• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

High pass filters on mains

CapMan

Active Member
Joined
Mar 18, 2022
Messages
167
Likes
392
Location
London
I use my Devialet Expert to implement high and low pass filters to mains and dual mono subs.

Having measured with REW I opted to roll off the mains (Harbeth P3ESR) at 120Hz.

The received wisdom is to roll off the mains 2nd order which when (theoretically) combined with the natural roll off of the sealed mains gives a combined 4th order roll off. With room gain in play the mains don’t measure at all like the theory and they have decent output down to 50hz :)

I therefore opted to implement a 4th order rolloff on the mains in the Devialet settings to better match the 4th order used in the subs.

Any wisdom from the experts on whether this is a good strategy? There are so many permutations of cross over frequency and roll off order that insights would be appreciated to narrow the field !

I can set 1st, 2nd, 3rd, 4th on mains and subs .

I think the answer will be - whatever measures best …

Thanks all.
 
Last edited:

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
6,941
Likes
13,714
Location
Monument, CO
Roll-off will be the electrical roll-off plus the acoustic roll-off of the speakers and room effects. I would use whatever sounds best, backed by measurements to identify problem areas like SBIR, room modes, and main/sub speaker interactions that cause comb filtering and peaks/valleys that may be hard to hear and identify without measurements to point the way. It is not unusual for the SPL level to be only 10 dB down or so an octave either side of the crossover, and that is about the half-loudness point (even more in the deep bass as the equal-loudness curves compress, i.e. are closer together), one reason I prefer to use speakers that reach deep enough that I can set the crossover to the subs at least an octave above the mains -3 dB point. Not set in stone by any means.

HTH - Don
 
OP
CapMan

CapMan

Active Member
Joined
Mar 18, 2022
Messages
167
Likes
392
Location
London
Roll-off will be the electrical roll-off plus the acoustic roll-off of the speakers and room effects. I would use whatever sounds best, backed by measurements to identify problem areas like SBIR, room modes, and main/sub speaker interactions that cause comb filtering and peaks/valleys that may be hard to hear and identify without measurements to point the way. It is not unusual for the SPL level to be only 10 dB down or so an octave either side of the crossover, and that is about the half-loudness point (even more in the deep bass as the equal-loudness curves compress, i.e. are closer together), one reason I prefer to use speakers that reach deep enough that I can set the crossover to the subs at least an octave above the mains -3 dB point. Not set in stone by any means.

HTH - Don
Thanks Don - will let you know how I get on.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
6,889
Likes
7,101
Location
Brussels, Belgium
It's not very clear what the settings you ended up using are.

The guidelines you shared are there to make sure of two things:

1) No speaker is being disproportionally driven compared to the other

and

2) there is no +6dB anechoic boost at the crossover frequency.

No 2 is crucial if you're crossing at 100Hz or higher as it might cause you to localize the subwoofer or cause some weirdness.

My go to ACCOUSTIC crossovers are 8th order at 100Hz and 4th order at 80Hz (75Hz if you have single digits flexiblity), otherwise you're going to localize the subwoofer.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
6,941
Likes
13,714
Location
Monument, CO
Thanks Don - will let you know how I get on.
For listening tests, pink noise (not too loud!) can be useful. I used to compare with headphones, but of course they aren't guaranteed flat either, and acoustic memory is short... If measurements identify a peak or null, you can listen for that, then see if you can hear the changes. A piano (could be online now) or audio oscillator (ditto) provides a useful way to correlate pitch (sound) to frequency.
 
OP
CapMan

CapMan

Active Member
Joined
Mar 18, 2022
Messages
167
Likes
392
Location
London
It's not very clear what the settings you ended up using are.
Current settings are 120hz / 4th order on subs and 120hz / 4th order on mains

Have dual mono subs - front corners to try and reduce risk of localisation.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
6,889
Likes
7,101
Location
Brussels, Belgium
Current settings are 120hz / 4th order on subs and 120hz / 4th order on mains

Have dual mono subs - front corners to try and reduce risk of localisation.
How flexible is the Devialet expert bass management?

Looks like the Harbeth P3ESR is following an Linkwitz-Riley 12db/oct at 80Hz.

Measurements from stereophile's review, their bass is usually off though so not very confident with that.
1659445697878.png
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
6,889
Likes
7,101
Location
Brussels, Belgium
Wait you don't have to relly on shady stereophile measurements, Since this is a sealed speaker all you got to do is do a REW measurement (No EQ, no Room correction, no crossovers) with a microphone a cm or two away from the woofer's dust cap and you will get something more accurate than stereophile.

Here is a post explaining the whole process. You only need the data from 20Hz to 200Hz by the way, you don't need to do the whole protocol for the full range response.

 

sigbergaudio

Major Contributor
Manufacturer
Forum Donor
Joined
Aug 21, 2020
Messages
1,479
Likes
2,625
Location
Norway
4th order both ways like you're doing now will likely work well. If you can measure you will know for sure, otherwise trust your ears :)
 
OP
CapMan

CapMan

Active Member
Joined
Mar 18, 2022
Messages
167
Likes
392
Location
London
How flexible is the Devialet expert bass management?
Hi - thanks for the input.

I can set :
- low pass frequency
- slope - 6,12,18,24
- delays on mains and subs

I can run the subs individually in R/L mode or summed to mono (which is what I have done).

The subs have been aligned using REW to get flattest summed response at the LP.
 

Scrappy

Active Member
Joined
Aug 4, 2022
Messages
189
Likes
145
I had my rig (Fostex PX5 over Zenith Allegro 10’s per side) crossed at 160hz for stereo full-range. Hipassed low boxes at 45-50? Can’t recall. Was nice in a small room, bigger room now, running lows below 80 mono, got some work to do.

Anyhow, what I hear from top-notch guys is 4th order Linkwitz-Riley for everything due to phase response. I test processing amps all day for Redacted and it’s Redacted. I remember fu ing around with Bessel, for some reason about compromise chart from a BSS whitepaper.
 

GalZohar

Senior Member
Joined
Aug 8, 2021
Messages
428
Likes
203
If you aligned the subs with each other, you should now also try to align them with the mains at the crossover region.

If you measure just the subs, then just the speakers, then you can try to sum them in REW with various delays. You can also see if your sum (when they work together) is not any less than any of the single measurements, which indicates a serious cancellation at that frequency.
 
OP
CapMan

CapMan

Active Member
Joined
Mar 18, 2022
Messages
167
Likes
392
Location
London
If you aligned the subs with each other, you should now also try to align them with the mains at the crossover region.

If you measure just the subs, then just the speakers, then you can try to sum them in REW with various delays. You can also see if your sum (when they work together) is not any less than any of the single measurements, which indicates a serious cancellation at that frequency.
Hi - after volume matching, my process has been to align the subs to get the flattest response using REW , in practice this meant polarity inverting the left sub and also running it 50 degrees out of phase with the right sub.

I used REW EQ and applied some modest 2 band EQ to the subs using their onboard equaliser . I then re measured the subs with the EQ.

I then summed left and right main and time aligned with the summed and EQd subs to get flattest response across the cross over region.

I double checked that the time alignment (delayed mains by 5ms) made sense from an impulse reponse perspective which it did (the rise on the subs was around 5 seconds ahead ) - but I appreciate this part is a little open to debate.

On the basis of less is more I have not applied further EQ on the mains - I am happy enough with the overall response .
 

GalZohar

Senior Member
Joined
Aug 8, 2021
Messages
428
Likes
203
You may want to align each speaker separately, as well as the center if you have one (and if you don't, maybe still check the "center channel").

I'm also not sure about the impulse being out of alignment, but in any case I know that dips in the crossover region sound bad.

When it comes to filter slopes, from what I understand the recommended approach is to have 24db/oct rolloff on both mains (highpass) and sub (lowpass), but this should be the result with the filters applied, therefore the actual filters needed to achieve this depend on the unfiltered frequency response. For example, Denon (and seems like many other) receivers apply 24db/oct for the sub (expecting it to naturally roll off at much higher frequencies), and only 12db/oct for the speakers, expecting the speaker to roll off naturally at 12db/oct, or at least be equalized by the room correction system to do so. Unfortunately in Denon changing the crossover doesn't automatically change the rolloff frequency of Audyssey, which can only be changed with MultEQ-X, and yet I haven't found anyone who tested how much that actually matters. If you're using manual EQ you could try messing with various filters to try get as close to a (measured) 24db/oct rolloff for your mains (with all filters applied).
 

FeddyLost

Addicted to Fun and Learning
Joined
May 24, 2020
Messages
712
Likes
509
I think the answer will be - whatever measures best …
There are lot of reasons to make high-passed mains.
In case of small sealed monitors with 83 Db/[email protected] sensitivity, I'd first check worst case of your SPL and measure distortion figures that you have.
In case of really big and heavy sub you can have issues with subjective "speed" of midbass: not every 15" long-thrower can handle 120-150 Hz nicely.
Also, speakers might be positioned suboptimal in terms of room modes, so it might be wise to cut off some problematic frequency.
Then you can try some sumilations EQing sub+sats in REW and decide what looks best. Then measure result and try to live with it ...
Different solutions usually mean different compromises. IMO "super flat at any cost" is not an option in real room and looks more like some surprise, that like ultimate goal.
If your subs can handle 120 Hz nicely, I'd also try 2.2 if resulting FR and decay will be acceptable.
 
Top Bottom