Crossover calculations (for RS150)

[jeff_free69]

This is a spinoff of another thread i started regarding downsizing my mains. That one has gone thru some interesting permutations and now my main question is more focused and I'm looking to resolve it immediately so i can get this up and running by memorial day:

I want to build a 4th order, LOW pass filter for 1kHz

if I use the the Madisound crossover calculator with parameters -6 DB @1k, 4th order (http://www.mhsoft.nl/CrossoverNetwor...udspeakers.asp)
it presents values for several options for filter types:
"Critical" , "Bessel", "Butterworth", "Chebyscheff"

I assume butterworth is the way to go. Is that correct? (Cheby is clearly not the one since it gives a hump)
I've seen other charts that give somewhat different values, but I am not sure who to believe.

BTW this is for the RS150 (Dayton 6" alum woofer) in an .25 cuft box, sealed (up against the wall)

[ThomasW]

I assume butterworth is the way to go. Is that correct?

The 'way to go' is choose a crossover topology appropriate for the impedance and frequency response measurements taken of your drivers mounted in the baffle where they're being used.

Those measurements are then imported into something like Jeff Bagby's Passive Crossover Designer...

http://audio.claub.net/software/jbagby.html

[Curt C]

Thomas as usual is right on.

Textbook calculators will fare poorly in practice, as they assume a constant impedance and a ruler flat driver response from DC to lightspeed. If you know the drivers inductance, you can implement a zobel that will linearize the impedance, but the resultant transfer function will still be affected by the actual response of the driver. Often only a 2nd order electrical filter is required to obtain a 4th order acoustical response, due to the natural roll off of the driver. Of course if that driver has break up modes higher in frequency, they need to be addressed separately with notch filters. Textbook calculators also don't compensate for baffle step and other diffraction effects, which will be very important with that driver's passband.

If you have the drivers .frd and .zma files, then by all means plug them into Jeff Bagby's programs and find out what you really need to have. Likely you will find the textbook calculations won't even be close.

C

[jeff_free69]

Often only a 2nd order electrical filter is required to obtain a 4th order acoustical response, due to the natural roll off of the driver. Of course if that driver has break up modes higher in frequency, they need to be addressed separately with notch filters.

My understanding is that the RS150 is notorious for its upper mid breakup, normally requiring a notch, but others on the forum suggested a 4th order electrical at 1k would also accomplish that (while mating with an existing dome mid)

If you have the drivers .frd and .zma files, then by all means plug them into Jeff Bagby's programs and find out what you really need to have. Likely you will find the textbook calculations won't even be close.

If we're talking about the actual drivers that I (will) own, thats probably not going to be practical for the immediate future. (I'm not even sure what these files are yet, let alone how to generate them). Or can I use the general parameters for this model?

As stated in my original post, i accept that this solution will not be the ultimate, and will have to have some compromises.

Thanks for your patience...

[dumaresq]

Along the lines of this thread, if someone wants to "get into" designing custom crossovers what equipment is he going to need (software and hardware), and how much math knowledge is required? I have a speaker design in my head that isn't even close to anything that I've been able to find so far... I'm not saying I am going to do it... it seems like too much work, but I'd like to know what is required.

[ThomasW]

Along the lines of this thread, if someone wants to "get into" designing custom crossovers what equipment is he going to need (software and hardware),

This has been asked and answered numerous times, using the forum's search function will get you to those threads.

[Curt C]

My understanding is that the RS150 is notorious for its upper mid breakup, normally requiring a notch, but others on the forum suggested a 4th order electrical at 1k would also accomplish that (while mating with an existing dome mid)

Ah sorry, I didn't know which driver we were speaking of. PE has Clio files on their site for the RS150:

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

You can cut and paste these into Notepad, save them, and then import them into Box Modeler and PCD. The learning curve will be well worth the much more accurate results you will get.

Yes, the RS150 needs some help up high. The most cost effective way to do this is to design your regular LP filter (Let's say for the purposes of example that it will be a second order with a series inductor and a shunt capacitor across the driver.) Once you have that close, add a small value cap (maybe 0.47 to 1 uF) in parallel with the series inductor. This will cause the response plot to have a sharp dip at some frequency close to the break up mode peaks. Vary the cap value until the dip is centered on the peaks. Add a resistor in series with the cap if necessary to smooth the response.

This topology resembles an elliptical filter, but you can also think of it as a series notch filter that shares a component with the low pass filter.

C

[dumaresq]

This has been asked and answered numerous times, using the forum's search function will get you to those threads.

hehe I fail

[Dennis H]

What Curt said.

I would add that the files at PE are measured in an infinite baffle and the response will be different in a small box. If you are going to use Jeff B's PCD for the crossover (highly recommended and free), you should download his other tools as well. In particular, his frequency response modeler will let you add baffle effects to the PE curve -- highs go mostly forward but lows go in all directions so the curve in a real box shows a droop in the bass that needs to be compensated for in the crossover.

Hey, this is a lot to chew on and there's a learning curve, but you're a sound pro and I think you'll enjoy learning how to do it right.

Edit: oops I just remembered you're using these with a TV. If they are right up against the wall, you probably won't need any baffle step compensation so you can pretty much ignore this post.

[jeff_free69]

THANKS! - now we're getting some place . I'll download that stuff to my laptop and play with it this weekend.

oops I just remembered you're using these with a TV. If they are right up against the wall, you probably won't need any baffle step compensation ...

Yes (they will be flush to the HD panel).
Not only that, but one speaker sits on a shelf and the other is on a stand but adjacent to a bay window; you guys are good, but i doubt anyone has a calculator for that! (Ok, you can stop rolling your eyes now - I told you it was a compromised situation and I truly appreciate your input

[Dennis H]

I'll download that stuff to my laptop and play with it this weekend.

You'll probably need your Excel disk. The programs need Excel for Windows with VBA and the Analysis Toolpack installed. They aren't installed by default.

It's likely in the manual but FRD files are the frequency response and ZMA files are the impedance from PE. Edit the PE files in Notepad to remove TS parameters, etc. so you just have columns of numbers -- frequency, dB or ohms, phase.

[jeff_free69]

Thanks all, your advice on Bagby is right on:
- installed the analysis pack
- downloaded the files from PE; renamed and edited (one also had stuff at the bottom that i missed the first time, causing a divide by zero error on the textbook calculation)

I loaded the RS150 files along with the Null tweeter (since I haven't a clue what the mid dome parameters are) and have my purple target line set to Lr4 @ 1000. Sure enough, the textbook 4th order network model comes up a little shy (ie it looks more like an 850hz filter, plus it had a big hump around 200hz).

So I guess its now a matter of plugging in the values, using available parts , to get the blue modeled output to track the purple target. I've gotten reasonably close, but there is one thing I'm stuck on - there seems to be about a +1 db hump around 200 hz . What i find interesting is that it implies there is actual gain here, over the raw driver response (ie the blue line is higher than the black line) .

What am I missing and how do I get rid of the hump (assuming this is not something I can ignore) ?

On the other hand, this is only really apparent when zoomed in on both freq & Db scales, without those on, its really a pretty close match. It seems for a given inductor value (eg 1.8 mH), lowering the DCR from .47 to .17 makes the hump a little worse. But I am unsure if that changes the shape at all or just raises the overall level evenly.


I haven't yet tried something like a second order plus a notch - this program seems pretty comprehensive and I've ran out of brain power for today.
(Too bad it can't do magic by letting you draw the desired output and generate the values for you , along with a PE parts list, and tea and crumpets

[jeff_free69]

bump-

Anybody want to save a semi-newbie from any possible blunder ?
about to add this to my PE order for RS150 + .25 cuft cab:

L2 = 1.8 mH (DCR .17) Jantzen 1.8mH 15 AWG P-Core Inductor
C2 = 36 uF 2X Dayton DMPC-18 18uF 250V Polypropylene
L3 = 1.0 mH (Dcr .13) Jantzen 1.0mH 15 AWG P-Core
C3 = 8.2 uF Dayton DMPC-8.2 8.2uF 250V Polypropylene

[augerpro]

Did you add baffle step to the driver responses you downloaded from PE? I think you can this with some of the FRD tools.

[jeff_free69]

Did you add baffle step to the driver responses you downloaded from PE? I think you can this with some of the FRD tools.

not sure I understand that - i am assuming that since this will be up against the wall I can ignore BSC. Especially since I am mating it with a pre-existing mid dome and ribbon that is not getting redesigned XO (a true frankenstein project, i know

Eventually I anticipate replacing that with just a simple seas tweeter and then I'll be more thorough (and learn these dang programs). BTW PE site has been down all day, giving me the opprotunity to reconsider the whole project and do that now.

[Dennis H]

Eventually I anticipate replacing that with just a simple seas tweeter and then I'll be more thorough (and learn these dang programs). BTW PE site has been down all day, giving me the opprotunity to reconsider the whole project and do that now.

Uh-oh, we may have created a monster. Do you have a flat mic in your pile of pro gear? The next step is taking your own measurements using free or pretty cheap software. Of course you could just go with one of the proven designs but that's not as fun.

[rc white]

Looking at the frequency response of the rs150-8 a simple 4th order filter is not going to give you a result that resembles a fourth order low pass characteristic to any extent that I would consider good enough.
Unless you want to embark upon a complex crossover design I would if I were you choose a much better behaved poly or paper coned driver.
rcw

[Mudjock]

Let's clarify a few things...

1. You've stated that you would like to use a 4th order filter at 1 kHz for the RS150. IIRC, you were planning to build an MT with the Seas 27TDFC. 1kHz is lower than you would want to cross the 27TDFC - and lower than you need to cross the RS150. Anywhere between about 1.5 kHz and 2 kHz for a crossover point will get you into the comfortable operating ranges of both drivers (assuming 4th order slopes on the frequency response with the crossover applied to the actual driver responses).

2. There are at least two approaches for dealing with the RS150 low pass filter that have been "proven" effective (built and peer-reviewed with good results). One is the approach that Curt Campbell suggested. If you search my posts, you will find a design for an RS150-Seas 27TBFC/G MT in a 0.25 cu. ft. box that uses that approach (a second order electrical filter with a small inductor in series with the cap chosen to create a deep notch filter centered in the cone breakup region.) I designed that MT to run ported with a broad, shallow bump created by the box tuning to fill out the baffle step compensation. If one plugged or omitted the port and placed it next to the wall, it would be close to optimal as a sealed design. I can try this if you are interested.

The other is the approach used by Jeff Bagby in the "Dreydels", cjd in his MT and MTM, and in my RS150-Neo3PDR. These use a third order electrical filter to create a 4th order rolloff in the response. Jeff's box size was also 0.25 cu. ft. and his port tuning was similar to mine - so his design might also work sealed and close to the wall.

Beyond that, cjd and I have both messed with series crossovers for the RS150. If you are interested in messing with those, google "acoustic reality series crossover" as a starting point. I wound up using a series LC across the terminals of the RS150 to get a clean 4th order rolloff on the RS150. It was promising, but I wound up not using it because the tweeter I was working with (the Neo3PDR) was not cooperating. A Seas 27 series tweeter or Vifa XT25 would be a better choice for series crossover implementation.

I dont' want to discourage you in any way from doing your own experimentation - but I did want to summarize what's been done with the RS150 so you have a good foundation to move forward from...

[jeff_free69]

Thanks - I've already come to the conclusion that the interim approach may very well not be worth it...so it was a good thing the PE site went down yesterday

To address one comment: Now that I think about it I actually don'thave a laboratory flat mike in my cabinet. ALmost every mike (Neumann and AKG etc) is made to have its own special character typically with a rise around 4k to flatter vocals. I have no idea how flat my tube preamps are either (probably not too bad, but they definitely have a nice tube sound)

Anyway, so last night I'm staring at the HD and the very lovely shelving unit it resides in, and decided I should be able to hack off one of the decorative scrolling metal support brackets fairly easily using the good ol' sawzall (while the wife is out of town of course). This has been one of the main constraints of the whole project , so good riddance (to the bracket - not the wife LOL !

This will open the area where the left speaker will be to allow up to a full 24" height. So now I can fit in an RS 150 based design for MT or even MTM ! (The Rs150's will be looking very sleek in a glossy black snuggled up next to the samsung)
Depth will still have to be limited to 10" and width to 7.5" though; perhaps not optimal but it will have to do. The largest prebuilt i could accommodate is the dayton .38 (14x8x10.5) . but that wastes 10" of new found height that might be used for an MTM, so I may be making MDF dust soon.

I found some of the designs you referenced (eg cjd, etc), they look excellent and I'm sure one of them is right for me. As noted they will be on wall, but I'm still unsure how to figure how BSC is affected by the huge 46" sheet of LCD glass the speakers will be snuggling up to. It seems from my experience this can reinforce the mid/hi's, so shouldn't I need to control reduce this?

I came to this conclusion when I put my Zaph Sr-71s in my studio which has 2 x 22" computer screens between them. I originally listened without them there and following john's suggestions for close-in monitoring, tried all the tweks like reducing the BSC. This was fine until i put the computer monitors back up and a keyboard on the desk, and then it was way too forward and bright! Ultimately I settled on making only a very slight tweak to raise the tweeter level (turns out all I had to do was drop an R from 4 to 3.7 ohms

By tomorrow I'll figure which I want MT or MTM, in any case your thoughts on BSC here would be appreciated...

OH one more basic question - will MTM design give me a more controlled dispersion pattern? the right speaker is adajcent to a very lovely bay window (which will not be getting hacked out) - can you say "parabolic reflector" ? My existing ribbon tweeters help control that, but I'm noticing that the SR71s have a very wide soundstage and seem to not. maybe MTM is a little better than MT here?

[ThomasW]

OH one more basic question - will MTM design give me a more controlled dispersion pattern?

No it's not 'more controlled' it's just a different dispersion pattern

[Curt C]

An MTM on a flat baffle will always have a symmetrical vertical off-axis response, regardless of the difference in acoustic centers or crossover topology. It will radiate perpendicular to the baffle axis.

An MT on the other hand, will likely have the main lobe tilted vertically to some degree. How much, and which direction will depend on the driver relative acoustic centers and also the crossover topology. Obviously this is not a show-stopper, as many excellent TM and TMW designs exist. It's just another factor to consider in the design process.

C

[fjhuerta]

Did you add baffle step to the driver responses you downloaded from PE? I think you can this with some of the FRD tools.

Since the woofer will be used mounted to a wall, maybe baffle step won't be so much of an issue...

[jeff_free69]

An MTM on a flat baffle will always have a symmetrical vertical off-axis response, regardless of the difference in acoustic centers or crossover topology. It will radiate perpendicular to the baffle axis.

('m bouncing around between a couple of threads here...)
- I mocked up a cardboard cutout of an MTM in the pE .75 curved box and was excited to see that it will shoehorn nicely into my space! Well, kind of.. there is a height limitation and the top woofer will be at ear level. i sit 6-8 ft away and it just doesn't look like a good placement.
a showstopper? just tilt it up ?

[jeff_free69]

OK - decision made and parts ordered!

Going with the RS150 + 28A in .25 curved box, using cjd's crossover.

Not cheap (about 450 incl shipping) , but this seems to be a low risk option. HD looks lovely close to the wall, so small is the way to go for now. If I decide to go bigger, they can always be used elsewhere or just addon another M.