How do you set your high/ low pass filters, for sub speaker integration?

[dualazmak]

Hello OP @My adventures in stereo,

In my multichannel DSP-based multi-driver multi-amplifier fully active stereo project, I assume these posts would be of your interest and reference;

- Measurement of transient characteristics of Yamaha 30 cm woofer JA-3058 in sealed cabinet and Yamaha active sub-woofer YST-SW1000: #495, #497, #503, #507

- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-1_ Precision pulse wave matching method: #493
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-2_ Energy peak matching method: #494
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-3_ Precision single sine wave matching method in 0.1 msec accuracy: #504, #507

I recently wrote here that;
Even very advanced XO/EQ/delay/measurement software would not be always "accurate/reliable enough" in assessing/tuning "time alignment of air sound"; one example case can be found here on my project thread; the story in that post strongly suggests that we do need "validated independent" measurement system.

You would please find my latest system setup here.

Present DSP EKIO's XO/group delay configuration in my setup and the entire signal path;

The actual air sound Fq response measured at my listening position having 0.1 msec precision time alignment achieved between all the SP drivers including sub-woofers as well as L-SPs to R-SPs;

As for the safe and flexible on-the-fly relative gain control, please refer to here and here.

 

[My adventures in stereo]

re time aligned after applying peq and crossover
subs left and right crossed over at 120 hz 6 db roll off, rear 80 hz at 6 db
Mains 80 hz at 12 db
variable smoothing applied

 

[ebslo]

Thanks for the detailed, insightful reply, and for introducing certain new concepts to me
1. SHD 4 channels - 2 channels to the fronts, one to RT front sub, one to Lt Front sub, out from Lt sub to Rear Sub
2. Mains Sum - Not sure what this is & how to check
3. I use the REW alignment tool, as per Jeff Mery youtube video, have not realigned after changing eq filter & cross over settings ( not sure how to redo this, as the delay is already applied, in SHD)
4. I will try the sweeps you posted

1. So in the SHD matrix mixer I presume you mix both L and R into each of the sub outputs? I wonder what that does to low frequency localization (ie. the test I referenced in 4.) If this configuration loses too much stereo effect between 80 and 150ish Hz, you could try running the front subs just as L and R respectively, then run the output of both to the rear sub (since it has L and R inputs). That way stereo effect would be preserved above 80Hz. You would just have to do all the sub EQ in the subs instead of the SHD, but you have that restriction on the left and rear subs already.
2. Don't worry about this too much, more just a point of interest. Since the SHD doesn't have FIR's, the only way to fix any issues is with all-pass filters which are quite tricky and I don't know of a tool to make it easier. Anyway, if you're interested, select just the left and right mains measurements in the REW "all-SPL" display. Open the menu above the upper right corner of the graph and click "RMS Average". De-select the RMS average that was generated, then click "Vector Average" from the same menu. Now select the "RMS Average" and "Vector Average" plots and de-select everything else. Any divergence between "Vector Average" and "RMS Average" means the mains have phase mismatch at that frequency (usually due to different room interaction). Note for this to work, the measurements must have been taken with a timing reference (acoustic timing reference is fine).

re time aligned after applying peq and crossover
subs left and right crossed over at 120 hz 6 db roll off, rear 80 hz at 6 db
Mains 80 hz at 12 db
variable smoothing applied

That looks like it's coming together nicely

 

[My adventures in stereo]

1. So in the SHD matrix mixer I presume you mix both L and R into each of the sub outputs? I wonder what that does to low frequency localization (ie. the test I referenced in 4.) If this configuration loses too much stereo effect between 80 and 150ish Hz, you could try running the front subs just as L and R respectively, then run the output of both to the rear sub (since it has L and R inputs). That way stereo effect would be preserved above 80Hz. You would just have to do all the sub EQ in the subs instead of the SHD, but you have that restriction on the left and rear subs already.
2. Don't worry about this too much, more just a point of interest. Since the SHD doesn't have FIR's, the only way to fix any issues is with all-pass filters which are quite tricky and I don't know of a tool to make it easier. Anyway, if you're interested, select just the left and right mains measurements in the REW "all-SPL" display. Open the menu above the upper right corner of the graph and click "RMS Average". De-select the RMS average that was generated, then click "Vector Average" from the same menu. Now select the "RMS Average" and "Vector Average" plots and de-select everything else. Any divergence between "Vector Average" and "RMS Average" means the mains have phase mismatch at that frequency (usually due to different room interaction). Note for this to work, the measurements must have been taken with a timing reference (acoustic timing reference is fine).

That looks like it's coming together nicely

1 I mix both L&R to L&R subs & since the Rear sub gets the L Sub signal, it is getting the same common feed
Tried the sub localisation test , no issues detected
2 Will give, what you suggested a try

I am really enjoying the result of todays calibration, thanks for your support

 

[Hexspa]

Off topic but how you like those SVSs? I’ve been debating whether to go Neumann KH 750 or SVS SB2000. This is for a home studio and I intend to get at least two. SVS is unbalanced but about 1/3 price for non-pro (would use minidsp and rew).

 

[My adventures in stereo]

Off topic but how you like those SVSs? I’ve been debating whether to go Neumann KH 750 or SVS SB2000. This is for a home studio and I intend to get at least two. SVS is unbalanced but about 1/3 price for non-pro (would use minidsp and rew).

I like the SVS PB 2000 Pro, it gives me sufficient headroom to work with
The PEQ is very useful too, to fine tune after REW calibration
The phone app makes working with 3 subs really easy

 

[My adventures in stereo]

did a recording today

 

[My adventures in stereo]

moved the front left sub , to position in picture above and recalibrated

 

[Hexspa]

Looks good. Apparently that dip at 182Hz has a quarter wavelength of 18.5" (47cm). It looks like your speakers are up about that high. How far are they from the rear and side walls? If equidistant, you can probably jigger their position to minimize that aspect, if you want. I'm not seeing the whole response so I can't say it's an issue.

Cool setup though. Gotta get some mood lighting in there. Steely Dan under 5600K florescents is a thing but maybe not all the time.

 

[Tom C]

Looks like you got there.
‘What else can that 55Hz peak be but a room mode?
The SoundStage! measurements show the low end dropping fast below about 85Hz. 100 Hz should make a good high-pass filter for these. I’d consider even higher, given the rate distortion is rising below 200Hz.

 

[Tangband]

I have been setting my low pass & high pass values to 80 hz -12 db on my minidsp systems
Noticed that on measuring through REW, this value causes a significant loss of gain from 50 to 150 hz, compared to unfiltered curve

I played around with values, setting high pass to 80 hz at -6 db and low pass to 50 hz at - 6db
Gave me the least destructive result

So my recommendation based on the above is not to blindly, set your high pass & low pass to the default 80 hz, measure & tweak instead

My gear

minidsp SHD, REW for measurements
Nad C 298 power amp
Revel M 126 Be
SVS PB 2000 Pro x3

Are you aware that slopes in the crossover ( second order , 12 dB/oct ) demands switching polarity + and - on either subwoofer or main speakers ?

A very good integration can be had with 18 dB/ oct high pass ( main speaker ) and 30 dB/oct low pass filtering ( subwoofer ) if the main speakers are closed.
In this way , the acoustical power response will be slightly better than using a 12/24 dB/oct THX crossover .

Using 6 dB slopes for subwoofers are not a good idea.

You should get away from the 80 Hz crossover dogma - in every room and case there are different needs. You could try anything between 60 Hz to 100 Hz .

Plug your Revels with socks so they are closed for better integration . After this is done , start experimenting with a THX crossover 12 dB/oct for the revels and 24 dB/oct for the subwoofers. Then try 18/30.
75 Hz LP/HP is a good starting point.
No polarity switch needed.

Do both measuring AND listening to real music . If the measurements look good but the sound with music sounds awful, trust your ears.

 

[My adventures in stereo]

Looks good. Apparently that dip at 182Hz has a quarter wavelength of 18.5" (47cm). It looks like your speakers are up about that high. How far are they from the rear and side walls? If equidistant, you can probably jigger their position to minimize that aspect, if you want. I'm not seeing the whole response so I can't say it's an issue.

Cool setup though. Gotta get some mood lighting in there. Steely Dan under 5600K florescents is a thing but maybe not all the time.

I will try moving one of the main speakers, a bit from the wall
Thanks

Looks like you got there.
‘What else can that 55Hz peak be but a room mode?
The SoundStage! measurements show the low end dropping fast below about 85Hz. 100 Hz should make a good high-pass filter for these. I’d consider even higher, given the rate distortion is rising below 200Hz.

The 55 hz is a room mode indeed
I have the mains crossed over at 80 hz 12 db roll off, front L& R subs 120 hz with 6 db roll off
Rear sub 80 hz at 6 db roll off
I will try setting the front mains to a 100hz 12 db roll off

Are you aware that slopes in the crossover ( second order , 12 dB/oct ) demands switching polarity + and - on either subwoofer or main speakers ?

A very good integration can be had with 18 dB/ oct high pass ( main speaker ) and 30 dB/oct low pass filtering ( subwoofer ) if the main speakers are closed.
In this way , the power response will be slightly better than using a 12/24 dB/oct THX crossover .

Using 6 dB slopes for subwoofers are not a good idea.

You should get away from the 80 Hz crossover dogma - in every room and case there are different needs. You could try everything between 60 Hz to 90 Hz . Plug your Revels with socks so they are closed for better integration . After this is done , start experimenting with a THX crossover 12 dB/oct for the revels and 24 dB/oct for the subwoofers. Then try 18/30.
No polarity switch needed.

Do both measuring AND listening to music . If the measurements look good but the sound with music sounds awful, trust your ears.

I have set polairty based on measured response and on listening
The main speakers are ported
the measured and listening responses seem to give best results with a 6 db roll off in my room

I agree that the 80 hz, should not be used as a hard rule
I have the mains crossed over at 80 hz 12 db roll off, front L& R subs 120 hz with 6 db roll off
Rear sub 80 hz at 6 db roll off

The present settings, is objectively & subjectively, the best I have got from this system

 

[Tangband]

I will try moving one of the main speakers, a bit from the wall
Thanks

The 55 hz is a room mode indeed
I have the mains crossed over at 80 hz 12 db roll off, front L& R subs 120 hz with 6 db roll off
Rear sub 80 hz at 6 db roll off
I will try setting the front mains to a 100hz 12 db roll off


I have set polairty based on measured response and on listening
The main speakers are ported
the measured and listening responses seem to give best results with a 6 db roll off in my room

I agree that the 80 hz, should not be used as a hard rule
I have the mains crossed over at 80 hz 12 db roll off, front L& R subs 120 hz with 6 db roll off
Rear sub 80 hz at 6 db roll off

The present settings, is objectively & subjectively, the best I have got from this system

You could always try plug your revels with socks to make them closed . The reason you have a dip in the power response with a 12 dB/HP crossover for your revels, is probably because you havent switched polarity ( +- ) on the revels . You shouldnt need to overlap the crossovers that much .

Nice setup on the picture

Trust your ears , if it sounds very good - it is good . This is just an internet forum and none of us has heard your system for real.

 

[Tangband]

One suggestion from me - try your system without minidsp in the chain ( unplug it ) and use plugged closed Revels without HP crossover and use the crossovers inbuilt in your SVS subwoofers . Do the tuning for the subs by ear. Start at 75 Hz

You might be surprised.

 

[My adventures in stereo]

You could always try plug your revels with socks to make them closed . The reason you have a dip in the power response with a 12 dB/HP crossover for your revels, is probably because you havent switched polarity ( +- ) on the revels . You shouldnt need to overlap the crossovers that much .

Nice setup on the picture

Trust your ears , if it sounds very good - it is good . This is just an internet forum and none of us has heard your system for real.

One suggestion from me - try your system without minidsp in the chain ( unplug it ) and use plugged closed Revels without HP crossover and use the crossovers inbuilt in your SVS subwoofers . Do the tuning for the subs by ear. Start at 75 Hz

You might be surprised.

Thanks
I will try reversing the polarity on the revels and running them sealed

 

[dualazmak]

Are you aware that slopes in the crossover ( second order , 12 dB/oct ) demands switching polarity + and - on either subwoofer or main speakers ?

The reason you have a dip in the power response with a 12 dB/HP crossover for your revels, is probably because you havent switched polarity ( +- ) on the revels .

I agree this is an important point if you (we) would use second order -12dB/Oct high-pass low-pass XO filters in DSP.

In my DSP setup, I decided to use -12dB/Oct XO filters simulating Yamaha NS-1000's original passive LCR network (please refer to my post here); I very carefully "inverted", therefore, the polarity settings in EKIO DSP configuration as shown below;

and,

Please note that the L&R sub-woofer YST-SW1000 has polarity (phase) inversion switch which can be controlled by IR remote controller. Furthermore, Be-tweeters and horn super-tweeters are "intentionally inverted" at their driver terminals as in Yamaha's original configuration.

These polarity settings have been very carefully and intensively confirmed by subjective listening tests and microphone measurement giving present best tuned Fq response at my listening position;

Please refer to my post here for the details of my latest system setup.

 

[Tangband]

This link below is good reading about different crossover slopes. Its important to remember that those rules are general with both active IIR dsp crossovers and passive crossovers . A 12dB/oct HP and LP crossover has one speaker inverted.

Also, a bass reflex loudspeaker falls with about 24 dB/oct below the tubes resonance frequency, and a closed box falls with about 12 dB/oct below the loudspeaker/box resonance frequency.

A closed boxed main speaker that falls 12 dB/oct below the box resonance frequency at maybe 80 Hz, can be crossed with a 12 dB/oct high pass electrical crossover . The combined result acoustically will be a 24 dB/oct high pass filter. This goes well combined with a 24 dB/oct electrically low pass subwoofer filter .

This is called a THX crossover ( 12HP/24LP = 24/24 acoustically ) and is popular in AVR:s.
In a home cinema setup, a good results can be had if all 5 main speakers are closed , and this is a demand for all THX certified systems.

With a 18HP/30LP crossover with closed main speakers , the resulting acoustically integration between subwoofer and main speaker will be a 30/30 dB/oct filter. This will give a slightly better power response in the room and many people prefer this compared to a THX crossover. I suggest everyone who has a dsp crossover to try this for perfect subwoofer integration.

Audio crossover - Wikipedia

en.wikipedia.org

Edit: its always the resulting acoustical slopes that matters, not the electrical slopes in the dsp or with calculation programs.

 

[My adventures in stereo]

so, did another run with the cross over suggestions above

1 sub and mains with no eq, no filters, ports plugged

2 Final with flat eq, mains phase inverted, time aligned
18 db roll off at 100 hz for mains
30 db roll off at 100 hz for subs
Had to invert the phase for the mains, to recover the gain loss noted previously (Thanks Tangband)
time aligned finally

3 Final with SVS eq

Final words
Sounds pretty good, good flat bass, no complaints about mid or treble

 

[dualazmak]

This link below is good reading about different crossover slopes.

Yes, that wiki page is really nice.

I also summarized on my project thread;
- Useful public domain documents and books for understandings on audio crossover: #132-#135
Let me copy-paste here again as follows;

https://en.wikipedia.org/wiki/Audio_crossover

In my post #120 and #129, I touched on this nice PDF document;
http://linea-research.co.uk/wp-content/uploads/LR Download Assets/Tech Docs/CrossoverFilters White Paper -C.pdf
which is very nice for having basic understandings on various filter types, i.e. Butterworth (BW), Bessel, Linkwitz-Riley (LR), Hardman and Linear Phase (LIR) filters. I learned a lot especially about the "Phase Issues".

One of the Senior Member of ASR and great contributor in audio science kindly informed me this document;
http://www.acourate.com/freedownload/XOWhitePaper.pdf
Yes, I know this is an introduction to Acourate(TM) Audio Toolbox. I found the contents are really nice for my basic understandings on crossover.

Many of you visiting here are already aware of this famous one;
https://www.amazon.com/dp/B01FURPS40/#customerReviews

And, not only about crossover but also about all aspects of acoustics;
https://www.amazon.com/Master-Handbook-Acoustics-Sixth-Everest-ebook/dp/B00O2A7GYW/ref=sr_1_1?dchild=1&keywords=Master+Handbook+of+Acoustics&qid=1590153537&s=digital-text&sr=1-1

User manuals of software crossover also contains nice info for our basic understandings at a glance, just for example;
http://downloads.lupisoft.com/ekio/EKIO_user_manual.pdf

Also suggested by @Burning Sounds;
http://www.linkwitzlab.com/crossovers.htm
http://www.linkwitzlab.com/frontiers_5.htm#V
and his (with Russ Riley) original paper - http://www.linkwitzlab.com/JAES/jaes_papers76.htm

 

[dualazmak]

Edit: its always the resulting acoustical slopes that matters, not the electrical slopes in the dsp or with calculation programs.

I fully agree on your fundamental and important comment and policy!

In my DSP-based multichannel multi-driver multi-amplifier stereo project, I too have been always sticking to subjective listening and objective "reliable and fully validated"measurements of real air sound at my listening position in my audio environments; this is especially true in my project since I have flexible relative gain control after the DAC for each of the SP drivers.

Not only for Fq response measurements but also time alignment measurement and tuning, we need to establish our own "independent and fully validated" measurement platform/methods as I recently pointed here; I wrote;
Even very advanced XO/EQ/delay/measurement software would not be always "accurate/reliable enough" in assessing/tuning "time alignment of air sound"; one example case can be found here on my project thread; the story in that post strongly suggests that we do need "validated independent" measurement system.