2nd Order 2 way crossover Questions

[BassAwdyO]

I am making a mini portable system similar to the poszuma parts express design, but just a MT instead of a MMT
Link here:
http://www.partsexpress.com/projects...uma/index.html

I am using a 3/4" neodynium
http://www.partsexpress.com/pe/showd...number=275-035
and also the 5" dayton 'fullranger'
http://www.partsexpress.com/pe/pshow...number=295-010


I decided on a 3khz crossover since
A. The midrange driver seems to suffer cone breakups past 3khz(judging by dayton's FR graph and the sample response provided by PE)
B. The tweeter's recomended XO is at 3.5khz, but since I'm going with a 2nd order filter it should be able to get down to 3khz without frying

I have a DCX and tried out crossed actively at 3khz with good results, so I'm pretty set on crossing there...

Anyhow, calculating the crossover components is where the trouble lies....

I really have no preference on what crossover alignment is used, I didnt find there to be much of a difference between LR, bessel, or butterworth when using the DCX, and since this isnt a project where the utmost in SQ is required... it just doesnt matter a whole lot to me!

I used the winISD passive network calculator and got

low pass 8ohm 3khz
9.4uF capactitor
.6mH inductor
High pass 4ohm 3khz
9.4uF capacitor
.15mH inductor

BUT when I calculate the crossover component values on websites which have crossover calcs on them I NEVER get those values no matter what alignment I use.

what I'm really wondering is, how do I calculate the crossover component values given the -3db point target and impedence

none of these calculators seem to tell where the -3db point is at. I need some mathematical equations and although I could probably figure them out on my own, I just dont feel like re-inventing the wheel.

Can anybody help me?

Thank you in advance!!!

[BassAwdyO]

OOPS, this should be in the loudspeakers forum!!! moderators help!


And BTW the really bad part of this whole thing is I already bought XO components for this project via PE and they'll be here wednesday! So you know... if the results are less than satisfactory I'll have to modify them... Even if things work out this knowledge is always useful for future reference

[Navy_44]

Xover calculators almost never give the right values of the xover components, some of the reasons are:

- textbook filters are designed with the assumption that the woofer has a flat SPL curve in the passband and some 1.5 - 2.5 octaves into the stopband. In most cases this is not true because of baffle step phenomenon (even if the driver has a flat SPL curve measured on infinite baffle). To compensate the baffle step the inductor is usually larger than what the textbook formula would give.

- non-flat impedance of the drivers (woofer impedance inductive rise in particular)

- misaligned acoustic centers of the drivers - because of this most properly designed xovers with 2nd order electrical filters have the drivers connected in normal polarity while the basic theory would suggest reverse polarity. The acoustic phase shift between the tweeter and the woofer mounted on a vertical baffle at 3k may be as high as 180 degrees (tweeter's phase being shifted "forward") so the xover becomes 4th order acoustic even if the electrical filters are 2nd order. In extreme cases (large woofer used with high xover point and a tweeter with high Fs) the shift may be even higher (for example 360 deg), in such cases 2nd filters should be connected in reverse polarity and the whole xover becomes 6th order acoustic.

- tweeters usually have higher SPL than midwoofers (especially if the baffle step is compensated), usually this makes the L-pad necessary

The whole matter is quite complicated and hard to explain in a single post. It is also hard to properly design an x-over without impedance and SPL measurements of the drivers in the enclosure used.

PS. In a properly designed 2-way TW x-over the drivers should meet at their -6dB points, not 3dB.

[richie00boy]

I would just simulate the crossovers you have bought parts for using 8 ohm resistors as the drivers.

Then you can run through the designs that the other programs give you and see the differences.

[BassAwdyO]

thanks for the responses guys.... I guess I'll just fiddle around with it until I get something I deem acceptable.... What I'd really like to find is a site which explaines LCR circuits in english. That is to say that they dont automatically assume you know what the variables in equations A B and C are and they explain how they got them....

[HiFiNutNut]

I don't have measurement equipment but I have had good success in designing XOs for my speakers, after numerous failures in a steep learning curve.

Simple calculators never work, because we will need to take the frequency response, phase and impedance that change with frequencies into account, which most calculators don't do.

I use the SPLTrace / SPLViewer to produce the FRD and ZMA files. The files are then regenerated with the minimum phase response with the FRC Excel spreadsheet (it would take quite a bit of effort to experiment it in order to learn how to use it). The FRD and ZMA files are then used in SpeakerWorkshop. After determining what XO to use, I create the Network Goals. With the help of the Goals, it is really easy to derive your optimised circuit. You must enter the correct data including the driver offset.

The Xdir and Edge programs are also extremely helpful, and in conjuction to SpeakerWorkshop, I have no doubt one can produce speakers with very good sound quality easily.

People may suggest not to rely on manufacturers' datasheets but to measure your run-in drivers after mounting them on the panel / box. This is true if you have measurement equipment, but for most of us we don't. With the help of the Edge program I guess one may be able to skip that step.

I found the FRD and ZMA files are highly accurate if you can take diffraction and baffle step into account (with Edge). I used this method to design a passive XO for the MTM part of my WWMTMWW and it sounds as good as it can get, and I have not heard anything with dynamic drivers better regardless of price. The XO designed in this way is faultless.

All of these software are free. They will take you some time to learn perhaps. Once you learn them, it is fun (actually addictive) to design speakers. I can't help thinking of designing a few more pair of speakers of various types.

I hope this helps.
Regards,
Bill

[Navy_44]

HiFiNutNut:

Simple question - how do you determine the acoustic offset between the mid and the tweeter?

[HiFiNutNut]

The centre of the voice coil is the approximate location I used for the estimate. See a picture of a typical driver construction to find out where a voice coil is located.

The accoustic centre changes with frequency so it is not a fixed number. Read the Loudspeaker Design Cook Book for more information.

So for a 5" driver, I set the tweeter offset to about 2-3cm. For a 6" driver, I set the tweeter offset to about 3.5cm.

As you can read from the LDCB, high order XO has less sensitivity to driver offset. This can be confirmed with SpeakerWorkshop.

If you use LR4, for example, when you model it, setting it to 2.5cm comparing to 3.5cm makes less than 1dB difference. Of course depending on the FR this may be important.

[Navy_44]

Have you confronted your methodology with some actual measurement data? If it's not too much work (or you have already done it) could you provide me with just the acoustic phase difference between an 18cm Seas midwoofer and Seas tweeter (both need to have been available in 2004 - I do not have the measurements of tweeters containing "22" or "29" and W18NX) on a vertical baffle, both flush mounted, at 3kHz? The comparison of re-generated and real world phase difference would be very helpful in deciding whether to learn the curve tracer software or not.

[HiFiNutNut]

Remember loudspeaker drivers are minimum phase devices. That is how the phase of the driver can be derived from the frequency response. You would not need to worry about the accoustic phase difference of your two drivers. SpeakerWorkshop will handle the phase relationships of the drivers superbly and you don't really need to worry about it. What you need to worry about is the driver accoustic centre (AC) offset (physical distance) by entering the TWEETER (not woofer) offset value.

I had the same question you raised. In the LDCB there was an example of calculating the AC. In one of the Example Speaker Project, it was shown how to calculate it, if I remember it correctly. I guess from those examples you could find out a typical 5" drivers would have a physical offset of just under 3cm.

Also when discussing the design of the Pheonix by SL, he mentioned that the Seas 8" and the Seas tweeter has an offset of 4cm.

Please check those carefully, my figures came from my vague memery. If you work out different figures, please post them here and send me a private email letting me know.

Comparing to the wavelength of your XO you can work out the phase angle difference.

Importantly, when I model XOs, I would not just check the network with a simple, say, 3cm offset but also check the tweeter with 2.5cm and 3.5cm offsets (knowing that it would never be out of that range). The FR must be acceptable for anyting between 2.5cm to 3.5cm (if my guessed offset is 3cm) then I would prototype it. The good thing is that with a high order network, the offsets are not making a huge difference. So this method is safe.

Regards,
Bill