X-over design (Formerly Help with LEAP sim)

[Forte_II]

I need help figureing out this LEAP sim. I am blind and of course it is in a pdf doc that I cant read with a screenreader... So if anyone with experience with Leap sims could please look at this and tell me what they think I would really appreciate it.
Edit: This has turned into a total redesign. Starting over from square one. So the next few posts arent relavent to the current discussion. Thanks to everyone who has and is helping me with my first Speaker project.

[cjd]

sure. what kind of information are you looking for?

The problem is probably not so much that it is a PDF as that it is a bunch of images in this PDF, nothing more. Regardless, I suspect there are a bunch of folks around here that would be happy to help out.

[Forte_II]

OK, I hope this works. Here is the pdf file.
take a look and tell me if it looks like a good x-over design. Should I have it built?

Image not available

What slopes and alignments are they suggesting? What is the x-over point? Is the phase and power good? Thanks in advance.

[cjd]

Unfortunately, when I try to download the PDF I get an error:

"User is not allowed to use direct links."

You are welcome to try e-mailing me. chris at eldamar dot net

C

[Warpdrv1]

I was able to grab a copy, and I have forwarded it to Chris for you Forte_II.

Warp

[Maximiliano]

Inspection of the primary inductor value and the response curves suggests that this crossover design did not consider baffle step loss. This crossover will make the speakers sound very bright unless this is for in-wall speakers.

Max

[Dennis H]

Chris, I got there with IE. It's too big to post here as an attachment so I temporarily posted it to my web space where you can grab it. Sorry, I can't tell if it's a good XO or not.

Edit, warp already sent it so deleting the link.

[cjd]

So there is perhaps 2dB of rise on the low end (perhaps an attempt at baffle step compensation?), but this will still sound quite thin I think. It appears to be a 2nd order Linkwitz-Riley acoustic slope at 2800 Hz.

The network, in case you've not seen it, is very basic. On the mid-woofers, we have .62 mH inductor, 12 uF cap, and a zobel of 6.8 uf cap and 4.7 ohm resistor. On the tweeter, it's an 8 uF cap, .3 mH inductor, and a 50 ohm resistor in parallel with the tweeter.

I must admit to being a little skeptical. They specify a 50ohm resistor across the tweeter drops the impedance very slightly and nets a slightly closer follow on the ideal L-R slope. Looking at the response it almost seems the response would be better without this resistor. Certainly its impact is very minimal. Without having access to phase data it's nigh impossible to really evaluate the truth to this.

I should add, I did some very quick modeling, but did not use data for the specific drivers you have here. I picked drivers with similar impedance characteristics and evaluated the transfer functions and impedance plots.

I have to admit being a little concerned because this almost looks like they've not used any phase data whatsoever on the driver response (if I do not load response data it models out with a very nice transfer function - soon as I load real drivers in I get some much lumpier response).

Thanks Warp (and Dennis) for your efforts. IE?! What the heck.

C

[Forte_II]

thanks. I hope it wasn't a waste of money and time... I look forward to your opinions. Thanks again.

[Forte_II]

Hi Cris. That sounds like a bad xover. I am not impressed so far. I cant beleave they charged 30.00 for that.

[Dennis H]

IE?! What the heck.

Yeah, I know. I run some stuff for work that doesn't seem to work as well with the others so I live with IE's quirks and make sure I keep all the latest IE exploits patched.

[cjd]

I write software for browsers.

We have to hack the code to get it to WORK in Internet Explorer. And we have to hack each version separately, because they're all broken in different ways.

Forte_II - it's probably as good as a lot of commercial crossovers out there. And for $30 it's not much more than I would expect. I estimate I spent about 40-50 hours (though it's really hard to quantify accurately when the process was often five minutes here, twenty minutes there) all told working on the crossover for Ryan's latest project (the Khanspires). Preliminary simulations, some box modeling, revised sims, measuring, more measuring, some modeling, tweaking, testing it out, more measuring, more tweaking... And, of course, it wasn't exactly short turn-around time.

The biggest issue here is that we simply can't begin to guess how well it will work because the PDF simply does not provide all the data.

On the flip side, it should not be all that expensive to put this one together and see how it sounds to you. I'd be pretty confident suggesting you could leave the 50 ohm resistor out for a preliminary build. I wonder if we can find someone local (or local enough) to you that could do some in-box measurements. Perhaps we can find data on the mid-woofers that could be used with data already available on the tweeter to do some more in-depth simulations with driver specific data.

C

[Warpdrv1]

I had no problem downloading with a current version of Firefox....

[Maximiliano]

Roman has frd and zma files for the midwoofer:

http://www.rjbaudio.com/Audiofiles/Driver%20FRD%20files.html

[Forte_II]

I have the pdf for the mids and the tweeter. I can do measurements. What do you want?

[Forte_II]

thanks Maximiliano . Anyone got Any tips on measureing in box response with True RTA? I have a CM100 mic and a M-audio Delta 44 sound card with a Behringer mixer/mic preamp?

[Forte_II]

pdf for mids
and Tweeter

[cjd]

Hmm. You could attempt Speaker Workshop for measurements, which is free. But a little obtuse. Measurements, in-box with the mic on the listening axis. Mid-woofers wired together. Tweeter. Separate measurements, without moving the box or the mic.

TrueRTA may do the necessary work as well, I don't know.

I want to make sure it's clear, integration may be fine with this crossover, and only a lean bottom end may be an issue. It's just not possible to tell given the data available in the LEAP PDF.

Let's see what can be done with the available files though.

Can you give us more information on the box you're using?

[Forte_II]

Box: Front baffle is 8-5/16 with a 45* bevel to 5-7/8 where the drivers are mounted. 18" high with a 3/4"bevel. The drivers are in a MTM arrangement set 1-3/4" from the top. Mids are 9-7/8" c to c. No driver offset was used. Box vol. is 0.47cuft and 60% fill with acoustic foam. These are the MTM of a modular MTMw system. I am still working on the bass bin. So performance below 300Hz is not an issue. The goal is to have tight fast prescise mid and real sounding highs.

I am running them with a CX 2310 set at approx 2200hz. This is a 24db L-R active x-over. Tweets are down 3db. I listened to them at 2800hz and they still sound good. Not sure if I like them crossed that high. more listening is required. I will get mesurements today. Thanks Cris.

[Maximiliano]

Hi Forte,

I performed a simulation of the LEAP crossover you posted, using modeled FR and impedance files. The modeled FRs of midwoofer and tweeter reflect box and baffle simulations based on the information you gave us. Acoustic minimum phase data were extracted from the modeled FRs by HBT. Modeled FRs has extended roll offs over extreme frequencies so the HBT computation may be reasonabley accurate. The result is attached to this post.

As cjd pointed out, the LEAP design doesn't seem to have used correct acoustic phase information. Phase is far misaligned, resulting in a wide 3 dB dip around the XO freq. The XO point is around 2.8 kHz. Did they intend to put a BBC dip? I don't know. They used some BSC, but only a little. Did you tell them that you're going to use this speaker only above 300 Hz? Perhaps that's why they used small BSC?

[Forte_II]

Thank you for doing that. How do I correct the phase problem? I am going to ask Madisound about that.
I didn't tell them it was going to be part of a 3 way. I wanted this to work as a 2 or 3 way system.
<edit> I won't have actual fr measurements until mid next week. I am waiting on some cables and decided to finish the cabinets while I wait. (Gloss black(

Hi Forte,

I performed a simulation of the LEAP crossover you posted, using modeled FR and impedance files. The modeled FRs of midwoofer and tweeter reflect box and baffle simulations based on the information you gave us. Acoustic minimum phase data were extracted from the modeled FRs by HBT. Modeled FRs has extended roll offs over extreme frequencies so the HBT computation may be reasonabley accurate. The result is attached to this post.

As cjd pointed out, the LEAP design doesn't seem to have used correct acoustic phase information. Phase is far misaligned, resulting in a wide 3 dB dip around the XO freq. The XO point is around 2.8 kHz. Did they intend to put a BBC dip? I don't know. They used some BSC, but only a little. Did you tell them that you're going to use this speaker only above 300 Hz? Perhaps that's why they used small BSC?

[Maximiliano]

Now attached is the simulation result of a crossover I designed. Let me describe it for you. In the LPF section for the midwoofers (connected in parallel), I used a .85 mH (.3 DCR) series inductor and a 25 uF shunted capacitor---simple 2nd order electrical. This is sufficient for obtaining LR4 acoustic rolloff. Zobel is not necessary.

Tweeter net is electrical third order with an LCR trap. The LCR trap is necessary, IMO, to suppress the high resonance of the XT25 at 500 Hz. The tweeter net uses a 13.5 uF primary series cap, a .15 mH (.15 DCR) shunted inductor, and a 30 uF series cap. The LCR trap uses 1.0 mH (.7 DCR), 100 uF, and 4 ohms. Tweeter SPL is a bit attenuated by a .75 ohm resistor before the network. This is adjustable by trying resistance between 0 and 1 ohm.

I didn't use a large amount of BSC since you're going to use this speaker above 300 Hz. The FR is pretty flat above 300 Hz. Below 300 Hz, it has a gentle rolloff down to about -5 dB at 100 Hz. Phase is well aligned around the crossover freq, which is about 2.1 kHz.

Hope this helps.

-Max

[cjd]

It probably requires someone to re-do the crossover taking phase integration into account - something Leap simply may not do.

Maximiliano has already done the hard part: pulling the data in, summing with baffle diffraction/step loss, hilbert transform, etc. Should be no difficulty at all for him (or someone else) to put together a different crossover option. I was planning to spend some time this weekend going through this process - if Maximiliano will share those files it'll short-circuit the process even more.

C

[Maximiliano]

Thank you for doing that. How do I correct the phase problem? I am going to ask Madisound about that.
I didn't tell them it was going to be part of a 3 way. I wanted this to work as a 2 or 3 way system.
<edit> I won't have actual fr measurements until mid next week. I am waiting on some cables and decided to finish the cabinets while I wait. (Gloss black(

I think I should correct what I said. The phase of the LEAP design is not far misaligned, but somewhat misaligned. About 50 degrees off at the xo point. But this becomes worse as the frequency goes down, and results in a 3 dB dip between 1.5 kHz and 3 kHz.

[Forte_II]

I really appreaciate you guys doing this for me. Thanks again.
Is LEAP available as a demo? I don't want to buy something I cant use with my screenreader.
Like I said earlier I will have actual fr measurements mid next week.

[Maximiliano]

One thing to add. I used an about .65" Z offset for the midwoofers (i.e., the woofer's acoustic center is .65" behind the tweeter's on the baffle) in PCD.

cjd, do you need box and baffle effect added frd files or raw ones? What is your email address?

-Max

[ThomasW]

Randy,

Don't concern yourself with LEAP. It's extremely expensive, and it's unnecessary for your project.

[cjd]

I would be very much surprised if any speaker design software at all were particularly useful with a screen reader. Unfortunately, a very large portion of the data we can interpret at this level is visual.

Max, whatever you've got could cut time off the process I was going to take. For that matter, you could very possibly take this into your own hands and tweak things till they're right.

e-mail is in my profile - chris at eldamar dot net

[Forte_II]

Ok, I blew the tweeter with some feedback. Hit the wrong button on my mixer. So Tweeter response will have to wait. Here is the in box FR
for the Vifa PL14's No x-over full range test signal. I used the test chirp in True RTA. It goes from a low to high freq. in a split second. On axis 1 meter.

[Forte_II]

I got a 1/24 oct response exported to a excel file. can someone turn it into a graph? and maybe give me some feedback... This is with the speacker in its normal location in the room mic 1 meter on axis. Full range signal, no filters.
You will need to save target as and then change the file extension to .xls it wouldn't let me upload it as a spreadsheet.

[Dennis H]

Here's the graph as a gif file. I included a smoothed line, roughly 1/3 octave. It looks pretty flat from 100 to 2000 except for the BSC dip at the lower frequencies. The big notches between 100 and 300 are most likely floor bounce. I've also attached a zipped version of the xls with the graph and the extra smoothing in a new column.

[Forte_II]

Thanks, So what do you think of the FR in respect to the box alignment I am trying to learn all I can...So can I discern anything from this curve? Bigger or smaller box vol., More or less acoustic material? Anything about the upper response from 2k to 6k? I have some ideas but I would like to hear from people with experience.

[Forte_II]

Hello...heellloo...hheellloooo...anybody...body... out...out...there...there?

[ThomasW]

Not sure what to say that hasn't already been covered. The pdf of the woofer shows a clear cone mode at 1500Hz.



Tweeter looks okay down to 750Hz, obviously don't run it that low..



Since you're using the Behringer active crossover I'd try crossover points starting at 1700Hz and go up from there. Be careful not to go loud at 1700Hz.

Given that these little midwoofers don't have a ton of Xmax it's probably best to keep the box you have, since you're going to add a woofer eventually.

[Maximiliano]

Thanks, So what do you think of the FR in respect to the box alignment I am trying to learn all I can...So can I discern anything from this curve? Bigger or smaller box vol., More or less acoustic material? Anything about the upper response from 2k to 6k? I have some ideas but I would like to hear from people with experience.

Looking at the FR, not much to tell about the box properties. IIRC, you used a good volume for sealed application of those midwoofers. In a closed box, it is generally good to fill the cabinet %100 except right behind the woofer.

The measured high frequency response rolls off much faster than the modeled one. I don't know why. Due to your measurement setup? Room?

Have you tried the xo I modeled for you? If you have a Behringer DCX, you may be able to try the transfer functions I modeled (I mean something similar), and compare it to the Madisound one whose transfer functions I also gave you in one of my above posts.

Max

[Forte_II]

Sorry Max No DCX. Just a CX2301... I have the x-overs from the Madisound sim... As soon as the new tweeters come in I will get measurements of the tweeter response and then the system response with the Madisound x-over.
I am just trying to learn all I can. As for the fill, it is just like you said. fill everywhere except behind the woofers. I figured approx. 60-70% fill.
Thomas, I appreaciate the help... But if it isn't in ASCII text I can't see it. So I am not seeing this thread the way you are. read it without the attachments that don't use text and that is what info I have so far. Pretend this thread can't have pics and just explain things. I know this is probly a pain in the A$$...but it is all I can do.

[Maximiliano]

As soon as the new tweeters come in I will get measurements of the tweeter response and then the system response with the Madisound x-over.

If you can afford, try my crossover, too. I'm pretty sure you'll notice a difference.

Max

[ThomasW]

The pictures were actually for other readers benefit. They're just screen shots of the frequency response plots from the Madisound driver pdf files.

The pictures were being used to reinforce the text I posted regarding woofer's 1500Hz cone breakup, and the fact that the tweeter is basically flat to 750Hz.

[Forte_II]

I just wanted to see what the cheap thrown together Madisound ones would sound like. Then I was going to use one of your guys designs.
To be honest I was excited when Cris said he would work on it... But he is a busy guy. Max how much experience do you have designing X-overs? Do you need any more info? Is the one above a prelimary design or the final version? Thanks again Max.

[ThomasW]

To be honest I was excited when Cris said he would work on it...

Perhaps it wasn't obvious but...

Once Chris saw the crossover work Max completed, I thought Chris handed off the design work to Max.

[cjd]

I'm still alive, just had to focus on finishing off a home improvement project. This is potentially train-time work for me though, so I'll try to do some work there.

[Forte_II]

Thomas, thanks for the help. I was refering to the statement "Not much else to say that hasn't already been said above". Unfotunatly I am lost except to say Max has a x0over for me that will work better and the Madisound is not very good since they didn't account for phase. I wam not sure why one is better than the other. I wish I could see the scematics. Oh well... Thanks anyway.

[Maximiliano]

Max how much experience do you have designing X-overs? Do you need any more info? Is the one above a prelimary design or the final version? Thanks again Max.

I built three designs of mine. All 2-ways. One is MTM. I used simulation softwares to design XO's and then fine tune them by my ear. All worked pretty nicely. Fine tuning was not a complete change of the XO designs. Just padding resistors for tweeter level and inductor unwinding for BSC adjustment. I'm pretty sure that my xo will work better than the Madisound's. But who knows? The latter may sound better to your ear

If you want to use the design without a bass module, you'll definitely need more BSC.

Max

[cjd]

Even with a bass module it'll probably want more BSC assuming the bass is being crossed in actively.

C

[Forte_II]

Cris and Thomas you guys got in before I posted.
Thomas, I was not sure if Cris was going to look at this or not. It did kind of sound like he handed it off to Max. But then he asked Max for files . So...
Chris, Any input you feel like giveing is welcome.
Max, I hope I didn't offend... I didn't mean to. I am 199% sure yours is better than the Madisound. I am jsut tryign out things and seeing what differences I can find and what does what better.