help, brain is melting

[jberg]

hello

I built myself a set of little speakers using the HiVi B3N and the Dayton ND20FA-6. I crossed them at 4000 Hz using a 2nd order filter and an L-pad to push the tweeter 10 dB down

But whenever there are a lot of esses involved it feels like my brain is melting and starts leaking out of my ears.



I think they look very pretty and it would be a shame to have to destroy them.

pardon my english, I am dutch

[Johnloudb]

Attenuating the tweeter by 10dB is a lot, if you mean like -10dB with respect to a 1kHz reference. Or did you mean -10dB with respect to the tweeters sensitivity?

I think most people on this board will want to see measurements and the crossover layout. I guessing you don't have elaborate testing equipment, but if you had just a test CD with test tones and a dB meter you could make some rough measurements of the frequency response. Or you could get some testing software and a mic.

Then, you could voice the speaker, once you have an idea where you're at with frequency response.

Without some kind of measurements you're lost. Do you have a schematic layout of the crossover?

[jberg]

-10dB because the tweeter is 10 dB more sensitive then the b3n
I used a simple LC second order filter for both the tweeter and the woofer, I cant remember exactly what values i used but i wil screw them open and take a look

[TacoD]

Textbook filters will not work as default. As John said measuring the actual response is needed to see what is happening.

There is a possibility that the harsh sounding edges are not from the tweeter but from the woofers resonance.

[augerpro]

That has to be the best looking mini speaker I've seen! Any pics/info of the construction?

As far as teh sibilance the tweeter does have a rise from 5khz-10khz that may be causing this. And hey, it's $5 tweeter. Now the B3N from my memory also has some breakup in this region, 4khz may too high to cross, although I probably wouldn't go under 3.5khz or so. What values are the components?

[Hdale85]

Those would make some nice little computer speakers if someone could take measurements and make a proper crossover

[jberg]

ok my crossover contains the following
for the woofer C=6,8 uF L = 0,47 mH
for the tweeter C=4,7 uF L =0,17 mH

to attenuate the tweeter i used 2 resistors, both 4 Ohm, one serie, one parallel

[banditto1969]

What tweeter mount is that. They do look nice.

[benchtester]

Nice looking speakers.

It sounds like there is a problem in the 8Khz range. It could be the tweeter or the woofer. The B3N has the primary breakup at this frequency.

Maybe you have seen this, but Zaph made a full range speaker with the square flange version of this driver. This link is his very good technical write up:

http://www.zaphaudio.com/audio-speaker18.html

His tweeter section also has data on your speaker.

My suggestion is listen to the woofers alone (disconnect the tweeters); and then listen to just the tweeter (this will sound strange). Then you can determine if one or the other is responsible for "melting you brain". I hate it when that happens! :lol:

[jberg]

What tweeter mount is that. They do look nice.

just two pieces of mdf glued together and sanded down to the right form (parabolic)

[Mudjock]

That's a great looking design, I would recommend against destroying them :T Parts Express has data files for both the woofer and tweeter, so it wouldn't be too hard to apply baffle diffraction effects and simulate what your crossover is doing, then come up with some sort of fix to get you close. I will work on it this weekend.

[JonW]

jberg,

Those look excellent! :T Really pretty and cute. The crossover experts here will help you out. You've got 2 drivers, so with the right data and a little effort they should sound OK.

[jberg]

my knowledge of crossovers is very limited, so if anyone has any ideas i will gladly try them out

i calculated mine with winisd

[Bent]

I highly doubt WinISD would take into account the cone breakup that the B3N has at 8 khz - nor would it tell what the tweeter would do. Zaph's filter is designed to alleviate that (at least on the B3N...), but at the expense of some sensitivity - which the B3n is shy of in the first place.
They are power hungry, and need to be crossed quite high in the first place, try not to run them any lower than 120 hz - and if you intend to make them play loud then cross them no lower than 150 hz or you'll run out of excursion quickly.

But,
don't give up on the design - they look like a million bucks.

[Mudjock]

Okay, I found the data for the B3n and ND-20fa from the P.E. website. I assumed you had a 7 inch tall, 5 inch wide baffle (probably not perfect but hopefully close enough to get a clear picture of what's going on). I ran the Baffle Diffraction Simulator based on that assumption and the approximate driver positions from the picture. I then combined the predicted baffle responses with the infinite baffle driver responses from P.E. to get the final predicted response. I then loaded those into Passive Crossover Designer and applied your crossover. I assumed that the acoustic center of the B3n was about 20 mm behind the tweeter (which shouldn't be too far off). The resulting response is shown in the first graph.

The problem is pretty obvious - a very pronounced peak at 3.5 kHz (most sibilance problems are caused by peaking in the 2 and 5 kHz range). This is really a problem with the woofer filter.

The second picture is the simulated result of applying a minor fix, actually leaving all existing components in place and adding a zobel across the woofer (10 uF capacitor and 15 ohm resistor in series across the woofer terminals). This will be smooth through the treble, but will probably sound very thin through the midbass.

The third picture is the response from a new crossover which will give a lot more body through the midbass region by accounting for baffle step losses. The woofer circuit uses a much larger inductor (1.8 mH). The parallel capacitor in the woofer circuit (5 uF) has a two ohm resistor in series with it to help shape the rolloff. The tweeter circuit is adjusted with a 2 uF capacitor and a 0.5 mH inductor (your 0.47 mH inductor from the original design would be close enough here). You will also bump up the series resistor to 8 ohms, but leave the 4 ohm parallel resistor alone. The 8 ohm series resistor is the one to adjust if you find the treble too bright (increase the resistor value) or too dull (decrease). This will depend on room acoustics and individual taste.

I can't promise that this will be the optimal crossover for this design, but I can promise that it will be a major improvement and should address the brain melting effect of the original.

If you have any questions or would like to see a diagram of the revised crossover, let me know.

[Mazeroth]

I just wanted to say I love this board. People always willing to help others out. :T

[jberg]

Mudjock, ýoure the best,

the baffle is 5.1 tall 3.9 inch wide, does that change anything??

[Winter]

You may have several issues causing the emphasis in the "esses", otherwise know as sibilance, has Augerpro and Mudjock have mentioned. While the mounting of your tweeter is great visually, that sharp edge at a constant distance from the dome tweeter on the top half is causing considerable diffraction, probably in the 5 to 8 kz region. Standard diameter round tweeter flanges that are surface mounted (not recessed) typically have diffraction peaks in the 4 to 5 kHz range. As this tweeter has a smaller flange, the effected frequencies should be a little higher. A radiused contour at the flange edge of 3/4 to 1 1/2 inches would be helpful.

[Mudjock]

I reran the simulations with the real baffle dimensions (5.1" tall by 3.9" wide).

The simulation of the design you have now (first graph) really doesn't change much. The peak is still around 3.5 kHz. The simple fix (second graph) changes a little, the zobel retains the 10 uF capacitor, but the resistor is now 8 ohms. I also had to make some adjustments to the more comprehensive revision (last graph). For the woofer circuit, the inductor is still 1.8 mH, but the parallel capacitor is now 6.8 uF and there is a 4 ohm resistor in series with it. For the tweeter circuit, the 8 ohm resistor becomes 10 ohms, but everything else stays the same from my previous simulation.

Again, let me know if you have further questions or if you would like to see a schematic of the crossover. I think the simple fix might be an option if these will be used on a desk (i.e. computer speakers). Otherwise, the comprehensive revision will work significantly better.

[jberg]

thanks alot mudjock, huge difference, youre the best :T :T :T

[Mudjock]

jberg,

I'm glad the suggested mods took you in the right direction. :T Out of curiosity, which one did you perform - adding just the zobel across the woofer or the full modification?