asymetrical xover fix for verticaly displaced drivers!?

[Guest]

I know, reads like a tabloid headline. I just stumbled upon a DIY site that suggests a 12db crossover on the woofer coupled with a 18db filter on the tweeter helps compensate for (better phase/summation) the vertical displacement of drivers on the front baffle. I have been under a rock for a few years but this is a new twist for me, can anybody shed some light on this for me, thanks in advance.

[cjd]

Dunno, around here we like to use things like measurements and simulations to design our crossovers. Certainly, the resulting crossover may be an uneven slope when you look at the actual transfer function, and there are other uses for this kind of thing.

Remember you get an additional 90 degrees of phase shift for every "order" increase - not sure that this is an accurate shift to account for driver offset induced phase alignme not differences, but hey. At any rate, it sounds like something that might be put forth by someone that does all textbook crossover work and tweaking by ear... Something I have yet to hear deliver results.

C

[Shonver]

This is quite common. Many 2-way designs incorrrectly claim to use Linkwitz-Riley crossovers. But the original design assumes zero acoustic offset. This is best achieve by aligning the drivers acoustically in the physical domain (thus sloped baffles). Seeing that the abovementioned designs don't employ acoustic alignment or electrical delay, they are in essence modified L-R designs. And it is indeed achieved by skewing the rolloff curves (perhaps the L-R should be omitted altogether).

I figured this out only after agonisingly and painstakingly trying to do it "right", needless to say, without much luck.

I would like to hear the final word on this from the resident gurus.

[JonMarsh]

A lot depends on what your choice of crossover frequency is versus the acoustical offset of the drivers...

The higher the crossover freuqency, the larger the net phase shift in the crossover region- with typical driver mounting and 3 kHz crossover, it's the difference in the depth of the acoustic center that wreaks havoc.

OTOH, those nutballs among us who like tweeters that can handle lowish crossover points and make careful acoustic measurements, find that just a little bit of management of the transistion region phase and amplitude goes a long way to solving these issues...

The NatalieP is a "classic" example of the current nutball approach (I have it on reliable authority that ET is in the Correllia region and won't likely ever see this posting by the time he get's back), using an LR3 acoustic target, doing so with a somewhat unconventional (to put it charitably) network approach, and realizing nice summing in the transistion region and a very deep notch with the drivers out of phase.

I mean, seriously, what more can you expect out of a low buck crossover? Free latte's? An improved sex life? (well, with better tunes at home, the latter might be possible...

[PatG]

No, the speaker projects have certainly NOT improved my sex life, quite the contrary! I am attempting to crossover this latest project (TM mini-monitor) at ~2.3kHz but had originaly intended to use nothing steeper than a 12db electrical slope (I'm using an XT25 for the highs). I was a little confused by the article because the author did not mention the importance of xover frequency and driver spacing, both having an influence on phase characteristics...but thats why I'm here. I will have to check out the NatelieP thread(s) and see if I can make some sense of the crossover topography.

[cjd]

I'm not sure the Natalie P is the place to start...

What's your process?

[JonMarsh]

No, the speaker projects have certainly NOT improved my sex life, quite the contrary! I am attempting to crossover this latest project (TM mini-monitor) at ~2.3kHz but had originaly intended to use nothing steeper than a 12db electrical slope (I'm using an XT25 for the highs). I was a little confused by the article because the author did not mention the importance of xover frequency and driver spacing, both having an influence on phase characteristics...but thats why I'm here. I will have to check out the NatelieP thread(s) and see if I can make some sense of the crossover topography.

Consider this- the Modula MT has a 1200 Hz crossover frequency (unusual high slope crossover, too). Many MTM with similar drivers have 2.4 kHz to 3 kHz. At 2.4 kHz, the net phase shift between drivers due to the acoustical center offset in depth is twice what it is at 1200 Hz. So, if at one frequency you have 30 degrees of excess shift, at the other you have 60 degrees.

2nd order slopes usually require high crossover points. Conventional 12 dB Butter worth have off axiss lobes, too. The problem with LR2 or BW2 is that 12 db/octave is the natural roll off rate of the driver, and with tweeters it isn't easy to get an FS and flat response to at least 2 octaves below the chosen crossover, without making the crossover frequency some what high. The XT25 is good from this viewpoint, extended bottom end, as Steen Duelund intended (I've heard from reliable sources that he had some input to that driver design), what you really want is a network design taht does phase tracking through the crossover region, keeping the drivers in relative phase, and then you have to adjust the transistion band behavior so that the drivers are synced up with the actual acoustical offset. Good test of whether you've got it right is the old reverse null.

And yeah, I agree with Chris, the Natalie P is not a "starter" crossover for getting a handle on these concepts for the first time- it's a bit unconventional in topology in order to implement some specific things with minimum components.

~Jon

[jkrutke]

I thought I'd post this since it seems relevant. It's a system I did using the XT25 tweeter, LR4 at 2250Hz, asymmetrical slopes, with the woofer being a bit shallower than the tweeter.

The XT25 was electrically 3rd order, with an LCR notch across the tweeter terminals, one resistor in series with the single inductor, and one other single resistor in series for padding purposes.

[Shonver]

I thought I'd post this since it seems relevant. It's a system I did using the XT25 tweeter, LR4 at 2250Hz, asymmetrical slopes, with the woofer being a bit shallower than the tweeter.

The XT25 was electrically 3rd order, with an LCR notch across the tweeter terminals, one resistor in series with the single inductor, and one other single resistor in series for padding purposes.

Hi John

For clarity: Did you design the HF section for LR4 acoustic @ 2250Hz, then optimise the LF section to suit (i.e., achieve flat response)?

[JonMarsh]

I thought I'd post this since it seems relevant. It's a system I did using the XT25 tweeter, LR4 at 2250Hz, asymmetrical slopes, with the woofer being a bit shallower than the tweeter.

The XT25 was electrically 3rd order, with an LCR notch across the tweeter terminals, one resistor in series with the single inductor, and one other single resistor in series for padding purposes.

Which system is this that you did, and how do you think it sounds? I couldn't identify it from your front page list of projects- didn't see anything with an XT25.

~Jon

[jkrutke]

For clarity: Did you design the HF section for LR4 acoustic @ 2250Hz, then optimise the LF section to suit (i.e., achieve flat response)?

Honestly, I don't remember my order of operations for the design of this one, but normally I'll design the simpler part of the filter first. In this case, probably the woofer. If I decide ahead of time that I am going asymmetrical to deal with the acoustic center difference, I'll then decide which response works better depending on the drivers - shallower on the woofer or steeper on the tweeter. In this case, I went shallower on the woofer, and my first run through probably had the optimizer set to 3rd order butterworth with Fc shifted to put the 6dB down point where I need it for an LR design. Then I fit the tweeter to that and tweak onwards from there.

Which system is this that you did, and how do you think it sounds? I couldn't identify it from your front page list of projects- didn't see anything with an XT25.

I didn't publish this one anywhere. Only a few of the designs I do show up on my web site. Some I deem not worthy for some reason or another, some I just lack the time to post, and a couple here and there are under contracts with confidentiality agreements. I've done several designs using the XT25. One on my web site is the ZD5. I love the XT25. It's very clean and easy to work with, and it's higher 2nd order HD is a good trade for almost a complete lack of tall order harmonics. That distortion profile sounds very natural and non-fatiguing. I've never been unsatisfied with the XT25's performance, and the top end off axis response is merely a placement limitation. (must toe in or design in a rising response to compensate)

[JonMarsh]

I've done several designs using the XT25. One on my web site is the ZD5. I love the XT25. It's very clean and easy to work with, and it's higher 2nd order HD is a good trade for almost a complete lack of tall order harmonics. That distortion profile sounds very natural and non-fatiguing. I've never been unsatisfied with the XT25's performance, and the top end off axis response is merely a placement limitation. (must toe in or design in a rising response to compensate)

Interesting comments. It reinforces my expectation that the BMS4540 will sound fairly good in a system, as it has a similar distortion spectra in some ways (dominated by 2nd order, extremely low higher order products), but in a DDS ENG-1 waveguide or similar, good dispersion can be obtained (measured) to beyond 30 degrees off axis at 25 kHz.

Thanks.

~Jon

[fjhuerta]

This thread is interesting. Lately, I've been finding out I haven't been able to create "textbook" crossovers (2nd order woofer, 3rd order tweeter, for example) that work at all. Most of the time, I have issues with phase integration. Asymmetrical crossover design may help with that, although I'm not sure as to how to do it. Do you create a target for the tweeter / woofer and then optimize the other driver?

Fortunately, I've found a sort of topology that lets me get to LR6, and all it needs is phase inversion of the tweeter. I've used it 3 times, and it hasn't failed me... yet