REW crossover slope oddity

[JasonB]

I noticed something that seems a bit odd when switching the target crossover slope fom 12 db to 24 db per octave in REW. The 24 db per octave slope starts rolling off sooner than the 12 db per octave slope. In my mind the opposite should be true since with a shallower slope there will be more overlap between the sub and the mains than with the steeper slope, therefore the 12 db per octave slope should start to roll off sooner. Am I missing something?

Thanks,

Jason

[brucek]

There's a known bug that will be fixed in the next release.

The temporary fix is that when you switch between 12dB/octave and 24dB/octave you have to give a toggle to the crossover frequency (down and up) thumbwheel to correct the problem.

[Wayne A. Pflughaupt]

Not sure I get where you are going. Perhaps you should re-phrase the question, since “sooner” suggests a time element (i.e., “Since we left before they did, we will get there sooner”). Frequency response and crossover filtering has no time element.

Also, can’t tell from what you wrote which slope you believe is shallower and which is steeper.

Switching between the two filter targets I can assure you that REW's presentation is correct. Help us understand what your understanding is, and maybe we can explain things so you'll understand them.

Regards,
Wayne

[JasonB]

Wayne,

I hope this is a clearer explanation....

My issue is not with the slopes themselves, I believe those are correct. My issue is where the initial roll off starts. An 80 hz 12 db per octave low pass filter should start to roll off at lower frequency than an 80 hz 24 db per octave filter. The crossover slopes on the REW are the opposite, the 24 db per octave low pass filter starts rolling off at a lower frequency than the 12 db per octave filter.

Also, if the crossover slopes are based on 2nd and 4th order Linkwitz Riley topologies they should both be down 6 db at the crossover frequency and neither are in the REW

Thanks,

Jason

Wayne A. Pflughaupt wrote:

Not sure I get where you are going. Perhaps you should re-phrase the question, since “sooner” suggests a time element (i.e., “Since we left before they did, we will get there sooner”). Frequency response and crossover filtering has no time element.

Also, can’t tell from what you wrote which slope you believe is shallower and which is steeper.

Switching between the two filter targets I can assure you that REW's presentation is correct. Help us understand what your understanding is, and maybe we can explain things so you'll understand them.

Regards,
Wayne

[JasonB]

Bruce,

The fix appears to correct the 24 db per octave slope so the it's 6 db down at the crossoever point but it still starts to roll off at a lower frequency than the 12 db per octave slope.

Thanks,

Jason

brucek wrote:

There's a known bug that will be fixed in the next release.

The temporary fix is that when you switch between 12dB/octave and 24dB/octave you have to give a toggle to the crossover frequency (down and up) thumbwheel to correct the problem.

[brucek]

Also, if the crossover slopes are based on 2nd and 4th order Linkwitz Riley topologies they should both be down 6 db at the crossover frequency and neither are in the REW

I believe the targets are complient with the THX recommendations for bass management filters, with the 24dB/oct being consistent with a dual butterworth slope (same as LR) at -6dB at the corner frequency and that the -12dB/oct slope matches a simple butterworth at -3dB at the corner? I may be wrong about this, but I remember JohnM (author of REW) discussing this before.

With regard to the mains rolloff, the idea is that the combined effect of a main speaker's own roll-off together with the 12dB/octave roll-off of the bass management filter will give an effective roll-off that matches the 24dB/octave applied to the sub. John added the 24dB/oct for the limited range mains to show the combined effect.

The -24dB/oct slope certain does cross at -6dB. It is down from 75dB to 69dB at 80Hz = -6dB, while the -12dB slope is down to 72dB = -3dB@80Hz......


80-12-24 copy.jpg


brucek

[Wayne A. Pflughaupt]

JasonB wrote:

Wayne,

I hope this is a clearer explanation....

My issue is not with the slopes themselves, I believe those are correct. My issue is where the initial roll off starts. An 80 hz 12 db per octave low pass filter should start to roll off at lower frequency than an 80 hz 24 db per octave filter. The crossover slopes on the REW are the opposite, the 24 db per octave low pass filter starts rolling off at a lower frequency than the 12 db per octave filter.

Also, if the crossover slopes are based on 2nd and 4th order Linkwitz Riley topologies they should both be down 6 db at the crossover frequency and neither are in the REW.

Thanks Jason, that clears it up nicely. Phrased like that it appears you know a lot about crossovers – more than I do apparently (not a hard status to achieve, I assure you, as you probably noticed from my misstatements in my previous post). I had to go searching for confirmation of your assertion about the 12 dB filter starting at a lower frequency than the 24. I came up with this:

Linkwitz-Riley alignments.svg[1] a.jpg

I think brucek has already ably answered your question; I dont think REW claims to be showing Linkwitz-Riley alignments.

Regards,
Wayne

[JasonB]

Bruce,

You are correct, the 24 db per octave slope is -6 db at the crossover frequency. It's the bug in REW that you previosly mentioned that threw me off there. If you switch from 24 db to 12 db and then back to 24 db the slope shown is a bit off until you adjust the frequency down one click and up one click as you suggested in your post above. I had been switching back and forth between 12 and 24 before I looked more closely at the slopes.

12 db per octave Butterworth slopes are down only 3 db at the crossover frequency, so the slopes are correct if that is what they are based on. But the high pass and low pass do not sum flat on a Butterworth topology, there will be a 3 db bump at the crossover frequency. That's why I made the assumption that both the 12 db and 24 db slopes would be based on Linkwitz Riley topologies which do sum flat.

I also agree with your statement about the roll off of the mains combining with the 12 db per octave high pass slope of the bass management to produce a 24 db per octave slope. Most people will never target a 12 db per octave low pass on there subwoofers, the only exception might be those running dedicated two channel systems without bass management or crossovers on the mains.

Thanks,

Jason

brucek wrote:

I believe the targets are complient with the THX recommendations for bass management filters, with the 24dB/oct being consistent with a dual butterworth slope (same as LR) at -6dB at the corner frequency and that the -12dB/oct slope matches a simple butterworth at -3dB at the corner? I may be wrong about this, but I remember JohnM (author of REW) discussing this before.

With regard to the mains rolloff, the idea is that the combined effect of a main speaker's own roll-off together with the 12dB/octave roll-off of the bass management filter will give an effective roll-off that matches the 24dB/octave applied to the sub. John added the 24dB/oct for the limited range mains to show the combined effect.

The -24dB/oct slope certain does cross at -6dB. It is down from 75dB to 69dB at 80Hz = -6dB, while the -12dB slope is down to 72dB = -3dB@80Hz......


Attachment 4650


brucek

[JasonB]

Wayne A. Pflughaupt wrote:

Phrased like that it appears you know a lot about crossovers ...

Wayne,

Looks can be deceiving!!! I know a tiny fractrion of all there is to know about crossovers. The chart you posted is great, it illustates my point perfectly. In order to sum flat both the 12 and 24 db per octave slopes should be -6 db at the crossover frequency and the 12 db per octave slope should start to roll off further away from the crossover point that the 24 db per octave slope.

If you were to cheat and adjust the crossover frequency of the 12 db per octave slope in REW until you reached - 6db at the crossover frequency the initial roll off may be correct for the intended crossover frequency.

I hope it doesn't seem like I am nit picking, REW is an awesome tool. I couldn't imagine trying to set up the BFD without it!

Thanks,

Jason