dayton ND20/hi-vi B3S slim TMM design

[gonar]

so I had some spare parts lying around, and I thought it was time to start getting my feet wet with designing my first speaker.

as a background, I have built 4 subs, 2 HT oriented, 2 music oriented. I am 3/4 of the way through building the Zaph B3S and am immediately following that with the BAMTMs (parts in house, ready to go...)

anyway, I had some extra B3Ss and thought it would be fun to try my hand at building a pair of MTMs with them.

I know this is _very_ similar to the BoseBuster project in this forum, but I want to learn this stuff for myself.

my goals for this project are decent sound, minimum cost and a simple low parts count xover.

so I have 4 B3S and 2 Dayton ND20FB tweets. I'm thinking I want to run a 3000-3500 2nd order xover. I would like to avoid a notch filter to tame the 8300Hz(i think) cone resonance of the B3S.

The B3S is rated 81Db 1W/1M (8 Ohm) so 87 1W/1M for a pair in parallel right?
the ND20FB is rated 92 2.83V/1M (4 Ohm) so 89 1W/1M right?

these will probably end up as stereo speakers for our second TV with a small sub for bass reinforcement, however I am more interested in music reproduction than HTish attributes...

I'm a bit worried about how much BSC I should use and how to minimize component count....

anyway, any tips people want to share, I'd really appreciate.

edit: update, planning slim TMM floorstanders for moderate level TV viewing and music

[max_andrews]

try this

This is for an MTM with 2 dayton PA130's and the ND20 tweeter. It's designed for use on a 6.5" wide baffle with the tweeter placed 1" from the side and both the drivers very close to it. Crossover point is about 3200hz. You may need to tweak things a bit if you use a smaller baffle but not by much. The tiny .025mh inductor in series with the shunt capacitor in the woofer circuit gets rid of the breakup. Of course you will need to add an lpad to the tweeter since it's a lot more sensitive than the B3 drivers. Oh and that 18uf should be a 17uf, and you can use cheap NPE caps all around. All inductors are 18 gauge jantzen air core

[Dennis H]

Max, I'm glad you like your new theoretical design that you haven't built yet but this is at least the third thread you've posted it in. Gonar specifically wants to use different drivers so your post isn't really all that helpful.

[max_andrews]

Just offering it up as a starting point if he wants, since the XO point and impedance is the same, and the tweeter is the same. He wanted to avoid a full notch filter and this was why I posted this, because I had the same issue and found the .025mh inductor was effective in reducing it (at 8khz, right where the B3s breakup is as well). Frankly, I fail to see the point of your post other than to chide my design for being theoretical, which I do not deny. I've been helped out many times by people posting their crossover topologies to solve similar issues, so I thought I'd do the same.

[gonar]

I'd read through the PA130 thread, looks interesting. I think the Bose Buster thread is closer to what I'm doing, but not an exact match.

I'm not settled on the driver placement, whether I want to do a MTM or TMM. I was thinking appearance wise, MTM with the B3S's flanges rotated 45deg and almost touching, with the tweet offset to the inside might be fun...

I would like to do some simulations of the crossover.

I would like to try a free tool before I buy one. I would appreciate a recommendation for a good beginner tool.

as far as speaker measurements, I don't have the equipment to measure my own, but PE has the CLIO files for the ND20, but nothing on the B3S.

if somebody has files for the B3S that would be great.

[fjhuerta]

Just offering it up as a starting point if he wants, since the XO point and impedance is the same, and the tweeter is the same. He wanted to avoid a full notch filter and this was why I posted this, because I had the same issue and found the .025mh inductor was effective in reducing it (at 8khz, right where the B3s breakup is as well). Frankly, I fail to see the point of your post other than to chide my design for being theoretical, which I do not deny. I've been helped out many times by people posting their crossover topologies to solve similar issues, so I thought I'd do the same.

Considering the drivers are completely different, baffle size isn't the same, etc... I'd say your post may cause more harm than good.

The B3s are easy drivers to work with, since their break-up point is high. But without in-baffle measurements and impedance plots, it's not a good idea to think that since a notch filter works with one set of drivers it will work with others.

[gonar]

I'm finding speaker workshop hard to use, trying to set up my networks, and I can't seem to get it to let me wire two drivers in parallel.

is there a more intuitive/better tool than speaker workshop?

[Nemophyle]

nothing free i fear

i think speaker workshop is indeed hard to use at first, but very powerfull once you get used to it. I remember there were some tutorials for it somewhere, let me wander around, or maybe someone else knows where they are.

[gonar]

ok, figured out how to connect the drivers in parallel, and how to import CLIO data.

but, hmmm, when I set up a network for the tweeter and specify a 3khz 2nd order high pass, it seems to be making a 1500Hz filter. I'm guessing I havent filled out the driver specifications properly....

does anybody have SW .FRD & .ZMA or CLIO data files for the B3S?

[Dennis H]

Try Jeff Bagby's Passive Crossover Designer. It's about as intuitive as it gets. You need Excel with VBA and the Analysis Toolpack installed (they aren't installed by default). Also read the help file carefully as I think the Z-axis direction for the drivers is the reverse of some other packages (forward vs backward for +Z).

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

[chrisn]

Curt C has a design using the round flange version of the HiVi and the front mount neo: http://www.geocities.com/cc00541/Scrappy.html

[69Stingray]

I found using the FRD tools very useful. Use Jeff PCD first, then experiment with speaker workshop. Use the target function in PCD to ensure you get the proper slopes. Pick an X-Over point that is within one wavelength of the center to center driver spacing & within the "distortion friendly" zone of the tweeter. If the X-Over point is low enough, you may not need the notch filter. I forget what freq. the notch is at. If you can get that freq. 20 dB "down", I think you will be all set without. Review Zaph's distortion curves also.

In regards to speaker workshop, keep practicing, it took me forever to get a crossover to build correctly.

http://www.rjbaudio.com/Audiofiles/FRDtools.html

[gonar]

I've just about decided I want to make these as a TMM layout in a slim tower.

I figure ~30" high tweeter, woofers mounted as close together as the frames will allow.

I will probably keep the baffles down to about 5 inches and use some form of outrigger for stability.

the will be used for stereo sound for a corner mounted flat-screen, and so will be somewhat corner loaded, as well as positioned close to the wall. they will also be supported by a small sub (hidden in the corner behind TV)

a while back, Zaph was talking about a similar design, although I think he was talking TMMMM. Zaph, if you're out there, I'd love to hear what ever became of that design...

[gonar]

progress report

been fiddling with PCD 6.12 and speaker workshop(SW).

I find PCD a lot easier to use, but like the ability to create a graphic of the xover in speaker workshop.

I've found that the predicted frequency response differs greatly between the two programs.

an xover designed in PCD that gives me +-3db 80-20k gives me +-6db in SW.

I've tweaked the xover to use the cheapest possible components off the shelf from PE (20ga air cores, electrolytics, basic non-inductive resistors).

here's the freq response from PCD and the xover layout from SW.

if somebody could check my work I would really appreciate it!

edit: the DCRs for the inductors are .23 for the .15mH, .29 for the .20mH and .53 for the .60mH

[Dennis H]

I've found that the predicted frequency response differs greatly between the two programs

Double check the (+/-) Z dimension of the drivers. I think PCD is the opposite of most programs.

[gonar]

I haven't played with the Z or X location of the drivers, only the Y in PCD.
I haven't done anything with speaker placement in SW.

the curve shown is with all speakers at 0,0,0 (in PCD).

I would love to simplify this crossover, and move the crossover point up (it's currently ~1900, probably too low for this tweeter, even for low-moderate volume use).

I modeled the "scrappy" crossover mentioned above, and it didn't look good at all in PCD (which doesn't necessarily mean anything, just that in PCD it didn't look good)

[gonar]

got a new xover using 4th order filters. effective crossover is wierd in that the down slope of the B3s coincided with a natral dip in the tweeter's response, resulting in a crossover that crosses over a range of 3600-3900, and gives an interesting inverted horseshoe double null over that freq band with polarity reversed.

on the other hand, the summed freq response looks real nice (+- <2dB 80-20k) and phase isn't bad....

is this an inherently bad thing? below is a pic showing a zoomed in view of the xover area.... (each horizontal line is 5dB)

[gonar]

well, I've been continuing to play with the crossover, and now have a really nice response curve from 100-15KHz with a really nice deep single null.

however, the rising response of the tweeter makes me think I would like to shape it's response a little by knocking it flat above 15k. this would allow me to boost overall tweeter response .5-1dB and give me _really_ flat response.

I'm waiting for my copy of loudspeaker design cookbook from madisound (on sale for $11!), but would like to be able to make progress on this.

can anybody give me some tips on response shaping networks? throwing a cap on the l-pad shunt (as Zaph did in his ZR-71 design) doesn't seem to have any effect.

also, I would love to simplify these high component count 4th order filters (10 components and counting), are there lower component count 4th order topologies out there than the standard lclc and clcl ladders? one of the goals was to keep this cheap, and high component count doesn't help with that.

of course one of the problems is the natural driver curves are not naturally falling off anywhere near the xover point.

[chrisn]

I took a stab at it and this is what I came up with: 9 components with the drivers centered on a 5.5X30" baffle as a TMM. The "schematic" is below:

Highpass: 7ohm---4.8uf---5.6uf------+
l
.15mh


Lowpass: .70mh-------------------1.8-2.2mh-----+
l l l l
10uf l Upper Lower
l 4uf B3S B3S
3ohm l





Hope this makes sense and is of some help. The reverse null can be made deeper by adding a 12-15ohm resistor on the 4uf shunt cap.

Good luck!

[chrisn]

the schematic does not appear as I had typed, my apologies.

[gonar]

thanks. so that's a 2.5 way filter filter.

I think I grok the tweeter filter, standard 3rd order with level trim resistor before filter.

the low range filter is not quite making sense to me.

[chrisn]

I drew up a schematic, attatched below. The large inductor can vary between 1.80-2.20. Hope this helps