What crossover-topology?

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

I've finished my big SB-Acoustics Satori project a while ago. With TW29B-B tweeter, MR16P-4 midrange and two WO24P-8 woofers in parallel in a MTWW configuration, actively tri-amp driven by a Hypex FA123. I've got the crossover points in the DSP set at 300 Hz and 3 kHz with a 1st-order roll-off. I made the frequency response completely flat, no BBC dip or any of that. I don't like the idea of changing the frequency response to make something seem to sound good because a given frequency response profile doesn't work as good for one music genre than the other. And also I don't like to change it in the first place, I want to hear things the way they are, no wizardry.

The speakers sound amazing, like high-end worthy, for easier, more relaxed and well-recorded music like acoustic music and other kinds of well-recorded music. For example albums like Anouk - Update, Nils Lofgren - Acoustic Live, Malia - Convergence, Gipsy Kings - Bem Bem Maria. You know, the beautiful kind of music, which they often like to play at hifi shows. Though with more complicated, not as well recorded music, like any rock, pop and alternative music like Golden Earring or Supertramp and all the other greatest music ever made, the speakers do get pretty harsh pretty quickly.

What do you think it will be? I think there's two reasons this could be. Either my speakers are SO good and revealing with their flat response to the point that when you play anything through it that isn't recorded very well, they shove it right in your face. Or, the 1st-order roll-off is a little to much even for these drivers. the midrange really is crossed quite at the limits of where it would like to be crossed. It doesn't like to do much below 300 Hz and it starts to roll off above 3 kHz. The roll-off above 3 kHz shouldn't really cause any harshness but maybe the very wide 1st-order roll-off below 300 Hz is a little much.

What do you think?
 
Sonce is probably correct, however there could be a second confounding factor. What's the frequency response at your listening position?

A lot of music was mixed expecting a continual roll off from the low mid-bass all the way up. The Harman Kardon target curve and the like. I've found music is usually way to grating unless I've got something similar to that curve at the listening position.
 
Sonce is probably correct, however there could be a second confounding factor. What's the frequency response at your listening position?

A lot of music was mixed expecting a continual roll off from the low mid-bass all the way up. The Harman Kardon target curve and the like. I've found music is usually way to grating unless I've got something similar to that curve at the listening position.

Well the tweeter is crossed at 3 kHz. Considering this tweeter one of the most capable on the market with a big size of 29 mm, has some of the lowest distortion on the market and is can actually be used happily to under 1,8 kHz I don't expect it to have much trouble I'm not saying your wrong, I believe it maybe does have a hard time after all.
I think I'll make the low-pass roll-off of the midrange and the high-pass roll-off of the tweeter 2nd-order. That might be a good shot. Crossing even higher is not an option.

When tuning the frequency response I put the speakers in the room and put the measuring mic at the listening position. I the DSP corrected the speakers to be totally flat.
 

Attachments

  • Meting grafiek 1.jpg
    Meting grafiek 1.jpg
    68.5 KB · Views: 449
Last edited:
Do your slopes refer to the electrical or acoustic responses?

Equalizing bass at the listening position is fine, though for mids and highs it is better to use anechoic measurements. Otherwise there is a risk of filling up dips that to not exist when the individual sounds reach your ears.

A quick recipe:
- Let the acoustic 'response' of the individual drivers follow a fourth order Linkwitz-Riley response at their operating frequency band + a few octaves outside of them. 'Response' refers to some weighted combination of on-axis and off-axis measurements including equalization, as measured under anechoic conditions.
- Apply delay to the individual drivers to time align them.
- Measure the complete loudspeaker at your listening position and equalize low frequencies < 300 Hz.
- Add equalization for personal preference.
 
Last edited:
Well the tweeter is crossed at 3 kHz. Considering this tweeter one of the most capable on the market with a big size of 29 mm, has some of the lowest distortion on the market and is can actually be used happily to under 1,8 kHz I don't expect it to have much trouble I'm not saying your wrong, I believe it maybe does have a hard time after all.
I think I'll make the low-pass roll-off of the midrange and the high-pass roll-off of the tweeter 2nd-order. That might be a good shot. Crossing even higher is not an option.

When tuning the frequency response I put the speakers in the room and put the measuring mic at the listening position. I the DSP corrected the speakers to be totally flat.

There's your problem. Music isn't mixed well for that. Your highs are way too strong compared to the bass. Read up on target curves. They're there for a reason and are the assumed response for listeners. You want the speaker to be able to reproduce flat with low distortion, but at listening position that isn't actually what you want at all. You want a rolloff from about 2Khz of 1-2.5 dB per octave. I'm not surprised one bit things sound harsh to you.
 
I made the frequency response completely flat, no BBC dip or any of that.
Your highs are way too strong compared to the bass. Read up on target curves.
I the DSP corrected the speakers to be totally flat.
I concur with mushroommunk, I assumed that your notion about "flat DSP" specifically means no BBC dip or something similar. But, your measurements are really flat, which subjectively means pronounced highs, especially when playing louder.
Anyhow, it is a good idea to check the tweeter distortion below 3 kHz (with the DSP filter ON).
 
I looked up the frequency response of some speakers at Stereophile, they are kind enough to make in-room measurements of every speaker they review.
I looked up some very good speaker like the Focal Kanta No.2, Focal Sopra No.3 Paradigm Persona H9F, KEF Reference 5 and KEF Blade. Then I looked up some of the best speakers in the world like the Tidal Akira, Rockport Technologies Avior II, YG Acoustics Carmel 2. And they all had very flat in-room responses, some even flatter than mine, some even with slightly raised highs. And I think it would be unacceptable from those speakers to be harsh. So I think I'll start with taking the slopes from 1st to 2nd-order. I'll see from there what happens.
 
I looked up the frequency response of some speakers at Stereophile, they are kind enough to make in-room measurements of every speaker they review.
I looked up some very good speaker like the Focal Kanta No.2, Focal Sopra No.3 Paradigm Persona H9F, KEF Reference 5 and KEF Blade. Then I looked up some of the best speakers in the world like the Tidal Akira, Rockport Technologies Avior II, YG Acoustics Carmel 2. And they all had very flat in-room responses, some even flatter than mine, some even with slightly raised highs. And I think it would be unacceptable from those speakers to be harsh. So I think I'll start with taking the slopes from 1st to 2nd-order. I'll see from there what happens.

If you look carefully, it isn't his listening position. He's using several tricks to get in-room measurements as close to anechoic as he can. (Speaking of stereophile here, for example the Focal Kanta No.2 for example)

"Fig.3 Focal Kanta No.2, anechoic response on tweeter axis at 50""

Speaker measures flat, but that isn't what he hears at listening position unless John Atkinson lives in an anechoic chamber, but then he wouldn't need to be merging nearfield and farfield responses like he does.
 
I looked up the frequency response of some speakers at Stereophile, they are kind enough to make in-room measurements of every speaker they review.
I looked up some very good speaker like the Focal Kanta No.2, Focal Sopra No.3 Paradigm Persona H9F, KEF Reference 5 and KEF Blade. Then I looked up some of the best speakers in the world like the Tidal Akira, Rockport Technologies Avior II, YG Acoustics Carmel 2. And they all had very flat in-room responses, some even flatter than mine, some even with slightly raised highs. And I think it would be unacceptable from those speakers to be harsh. So I think I'll start with taking the slopes from 1st to 2nd-order. I'll see from there what happens.
Please note that Stereophile, unfortunately, do not publish in-room response for all reviewed loudspeakers. There are anechoic measurements for Focal Sopra No.3 and Focal Kanta No. 2, but no in-room responses for them.
KEF Blade, KEF Reference 5 and Magico S5 mKII have nice downhill in-room response above 2 kHz (textbook target curve). Rockport Technologies Avior II is similar, except slight peak at 3 kHz.
On the other hand, Tidal Audio Akira and YG Acoustics Carmel 2 indeed have flat in-room responses. I bet I will not like their sound. After many loudspeakers I measured, I am assured that in-room response is of paramount importance.
 
Okay, well I don't have a anechoic chamber to my disposal but I've got quite some damping panels and other material in my room to control the sound so maybe if I pack al of it around the front of the speaker should be able to make somewhat anechoic measurements. Or maybe put them outside, point them up to the sky, lay some damping on the ground and place the mic above them. It's summer so that should be doable, except lifting these 65 kg closets outside. Though to control the DSP I need a Windows laptop, which I don't have so I'll have to see how I'm gonna get around that.
 
Last edited:
KEF Blade, KEF Reference 5 and Magico S5 mKII have nice downhill in-room response above 2 kHz (textbook target curve). Rockport Technologies Avior II is similar, except slight peak at 3 kHz.

Ahhh, I see!. They do indeed roll off quite a bit, and the KEF and Magico speaker are on some of the most revealing speakers around in terms of their top end, which is what I'm after with my beryllium tweeters.
 
diyAudio Moderator
Joined 2008
Paid Member
the speakers do get pretty harsh pretty quickly.
I wouldn't be looking at non-linear distortion or crossover slopes.

The top end rolloff mentioned is worth pursuing. It may be only part of the problem.
I've got quite some damping panels and other material in my room to control the sound
Have you considered reflections from the cabinet?
 
What do you mean? The baffle diffraction?

I've got one mid/high frequency absorbing panel in-between the speakers against the front wall to get rid of the most direct reflecting backwards traveling waves. It got rid of some excess highs reflecting from the front wall (not that there aren't any now) and it heavily improved the soundstage.
Further I've got a small block of glass wool isolation material in each corner to control the bass. The really tightened up the bass, made the punches quicker and got rid of some peaks caused by resonances in the corners.
 
That of the tweeter or the entire system?

For the tweeter itself. It's at it highest at 1,1 kHz at 1,8 % gradually going down without any peaks to under 0,2 % from 5,5 to 8 kHz and then gradually rising to 1 % at 10 kHz and then slightly lowering again to 0,8 % at 20 kHz.
This is at 94 dB.

For the entire system, this was the last in-room measurement I made at 75 dB. I don't know if it's any accurate or reliable though.
 

Attachments

  • Schermafbeelding 2019-04-29 om 22.50.28.jpg
    Schermafbeelding 2019-04-29 om 22.50.28.jpg
    411.1 KB · Views: 197
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.