SB Acoustics monitors build

[rpb]

For damping, I like about 70% wall coverage with thick fiberglass like this. I prefer 2" depth in about 50% of the box. The more the better.

[rpb]

Is it just felt on the side walls? If so, I'd add a lot of the acoustic foam there.

What exactly is that acoustic foam?

[DeZZar]

Its from a local electronics supplier - described as "650gsm Acrylic Speaker Dampening Material"

[rpb]

Can they supply absorption data? Nothing beats fiberglass that I'm aware of.

[DeZZar]

Are you sure it's the tweeter misbehaving? Try a resonance trap, some tweeters just really don't tolerate the Fs getting tickled. Try 2.0 mH, 22 uF, 2.0 ohm trap on one speaker and see if that makes a difference. Model it first, that is for a ScanSpeak 9130 but the Fs is similar to the SB29 you're using and should get you close.

Off axis modeling is essential, off axis flares in the 2khz area can really make for a hard/harsh sounding speaker.

Not absolutely certain no. However, in a lot of testing I would listen to either just the woofer or tweeter. The woofer got a little 'honky' with a sort of ring to it in some cases which resulted in trying to further smooth out its knee and bring down the lumps around 1k. Female vocals, tenors etc just sounded a little harsh at the top end of their scales. Once that was taken care of the same content sounded spot on - even smile inducing for my neighbor (who these are for) and a very nice warm yet detailed sound. The tweeter mostly added only ambience to a lot of this music. However, switching it up to some metal tracks is where things were getting a bit bright/hot and I again though maybe the woofer. But in this case when playing the tweeter on its own it definitely seemed to be the contributor - where before when addressing the woofer - the tweeter barely added anything (with different musical content).

Based on the models for these current crossovers there does appear to be an off-axis increase in output between 2-3K. I'm not sure at this point what I might be able to do about that - however - for the meantime I've just going to get both boxes caught up to the same point and then resume modelling and testing.

I'll try the resonance trap (notch?) and see how that goes as well.

[rpb]

Not absolutely certain no. However, in a lot of testing I would listen to either just the woofer or tweeter. The woofer got a little 'honky' with a sort of ring to it in some cases which resulted in trying to further smooth out its knee and bring down the lumps around 1k. Female vocals, tenors etc just sounded a little harsh at the top end of their scales. Once that was taken care of the same content sounded spot on - even smile inducing for my neighbor (who these are for) and a very nice warm yet detailed sound. The tweeter mostly added only ambience to a lot of this music. However, switching it up to some metal tracks is where things were getting a bit bright/hot and I again though maybe the woofer. But in this case when playing the tweeter on its own it definitely seemed to be the contributor - where before when addressing the woofer - the tweeter barely added anything (with different musical content).

Based on the models for these current crossovers there does appear to be an off-axis increase in output between 2-3K. I'm not sure at this point what I might be able to do about that - however - for the meantime I've just going to get both boxes caught up to the same point and then resume modelling and testing.

I'll try the resonance trap (notch?) and see how that goes as well.

This sounds like insufficient damping to me. See added comment in above post.

A thin layer of felt probably does very little below 2k.

[DeZZar]

I'll give that a go!!

[rpb]

Leave a 1" to 2" airspace behind the foam on one, or both sides. That will increase effectiveness.

[Steve Lee]

We will never get away from the need to EQ.

There are no perfect speakers/rooms for all audio . . .

Good luck, DeZZar - I gave up trying to mix for everyone else . . .

[DeZZar]

I've never really used EQ - I mean I use it sometimes to help in development - add some filters etc to test speakers out but not for my main listening system. I've just concluded over the years that some songs just wont suit some speakers particularly when you're talking high resolution systems. The old 'bad recording made worse' is definitely true. In this case I just want to make sure there are no technical mistakes in the implementation! Few more iterations to go I think! :D

[DeZZar]

Thank you to everyone for the suggestions. I've worked up a few more models and I think I'm going to try the one below. Thanks to rpb and Wolf for getting me to re-consider where the crossover point could be. I really didn't believe a higher point was going to be possible but working up a little higher has actually provided the best modelled result so far. As soon as I can I'll throw this one together and see how well it translates into reality.

This is around 2.6K actual crossover point. It looks as though its crossing at 3K due to the diffraction null that's occurring at 3K.




I didn't confirm this with measurements but I think this was probably a contributing factor in the brightness on one of the earlier attempts. Thoughts?
(in terms of how smooth this response looks I note that the modelling tends to be a little more extreme looking than what it actually measures like)

[Wolf]

Uh- that will definitely get fatiguing! I'm sure that was the cause.
Wolf

[Steve Lee]

DeZZar, not sure where that 2.4 KHz peak came from but it wasn't in your prior postings.

Yes - that specific frequency is a really nasty peak to have - Cybill-Aunt as hell, too!

[DeZZar]

It appears in the off-axis modelling. I hadn't done any actual off-axis measurements on the early prototype crossovers so I hadn't confirmed it was the case, but certainly sounds like it was there. Once I can try a higher crossover point per below I'll run the full off-axis measurements and confirm.

[davidroberts]

Several years ago I convinced Dennis Murphy to help me with a set of speakers. At the time there was a lot of buzz going around about the high order filters so we decided it would be fun to publish the measurements and allow others a shot at the xovers. Many known designers chipped in and posted their results. Funny thing is that although Dennis' design had the most shallow slopes it was also the most preferred when compared. Long story short, sometimes you have to throw the measurements out the window and trust your ears.

[DeZZar]

Long story short, sometimes you have to throw the measurements out the window and trust your ears.

Your ears are definitely the final judge but sometimes when trying to triage issues in the middle of design you need the measurements to tell you what you are hearing (this is a perfect example of that).

Shallow/Steep slopes etc....my view is there are no rules here - everything has to be application specific. Shallow slopes wont work here - but might with other drivers and baffles, and so on. To follow a rule might result in sometimes awesome, but sometimes disaster. Its why my advice is always build a kit if you're thinking you can use some calculators and general rules of thumb to design from scratch!

[Steve Lee]

"Build a Kit"

Damn sound advice!!


Pun intended.

[Wolf]

I find the really steep filters like the DSP/48 really sound like one driver stops and one picks up. In shallower targets, this is not such an issue and the drivers blend really well together.
I say- steep enough to do what is necessary for blemishes should be the MO. This does not mean you cannot experiment, but with regards to cost, lesser orders are usually less costly.

Wolf

[DeZZar]

Definitely agree - the steeper the slope the more it becomes obvious the sound is coming from distinctly different points on the baffle, depending on the content.

The model above I am yet to try, although 4th order electrical on the woofer, it actually only results in around 13.6db/octave over the crossover region (2-4K). On the tweeter the third order yields 11.3db over the same octave. Is this where we say "fourth order electrical, 2nd order acoustic" ? So then 3rd and 4th order electricals in this case have yielded second order slopes on both drivers.

Keen to see how this sounds anyway!!

[DeZZar]

After a decent few rounds of re-modelling, re-working, measuring, rinsing and repeating I am happy to present the results on this build.

In the end I had the best results with a fourth order electrical all round (tweeter and woofer).

Here is the final crossover schema.


Yes its a fairly high parts count crossover but overall fairly simple and the smaller sized caps on the tweeter provide an opportunity to spend a little more here if so desired.

All in all this woofer isn't the best behaved in terms of upper octave performance and I would have to admit the baffle and resultant tweeter positioning aren't optimal either - however, the point is that an older style "box speaker" design was the order of the day. Crossover lands right on 2.9K in the end which helps dramatically with off-axis performance.

All of the following measurements were taken @2.83volt with mic @ 850mm distance on tweeter axis.
REW was not SPL configured so SPL shown is not absolute.

Here is the final full spectrum result:

Here's the result with the woofer and tweeters individual measurements and an inversion of tweeter polarity. Everything looks to be well behaved. The troublesome parts of woofer breakup have been pushed -50db down. The fourth order electrical is giving us an acoustical slope of just under 18db per octave on the woofer, and 20db on the tweeter.



Sanity checking the voicing here shows that on the chosen -5db gradual slope out to the top end, the performance stays well within +-2.5db from 42Hz all the way to 20K.



Horizontal Off-Axis performance shows that the previously encountered flare around 2-3K has been eliminated in terms of seeing a significant spike up in off axis db, although I do note that the 2-3K region stays about the same level all the way out to 45 degrees. From listening so far this certainly doesn't seem to be problematic at all and in terms of the overall off-axis performance we've got the sort of roll off you would expect all the way out to 90 degrees.

[DeZZar]

0-60 Off-Axis going Up



And 0-90 coming down

[rpb]

For anyone that doesn't know, those two dips near 3k are the result of the x-over, and are normal. It's due to the relative distances from the drivers to the mic changing. This is the same reverse null that occurs on axis when you swap polarity.