you guys have got the direction right... get one driver to cover as much of the spectrum as possible. Wish one driver could do it all. Without a horn. And without the expense of those Lowther drivers.

Tarun

 

[Vikash]

Hindi film music

Hehe, re-creating the atmosphere of a song that's been sung on 16 different hills in 4 countries cannot be easy.

But in all seriousness, I see how it would be difficult to reproduce
some of the live tracks by AR Rahman for example without extending all the way up to the very high frequencies. Not sure how well a single full-range driver would stand up.

And without the expense of those Lowther drivers.

I feel Steve on your case any second now...

 

[ronc]

Full range single driver horns.

I have been designing and building speakers for over 30 years( yea i am an old fart).Been the multi way for the majority of that time. About 2 years ago i went to full range single drivers as i heard a set of high eff. single driver horns and the sound blew me away.I have also built full rangers with helper tweets and found that b-amping (with gainclones) can make a smaller speaker than a full rnge horn. If you roll in a tweet when the HF slope starts to drop the sound is very coherent.Its also easy to roll off a sub , i always design the GC to roll off the sub below 150Hz,an use a higher frequency full ranger driver for mid bass to HF range.My present horns (see below) have an F3 around 53Hz and extend to 20KHz.As i live in an apt and am not really into kidney shaking bass i find them more than adequate.If i ever wanted a lower LF response i would just build in another gainclone and roll off a sub at the front end of the amp.
ron

http://community.webtv.net/roncla/FOSTEXfe103ehorns

 

[tcpip]

Re: Full range single driver horns.

have been designing and building speakers for over 30 years( yea i am an old fart).

You shouldn't say such things; others will feel hurt. Navin here has been building speakers and amps from the time he was eleven. He must have clocked thirty years too. Why do you think people like me, who built their first speakers only eighteen years ago, call ourselves freshies?

http://community.webtv.net/roncla/FOSTEXfe103ehorns

Lovely pictures. Specially like the pictures of the pine planks, with those knots. Wish I could hear them (speakers, not pine planks). Got a bit confused with some of the horn pictures though. Are they kept lying on their sides or something?

Will wait to hear more reports about these horns' performance.

Tarun

PS: How do you find the sound of the gainclones? Do you feel the highs clean enough for your sensitive and clean-sounding horns? Or do the Fostex drivers roll off the highs early enough to hide any glare from the amps?

 

[tcpip]

Hehe, re-creating the atmosphere of a song that's been sung on 16 different hills in 4 countries cannot be easy.

You missed out the five different changes of garb during a single song for the leading pair as they run around trees/lampposts.

But in all seriousness, I see how it would be difficult to reproduce some of the live tracks by AR Rahman for example without extending all the way up to the very high frequencies.

And now they've started playing with the low end too... see the first song of "Lagaan", where they dance with joy on seeing the rain clouds. Almost demands subwoofers.

But apart from the recent generation of HIndi film music, why is the rest of the archival collection, all those 100,000+ songs, so badly recorded? Why are they so, well, plain muffled, and lacking in definition, bass extension, poor on PRAT, etc? Know it's off-topic, but couldn't help wondering. I'm sure that any Indian recording studio thirty years ago used decent quality reel-to-reel recorders, and practically any well-maintained reel-to-reel will record better than this, won't it?

Is this seen in old footage of other music too, e.g. fifties and sixties jazz archives of Columbia, say?

Tarun

 

[tcpip]

Generic active equaliser?

Steve, SY, and Ken,

Thanks very much for your posts. I know I took time to respond, but I was trying to absorb what you'd written.

Pooling together what all of you have written, one idea that strongly suggests itself is that the equalisation fine-tuning phase can be made easier. It cries out to be done with a generic high-quality equaliser. You can then twiddle knobs instead of replacing resistors and caps.

The idea that comes to mind is to use the following strategy: Use a

• digital equaliser/XO (e.g. Driverack/Behringer), OR
• an analog parametric equaliser, together with analog active XO

to set up a first-cut version of the eq+XO. Then strart tuning using either the analog par-eq or the digital eq, and get your frequency response (and perhaps channel delays) correct. Finally, pull out the par-eq or digital eq out of the circuit and reproduce the same thing using a fixed active eq + XO circuit. (This last step is done to reduce the cost of the system. The par-eq or the digital eq box then becomes rentable or reusable.) Once the dedicated circuits are done, you can get down to component-level tweaking, i.e. replace 5532 with OPA2134, etc.

SY, this is why I was asking about the 10-band par-eq. The number of bands is not important (I too have never used one), but the idea was that one should be able to use a generic equaliser of good enough quality to get the full eq done quickly. This step should not demand any circuit mods.

Tell me, is this doable?

Secondly, how difficult is it, do you think, to precisely convert a particular set of settings of the Behringer or Driverack into a dedicated active filter circuit? That way, the tunable nature of the digital eq or par-eq box will be used only for making the first prototype, and subsequent replicas of the speaker system (assume more than one, to satisfy the clamour of thirsty fans, family and friends, etc) can be built at low cost using inexpensive active filter circuits dedicated to that speaker design.

Tarun

 

[pinkmouse]

Tarun,

given the price and facilities of the Behringer unit, I suspect it would be hard to build something to match for cheaper than you can buy it. Your approach was very valid, (and I was thinking of the exact same thing), when active digital system controllers were £1500 or more, like the Omnidrive or XTA.

Considering you may need to duplicate a stereo 3 way xover, with individual LP/BP/HP, individual delay settings per driver, 3 or 4 bands of stereo eq, and put it all in a nice box with traffos, connectors, and then debug it, it would be hard to match.

But, if you want the fun/frustration of doing it yourself, then go for it by all means.

 

[Vikash]

Hi Tarun,

But apart from the recent generation of HIndi film music, why is the rest of the archival collection, all those 100,000+ songs, so badly recorded? Why are they so, well, plain muffled, and lacking in definition, bass extension, poor on PRAT, etc? Know it's off-topic, but couldn't help wondering. I'm sure that any Indian recording studio thirty years ago used decent quality reel-to-reel recorders, and practically any well-maintained reel-to-reel will record better than this, won't it?

Well you have to look at the largely ignorant mass market it was aimed at. It's a shame that all we can do now is take these classics through a synthetic restoration stage. My dad will tell you how they only cared about the moving lyrics back then, and us kids nowa days...bla bla bla.

As you said it's off topic, so I won't start a philosophical rant on the subject. But I think all the nasties in the old recordings paint a picture of the time (for all genres) - and that's an important part of it to some people.

given the price and facilities of the Behringer unit, I suspect it would be hard to build something to match for cheaper than you can buy it. Your approach was very valid, (and I was thinking of the exact same thing), when active digital system controllers were £1500 or more, like the Omnidrive or XTA.

Considering you may need to duplicate a stereo 3 way xover, with individual LP/BP/HP, individual delay settings per driver, 3 or 4 bands of stereo eq, and put it all in a nice box with traffos, connectors, and then debug it, it would be hard to match.

My knowledge is minimal on this subject, but perhaps this is of interest: $30 line level crossover/dsp thread.

 

[mhelin]

This one will be only $5 (+ADC, DACs, crystal and an EEPROM chip, and some analog components). You could do a 1Ktap FIR with it, for an xover you need not so much. For 3-way propably max. 100 taps for each band. The EVM is also quite inexpensive, about 230 EUR. Just add DAC boards for it (it comes with only stereo in/out).

 

[SY]

Tarun, an analog eq will not give you flexibility in slope, nor flexibility in the bandwidth and position of the boost/cut. It also will be limited in the amount of boost and cut. Parametric eqs can do better, but are a royal pain to set up; position, bandwidth, and degree of cut are rarely calibrated (and can interact), so there's another iterative measuring process.

Digital is the only way to go, as a practical matter- and that's the way I'll go next time. If you want to convert the eq/xo settings you finally arrive at to an analog circuit, that's not hard to do at all, and it frees up the digital unit for the next design.

OT: I heard Hindi rap for the first time about a year ago. I congratulate the artists, who finally gave me something to make fun of that's even worse than French rap.

 

[angshudas]

Active XO

Tarun,

Though bit late on this thread, I would like to share my views on XO design. Most of it has already been said, so pardon me if I sound repetitive.

Designing XOs become easier if you have the proper tools. You need means to measure SPL and impedances of your drivers. Radio Shack SPL meter is not adequate. You need to measure anechoic in room measurements at least 1-2m from the drivers.

If you are designing passive XO then you need to have some optimising software, I use CALSOD. I do not know how people ever design manually by listening, I don't think I can do that.

If you are planning active XO, then you can get away with the Optimising software. If you find your drivers overlapping by 2 octaves either side (this you already know) then you can use classic filter topology. I prefer LR4, you are -48db down 2 octaves away, so the acoustic slop will be quite accurate. You also know that it is the acoustic slope we are talking about, the electronics and the driver output should give you the target acoustic response.

For time delay adjustment I do the following. Finding the acoustic centers is bit complicated. You can use CALSOD to do that. I measure the distance from the top plate of the magnet to the flange for both the midrange and tweeter. Subtract to get the offset distance. Calculate the time delay. In use a variable time delay circuit which I can tweek..

I reverse the tweeter polarity and take far field SPL measurements. You should get a notch at the XO frequency. The deeper the notch the better your XO is adding. I vary the delay so that I maximise the notch. Once a null is reached, I put back the tweeter in the correct polarity.

I take far field SPL readings at different angles and see whether the response is smooth. there should not be any bumps in the off axis XO point.

If you are designing passive XO it is not necessary that your drivers overlap 2 octave either side. The optimiser adjusts the component values to give the desired roll off characteristics. Many a times I have been lucky to get LR4 slopes with just a single L and C.

I am attracted towards line level passive XO because you can use your optimiser as well as drive the individual drivers with separate power amps. You get the advantage of both passive and active XO.

Angshu

 

[ronc]

Kinda OT fer this thread-gainclone/horns

The sound of the gainclones is by far better than anything i have ever built or heard.I know yall may think this a bold statement but there it is.I use to be a very deep marantz and tube buff, but after fine tuning the GC i have found no better sounding amp and its cheap also. When you get into die hard GC forums you find so many ways to tune the GC to the performance that you desire, i run a LPF on the input and with battery power the sound is exceptional.IMHO the GC has dynamics that is startling and combined with the greater dynamics of the horns makes for a combo that gives a presentation and 3-d image that is almost scary.I had a friend with the B&W/Krell combo give it a listen and his jaw hit the floor.He did state that his bass went deeper(hay its a 4" driver) but, as stated, i run no sub but he also stated that the dynamics and accuracy by far surpassed his system, all with around 20/watts a channel.
ron

 

[sfdoddsy]

Tarun,

Do Option A with the digital crossover.

Firstly for the reasons above, but otherwise for the horribly obvious one that most analog active crossovers and EQ do not have accurate markings for what you are doing.

You will find, should you try it, that you are guessing what the crossover frequency is, the EQ frequency, the Q and much more. And if a knob gets bumped, prepare to start again and measure from scratch. Take it from me, it's a bitch.

A digital EQ/crossover is precise and repeatable.

Should you wish to then convert to a custom analog, at least you will know what you need.


Cheers

Steve

 

[MarkMcK]

Great question. It made me think in different pathways.

There is an analogy in the world of passive speaker level crossovers. Passive crossovers that you buy as crossovers are usually simple, often inexpensive, and not designed with any particluar driver in mind. Instead, they are math based designs assuming perfect loudspeaker drivers.

Electronic crossovers, designed as free-standing products are usually designed without specific drivers in mind. As such, they too are designed assuming perfect drivers. Because electronic crossovers occupy a higher price point in the capitalist marketplace, they can offer more adjustment possibilities (options of slopes and frequency knee points). Testing (and designing) will still have to be done to figure out what settings (if any) will fit with the drivers chosen for the final loudspeaker.

Since perfect drivers do not exist, I have to wonder if such crossovers, passive or active, are the best approach to the complex problem of making imperfect loudspeaker drivers sound natural when assembled into a loudspeaker.

While using high pass filters on the high frequency drivers will make it more likely that those drivers will continue to work (not overheat and fail), would using one of these generic designs make a multiple driver loudspeaker sound better than not using a crossover? In all honesty, I do not know. I would, however, tend to doubt the quality of the outcome.

My own perspective is that loudspeaker design must be wholistic and account for as many variables as possible. You can make active designs that are tailored for specific drivers. They are just more difficult to design and assemble. It requires more technical know how and fabricating capabilities than most people have in their workshops. It also requires more time and a greater investment in a project that may prove to be inadequate. As a result, few people pursue this design path.

To sum, the answer to your question is not simply technical, it also has a marketing and a fabrication dimension.

 

[Ken L]

Tarun,

Do Option A with the digital crossover.

I agree with Steve -


I always hesitate to strongly suggest what I am personally doing, there are so many ways of achieving a goal.

If you're biamping or triamping, the digital crossover is a no brainer.

It is relatively new for this use and has not gained widespread acceptance.

Pros have been using them for a long time.

With the cost dropping so low they now become feasible for home use.

Regards

Ken L

 

[tcpip]

Value for money: a digital XO?

given the price and facilities of the Behringer unit, I suspect it would be hard to build something to match for cheaper than you can buy it. Your approach was very valid, (and I was thinking of the exact same thing), when active digital system controllers were £1500 or more, like the Omnidrive or XTA.

Pinkmouse,
Can't agree on this one. I'm far more low-tech than you guys... I can feel happy with an active XO on prototyping board. I can feel thrilled with the sound of a well-designed NE5532 circuit. My feeling is that just the parts cost for a 3-way stereo active XO, plus about three to four bands of eq plus one delay block per channel, will altogether require probably 50 dual-opamp chips for both channels. That's about USD 25.00 or so here (NE5532). Add passives and simple PSU, I don't think I can exceed USD 100, everything put together. I'll scrimp on the chassis (don't want to blow $200 on a super-duper brushed aluminium chassis if I'm funding my own hobby)... That makes it much cheaper than even a discounted price of a Behringer ($350?)

Of course, the time I put in will be quite a bit. I may still go with the digital XO finally, someday, when I decide to build the stuff. (No money for $350 now... no way!) But I first wanted to know whether the approach works at all or not.

What I'm still not sure about, after hearing from you, SY, Steve, and others, is whether a custom-built analog circuit equivalent will be an accurate enough replica of the Behringer's settings. Will it?

Tarun