Bandpass formulas dilemma - newbie Q

[grantnsw]

Hi! Its great to discover this wonderful group.
I'm a new member, (beginner) reading Vance Dickason 1997 'cookbook'.

I'm wondering if someone could steer me in the right direction please?
I want to design a 2nd order Linkwitz-Riley passive 3-way crossover for some Vifa drivers I have lying around: D25 ferro-fluid dome tweeter(6ohm), D75 dome midrange and P25 10" poly woofer. (I realise 3-way designs are problematic; especially for a newbie). The D75mid is not very good? - the 60deg off-axis response is down 9db at 3000Hz, compared to on-axis.

Looking at the freq. response graphs, x'over freqs of 800 and 3000Hz seem appropriate?

A LR-2 high pass for the tweeter seems straightforward , plug Fh=3000 into the equations to get L and C. (The attenuation at resonance should sufficiently reduce the power to minimize distortion, I hope).
Similarly for the woofer low pass FL=800, the attenuation higher than one octave above this should minimize unwanted summation; (plus an RC impedance equalizer).
Seems reasonable so far?

My problem is for the bandpass midrange. I assume the highpass section will be set at 800Hz to allow the higher frequencies through. Is this correct?
I also assume the natural roll-off of 5dB/octave (below 800) combined with the theoretical(?) attenuation of 12dB/O for the LR-2 circuit will sufficiently reduce the power at resonance.
Similarly the lowpass set at 3000Hz will let lower frequencies through?
This seems reasonable to me, or is my thinking fundamentally flawed?

The reason why I'm asking is that the 'cookbook' offers APC (A) and (B) formulas for 1st to 4th order. The (B) being for x'overs at 625 and 5000Hz. With Fm=sqrt(Fh X FL) being the freq. used to calculate BOTH the highpass and lowpass midrange component values! In the case above, Fm=sqrt(800X3000), i.e. 1550Hz.

Could one of you learn-ed folk clarify this for me please? Any help or
reasonable suggestions would be very much appreciated.
One last question (can't help myself!) would I be better off 'junking' the D75's and buying 5or61/2inch woofers instead; with the 10" wired as a sub?
These questions probably seem incredibly nai-ive to you, but thanks for patiently reading it. best wishes, Grant
(I searched but found nothing appropriate)

[richie00boy]

Your thinking seems pretty sound so far. You seem to be going some way to taking acoustic responses into account to arrive at final target responses, which is good. Have you taken level matching into account though?

The A and B formulas for bandpass are for when the Fl and Fh are sufficiently apart not to interfere with each other, and when they are so close that they interefere with each other so some fiddling is required. I would strongly suggest you try to make them not interfere with each other.

[e-side]

Originally posted by grantnsw

A LR-2 high pass for the tweeter seems straightforward , plug Fh=3000 into the equations to get L and C. (The attenuation at resonance should sufficiently reduce the power to minimize distortion, I hope).

(...)

My problem is for the bandpass midrange. I assume the highpass section will be set at 800Hz to allow the higher frequencies through. Is this correct?
I also assume the natural roll-off of 5dB/octave (below 800) combined with the theoretical(?) attenuation of 12dB/O for the LR-2 circuit will sufficiently reduce the power at resonance.
Similarly the lowpass set at 3000Hz will let lower frequencies through?

The reason why I'm asking is that the 'cookbook' offers APC (A) and (B) formulas for 1st to 4th order. The (B) being for x'overs at 625 and 5000Hz. With Fm=sqrt(Fh X FL) being the freq. used to calculate BOTH the highpass and lowpass midrange component values! In the case above, Fm=sqrt(800X3000), i.e. 1550Hz.

Hello Grant

When using those formulas be sure you use the impedance of the driver at the chosen cross-over-frequency, or use a Zobel network to keep the impedance constant.

The impedance of a driver varies with frequency. Also, because the voice coil acts like an inductor the impedance rises when frequency increases. Therefore you should measure the impedance at your chosen cross-over-frequency to get useable inductor- and capacitor-values.

I haven't built a three-way system already, but the solution you give for the midrange x-over seems correct te me. You have to calculate the center frequency of the midrange's passband, so in your case it becomes SQRT(Fl*Fh) = SQRT(800*3000) = 1550 Hz, as you already mentioned.
The frequencies you used for the lower (Fl) and higher (Fh) midrange cutoff frequencies are correct.

I'm not sure how to wire the woofer, mid and tweeter with those LR2 cross-overs, but make sure that the drivers are in phase. Also, you should take the distances between the drivers into account, these should be less than 34400/(crossover-frequency)/4 cm. Because this is often impossible, just keep the distances as small as possible.
You might search for baffle step too, you may adjust the x-over-frequency between the woofer and midrange so that only the woofer 'suffers' from it.

Good info about passive crossovers

Hope this helps.

Erwin

[Ron E]

I'll tell you right now, your chances of designing a three way crossover properly with cookbook formulas is next to nil.

They make decent starting pooints, but with real drivers the sum is usually not very flat. If you want to design a three way, get speaker workshop.

[sreten]

Hi,

just backing up RonE somwhat :

The dome midrange will almost certainly need a conjugate impedance
network for its fundamental resonance (depending on c/o frequency).
Croosovers have target acoustic not electrical slopes, without full
measuring and simulation software 3-ways are almost impossible to get right.



shown is the response of a 3-way with a "stock" crossover, from :

http://www.speakerbuilder.net/web_fi...a/lyramain.htm

/sreten.

[grantnsw]

Thank you very much for your helpful replies richie00boy, e-side and Ron E.
I just found an old Vifa brochure that recommends these 3 drivers in a 3-way! (with a DCO 3/120?-X'over).
Maybe I should see if Vifa(now Peerless?) Australia still has circuit of this crossover?

Otherwise, (thanks Ron E) it seems that the 'best' initial design, however, will only be a very crude starting point, requiring lots of trial and error experimentation and the purchase of some costly test equipment. So maybe I should shelve this idea, right now, and concentrate on finding a well accepted 2-way design using quality 5 or ~6inch woofers suitable to my D25 tweeters instead. Maybe Peerless/Seas/Audax etc are affordable? (I could keep the P25's, 10inch, as side-fired subs? or even an isobaric? <grin>)

Anyway, firstly to richie00boy,
level matching via voltage dividers for tweeter and mid seems necessary.
Yes FL and Fh do seem too close. Freq. response graphs show ~90dB for each driver in their usable ranges, so the tweeter would only be 18dB down (or less) at 1500Hz resonance - not good enough? Solution, go to higher order and/or move x'over to 5Khz. BUT, D75mid looks fairly useless above 3Khz due
to seemingly poor dispersion characteristics off-axis.

hi, e-side,
I think I now know Zobel means RC impedance equalization and that impedance measurement is mandatory. Maybe I will find a local audio club who has the gear! Also, I can only assume that for a 3-way LR-2 the mid needs to be reversed. And thanks for the link, have just started to read.

hello Ron E,
Understood, thank you! I do have latest Speaker Workshop but have downloaded the manual twice and unzipped only to get 'unable to load graphics conversion' in wordpad/write. So haven't used the prog yet.
btw, I like the 'quote'.

Thanks everyone, I haven't totally given up on this idea just yet, but it is looking a tad too difficult, unless I
can find the circuit schematic for the above recommended x'over or a local 'audophile' with some test gear or get the SpeakerWorkshop manual. best wishes, Grant

[Ron E]

The speaker workshop manual requires MS Office 97+. It has embedded pictures, spreadsheets, and programs.

I downloaded it some time ago and you can at least view the text and pictures with Open Office, which is free. I toyed with the idea of editing (Paul Butterfields???) nice, but long, manual into a reader's digest version with just brief how-to's rather than the long explanations, but I don't have time....

[grantnsw]

Hi Sreten,
(I wonder if anyone is still reading this thread?)

re the Lyra, those Dayton drivers may not be too dissimilar to my Vifa's? I have just downloaded some info from Partsexpress, so
just maybe a slightly modified x'over will work? Any ideas would
be very much appreciated.... thanks Grant

[sreten]

Originally posted by grantnsw
Hi Sreten,
(I wonder if anyone is still reading this thread?)

re the Lyra, those Dayton drivers may not be too dissimilar to my Vifa's? I have just downloaded some info from Partsexpress, so
just maybe a slightly modified x'over will work? Any ideas would
be very much appreciated.... thanks Grant

Hi,

The design process of the Lyra is the sort of thing you need to follow.

Personally for the Lyra design I would have offset the mid/treble
drivers to give handed pairs and possibly smoother response.

You would need to carefully check the similarities of the drivers
before you could consider a modified version of the Lyra c/o,
but the article does indicate the direction the crossover would
go from starting with a "stock" crossover.

specifically note R2 and R3 which are used to set levels to the
bass unit (typically -4dB to -6dB of bass unit level due to BSC)
may need different values and thus the rest would need scaling.

/sreten.

[grantnsw]

Thank you very much Sreten!
I have laterally offset the tweeter and mid towards the outside edge of the boxes
and kept the vertical distance on the baffle to a minimum as well.
The drivers do look remarkably similar in response characteristics , the Vifas
being more uniform in the 'working range' with less dB variation, eg:

Vifa D25AG-35-06 tweeter 90dB (+1) 2000 to 11000 Hz (off axis probs at 20K!*)
Dayton 275-010 tweeter 89dB (+- 2) 2000 to 11000 Hz

Vifa D75MX-31-08 dome mid 90dB (+-2) 500 to 5000 Hz (cone problems @6K)
Dayton 285-010 dome mid 89dB (+-3) 500 to 5000 Hz

Vifa P25WO-00-08 10" woof 88dB (+-2) 150 to 1000 Hz
Dayton 295-315 10" woof 86dB (+3-2) 150 to 1000 Hz.
The roll off rates look similar too, so the Lyra x'over seems quite usable, except that
PartsExpress says the Dayton tweet is 8ohms and mid is 6ohms? Unlikely.

I would do mid/tweet attenuation by voltage dividers and keeping a constant impedance
of 6(tweet) and 8ohms. So things are looking up!
A 625 and 5000Hz split yields a Fm=1768, but the values I get for 3rd order tweeter,
and two 2nd orders for mid bandpass and also 2nd order for woofer are nothing like
the Lyra values! and I'm sure I have followed Mr Dickason's equations to the 'letter'!
So, I unfortunately don't know where to go from here?

But thanks again for reading all this and kind regards, Grant.
(btw, vifa specs at www.solen.ca/vif.htm)