48db/oct crossovers causing listening fatigue?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,

I see I have started an interesting discussion here :). Thanks everyone for effort. A quote from a link in previous post :
Linkwitz-Riley Crossovers: A Primer :

"The caveat, though, is an increased difficulty in designing good systems with sharp slopes. The loudspeakers involved have differing transient responses, polar patterns and power responses. This means the system designer must know the driver characteristics thoroughly. Ironically, sometimes loudspeaker overlap helps the system blend better even when on-axis amplitude response is flat."

It is interesting because I have assumed in the past the steeper slope the better. That assumption has changed when I have listened to various configurations, as playing with miniDSP is so easy, and you can load all configuration at once. 48db vs 24db filters make a very audible difference all else being equal.
Still I believe if filter order is too low then distortion, cone break ups and resonances can come into play, so 24db/oct is now my choice.
Also its good hear I am not the only one having listening fatigue from using 48db/oct filters.

Regards,
Lukas.
 
Last edited:

ra7

Member
Joined 2009
Paid Member
This is silly, really. It does not matter what electrical filters you are using. What matters is the acoustical target. For a woofer crossing at 700 Hz, you may have an acoustical LR4 target, but it is very likely you would need a 400 Hz BW 3rd order electrical filter to get there. However, on the tweeter you may need an 2nd order LR at 1000 Hz to get to the acoustical target. You cannot simply flip from 8th order to 4th order and proclaim one better than the other. Proper on-axis summing with good phase overlap through the crossover, matching the directivity to ensure smooth transition, ensuring the drivers are not running into distortion outside their passbands, and notching out anomalies are all things that need to be considered.
 
This is silly, really. It does not matter what electrical filters you are using. What matters is the acoustical target. For a woofer crossing at 700 Hz, you may have an acoustical LR4 target, but it is very likely you would need a 400 Hz BW 3rd order electrical filter to get there. However, on the tweeter you may need an 2nd order LR at 1000 Hz to get to the acoustical target. You cannot simply flip from 8th order to 4th order and proclaim one better than the other. Proper on-axis summing with good phase overlap through the crossover, matching the directivity to ensure smooth transition, ensuring the drivers are not running into distortion outside their passbands, and notching out anomalies are all things that need to be considered.

If you are not listening to your speakers on-axis, is it still best to aim for a flat on-axis response? Would you definitely not want to achieve a flat response at the angle you will be listening at? (That's a genuine question, by the way!)
 

ra7

Member
Joined 2009
Paid Member
This is slightly OT, CopperTop, but I'll give you my take on your question. If you're listening at some angle and try to make it flat there, then there is a chance that the on-axis has a rising response. This is never good, IME. It gives a bright character to the sound, which at first seems nice, coz you can hear so much more detail, but eventually, it becomes fatiguing.

Now, if you have a constant directivity waveguide, where the off-axis response is nearly the same as the on-axis response, then you can make it flat everywhere by ensuring it's flat on-axis. But then you also need to match it to the response of the woofer, i.e., cross to it just when it starts becoming directional.

If you're doing conventional cone and dome, best to make it flat on-axis and have the off-axis be smooth replicas of the on-axis, except in the HF where a falling response is expected. The research says holes in the power response are acceptable, peaks are not. Also, flat power usually sounds bright.

Check out these measurements:

http://www.soundstagenetwork.com/measurements/speakers/revel_ultima_salon2/

http://www.soundstagenetwork.com/measurements/speakers/kef_201-2/

These will sound good, IME.
 
Last edited:
If you are not listening to your speakers on-axis, is it still best to aim for a flat on-axis response? Would you definitely not want to achieve a flat response at the angle you will be listening at? (That's a genuine question, by the way!)

Speaker is two port, in and out, if trying to modify response then measurement point must be chosen, and might as well be close to listening axis.

On the other hand, measurement of bare circular driver or such centered on circular baffle produces worst case circular diffraction pattern (Bessel function?). A measurement taken a few degrees off axis with well behaved driver is solution for all points in plane of measurement perpendicular to driver axis that are equidistant from intersection of plane and driver axis. Now for region +/- few degrees of measurement point an area is bounded in the plane with much greater surface area than circular spot about driver axis in same plane with same angular spread.

Maybe built in advantage in two-way when measuring normal plane bisecting acoustic centers of drivers?

I make point of not measuring exactly on driver axis.....
 
My recent experience is dealing between the mains and subs. I find the steeper the better. 4th or 6th electrical. From listening to others speakers, some of the best have had very shallow crossovers. I just have never found drivers I thought suitable for first order electrical. As was commented on before, it could also depend a lot if you are aligning the crossover with the acoustical rolloff. I don't always do that due to staying away from breakup or just too much distortion in the drivers lower end.
 
My recent experience is dealing between the mains and subs. I find the steeper the better. 4th or 6th electrical. From listening to others speakers, some of the best have had very shallow crossovers. I just have never found drivers I thought suitable for first order electrical. As was commented on before, it could also depend a lot if you are aligning the crossover with the acoustical rolloff. I don't always do that due to staying away from breakup or just too much distortion in the drivers lower end.
There is a lot of personal preference on this apparently. Steep slopes in the area you describe certainly do the job but I find that the system as a whole lacks some dynamics as a result of the higher order filters. Bass lines and a pianos lower register sound thin(ner) to me. A third order elec. seems like a nice compromise to my ear ... even though the Bessel adds a slight bump here (around 116hz) ... I'm very OK with that.
 
Another subjective opinion.
While playing with the DCX2496, my impression was that the LR24 might be "better implemented" than the LR48. Something simular could be an issue with the Mini-DSP.
I have no proof, but I picked drivers with wide overlap for my first project so I could try different crossover orders, types and points.
 
Administrator
Joined 2004
Paid Member
What are you saying, Pano?
When choosing crossover points and slopes most designers talk about FR, power response, out of band, polar response - all of them important. But I never see discussions of harmonics.

A lot of what gives different materials their "sound" - paper, metal, plastic, kevlar, etc is the harmonic structure. If you truncate those harmonics too soon or in the wrong place something will sound "off". The ear expects to hear certain things, if it doesn't they will sound odd.

Not something you read much about, but it's an important consideration when picking drivers and designing crossovers.
 
Another subjective opinion.
While playing with the DCX2496, my impression was that the LR24 might be "better implemented" than the LR48. Something simular could be an issue with the Mini-DSP.
I have no proof, but I picked drivers with wide overlap for my first project so I could try different crossover orders, types and points.

I've worked with DCX2496 a lot, and have captured IR for both filters; very text book. Each when properly aligned sounds near identical at moderate levels with music program. Stress becomes apparent at lower power levels with LR24.
 
Post 42 by Ra7 so far is one of the few posts that gives the correct picture.

Most of the posters here seem to think you can treat XO filters as separate beasts and make far reaching statements about the audibility of electrical transfer function as such. That is simply not correct.

Loudspeaker and XO should be treated as one. It is the combined filter plus driver acoustic output that matters. Only that way will the drivers sum correctly. Often the XO also has an equalizing function for baffle step and driver peaking. There should not be much of a difference between the transfer functions of active vs.passive.

Sometimes even a properly optimized 2e order electric will produce an acoustic LR4. There is no way a designer can evade measuring drivers in the enclosure that will be used and import the results in proper XO software, such as Leap, Calsod, LspCad. Standard calculators are completely useless here.

Simply playing with textbook filters of your DCX or MiniDSP will get you nowhere, because the summed SPL is all over the place. Once you have obtained flat on axis on 1 meter or so, and have also obtained a deep nul at the XO by reversing polarity of one of the drivers, you are on safe ground and have a starting point for any discussion.

Regards,

Eelco
 
True, there are a lot of things that change with changed crossover slopes (order), but I almost always found that 8th order slopes sounded worse than 4rth (and 4rth from 2nd), in middle frequencies (not in a sub), and with drivers that were working well within their “comfort” zone, even with matched driver directivities at the point.
Reasons could be (among many): differences in phase response (gr.delay), in dynamic response regarding power margin / distortion rise vs power / flux intermodulation in the v.c.-magnet assembly etc. (abrupt changes vs progressive integration), abrupt vs progressive change in the height of the sound source.
 
There is a lot of personal preference on this apparently. Steep slopes in the area you describe certainly do the job but I find that the system as a whole lacks some dynamics as a result of the higher order filters. Bass lines and a pianos lower register sound thin(ner) to me. A third order elec. seems like a nice compromise to my ear ... even though the Bessel adds a slight bump here (around 116hz) ... I'm very OK with that.

I should mention, I usually cross at about 60 to 65. That leaves 99% of the music in my main 6 1/2 inchers. My experience is any lack of punch in the bass is far more likely limitations in the upper bass. Frequently that may be a dip caused by sub phase-time mismatch causing a dip. For me, sub position and step crossover avoid that.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.