Crossover help

[AF2000]

Hello. I was hoping to build a set of speakers with a 2 way crossover. The mid that I want to use is 8 ohms, and the tweeter is 4 ohms. I have found premade crossovers that require a 8 ohm tweeter with either a 4 or 8 ohm mid, but nothing the other way around. I want the crossover point around 3500 hz and I want the impedance seen by the amp to be 8 ohms.
I know there are calculators I can use to determine the components of the crossover, but how do I know what the overall impedance of the speakers will be?
Thank you in advance!

 

[jaxvon]

Hello. I was hoping to build a set of speakers with a 2 way crossover. The mid that I want to use is 8 ohms, and the tweeter is 4 ohms. I have found premade crossovers that require a 8 ohm tweeter with either a 4 or 8 ohm mid, but nothing the other way around. I want the crossover point around 3500 hz and I want the impedance seen by the amp to be 8 ohms.
I know there are calculators I can use to determine the components of the crossover, but how do I know what the overall impedance of the speakers will be?
Thank you in advance!

You could always avoid this trouble by using an active digital crossover, giving you perfect integration and EQ options to precisely correct response and/or shape it to your preference. A Behringer DCX2496 would do nicely. That, and I'm too lazy to do the math.

 

[Spkr_Bldr]

There are a lot of reasons not to use those 'premade' crossovers. First you have to understand how they come up with impedance ratings for a driver - it's an average of a complex curve. Let's say you have a woofer rated at 8 ohms - it'll probably have something like a 6.5ohm min that will cover a range from like 150hz to 1500hz. Below that it will rise sharply to like 25 ohm, and above that it will rise also although less sharply, probably to around 12 ohm by 10khz. And the values of caps/coils you use to achieve a specific crossover point is 100% dependend on the actual impedance based on that curve, not it's rating.

Those premade crossovers also dont' have zobel networks, so your behaviour out of passband will be very unreliable. You'll also need to build in some baffle step compensation, and you'll need to level match between the tweeter and woofer.

Now, it may sound hard, but it's really not. You don't need to be a rocket surgeon to build excellent custom speakers, but there's a foundation of knowledge that you do need before beginning.

If it's something you want to learn more about and maybe try, I'd suggest signing up on diyaudio.com and introducing yourself and a n00b in the Loudspeaker section. It wasn't more than a couple years ago I myself was a n00b over there, and now I've built speakers that were widely acclaimed at RMAF, and working on launching a full line of ultra high-end speakers that I've designed from the ground up.

 

[jinjuku]

You could always avoid this trouble by using an active digital crossover, giving you perfect integration and EQ options to precisely correct response and/or shape it to your preference. A Behringer DCX2496 would do nicely. That, and I'm too lazy to do the math.

Wouldn't this approach require him to bi-amp the setup?

 

[jaxvon]

Wouldn't this approach require him to bi-amp the setup?

It would, but for a small two-way setup it would not be very expensive and the added headroom is always good.

 

[Spkr_Bldr]

Honestly the only skill needed in passive design that's not for active design is soldering. You need a foundation of knowledge to successfully pull off either implementation. Proper measurements and ability to interpret them are crucial to either. But don't fret, it's really quite easy to get a good measurement rig up and running ... and all the information you need to get rolling is at your fingertips.

The best cheap setup to cover all the bases is an ECM8000 mic, good soundcard with RCA connections in and out, something to provide phantom power to the mic, and a free copy of ARTA. For measuring phase/impedance, it's hard to beat the Woofer Tester.

 

[TLS Guy]

Hello. I was hoping to build a set of speakers with a 2 way crossover. The mid that I want to use is 8 ohms, and the tweeter is 4 ohms. I have found premade crossovers that require a 8 ohm tweeter with either a 4 or 8 ohm mid, but nothing the other way around. I want the crossover point around 3500 hz and I want the impedance seen by the amp to be 8 ohms.
I know there are calculators I can use to determine the components of the crossover, but how do I know what the overall impedance of the speakers will be?
Thank you in advance!

What drivers are you planning to use?

 

[AF2000]

Thanks for all the help so far everyone. Here are the drivers I want to use:
(sorry, I'm too new here to add links yet)

Eminence LA6-CBMR 6-1/2" Sealed Back Mid Line Array Series (Parts Express)
and
AR 3/4" Shielded Soft Dome Tweeter (Parts Express)


I do already own a pair of 12" JBL subs that I'm not using right now. (GT1241D). I'm thinking of possibly trying a three way set-up now. If I run the sub(8 ohms) and mid(8 ohms) in parallel and add the 4 ohm tweeter in series I should get a 8 ohm total load.

I'm still confused though by the effects of the crossover components on the impedance of the speakers. I have used online calculators to determine what components to use, but I'm not sure what my overall impedance would be. I know how to calculate it without a crossover, do the same rules apply despite the crossover?

Here is what I got from the calculator:

1st Order Normal Polarity

3800 Hertz / 475 Hertz

4 Ohm Tweeter / 8 Ohm Mid / 8 Ohm Woofer
0 db Bandpass Gain, Spread = 8 : 3 octaves
Parts List
Capacitors
C1 = 10.46 uF
C2 = 47.17 uF
Inductors
L1 = 0.3 mH
L2 = 2.68 mH



Thanks again, I appreciate everyone's help!

 

[TLS Guy]

Thanks for all the help so far everyone. Here are the drivers I want to use:
(sorry, I'm too new here to add links yet)

Eminence LA6-CBMR 6-1/2" Sealed Back Mid Line Array Series (Parts Express)
and
AR 3/4" Shielded Soft Dome Tweeter (Parts Express)


I do already own a pair of 12" JBL subs that I'm not using right now. (GT1241D). I'm thinking of possibly trying a three way set-up now. If I run the sub(8 ohms) and mid(8 ohms) in parallel and add the 4 ohm tweeter in series I should get a 8 ohm total load.

I'm still confused though by the effects of the crossover components on the impedance of the speakers. I have used online calculators to determine what components to use, but I'm not sure what my overall impedance would be. I know how to calculate it without a crossover, do the same rules apply despite the crossover?

Here is what I got from the calculator:

1st Order Normal Polarity

3800 Hertz / 475 Hertz

4 Ohm Tweeter / 8 Ohm Mid / 8 Ohm Woofer
0 db Bandpass Gain, Spread = 8 : 3 octaves
Parts List
Capacitors
C1 = 10.46 uF
C2 = 47.17 uF
Inductors
L1 = 0.3 mH
L2 = 2.68 mH



Thanks again, I appreciate everyone's help!

This system is not constructable. Your Eminence midrange has an Fs of 450 Hz. It will have to be crossed over around 900 to 1000 Hz. There is no way that JBL driver will go up that high it is a sub driver.

Next advice do not start with a three way for you first project. Three ways are highly problematic and few turn out well even in the hands of professionals. Start with a two way.

 

[AF2000]

Is it possible to build an 8 ohm two-way with that mid and tweeter?

 

[Spkr_Bldr]

Well, that would be a tough project ... really tough. The mid/woofer transition would be nearly impossible as TLS says. And the woofer sensitivity would be WAY down from the mids, you'd have to use a plate amp or some other active solution for the woofer.

You say you want to use those drivers, why? If you don't own them yet, I'd suggest totally scrapping that idea and start with something else. If you just want to build something, I'd say start with a proven and published design ... or buy a kit from somebody like GR Research.

 

[TLS Guy]

Is it possible to build an 8 ohm two-way with that mid and tweeter?

No it is not. If you use a mid range and tweeter, then you need a woofer (not sub) as well. Then it is a three way.

You could build an 8 ohm speaker using a bass/mid and tweeter though.

 

[TLS Guy]

Is it possible to build an 8 ohm two-way with that mid and tweeter?

I have modified a design I did for a novice builder a little while ago. It was done to get an experienced builders novice brother in law started. They were very pleased with it. They used a tweeter that is NLA because they had it. However the design was easily modified to work well with this tweeter.

This is a good starter project and will not upset or destroy any amp. It will yield superior results and if you build subs for your JBLs will make a first class system.

I have called the design NASP for Noob audiophile speaker.

I have posted the design on my web site.

http://www.drmarksays.com/

 

[AF2000]

Well, that would be a tough project ... really tough. The mid/woofer transition would be nearly impossible as TLS says. And the woofer sensitivity would be WAY down from the mids, you'd have to use a plate amp or some other active solution for the woofer.

You say you want to use those drivers, why? If you don't own them yet, I'd suggest totally scrapping that idea and start with something else. If you just want to build something, I'd say start with a proven and published design ... or buy a kit from somebody like GR Research.

The reason I like the Eminence is because of its high efficiency. The AR tweeter was the closest match in efficiency at a good price that I found. And the whole system was under $200.

I currently have a 5.1 set-up with Infinity speakers and a 1000w powered Cerwin-Vega sub. I wanted to replace my main speakers because they are the weak point in the system. They don't have near the clarity of the CC-3 center channel I'm using. 10 years ago a family friend helped me build them, but we used Infinity Reference Series car speakers.

Thank you again to everyone here. This seems like a great board with really friendly people. I think I'll go back to the drawing board!

 

[TLS Guy]

The reason I like the Eminence is because of its high efficiency. The AR tweeter was the closest match in efficiency at a good price that I found. And the whole system was under $200.

I currently have a 5.1 set-up with Infinity speakers and a 1000w powered Cerwin-Vega sub. I wanted to replace my main speakers because they are the weak point in the system. They don't have near the clarity of the CC-3 center channel I'm using. 10 years ago a family friend helped me build them, but we used Infinity Reference Series car speakers.

Thank you again to everyone here. This seems like a great board with really friendly people. I think I'll go back to the drawing board!

Those Eminence belong in the cheap and nasty category.

If you build the speakers I linked you to, you won't spend a lot over $200, at the most $300 for crossover parts and building supplies. I can assure those speakers will not let the side down. It will be a fun project!

 

[Spkr_Bldr]

I have modified a design I did for a novice builder a little while ago. It was done to get an experienced builders novice brother in law started. They were very pleased with it. They used a tweeter that is NLA because they had it. However the design was easily modified to work well with this tweeter.

This is a good starter project and will not upset or destroy any amp. It will yield superior results and if you build subs for your JBLs will make a first class system.

I have called the design NASP for Noob audiophile speaker.

I have posted the design on my web site.

http://www.drmarksays.com/

Those are nice drivers, but it seems like some odd design choices - at least from my perspective. Why would you go high order on the woofer? The peerless poly woofers are smooth as can be, no breakup of any sort, and hence can be used with a simple 1st order and a Zobel. The rising response is due to the low inductive rise, because of the good motor with plenty of copper - the hump around 3300 hz is not a breakup.

Personally (not that it means anything) but I'd think a 1st order Series network around 2400hz would be perfect for those drivers ... or Parallel network if you're wanting to make it n00b friendly.

 

[TLS Guy]

Those are nice drivers, but it seems like some odd design choices - at least from my perspective. Why would you go high order on the woofer? The peerless poly woofers are smooth as can be, no breakup of any sort, and hence can be used with a simple 1st order and a Zobel. The rising response is due to the low inductive rise, because of the good motor with plenty of copper - the hump around 3300 hz is not a breakup.

Personally (not that it means anything) but I'd think a 1st order Series network around 2400hz would be perfect for those drivers ... or Parallel network if you're wanting to make it n00b friendly.

That gives the smoothest response. The woofer needs a higher order of cut than the tweeter. If you go second order on the woofer and first on the tweeter, then the tweeter is driven to its resonance.

I played with this a long time, and I found that combination gives the smoothest axis ans off axis response.

It is still a pretty simple crossover. The responses of the two drivers with that circuit are extremely symmetrical.

A first order network for both drivers would be hopeless and sound awful.

First order crossover are so often "fools gold". The problems relate to getting a smooth crossover.
Also few drivers have the bandwidth to use them. Take tweeters for instance. With only 6db per octave roll off you have to start the crossover at 5KHz, otherwise the tweeter has too much power in its out of band response. You need to have tweeters about 24db down at resonance ideally, if you want nice smooth sound. The use of first order filters, actually ends up making for the most complex circuits, as Thiel found out and so did I in the eighties. I did a first order crossover design and it took 10 years R & D. When all is said and done the speakers are useful as monitors where the listening is predominantly going to be done by one listener, because despite all the apparent advantages you end up with a very suboptimal lobing pattern. In other words very much a sweet spot speaker. This is obviously less than ideal for HT.

Not that you should be Cavalier about phase and time, I don't believe. Unlike WmAx I do believe you should minimize the trespass you make with phase and time, compatible with the smoothest mid band response. I think you will have to agree though the phase response of those drivers and that crossover are far from extreme.

There is nothing wrong with being a full ranger at heart. I keep JW modules on hand for reference, and have a set in some TLs, I have used for years for a reality check.

One of the things that stays on my list to do, is start a JW website, less lessons once learned get forgotten.

 

[Spkr_Bldr]

That gives the smoothest response. The woofer needs a higher order of cut than the tweeter. If you go second order on the woofer and first on the tweeter, then the tweeter is driven to its resonance.

I played with this a long time, and I found that combination gives the smoothest axis ans off axis response.

It is still a pretty simple crossover. The responses of the two drivers with that circuit are extremely symmetrical.

A first order network for both drivers would be hopeless and sound awful.

We're going off track, and I apologize to the OP ... but just wondering.

Did you ever measure a finished pair? Why does the woofer 'need' a higher slope than the tweeter? And how can you say a 1st order network would be hopeless and sound awful, when you didn't try one?

Just about every pulp or poly woofer I've worked with sounds and measures well with a 1st order, and if needed a Zobel to flatten the impedance and do some shaping. If you want to cross low, then put a 3rd order on the tweeter and they're both electrically in phase. If somebody gave me those two drivers that's where I'd start, crossed around 2000-2200.

I've modeled and built probably 40-50 designs now, and not one has been acceptable on my first attempt based on modeled response. And I even use a Woofer Tester to get my phase/impedance curve, and then model using the actual measured responses of the drivers on the baffle which they're going to be used ... and still modeling only gets you in the ballpark.

 

[TLS Guy]

We're going off track, and I apologize to the OP ... but just wondering.

Did you ever measure a finished pair? Why does the woofer 'need' a higher slope than the tweeter? And how can you say a 1st order network would be hopeless and sound awful, when you didn't try one?

Just about every pulp or poly woofer I've worked with sounds and measures well with a 1st order, and if needed a Zobel to flatten the impedance and do some shaping. If you want to cross low, then put a 3rd order on the tweeter and they're both electrically in phase. If somebody gave me those two drivers that's where I'd start, crossed around 2000-2200.

I've modeled and built probably 40-50 designs now, and not one has been acceptable on my first attempt based on modeled response. And I even use a Woofer Tester to get my phase/impedance curve, and then model using the actual measured responses of the drivers on the baffle which they're going to be used ... and still modeling only gets you in the ballpark.

I extended my remarks in the other post. It has to be really exceptional drivers to work with first order crossover. Most of the time they are best avoided. My friend is out of state, so I have not heard them. However he has built a pair of my NFM-1 and he says their tonal balance is extremely close. He sent me some crude measurements, and they looked OK as far as it went.

I agree that you can NOT cookie cut speakers. Most of the time you end up making changes, but my track record is getting better. He is an experienced listener, and if those speakers sound tonally very close to the NFM-1, they can't be far adrift. I generally keep my modeling to myself, except subs, where I think you are pretty safe. This post is somewhere in between, but I felt I could risk it. It will be a hell of a lot better than what the OP had in mind.

The crossover is actually 2.2 KHz. The acoustic roll off of the drivers have different slopes.
If you look at the graphs, you will see that the combined slopes of the drivers and crossover slopes, gives a composite very symmetrical third order crossover centered at around 2.2 KHz. The outputs sum to a smooth response in the crossover region. If you can model a better crossover be my guest, but I'm certain you won't do it with first order low and high pass filters.

 

[Spkr_Bldr]

I extended my remarks in the other post. It has to be really exceptional drivers to work with first order crossover. Most of the time they are best avoided. My friend is out of state, so I have not heard them. However he has built a pair of my NFM-1 and he says their tonal balance is extremely close. He sent me some crude measurements, and they looked OK as far as it went.

About 1st order slopes, it seems you have some preconceptions ... maybe based on past experience. But my experience has showed a somewhat different result. I'm not a huge 1st order proponent, but I have pretty high confidence that one would work extremely well on the Peerless poly woofer. Your assertations that you need exceptional drivers just isn't true, a perfect example is just about every product that Danny Richie at GR Research puts out ... he uses 1st order on all his woofers which are far from exceptional, with a zobel that does shaping as well. My approach with paper/poly woofers is often similar to his, cross low with a 1st order on the woofer and 3rd order on the tweeter. Or with something as smooth as the Peerless Poly cones I'd at least try a 1st order on both woofer and tweeter. But a 3rd order on that woofer is just not needed.

I agree that you can NOT cookie cut speakers. Most of the time you end up making changes, but my track record is getting better. He is an experienced listener, and if those speakers sound tonally very close to the NFM-1, they can't be far adrift. I generally keep my modeling to myself, except subs, where I think you are pretty safe. This post is somewhere in between, but I felt I could risk it. It will be a hell of a lot better than what the OP had in mind.

Well, I have no doubt this would be far better than what the OP had in mind ... but also there are hundreds of published designs around that have all the supporting measurements to go with it. And I don't mean to take a shot at you by any means, but I'd never publish or suggest a design to anybody without building it, measuring it, and listening extensively.

If you can model a better crossover be my guest, but I'm certain you won't do it with first order low and high pass filters.

Honestly, with those woofers I wouldn't even bother with modeling. They're extremely easy to work with, I'd just put them in a box and put a guesstimate network on it and start measuring ... tweak, measure, tweak, measure.

And why are you so worried about having any tweeter output around the resonance? It's not like anything necessarily bad happens by doing so, but of course the only way you'd know is by first measuring said tweeter's frequency response, distortion behavior, and phase/impedance. I've done a 1st order network on a Vifa XT25 before centered at 2400hz, and even with it's huge resonance peak it performs very well. I mention that speaker specifically because it's at a friends how now replacing his Dynaudio Confidence C2's while they're being repaired, and he's shocked at how good they are in comparison to a $12K speaker. It's just a 2-way sealed bookshelf with about $300 of parts.