Crossovers in the 1KHz region

[angel]

Hi all..

From what I've read, conventional wisdom holds that crossover points should try to leave as much of the 150Hz-4KHz range as possible covered by a single driver, in order to gain coherence.

However, there is a gap around 1000-1500Hz where our ears have little directivity (interaural timing cues are most useful at low frequencies (especially <1KHz), whereas interaural amplitude difference cues are most useful at high frequencies (especially >4KHz), and neither is very effective around 1KHz-1.5KHz). Also, while our ears are fairly sensitive to level in this range, it is not nearly as bad as e.g. around 3-4KHz. And at this frequency, beaming is not nearly as much of an issue, hence the off-axis response should not differ much between the drivers.

It would seem that placing a very steep filter (4th order or higher) around 1250Hz could offer good imaging at the expense of little else.

Many tweeters, including e.g. the Stage Accompany ribbons, can be crossed over this low, if the crossover is steep. Notably, the Seas coax drivers, with their already excellent imaging, might be able to deal with this, since their 1200Hz resonance is well damped.

Opinions on this, or experiences with such an approach?

[Adam M.]

You would really need an robust tweeter to cross this low. SL crosses the orion at about 1400Hz if I remember correctly, but it is all active and I'm sure very steep. Otherwise, you will drive the tweeter into areas where it will distort very easily, even with a well damped resonance peak. Is it possible with the correct tweeter, definitely, but I think a more realistic target would be the 1.5Khz to 2Khz range. I find my designs often end up just above this, from about 2k to 2.5k...things just seem to fall in place in that range.

[Hennie]

angel,

Jon Marsh (http://www.htguide.com/forum/forumdisplay.php4?f=6) uses steep Cauer filters to implement crossovers at the frequencies you suggested. Surely this appraoch has benefits. Why not join there and discuss the matter with him? He is a prolific designer and is well respected.

[phase_accurate]

The SA driver is meant to be croosed in that area and i had the opportunity to hear an SA monitor that was driven actively. It sounded very clean at an insanely high SPL so one can assume that the "tweeter" didn't have to struggle.

IMO if done right (which accounts for almost anything in engineering) a crossover frequency of 1 kHz can be as good as any other one. But you need to use suitable drivers.

Regards

Charles

[Salas]

Any fairly large planar from china can be an inexpensive testbed for you. Or a BG if you like (they make em in Far East too I think).
I have crossed larger dome mid tweeters downthere (Audax, LPG) and I agree its an excellent crosspoint for great imaging.

[Pan]

The function of localization you refer to is between the ears, there is nothing that indicates that this findings would have any relevance for crossing between two drivers in the same speaker.

It´s a similar "false" logic that suggest that the ears are most sensible to crossing at 3-5k only becasue the spl sensitivity of the ear is highest there. Again there is nothing that suggest that this would be the case.

/Peter

[Grahamt]

Originally posted by Hennie
angel,

Jon Marsh (http://www.htguide.com/forum/forumdisplay.php4?f=6) uses steep Cauer filters to implement crossovers at the frequencies you suggested. Surely this appraoch has benefits. Why not join there and discuss the matter with him? He is a prolific designer and is well respected.

Jon and I just did a design using a Seas 27 TDFC crossed at 1200 Hz. Tweeter distortion is not a problem.

I cannot detect ANY lobing, off axis response is great and the upper midrange is very clear. I'm a big fan of low crossovers now.

[Hennie]

Hi Graham, good stuff you are doing on the other side. I have never done a passive or Cauer in that region, just an active 4th order a la SL, but just a little bit lower at 1200, and indeed it works well with a Seas 27 TDFC. The Cauer topology should do much better, but I can endorse what you've said. Indeed, very good clarity. A reduction in energy storage I suppose.

[AJinFLA]

Here's my design (Orion variant) which uses a waveguide loaded XT19 XO'd at 1.3k LR8
http://photos.yahoo.com/ph//my_photos

[angel]

Originally posted by Pan
The function of localization you refer to is between the ears, there is nothing that indicates that this findings would have any relevance for crossing between two drivers in the same speaker.

If you are familiar with the neural network model that is used to model this, it has some relevance. Also, unless you have headphones, the same signal is going to arrive at both ears with a difference in level and timing. Even with a single source.

Similarly, I think using a transconductance output with a lowpass on the input and a lowpassed voltage feedback to transition to voltage drive around 1250Hz would improve fidelity and imaging. The main issues with feedback, such as high order harmonics and phase turn become less relevant at high frequencies, due to the natural lack of phase sensitivity at these frequencies, as well as the harmonics ending up in a very insensitive region that is poorly covered by resonators and nerve cells. The main issues with low feedback, such as poor damping and level consistency, become less important at low frequencies, due to poor level sensitivity and level resolution, as well as harmonics due to current drive being lower.

It´s a similar "false" logic that suggest that the ears are most sensible to crossing at 3-5k only becasue the spl sensitivity of the ear is highest there. Again there is nothing that suggest that this would be the case.

I did not mean to imply that.