Help needed regarding choice of HPF

[K man]

As far as I've understood it, most AVRs' crossover is a 4th order Linkwitz-Riley LPF and 2nd order Butterworth HPF.

I guess this works well when you put the crossover point where the speakers natural roll off starts, but if you want to put it higher you get an increase of bass in the area between the speakers natural roll off and the crossover point.

For example I have a speaker pair that starts rolling off at 80Hz, but I want to put the crossover point at 120Hz.

To compensate for this I'm considering putting an extra HPF to my speakers, but I don't have much experience in calculating crossovers and would need some guidance here.

Just adding another 2nd order Butterworth HPF at 120Hz seems to still give me a boost at the crossover point, while adding a 2nd order Linkwitz-Riley HPF at 120Hz seems to make me loose a bit of higher frequencies.

Should I add a 2nd order Butterworth HPF, but at slightly higher frequency, such as 180Hz? Or how to best approach this?

 

[ZolaIII]

What do you use? I guess it's AVR. What's the sub/sub's and which speakers (including enclosure types)? Do you have UMIK or such?
Thing is they don't usually roll off exactly as declared (12 or 24 dB per octave for closed enclosure / ported designs).
I in REW define type of speaker I measure (sub or full range), then I individually measure them (matching left and right main with a PEQ) and separate sub/sub's. Then I PEQ the slope to have almost perfect shape and that's about it regarding crossover and matching in generally but I do it on a PC with infinite PEQ's and chenel remapping as suit's my needs (which is pretty much both limited more or less in closed designs either it's a standalone DSP or AVR).
Will be interesting for me to hear how other people do it on particular closed box designs (different AVR's and DSP's).

 

[K man]

I guess the essence of my question is: What HPF should I add on top of a 2nd order Butterworth to best match (get a linear sum) with a 4th order Linkwitz-Riley LPF.

The practicalities with measurements, PEQs, etc. comes later. I want to get the theory right first (or at least a good approximation).

 

[mdsimon2]

If you have the ability to implement biquads (any miniDSP can to do this) and you understand the roll-off behavior of your speaker you can arbitrarily change the roll-off behavior with a Linkwitz transform.

If you are starting from an 80 Hz BW2 roll-off (I assume your speaker is sealed?) and want to get a good sum with a 120 Hz LR4 low pass, you would use a Linkwitz transform to change the roll-off behavior to 120 Hz BW2 and then apply an additional 120 Hz BW2 high pass on top of that. In an ideal world this will give you exactly 120 HZ LR4 high pass behavior.

Linkwitz Transform

miniDSP is a leading manufacturer of Digital Audio Signal Processors for the HomeTheater, Hifi, headphone and Automotive market. Join our large community of Audiophiles, Engineers and DIYers using our innovative products.

www.minidsp.com

Michael

 

[K man]

Perfect. That was exactly what I needed. Thanks a lot.

 

[witwald]

Out of curiosity, I implemented a VituixCAD circuit model of the system implementing the Linkwitz Transform. I've assumed the subwoofer has a 2nd-order roll-off and a –3dB point of 30 Hz. The main speakers are assumed to be closed-box enclosures with a 2nd-order roll-off and –3dB point of 80 Hz. The circuit is shown below.

The resulting computed summed frequency response is shown below. Note that the subwoofer gain was adjusted to be +0.3dB to get the smoothest response. The results look excellent.

 

[dasdoing]

imo the king's way if you want the bias to be the mains is to find the exact rolloff in meassurments and than just use the corresponding opposite LPF; So you don't need a HPF at all.

 

[mdsimon2]

Note that the subwoofer gain was adjusted to be +0.3dB to get the smoothest response.

You shouldn't need to do any level adjustments. If you want to make the phase matching perfect apply a HPF equivalent to the sub roll-off (BW2 @ 30 Hz) to the mains. Although the difference in most cases is very small.

Michael

 

[mdsimon2]

imo the king's way if you want the bias to be the mains is to find the exact rolloff in meassurments and than just use the corresponding opposite LPF; So you don't need a HPF at all.

I use this approach myself but a lot of people are not fans of it because they think mains should be high passed to minimize distortion.

Personally I like the sound of a 2nd order x-over between my main/subs, not sure if this is because distortion gives a fuller sound or the phase distortion is less or something else. I do high pass my mains with a filter equivalent to my sub rolloff (30 Hz BW2) for phase matching reasons so the mains get a small amount of relief.

Michael

 

[dasdoing]

I use this approach myself but a lot of people are not fans of it because they think mains should be high passed to minimize distortion.

It depends. A bass bossted 2-way with 5" or 6" woofer I surely would want to high pass. A speaker that starts rolling off at 80Hz though? it surely has no boost whatsoever

 

[Salt]

In an ideal world this would give you an ideal solution.
In real life there is phase-relation and roll-off detoriation by filters.
To go out right, find the filter configuration that gives a flat response; then invert one speaker (no matter if sub or mains): if then there is a deep notch at croossing frequency, it's fine.
Return to former config and enjoy.

If you have the ability to implement biquads (any miniDSP can to do this) and you understand the roll-off behavior of your speaker you can arbitrarily change the roll-off behavior with a Linkwitz transform.

If you are starting from an 80 Hz BW2 roll-off (I assume your speaker is sealed?) and want to get a good sum with a 120 Hz LR4 low pass, you would use a Linkwitz transform to change the roll-off behavior to 120 Hz BW2 and then apply an additional 120 Hz BW2 high pass on top of that. In an ideal world this will give you exactly 120 HZ LR4 high pass behavior.

Linkwitz Transform

miniDSP is a leading manufacturer of Digital Audio Signal Processors for the HomeTheater, Hifi, headphone and Automotive market. Join our large community of Audiophiles, Engineers and DIYers using our innovative products.

www.minidsp.com

Michael

 

[mdsimon2]

In real life there is phase-relation and roll-off detoriation by filters.

I don't understand what you are getting at here, the behavior of the filters is easy to understand.

The biggest difference between the "real world" and "ideal world" is understanding the response of the speakers. The good thing is that in general the high pass behavior of a sealed speaker is minimum phase and it is relatively easy to make a near field measurement and model the baffle step contribution and add it to the measured response. In the "real world" you may need some additional EQ to get the speaker roll-off to fit a desired response. Ported speakers are more difficult to measure / model but not impossible.

Another good way to approach this issue is to measure the drivers, flatten their response with EQ well past their intended frequency window and then apply a crossover. Such an approach is described in the Grimm LS-1 white paper -> https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf. This approach is very flexible as once you have EQ'd each side of the x-over for flat response you can very easily change the x-over.

To go out right, find the filter configuration that gives a flat response; then invert one speaker (no matter if sub or mains): if then there is a deep notch at croossing frequency, it's fine.

YMMV but I would not trust in room measurements to tell me how a sub / main crossover is behaving as there is far too much room interaction. Better to measure / model it unless you can make ground plane measurements IMO.

Michael

 

[Salt]

I don't understand what you are getting at here, the behavior of the filters is easy to understand.

The biggest difference between the "real world" and "ideal world" is understanding the response of the speakers. The good thing is that in general a sealed speaker is minimum phase and it is relatively easy to make a near field measurement and model the baffle step contribution and add it to the measured response. In the "real world" you may need some additional EQ to get the speaker roll-off to fit a desired response. Ported speakers are more difficult to measure / model but not impossible.

Another good way to approach this issue is to measure the drivers, flatten their response with EQ well past their intended frequency window and then apply a crossover. Such an approach is described in the Grimm LS-1 white paper -> https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf. This approach is very flexible as once you have EQ'd each side of the x-over for flat response you can very easily change the x-over.



YMMV but I would not trust in room measurements to tell me how a sub / main crossover is behaving as there is far too much room interaction. Better to measure / model it unless you can make ground plane measurements IMO.

Michael

First: baffle step is out of low range in about 80-120 Hz (except for very large speakers, of course);
Second: theoretically is the one side, measured the other
Third: of course it all depends on the result at prefered listenening seat

"Another good way to approach this issue is to measure the drivers, flatten their response with EQ well past their intended frequency window and then apply a crossover": is what I said....

 

[mdsimon2]

We have the capability to model baffle step so you might as well, if it doesn't matter you will see that in your modeling. I find it most important in terms of level setting between the sub and main. This plot from @napilopez's excellent quasi anechoic measurement tutorial shows the significance of considering it on a 6.5" bookshelf speaker.

"Another good way to approach this issue is to measure the drivers, flatten their response with EQ well past their intended frequency window and then apply a crossover": is what I said....

I may have misunderstood you as you did not explain your technique in much detail. I assumed when you said "To go out right, find the filter configuration that gives a flat response" you were talking about in room measurements, if this was not the case I am sorry I misunderstood.

Michael

 

[Salt]

Misunderstood.
Baffle step is rising SPL due to cabinet width and woofer switching from omnidirectional to bidirectional, usually at lower mids.
Your graph gives bass response due to real SPL and that would be pretty poor with boom bass at 100Hz, or, as mentioned, room modes.
Modelling is the one, measuring and analysis the other point.
Trial and error it's always with loudspeaker, but they are worth it to work on it ... sometimes, at least.

 

[dasdoing]

it is true that the HP and LP will never create perfect crossovers.
one way is to EQ the total response later in the chain.
the more perfect way is to use the passes as targets in REW and divide the measurements by it. this will create perfect passes for a perfect crossover