need help with DIY speaker crossovers

[Ryan T]

Hey guys,


I've been designing and building some DIY speakers for the past 6 or 7 months. For crossovers I've just used a REALLY basic online calculator and I've gotten OK results

http://www.mhsoft.nl/spk_calc.asp

But I know this is not ideal for crossover design. So I'm looking to build a set of speakers and I want to do it right this time.

I have recently purchased a pair of infinity EMIT ribbon tweeters and I'm wanting to build a 2 way system using them. I really like the GR-research M130 midbass and I think it may go well with the EMIT tweeter. I know that the EMIT can be crossed as low as 3,500 Hz provided its a 3rd or greater order crossover. So thats the Xover point i'm thinking of going for. I've downloaded Speaker workshop and I'm planning to make a jig, buying a measurement mic and pre amp etc.

The problem is I cant seem to find any solid info on crossver design. I think SW lets you test the crossover and optimize them but I cant seem to find anything to give me the base values to start with (other than the site I mentioned above). I also dont have any of the T/S params for the EMIT and dont know how much of an issue that is. There is no mention of getting the TS params for tweeters in any of the walkthroughs for SW so I'm stumped.


I want to build a good set of speakers and I know the crossover is one of the most important factors so I really want to do it right this time. Can someone give me some tips on how the get started??



Thanks,
Ryan

[Saurav]

Well, since I'm the one who suggested you come to this forum and ask your questions here, I should be the first to reply I guess

Here's what I do with SW. First you'll need to measure your tweeters to find out what their response curve is. Let's say their output shows up as 90dB (or whatever). You then create a filter network with this driver, and put in something close to what you want. So say a 3rd order Butterworth high-pass at 3500Hz (you can specify that in SW and it'll insert the XO components for you, using the same formulae as the other software you're using). Then you tell SW to calculate the final output of this network+driver, which will not be what you want, because the driver's output isn't a straight line.

Then, what you do is create a target for your network. So if the tweeter is 90dB where it's flat, your target would be a curve that's at 90dB, and say a 4th order curve at 350Hz. Once you have a target, you can tell SW to optimize the network to get the final output close to the target.

This is where I go into guesswork and eyeballing. Maybe there is a better way to do it, but I don't know one. Basically, if it looks like the tweeter already has a 1st order-ish rolloff and you want a final 4th order rollof, try adding a 3rd order network and letting SW optimize that. If the output doesn't look good, try a 2nd order network or 4th order network too. Try changing the optimized values by hand to see the effect of increasing this inductance or reducing that capacitance. Sounds really hokey and time consuming, and it is, but it's fun and you sometimes learn something. I can hear the real designers here laughing as they read this

And once you have something that looks good in the simulation, build it and measure/listen. What looks best on paper isn't always what will sound best to you.

Hope that helps a little

[Ryan T]

I think I'm understanding a bit more . I read through that Claudio walkthrough. I think that before I have any really good idea what I'm doing I'm gonna need all the equipment and drivers in front of me. Then I can measure them and start messing with the different features etc.


For the tweeter I'm still a bit confused. Do I use that website I linked above to get a rough 3rd order Xover then measure the FR using a mic and SW then try and smooth it out?





Thanks,
Ryan

[Evil Twin]

Get a copy of Speaker Workshop. It's freeware.




Speaker Workshop


You'll use it to measure the drivers in your cabinets, using MLS to reduce room interaction. You don't need T/S parmeters for the EMIT; you're not designing a low frequency enclosure. The program you're using is asking for those so that it can estimate the LF impedance curve, which can interact with the crossover. There's no substitute for measured data. The EMIT will probably have a pretty flat impedance curve, most leafe/planar tweeters do. Look at the curves for the B&G Neo3; probably very similar, as dirver construction is quite similar. (Hint: Google is your friend).

With the measured driver/cabinet data, you can select acoutic targets for the high and low pass functions of your crossover after examination of the speaker response. The program has an optimizer which when you run it on each section will adjust component values to get the best fit to your target.

There's a nice demo of a complete project on the site, too.

You may have trouble matching up an EMIT to a midwoofer- the EMIT may not go low enough. If you don't get the results you expect when you finish your design, try the same one but with a SEAS 27TDFC- very nice little tweeter, more uniform dispersion than the EMIT, lower distortion, and costs only $28 in the US from Madisound.

It sounds like you'd benefit from more background knowledge on the design process. Download the manuals and demos from all the commercial speaker software makers, like LspCAD and Soundeasy, and READ, READ, READ. Also, "Loudspeaker Building 201", available at Parts Express, would be a good reference and tutorial.

Good luck, and most important, have fun!

[ThomasW]

You might to take a look at Mark K's testing of the M-130. That's not a driver I'd want to be using with a high XO point.

[Ryan T]

I've got speaker workshop and I've been messing around with it. I've also read through the CLaudio walkthrough and think I can fallow that pretty easy. I'll hunt around for those manuals and demos! I also may try that book.


I wasnt aware of any major problems with the m130. But the other drivers I was thinking of using are the dayton reference series.

http://www.partsexpress.com/pe/showdetl.cfm?&DID=7&Partnumber=295-362

I really like the sound of the AV 1's midrange and I beleive the crossover point is at or around 3,500 Hz. But the GR tweeter never really sounded "smooth" to me. It isnt bad but it tends to be a little hissy sometimes. So really my goal is to get clean midrange (like the AV 1's) and clean highs. I dunno if its even possible for my budget but i'd like to try!

So far with the limited testing i've done on the EMIT it is very clean sounding. But the problem with the EMIT is that it has to be crossed over above 3500 Hz. If there are some midbass drivers you would recommend I look at i'm open to changing out the midbass. The main factor is cost though . I was hopeing to keep the budget around $150~$200 for the two midbasses, crossovers and mic/preamp. I know that is quite a bit of stuff to cram in a small budget but I'm hopeing to get the Behringer EC8000 and the behringer pre amp on ebay for under $80.





Ryan

[JonMarsh]

Hello Ryan,

Part of the tonality of a driver is due to distortion products, both linear and non-linear. The EMIT is similar to other tweeters requiring high crossover points (small ribbons come to mind), in that they don't have much diaphragm excursion capbility, and they're really only suited for three way systems.

On the lower end, larger midwoofers have problems going up high- somewhere not far above 1500 Hz or so, most get into cone break up modes of one sort or another; not all of these are obvioius peaks (some show up in on axis amplitude as dips), but they all involve energy storage and degradation of sonics.

The AV1 tweeter not sounding smooth may just have likely been distortion products from running the woofer up too high. Think about it- 2nd harmonic of 2.5 kHz is 5 kHz; 3rd Harmonic is 7.5 kHz. A good test for this is to listen to the speaker with the tweeter disconnected, and see how much grunge or grit comes from the midwoofer. This is a standard test I do while tweaking my designs. You'd be surprised how many speakers don't sound like a nice clean roll off from a tone control on the top. If you check the link ThomasW provided, you'll see that there are some problems with the cone behavior of the M130 which can be seen in the near field behavior - this is due to non-pistonic cone operation. If you do ETC cuves on this driver, it will show some energy storage phenomena related to these frequencies. Now, you may not notice or hear this at first, but in A/B with a speaker that is operating pistonically, there is more coloration in the midrange, and it can't be EQ'd out by level, because it's as much time related (smear from energy storage).

My own personal preference is to only use drivers within their nearly pistonic range. This limits the best of 8's to about 1200-1400 Hz, the best of 7"s to about 1800 Hz, and a really good 5 you might push to 2500 or 3000- like the Eton 5-880 (newest versions).

Consider also that the dispersion of midwoofers narrows significantly with increasing frequency- for good off axis power response, I wouldn't run a 7" any higher than 1400-1800, for example.

If you're interested in the Dayton Reference series, take a look at the threads about the Modula MTM using the RS180- there might be some useful discussion in there for you.

If you really have your heart set on taking a 6-1/2" woofer up to 3.5 kHz, you might have better luck with a Seas CA18RNX. They're about $50 each from Madisound.

Sometiems the only way these concept really settle in is by trying them out, and hearing for yourself how things sound.

So, if you build these speakers and aren't happy with how they turn out, come back and talk to us. One option would be to turn them into a three way (add a midrange; expensive crossover wise); another might be to use a tweeter better suited to a two way, having low distortion, and extended, smooth response- like that 27TDFC I mentioned earlier.

I'm just suggesting you do some research, to get the most out of your driver expenditure. Sometimes it's better to approach a speaker project with a set of requirements in mind, rather than a fixed set of drivers.

Regards,

Jon

[Ryan T]

I dont have any major preference on woofer size or anything. I really have no problem taking my time on this build.

What I want to achive is a 2 way speaker that is really clear and accurate for vocals and insturments etc. Bass and high SPL's arent nearly as important. I would like them to extend down to 60~70 Hz but like I said clearity is the most important thing I'm going for.

So if it would be best to get a new tweeter and midbass that work well together I'm all for it. I knew coming into this I may not be able to build a nice set of speakers with the EMIT's but I went a head and bought them since they were only $25 for the pair. They were from some infinity RS5000 speakers which had the midrange and 10" woofer replaced with *gasp* radio shack drivers lol. But since I really only wanted the tweeters it was no biggie.

So I guess my two options are

A:
use the EMIT's and maybe some small 4" midbasses. Use SW to try and get a clean FR but not hope for much.

Or B:
Start from scratch with the 27TDFC and a slightly larger midbass like the dayton ref 5" or 6" and and shoot for a lower X over point (around 2,000 Hz ?) and go from there.

At this point i'm not opposed to starting from scratch with new drivers. I do think I want to do some sort of project with the EMIT's but I may just shelf them for now. Budget wise I guess I would belooking at around $180 or so for the tweeters, midbasses, and crossovers if I start over. I guess I have to decide if I'm happy enough with my AV 1's or if I want to set a much larger budget and design my own .

[Ryan T]

Well after messing around with my AV 1's I think I've narrowed it down to the tweeters being the main culprit. The only thing I didnt like about the AV 1's is that they sometimes have and sharp S's and T's. On a few occasions they sound a little tinny (but usually only on bad recordings).

My reference point is my HD-580 headphones in terms of tonality etc. I listened to just the midbass and while it did have a little bit of a "hissy" sound in the high freq range it was very light. Once the tweeters are playing I think it really overshadows the problems with the M130's. I dont know if its right to compare my HD-580s to the AV 1's but the 580's have some of the cleanest sound i've ever heard so I've been using that as a reference point. With the 580's there is a little bit sibilance but it sounds natural. While with the AV 1's it seems to hang longer almost. I suppose that could be because the midbass and the tweeter are hitting those higher freq together and causing problems. To my ears almost everything is perfect with the AV 1's. I just wish the tweeter was a bit cleaner. I dunno if i'm explaining this very well.





Ryan

[Ryan T]

I think what I'm gonna do is get the mic and pre amp for SW first since I need that no matter what drivers I choose. Then I'll barrow one of the M130's from my AV 1's and build a test network at 3,500 hz. I'll try it with the EMIT tweeter and see how it sounds. If it sucks I'm only be out the cost of one crossover and an enclosure. If I can tweak it to sound how I want I'll just purchase two M130's and build the second enclosure.

It'll probably be a little while till I get the stuff peiced together but in the mean time I'll try and read some more on crossover design. If you have any links that would be cool, otherwise I'll just start googlin. I really do appreciate your guys help!



Ryan

[GrahamT]

There are lots of good sites, madisound for one. Try searching here too. Jon mentioned the "Loudspeaker Building 201", no wait, that was Evil Twin? He hasn't been his "irascible" self lately. Something is probably going well at Coruscant. I noticed that an error was pointed out in AudioXpress (another good DIY resource, but usually over my head) that made me nervous about getting that book. I cant remember what the error was but my jaw dropped.

There are good books but Vance Dicakson, John L Murphy, Davib B Weams, Joseph D'Appolito and others. I like to follow projects and see why designers do what they do. Watching them and how they choose to solve problems is helpful. You can find a lot of these projects here, and at the other DIY audio sites.