need some x-over advise before I build... (1 Viewer)

[Ron D Core]

I designed this 4th order Linkwitz-Riley x-over for my MTM tower setup. How do these graphs look? The amplitude graph is mostly within a 3dB range, but the corner frequencies are half an octave apart. The Focal 7k4211's have a 6dB spike in the upper range and the corner frequency separation plus the impedance EQ's are all I could think of to take care of it without a notch filter. Is it a bad thing to separate the frequencies like that? Other than that I have a big problem with the system impedance. In the upper range its over 4 Ohms which is fine, but it dips to 2.4 Ohms around 50 and 500 Hz. Would it hurt anything to add a non-inductive 3 Ohm resistor between the amp and the x-over? Wiring the woofers in series messes up the amplitude response significantly and doesn't seem to help the impedance much anyway. The phases between the woofers and tweeter (Focal TC120TD5) are no more than 50 degrees apart where they are summing and right where they cross, they are about 35 degrees apart (when the polarity of the tweeter is reversed). Does this seem ok? One last thing, i've been looking at the Solen capacitors from PE for all the caps accept for the 80uF. Should I spend the money on them or would it be ok to use cheap caps and bypass them all with some .1uF's? Heres the setup as copied form x-over pro...
2-Way Crossover Network
Low-Pass (LP) Filter: 1 required
Type: 4th-Order Linkwitz-Riley
Desired Corner Frequency: 2000 Hz
High-Pass (HP) Filter: 1 required
Type: 4th-Order Linkwitz-Riley
Desired Corner Frequency: 3000 Hz
C1 = 5.1 µF, Solen MP
C2 = 8.2 µF, Solen MP
C3 = 39 µF, Solen MP
C4 = 9.1 µF, Solen MP
L1 = 0.2 mH, Air Core(16gauge), DCR 0.18 ohms
L2 = 0.9 mH, Air Core(16 gauge), DCR 0.44 ohms
L3 = 0.47 mH, CFAC(16 gauge), DCR 0.22 ohms
L4 = 0.22 mH, CFAC(16 gauge), DCR 0.13 ohms
Tweeter
Impedance EQ
Req = 8 ohms
Ce = 3 µF
Woofer
Impedance EQ
Req = 2 ohms
Ce = 80 µF

 

[ThomasW]

Did you use speaker data taken from your own measurements of the speakers mounted in the baffle you intend to use? Or did you use the mfgr's data when designing the XO?

 

[Isaac C]

I recall that XoverPro cannot accept measured impedance data. It derives the electrical impedance from driver specifications :frowning:

 

[Dennis XYZ]

Normally, the woofer and tweeter would be wired in phase with a 4th order crossover. Perhaps that explains why you had to shift the poles 1/2 octave.

 

[Jon Hancock]

If your crossover design program is using "simulated" impedance from T/S parameters, as opposed to a measured curve, there's a real chance for haviing "garbage in, garbage out". Looking at the impedance plot, your LF resonance (sealed?) is at about 18 Hz, not remotely in the ball park for the 7K4211's, which should be in 40Hz or so area. Another clue that you have problems with your design program is your statement

Wiring the woofers in series messes up the amplitude response significantly and doesn't seem to help the impedance much anyway.

Wiring the woofers in series will double the impedance, but it will also require a complete redesign of the LP crossover, as it will then be working into 14-16 ohms, instead of 6-8 nominal.


For the 4th order Linkwitz-Riley crossover, the response should be down 6 dB at the nominal crossover frequency. Your design seems to meet this criteria, but the "corners" or Q of the filters doesn't seem right for a 4th order L-R acoustical response. Did you design a 4th order Butterworth, then have to move them apart in corner frequencies to get them to sum correctly?

Also, keep in mind that in the real world, you have to take into account the time offset of the acoustical centers of the drivers. This usually means that for the woofer, which is "farther" away acoustically if both are mounted on a flat baffle, you have to use a softer and slightly shallower slope in order to provide some phase correction. If anything, your woofer slope looks steeper than the tweeter- this could be a problem when you assemble and measure the completed system.

Last, the Tc120 can be used lower than 3000 Hz with a 4th order Linkwitz-Riley crossover. On the otherhand, the Focal 7" kevlar drivers are not a good choice to run that high. Take a look at the measured impedance curve, even free air. The current version, the 7K4412, has ripples at 1200 and 2600. These correspond with significant cone breakup modes. Ideally, you'd want to stay below these modes. The lower mode doesn't have an amplitude peak associated with it, and only smears the sound somewhat; the other, of course, is more problematical. I'd recommend strongly against trying to run these up to 3 kHz.

Consider a 4th order crossover around 2.25 kHz; I think you'll get results you'll be more happy with. If you keep the woofers in parallel, which you should be able to as long as your amp is happy with loads in the 3-4 ohm region, I'd recommend going with at least the Solen AWG14 coils; the 16AWG have too much DC resistance.

Also, for $130 or so, a copy of LspCAD might be a very helpful investment, considering what' you've already spent on drivers. Good luck with finishing your project and realizing it's potential.

Best regards,

Jon

 

[Ron D Core]

in the driver database, the the anechoic frequency response is included with the driver's parameters, so its not simulated. The baffle step and room response was included in the calculations. But I think I'll try to lower the crossover to about 2.25khZ like you say and see how it turns out.

 

[Jon Hancock]

Yes, that may be Ron, but you can't do accurate crossover filter calculations if the impedance curves are wrong- so if you're using data supplied with the crossover design program, which has invalid impedance cuves, then inherently the crossover filter design will be incorrect. For example, if you halve or double the impedance of a driver, (such as by paralleling or series connection), you basically have to halve or double the size of the inductors and capacitors, as well as adjust resistor dimensioning.

Regarding the impedance plot above, I have to say it's suspect. Also, unless the "anechoic" frequency response it was measured using a box with the same dimensions you're constructing, the response, as affected by the baffle step, won't be the same. Plus, as I've pointed out, the LF impedance curve is not possibly real- these driveres don't have an Fs of 18 Hz. Perhaps something is wrong with the software, perhaps with their database. But I would be leary of basing design decisions on what you show above.

Have you already purchased the drivers?

 

[Ron D Core]

I haven't actually purchased them yet, but they are exactly what I'm looking for in a speaker. I would do the W-cone series but the price is alittle steep. I though about doing something like a Kevlar 7 for the midbass and a W-cone 6 for the midrange. I do want the Focals and I do want the kevlar cones for the midbass, I just really like the way they sound. My budget is $600 for the raw drivers. The MTM setup with the 7K4412's and tc120 is right in there. I know those drivers can sound great with the proper crossover design, but that's what I'm attempting to do with some help. So as far as impedance curves are conserned, would a woofer tester from PE do a decent job at measuring that? Should I build a test enclosure then do the measurements? I might be able to find someone with halfway decent testing equipment for measuring the frequency response in the boxes(ported). I'll have to agree with you that that 18Hz peak makes no sense. I'll mess with it some more and see what is wrong (if its not something wrong with the program itself). Thanks for the helping me out btw.

 

[ThomasW]

Ron
Jon's probably commuting home via the train right now. He's going to tell you not to get the 7K4411 drivers for a number of reasons.....
We have some interesting DIY projects in the works that are right in your budget, and will provide better performance.
Jon will either post info here or drop you an email regarding alternative designs.
Here's a hint, consider a MTM version of this, the tweeter shown is a Focal TC120, using the Vifa XT25 is another option

 

[Ron D Core]

That looks pretty cool, I thought about those Hi-Vi M-series too. I would definitly consider that driver combo for the price. I'll see what he says.

 

[Jon Hancock]

Hi Ron,
Back again- busy evening, didn't have a chance to go online (daughter is graduating this Friday from High School).
First, let me say I've looked at and used some of the Focal woofers- I actually prefer some of the V series over the K and W series, I'll explain why.
On the otherhand, I think there are better choices for you to consider, (which I'll get to), so in once sense, discussing these Focal drivers is more an intellectual exercise- but possibly interesting, anyway.
Link Removed
This is the V series 7" from Focal. Note that if you examine the impedance curve, you can just barely see a little bobble around 1200 Hz, and another one higher, around 3000. Overall response is pretty smooth, both on and off axis.
Link Removed
This is the K series driver. Pretty similar, but note that the impedance glitches are bigger in magnitude, the upper one is at a lower frequency (more mass, or less stiffness?), and effects in the amplitude curve are clearly visible.
Link Removed
W series driver, basically more of the same; not an improvement over the K series.
These impedance and amplitude glitches are clues to cone modes, which make the sound both a little edgy and less clear. It's not uncommon for someone to run a metal tweeter like the TC120 with a midwoofer driver with upper mid/lower treble breakup modes like these, and blame some harshness in reproduction on the tweeter. But the fault doesn't lie there.
I've used one of the 8" V series Focals successfully(a now discontinued model, and the replacement has really strange T/S parameters), but I was crossing it over about 1.25 kHz.
These Focal drivers are measured on an IEC baffle- that's a very large test baffle, so you don't see any effects from baffle step in the midrange.
Seas measures their midwoofers in boxes with small baffles, and so their published curves look funny in comparison to Focal or Eton, because they include the baffle step effect of the test enclosure.
If I was making a small MTM with dual 7's, the driver I would use is the SEAS Excel W0018.
Let's look at it's published plot:

In fact, I'm going to use these later this year to retrofit a set of speakers I designed with Eton 7's and the Focal Tc120dx2, which are discussed here, at Audiomatica.
The Seas driver has a clean impedance plot and response until it's primary resonance at a fairly high 5 kHz. It's feasible to use these up to about 2-2.5 kHz, with the right crossover. They're not inexpensive, at $143 each, but they're good value in comparison with the Focal drivers; actually, better value, I think.
The speaker Tom posted is a design we recently did, the third iteration of an 8" two way bookshelf. This version uses the TC120dx2 with the HiVi M8a; a fourth version is almost completed using a Vifa XT25, which costs about $25 less, and is easier to blend with the HiVi M8a, due to a more extended low end of the tweeter. The XT25 is the first "soft dome" I've heard that I like, but then, it's not really a soft dome.

The HiVi has a primary resonance at about 2.8 kHz. That can be seen in the impedance and amplitude measurements I made; posted below.

Impedance Plot HiVi M8a

Measured Amplitude response, gated MLS plot to exclude room contributions below about 200 Hz.
The HiVi M8a has T/S parameters ideal for tuning for bass extension for a smaller driver; relatively high Xmax (for an 8), 8mm. But sensitivity suffers somewhat because of the longer Xmax and stiffer cone; it's about 2 dB lower than the typical 89 dB for 7" and 8" drivers, which, after you factor in baffle step compensation (if it's to be used away from room boundaries) results in a net sensitivity of about 83 dB- but they go low - pretty flat to 30 Hz in the tuning we use when ported. With an MTM, you'd have a net sensitivity of 86-87 dB. Recommended power is 60 watts per channel and up- they sing very nicely on my Ayre V-5, a 150 watt/channel at 8 ohms NLFB amp.
For grins, I'm including a plot of the HiVi/focal two way ThomasW posted above; this was done in room with long gating, so it shows a lot of minor ripple due to wall and ceiling reflections (64 msec gating window necessary to go into the low end a bit), but it shows the basic integrity of the design.

It uses an unusual crossover design which I can't discuss on the board, because a magazine article is pending.
The impedance tester you mention from Parts Express might be very helpful.
Also, if you get LspCAD, Ingemar includes a program called Just MLS, which you can do MLS measurements with- not the best program, but free, and fairly adequate.
We've found a Behringer mic (ECM8000,

 

[Ron D Core]

Thanks allot for the info, I change my plans so often who knows what I'll end up with, but right now I'm seriously considering the MTM with the M8a's and the XT25. I know that XT25 is a good tweeter since my brother used a couple in his design. It seems to have excellent information retrieval. I was thinking of an EBS alignment (something with an F3 of around 30-40 Hz) which would blend allot better with the subs than the Focals would have. Do you have a site with the crossover you used for that two-way posted above? I want to try and stay away from something really complex, simply for cost reasons. Maybe I could play with bypassing the caps or something. Thanks again, I'll post another crossover when I come up with something that looks right.

 

[Jon Hancock]

Hi Ron,

I change my plans so often who knows what I'll end up with

Why do you think there have been four versions of this bookshelf?
The box tuning on all the version has been 32 Hz; it would be even a skosh flatter in room and go deeper with around 28-30 Hz. Then, there's no problem having good overlap and blending when you use an additional sub; standard active crossover filters really work on the premise that the sub and main speaker have at least an octave overlap. Then, a sub crossover at 60 Hz or 75 will work quite well.
Write me at the SBCGLOBAL.NET address (home) and I'll fill you in more about the design and how we can help. There's a couple other guys building an MTM version, too, plus I'm doing a new dipole speaker which uses these drivers in the 250 Hz up area, in an MTM.
Best regards,
Jon

 

[Danny Richie]

Well I'll be?
This has really been some good info.
Jon Hancock hit the exact things I was thinking when I began to read through Ron's post.
As someone that has been a Focal distributor and worked with those drivers extensively I know exactly the problems associated with those woofers.
It was the high frequency cone break-up and the just flat out un-natural sound of Kevlar and other exotic materials that drove my (at the time) business partner and I to have a custom woofer built that had a more natural sound and didn't require multiple components to control.
When we first began the Paradox-3 (which used our GR-165 woofers and the Focal Tc120Tdx2 tweeter) design we threw a cap on the tweeter and a coil on the woofers and just let them play awhile to burn in.
You can see the final design here: http://www.gr-research.com/paradox/paradox_3.htm
A second speaker was taken into the chamber and a 4th order network was designed for it based on its measurements.
Then we brought it out to listen to it.
My partner and I just looked at each other and said something like oooh, yuck, we have packed so many components into the signal path that a lot of the air, openness and dynamics were gone. Lots of components meant lots of insertion losses.
The first order design killed it.
We then tweaked the first order design.
The Tc120Tdx2 worked well with the first order network and so did our woofers.
A few measurements can be seen by clicking links at the bottom.

 

[Ron D Core]

I am really glad I posted this question. I learned some important things that probably would have driven me nuts if I hadn't taken them into consideration. Thats a good point about the 4th order vs first networks. I thought about that myself, I think I will try to come up with something like a 2nd order Linkwitz with decent components and see how it works.

 

[ThomasW]

Thats a good point about the 4th order vs first networks. I thought about that myself,

Don't get suckered by the 'less = more' theory of XO design. That's the same myth that creates nonsense statements like small woofers = fast bass.

A first order XO requires the tweeter to be used at a high XO frequency so as to not blow up the thing with LFE. The higher XO frequency then forces the midwoofer run higher and that = high frequency beaming.

IMO it's best to have a tweeter XO point that allows the tweeter to safely run as low as possible. That avoid issues with the midwoofer beaming, creates the most coherent soundfield and best sound quality

When done right, higher order XOs can sound very good, if not virtually transparent. But achieving that requires a skilled XO designer.

 

[Danny Richie]

Admin note: please present your side constructively, but leave the insults off this forum. I've deleted the insults out of this post. Do not continue with this line of posting on this forum.
So you think that the degrading qualities of components are not cumulative when more are used?
Each and every component you use in your crossover will create noise.
Ever notice the difference in the quality in caps?
I know you have. I saw you posting info on one of the other well know forums about a rather well known speaker company that used cheap electrolytic caps in their $6,000. design.
Those electrolytic caps do not dissipate very fast. They store energy. This residual charge creates smearing in the music.
This is also a cumulative effect.
If the signal to a tweeter passes through two of them the effect is worse. There is at least twice as much residual charge held.
Poly caps have better dissipation rates and the sound is of coarse much better. They still have a big effect on the signal though. The larger the cap value the more stored energy is held.
This is why by-passing with small values is popular.
Many film and foil type caps are better still, and not just in dissipation. There are many other things going on there as well.
The better caps don't really "sound better" because you are not listening to the sound of the cap. The better caps sound better because they insert less noise.
Last year I assisted one of my customers with a digital network he was working on.
I put together a pair of speakers for him to test it on.
The digital signal passed through a digital pre-amp that adjusted overall gain. Then it was sent through a crossover processor that separated the signal three ways and attenuated the amplitude appropriately to allow the drivers to "crossover" to one another. Then each of the three signals went through a D/A converter right before each power amp.
I can tell you for sure that I have heard our tweeter with lots of different types of capacitors on it and each one sounded a little different. Some sounded better than others, but none of them ever sounded better than nothing.
Also this statement you made "A first order XO requires the tweeter to be used at a high XO frequency so as to not blow up the thing with LFE. The higher XO frequency then forces the midwoofer run higher and that = high frequency beaming."
Is not necessarily true.
Just because a company states that they are using a first order network does not mean the speaker will have 6db per octave slopes.
Unless otherwise specified when a company states what order their network is they are referring to the electrical network not the acoustical slope.
Here are a few examples:
Take a look at this speaker here at Speaker City.
http://www.speakercity.com/GRProject...vProject.shtml
It is an incredible speaker. The reviewer freaked out at how good it sounded.
http://www.geocities.com/Vienna/Choir/4106/
It technically uses a first order network with the tweeter crossing to the woofers at 1,800Hz.
Hard to believe?
Take a good look. I kid you not, and I should know. I designed that speaker.
First of all some tweeters will play down low. And yes you are correct there are several advantages to letting the tweeter play as low as possible.
No matter how good or how fast a woofer is it will not match a tweeter. Kind of like comparing a smaller faster woofer to a larger one, eh?
The Revelator tweeter has no problem playing that low so long as you are not playing the hell out of it.
With 91db sensitivity there is no reason that in an average sized room one would ever need to put much power on it to reach unbearable levels.
Now look at the response of the individual drivers and scroll down to see the crossover design.
http://www.speakercity.com/GRProject/images/xover.jpg
Because there is a resistor in line with the shunt coil it is technically not a second order electrical network. Just like the woofer with its single inductor and impedance compensation network the tweeter network is very much the same.
Granted there is not much resistance there but to maintain the needed time alignment it was necessary.
Note the acoustic slopes are closer to 4th order.
Here is another good example:
http://www.gr-research.com/performer/av1plus.htm
Our A/V-1+ crosses at just below 2kHz with a first order network.
See graph of driver responses:
http://www.gr-research.com/performer/av1_plus_xover.htm
Note that the tweeter maintains a first order type roll off only until it crosses to the woofer. Then as impedance on the tweeter increase power is pulled off of it via a LR circuit, and it falls much more steeply.
How is power handling?
I have never had any come back.
And one day at Gary Dodd's we drove his A/V-1+'s with 400 watt, mono-block, Melos tube amps. We drove the hell out of them for several songs all the way through. They were ripping like you wouldn't believe, but had no sign of stress or break-up. No blown tweeters either.
Admin note, insulting comment deleted.

 

[Ron D Core]

Yeah I agree, and I'm not a skilled x-over designer. I think I will try a 2nd order Linkwitz Riley. Besides, If I mess it up, I won't be too upset since I didn't spend too much on components.

 

[Jon Hancock]

I think this thread has taken an unfortunate turn; the fun of DIY is that everyone can try out and experiment with what works for them.

Danny, of course, isnt' strictly speaking DIY; he's in this as a business, and I can understand from that viewpoint why he would be more sensitive to challenges to his design philosophy.

In the seventies I designed and built several speakers which would reproduce fairly high fidelity pulse waveforms on the design axis. These were three or four way systems, partly due to the limitations of available drivers, and also because of power handling and distortion issues which arise when using first order networks. I eventually abandoned their use, especially because of the well documented issues with polar lobing with this type of crossover using a W-M-T configuration. (see Small, etc.). WMTMW, on the otherhand, as Dunlavy uses, seem to work much better in that regard.

Regarding tweeter power handling, some of the most interesting documentation anyone's been willing to publish on the web is at Sigfried Linkwitz's site; it's recommended reading. Coincidentally, he does use high order crossovers himself with these same tweeters.

It's an interesting question whether the issues with low frequency tweeter power handling or capacitor quality are more telling in the final results of a design- that will have to be up to each designer. I don't have an axe to grind in this, though, because I'm just another interested hobbyist.

Cheers,

Jon

 

[Ron D Core]

I came up with this...It takes into account the simulated diffraction for the baffle I want to use. The impedence curves for the woofers without the box effects looks pretty much exactly like the plot in the manufacturer's manual untill I enter the box parameters. It changes to like the one below. The box is 3cu.ft. with 2 3" x 8" ports. I would tune them deeper, but I was shooting for a nicer looking group delay (assuming it matters). And would anyone know how to adjust the phase angle?