Subwoofer integration help request

[lawnchair04]

Hello all.

I added a subwoofer to my setup, and I am doing measurements with REW and a Umik-1 to help me set the crossover and generate some EQ filters. I was hoping I could get some input on the measurements I've made and see if anyone can provide their input or guidance on how to choose crossover settings based on these measurements.

My setup is a pair of KEF LSX II, with the LSX II sub out going to a KEF Kube 8 MIE. The KEF app has options for managing High Pass Filter, Low Pass Filter, sub gain, and polarity (which can be set to either 0 or 180). (I believe KEF uses -24dB slopes for the crossovers.) I will use Wiim Pro for EQ settings for music. Unfortunately speaker and sub positions are pretty much set, so I'm trying to do my best with digital adjustments. Here is the room layout--the red circle is the subwoofer:

I've taken a series of sweep measurements using different combinations of HPF, LPF, and polarity. I also took MMM measurements using some of those combinations. At this point I am mostly using the sweep measurements to try to figure out the best crossover setting.

I can see how different combinations have different effects. For example, the closer together the HPF and LPF are, the higher the SPL between 60Hz and 100Hz. And the higher the average of the HPF and LPF, the smaller the dip between 60Hz and 75Hz. But then there are some overlaps, so for example HPF=80, LPF=80 gives almost the same curve as HPF=90, LPF=85. At any rate, it looks like I am aiming to find the best situation in the 60-100Hz region, but I'm not quite sure what that looks like. (I've mostly dismissed the measurements with the polarity inverted, since they seem to cause more problems than they solve, but I'd be welcome to input on that, too.)

For example, here are the measurements for HPF=90, with LPF ranging from 75-90 (LPF=75 being the lowest null at ~75Hz, LPF=90 the highest):

Would I be better off with the biggest boost there, which is easier to EQ down? Or is something in the middle better, to get the best average between peaks and dips?

The KEF recommended settings for the LSX II+Kube 8 MIE combo are HPF=67.5, LPF 62.5. That's in green below, with HPF=80, LPF=75 in orange and HPF=90, LPF=85 in blue for comparison:

I can see arguments for any of these, but I'm just not sure where to go, so I'd love to get some input or guidance. I've attached some files with my measurements to make the files smaller. Measurements are named in the format "H# L# g+# pol#," where H is the HPF, L is the LPF, g is the sub gain (+3 or +5), and pol is the polarity (either 0 or 180). I've also attached the MMM measurements (which include some variations in gain and polarity, but aren't as extensive as the sweep measurements) in case they're of interest.

Thanks so much for any thoughts!

 

[kyuu]

Perhaps someone else more knowledgeable might chime in, but personally I'm a bit confused as to why you would set different LPF and HPF frequencies. Crossover filters, by my understanding, should have the LPF and HPF at the same frequency in order to sum properly. Unless the filters in the LSX use different slopes? But why would they do that?

In any case, your graphs seem to support my thinking: all you get with the widely spaced filters is less energy. The LSX benefits from a relatively high crossover to reduce distortion (it's a single, small woofer also acting as the waveguide for the tweeter after all). I would set the crossover as high as possible while minimizing the ability to localize the subwoofer.

In terms of EQing out the bass region, you generally benefit from having as much energy as possible. You can always cut out the extra energy with the EQ without any issue. Adding in missing energy is much more problematic.

 

[Keith_W]

Don't post 28 graphs in 3 MDAT's. Nobody has the time to look through all of them.

I wonder why you have a rising treble peak between 10kHz - 20kHz. There are 3 possible causes: (1) you measured so loud that you have copious harmonic distortion, (2) wrong calibration file for mic orientation, (3) terrible or faulty speakers. That said, it's missing in the MMM so either you are using the wrong signal for the MMM or it's some kind of local phenomenon.

How to choose a crossover point: take INDIVIDUAL measurements of left / right speakers with no subwoofer. Examine the graph to see what the natural roll-off is, and look for distortion. If you have a lot of low end distortion, consider a higher XO point. Otherwise, the XO point should be as low as possible to avoid hearing the position of the sub.

It is unusual to choose asymmetric crossovers as you have done.

Re: Polarity. Green is Pol0, red is Pol180. Polarity 0 is obviously better. Once you decide on correct polarity, delete all the inverted polarity measurements.

Re: Gain. It looks as if your subwoofer is too loud. But without individual measurements of speaker and subwoofer I can't tell.

 

[lawnchair04]

Perhaps someone else more knowledgeable might chime in, but personally I'm a bit confused as to why you would set different LPF and HPF frequencies. Crossover filters, by my understanding, should have the LPF and HPF at the same frequency in order to sum properly. Unless the filters in the LSX use different slopes? But why would they do that?

My understanding of the theory behind using different frequencies comes from an explanation using this image (from some KEF material):

Would definitely appreciate more insight on the accuracy of this idea.

In any case, your graphs seem to support my thinking: all you get with the widely spaced filters is less energy. The LSX benefits from a relatively high crossover to reduce distortion (it's a single, small woofer also acting as the waveguide for the tweeter after all). I would set the crossover as high as possible while minimizing the ability to localize the subwoofer.

In terms of EQing out the bass region, you generally benefit from having as much energy as possible. You can always cut out the extra energy with the EQ without any issue. Adding in missing energy is much more problematic.

Thanks! This confirms my thought process as well. KEF’s recommended settings have a 5dB gap between the filters, but my measurements suggested to me that that was probably the widest I’d want them apart. And one of my reasons for adding the sub was the take some of the lower-frequency load off the speakers, so to me a higher crossover makes sense. So far I haven’t been able to detect any subwoofer localization with any of the tested crossover frequencies, so it seems I’m still in a reasonable zone.

 

[kyuu]

Hm. It appears KEF isn't doing what I expected, which is that you'd be specifying the frequency where the filter is attenuating the response by 6dB for a 24dB/octave Linkwitz-Riley (and so get a flat sum with both filters set at the same frequency). I'm not sure why they would do the crossover settings this way. I'm not finding any resources for determining where to set the frequencies on the HPF/LPF to get a proper flat sum with a quick search. I'll have to look into this more, or hope someone else can chime in here.

 

[lawnchair04]

Don't post 28 graphs in 3 MDAT's. Nobody has the time to look through all of them.

Thanks for your feedback, and apologies for the overload of graphs. I can delete the pol180 file from the original post given your feedback on that below

I wonder why you have a rising treble peak between 10kHz - 20kHz. There are 3 possible causes: (1) you measured so loud that you have copious harmonic distortion, (2) wrong calibration file for mic orientation, (3) terrible or faulty speakers. That said, it's missing in the MMM so either you are using the wrong signal for the MMM or it's some kind of local phenomenon.

I hadn't even noticed that since I was focused on the bass region. I wasn't measuring very loudly, around 75dB SPL per REW's SPL meter and pink periodic noise (also used for the MMM measurements). Could it be caused by measuring the sweeps with the mic oriented vertically (using the 90-degree cal file)?

How to choose a crossover point: take INDIVIDUAL measurements of left / right speakers with no subwoofer. Examine the graph to see what the natural roll-off is, and look for distortion. If you have a lot of low end distortion, consider a higher XO point. Otherwise, the XO point should be as low as possible to avoid hearing the position of the sub.

Here are a pair of L (blue) and R (orange) sweeps (taken with 0degree mic orientation; no 10kHz-20kHz rising peak):

Distortion for L and R:

It is unusual to choose asymmetric crossovers as you have done.

As echoed by @kyuu above. I may be misunderstanding what the setting numbers mean. I am confused as to why KEF recommends such a wide gap between HPF and LPF, which are as much as 25dB different for some speaker+sub combinations. The recommended starting setting for LS50WII+KC62, for example, is HPF=70, LPF=45...

Re: Gain. It looks as if your subwoofer is too loud. But without individual measurements of speaker and subwoofer I can't tell.

Good to know. I wasn't sure how to level match, so I used the default +3. Since I'm going through the speakers to the amp for the measurements, I'm not sure if there is a way to test the sub individually.

 

[lawnchair04]

Hm. It appears KEF isn't doing what I expected, which is that you'd be specifying the frequency where the filter is attenuating the response by 6dB for a 24dB/octave Linkwitz-Riley (and so get a flat sum with both filters set at the same frequency). I'm not sure why they would do the crossover settings this way. I'm not finding any resources for determining where to set the frequencies on the HPF/LPF to get a proper flat sum with a quick search. I'll have to look into this more, or hope someone else can chime in here.

Yes, it is a bit confusing. I believe they do use 24dB Linkwitz-Riley crossovers in their active speakers/subwoofers, but on this page for example they explain that "With the HPF filter set to 70 Hz, all frequencies above 70 Hz are sent to the LS50 Wireless II at 0 dB attenuation," so it appears that the setting refers to the frequency where the slope begins.

In other words, it seems that in the KEF app, you are not setting a crossover frequency, i.e. the frequency at which the slopes meet, but the point above which (HPF) and below which (LPF) unattenuated frequencies are passed on to the loudspeakers and subwoofer respectively. So I guess that leaves it up to the user to either use the preset, or find the best combination using measurements or based on listening...

 

[lawnchair04]

I was able to take more measurements, with individual measurements for L, R, and subwoofer. I think these are better than what I posted before. And I think they confirm @Keith_W's suspicion that my sub gain was too high. All the sub measurements are taken at gain=0dB.

L and R at HPF=off:

Subwoofer at LPF=240 with L and R at HPF=off:

Subwoofer at LPF=62.5 with L and R at HPF=67.5:

Subwoofer at LFP=80 with L and R at HPF=80:

And with both filters at 90:

Perhaps the KEF recommended setting of LPF=62.5, HPF=67.5 (or a higher setting with the same 5Hz gap between filters) might actually be the best after all? It's the only one where the curves "cross over." Or is the extra energy caused by the overlap in the other examples preferred?

Thanks again for any attention and advice. I really do appreciate it.

 

[Keith_W]

Please post the MDAT so that I can take a look.

 

[lawnchair04]

Please post the MDAT so that I can take a look.

Here they are.

 

[Keith_W]

First thing I did was to get rid of all the measurements I did not need, then sum the left and right speakers with the 67.5Hz high-pass using REW's Trace Arithmetic A+B function. I renamed it to "L+R H67.5". The window above shows the summed L+R main speakers in yellow and the subwoofer with the 62.5Hz low-pass in green.

We can see a few things, there are a couple of nasty room modes at 85Hz and 120Hz from your main speakers alone, and from this it appears that your sub is not loud enough. But we'll see. Right click on the graph and choose the alignment tool.

Choose these settings. Note REW warning you that your sub measurement does not have a timing reference, so the outcome of the alignment tool may not be valid. Play around with the gain and polarity of the sub, then slide the "fine delay adjustment" slider and watch what happens to the frequency response.

If you have DSP, the goal is to get rid of as many dips as possible and ignore all the peaks because you will chop them off later. If you don't have DSP, your goal is to get the overall freq response as flat as possible. You said in your first post that you want to make some EQ filters, but you did not mention what DSP you are using. Assuming you are using a Wiim or MiniDSP or something that uses PEQ's, continue with these steps.

First, write down the interventions that you made - gain, polarity, and delays. You will have to manually enter them into your DSP unit later. Then click on "Aligned copy" for both sub and main speakers. Then go to "Trace Arithmetic" and sum the aligned copies of the subwoofer and main speakers with A+B. Then you can use REW's EQ function. I presume you know how to use this?

 

[lawnchair04]

Amazing, thank you so much for this walkthrough. To answer your questions, I'll be using a Wiim Pro to manage the EQ filters. And yes, I am familiar with the EQ function in REW, so I should be good on that step.

I was playing with the alignment tool, and came across something that was confusing me a bit. The summed version of the sub and mains without any adjustment except +3dB gain matches the polarity-inverted sweep of mains+sub I did the other day; a summed version of the sub with the mains delayed about -16ms (as you have in your example above) matches the uninverted sweep of mains+sub I did the other day.

Aligned sum with no adjustments except +3dB gain, compared to L+R+sub with polarity inverted:

Aligned sum with +3dB gain, a -16.59ms delay and no polarity inversion, compared to L+R+sub with no polarity inversion:

Would this seem to indicate that KEF's DSP is managing the delay automatically? Or am I misunderstanding something about the alignment tool's results?

 

[dualazmak]

Just for your fundamental and essential reference and interest, please visit my post here #3 on the thread entitled "Seeking advice on integrating two subwoofers with full-range stereo speakers with passive radiators".

My recent post here #1,009 on my project thread would be also of your interest, I assume.

 

[Keith_W]

Would this seem to indicate that KEF's DSP is managing the delay automatically? Or am I misunderstanding something about the alignment tool's results?

I would say it's fortuitous. Your KEF sub has no way of knowing what delay it should set.

Don't forget to take proper measurements. The measurements I was playing with yesterday did not have a timing reference so you can not use it for time alignment. I pressed ahead and showed you how to do it anyway because I didn't want to be too much of a pain in the butt by insisting that you redo them and repost the MDAT with the time reference. But you do need to redo the measurements before you use them to make any timing related adjustments.

Also don't forget to keep an eye on the impulse response when you are using the alignment tool. You want the start of the subwoofer impulse to match the start of the main speaker impulse as much as possible.

 

[lawnchair04]

I would say it's fortuitous. Your KEF sub has no way of knowing what delay it should set.

Don't forget to take proper measurements. The measurements I was playing with yesterday did not have a timing reference so you can not use it for time alignment. I pressed ahead and showed you how to do it anyway because I didn't want to be too much of a pain in the butt by insisting that you redo them and repost the MDAT with the time reference. But you do need to redo the measurements before you use them to make any timing related adjustments.

Also don't forget to keep an eye on the impulse response when you are using the alignment tool. You want the start of the subwoofer impulse to match the start of the main speaker impulse as much as possible.

Got it! Thanks for clarifying that. I will try to take measurements with a timing reference tomorrow and then play around with things some more. (Unfortunately with just the Wiim Pro for EQ I don’t think I actually have a way to adjust the timing on the speakers, but it’s still a great learning process.)

 

[Keith_W]

Got it! Thanks for clarifying that. I will try to take measurements with a timing reference tomorrow and then play around with things some more. (Unfortunately with just the Wiim Pro for EQ I don’t think I actually have a way to adjust the timing on the speakers, but it’s still a great learning process.)

As it happens I helped someone else with a Wiim set up his subwoofers via WhatsApp ... @Arianoxx . So I know for a fact that you can adjust timing on the speakers, because he sent me this screenshot:

 

[lawnchair04]

As it happens I helped someone else with a Wiim set up his subwoofers via WhatsApp ... @Arianoxx . So I know for a fact that you can adjust timing on the speakers, because he sent me this screenshot:

Alas, that feature appears to only be available on the models that have a subwoofer out like Wiim Ultra and Amp, which the Wiim Pro does not. I'm using the sub out from my KEF LSX II.

I took some time-referenced subwoofer measurements, but unfortunately it seems to have an unexpected effect. See the measurements below, one without a time reference (red) compared to one with a time reference (blue). For some reason including a time reference also makes the response quieter below 50Hz.

It also wasn't very consistent; four different measurements yielded delays of 5.9, 1.9, 1.6, and 1.3. Not really sure what's going on there. I was trying different things all morning and the problem was consistent. I assume it has to do with running the signal through the KEF LSX II and then to the subwoofer.

Still, it looks like the delay is probably only around 2-4ms, and adjusting for it doesn't do anything about the null at ~45Hz anyway, so I think I'll be able to live with it for now. Great experience to learn more about measurements and REW's features.

 

[Keith_W]

I took some time-referenced subwoofer measurements, but unfortunately it seems to have an unexpected effect. See the measurements below, one without a time reference (red) compared to one with a time reference (blue). For some reason including a time reference also makes the response quieter below 50Hz.
View attachment 460065

That is absolutely bizarre. I have never seen anything like this in my personal experience, nor have I seen it happen to anyone else. It's also suspiciously close to your crossover point. Are you SURE that the REW settings are correct?

It also wasn't very consistent; four different measurements yielded delays of 5.9, 1.9, 1.6, and 1.3. Not really sure what's going on there. I was trying different things all morning and the problem was consistent. I assume it has to do with running the signal through the KEF LSX II and then to the subwoofer.

Still, it looks like the delay is probably only around 2-4ms, and adjusting for it doesn't do anything about the null at ~45Hz anyway, so I think I'll be able to live with it for now. Great experience to learn more about measurements and REW's features.

I am not a big fan of how REW uses the timing reference. I asked John about it before, and he said that the acoustic reference is not part of the impulse response so I can't use it to determine the timing manually. So I have my own way of doing a timing reference which ignores REW's method. It's a bit too difficult to describe it in a forum post, but if you are really keen you can send me a PM and we can have a chat on WhatsApp and I can walk you through it.

FWIW, inconsistent subwoofer measurements are a fact of life, but I would not expect them to vary by so much. In my own experience, sub measurements vary by about 0.2ms or so. I prefer to read the impulse peak myself and determine the timing manually, rather than simply have REW report it.

 

[janbth]

Alas, that feature appears to only be available on the models that have a subwoofer out like Wiim Ultra and Amp, which the Wiim Pro does not. I'm using the sub out from my KEF LSX II.

I took some time-referenced subwoofer measurements, but unfortunately it seems to have an unexpected effect. See the measurements below, one without a time reference (red) compared to one with a time reference (blue). For some reason including a time reference also makes the response quieter below 50Hz.
View attachment 460065

It also wasn't very consistent; four different measurements yielded delays of 5.9, 1.9, 1.6, and 1.3. Not really sure what's going on there. I was trying different things all morning and the problem was consistent. I assume it has to do with running the signal through the KEF LSX II and then to the subwoofer.

Still, it looks like the delay is probably only around 2-4ms, and adjusting for it doesn't do anything about the null at ~45Hz anyway, so I think I'll be able to live with it for now. Great experience to learn more about measurements and REW's features.

You could also try an RTA measurement with REW as another test to see if you can resolve this discrepancy. A link I could find on this is: https://mehlau.net/audio/room-correction-peq/, but you could probably find others by searching.

By the way, I would recommend using Var smoothing in REW, in order to see the correct shape of the room resonances: https://www.roomeqwizard.com/help/help_en-GB/html/graph.html#top

 

[dualazmak]

I am not a big fan of how REW uses the timing reference. I asked John about it before, and he said that the acoustic reference is not part of the impulse response so I can't use it to determine the timing manually. So I have my own way of doing a timing reference which ignores REW's method.

This is for me too!

OP @lawnchair04,
You would please read again carefully my above post #13 and the linked post #3 on the thread entitled "Seeking advice on integrating two subwoofers with full-range stereo speakers with passive radiators".
I now attach the contents under the below spoiler cover.

Spoiler: Summary of critical factors in integration of subwoofer(s) and main woofers in our individual acoustic environments:

We have so many critical factors for crossover (XO) between subwoofer(s) (SWs) and main woofer(s) (WOs) in our audio system for better/best low frequency (Fq) sound reproduction in our own/individual listening environments including room acoustics. Of course, the final goal would be greatly dependent on your music listening personal preferences, and hence there would be no general/standard procedures and approaches, I assume.

The major factors would be;
A1. Room acoustic mode(s) including reflection, dispersion, absorption, standing waves, resonances, etc.
A2. Precision (1 ms precision) time-alignment (a kind/side of phase tuning) between SWs and WOs, at your listening position,
A3. Optimization of relative gains for SWs and WOs, should be flexibly controlled on-the-fly while listening to music,
A4. Optimal selection of XO filter type(s) (i.e. BW, LR, Bessel, etc.), XO Fq, slopes at both side, phase inverse or not, further specific EQs or not, etc.

Before starting your optimization/tuning exploration journey in this regard, you need to know/understand several features/aspects; the major points would be;
B1. "Major" low Fq sound reflective plane/wall in your acoustic environment (not always needed to be fully eliminated),
B2. Be aware of that we always have overlapped Fq zone where SWs and WOs sing together, whatever XO Fq and slopes we would use,
B3. Basic understandings on the physical configuration(s) of SWs and WOs, especially ported or sealed, difference in mass of moving parts, etc.,
B4. Difference in transient behavior (step response?) of SWs and WOs, both kick-up responses and fade-out patterns;
____Damping factor/performance of the amplifier(s) driving SWs and WOs more-or-less do "matter" for transient behaviors.

As you may well aware, many audiophile people use REW and/or similar advanced audio measurement/tuning software tools together with suitable measurement microphone(s) for the optimization. I too used wonderful REW during my early stage in my multichannel project as you can find my posts #17, #18, #20, #21, #22 on my project thread.

Because of various reasons, however, nowadays, I use REW mainly as validation and confirmation tool for my rather primitive but reliable reproducible understandable (to me) validated simpler measurement and tuning methods as shared below; If you would be interested, please read carefully these posts on my project thread;
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-1_ Precision (time-shifted) 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
- 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
- Identification of sound reflecting plane/wall by strong excitation of SP unit and room acoustics: #498

I believe the above linked my posts well cover almost all the aspects and tunings relating to above A1 through A4 and B1 through B4.

Let me emphasize that the use of rectangular-sine-tone-bust signals (8-wave, 3-wave, and even 1-wave) of various Fq and the analysis of the recorded air sound (by second independent PC) of these tone-burst signals using Adobe Audition (or Audacity) would give you really useful information on optimization of SWs and WOs. Especially the 3D (gain-Fq-time) color spectrum of Adobe Audition showing "3D sound energy distribution" is much useful (at least to me!) for observing and tuning the XO configuration for SWs and WOs (and other SP drivers). You can find typical example case in my posts #503 and #507.

If you would be seriously interested in using the test tone signal tracks I prepared and applied in these my measurements and tunings, please simply PM me writing your wish.

These posts would be also of your interest and reference;
- Perfect (0.1 msec precision) time alignment of all the SP drivers greatly contributes to amazing disappearance of SPs, tightness and cleanliness of the sound, and superior 3D sound stage: #520

- Not only the precision (0.1 msec level) time alignment over all the SP drivers but also SP facing directions and sound-deadening space behind the SPs plus behind our listening position would be critically important for effective (perfect?) disappearance of speakers: #687

- Reproduction and listening/hearing/feeling sensations to 16 Hz (organ) sound with my DSP-based multichannel multi-SP-driver multi-amplifier fully active stereo audio system having big-heavy active L&R sub-woofers: #782

- A nice smooth-jazz album for bass (low Fq) and higher Fq tonality check and tuning: #910, #63(remote thread)


Furthermore, I highly recommend you to establish your own consistent "reference/sampler music playlist" consists of tracks of various genres hopefully fitting well for your/our music preferences.

At least in my case, I have been using my own consistent "reference/sampler music playlist" consists of 60 tracks as I shared here #670 on my project thread, and also I have dedicated thread;
- An Attempt Sharing Reference Quality Music Playlist: at least a portion and/or whole track being analyzed by 3D color spectrum of Adobe Audition

You would please find details of my latest audio setup, well covering all of the above mentioned topics, in my post here #931 on my project thread.
- The latest system setup of my DSP-based multichannel multi-SP-driver multi-amplifier fully active audio rig, including updated startup/ignition sequences and shutdown sequences: as of June 26, 2024: #931
Edit:
- The latest Fq-SPL (re-confirmation) of multiple amplifiers SP high-level output signals and that of room air sound at listening position: all measured by “FFT averaging of recorded cumulative DSP-processed flat white noise” (as of June 8, 2025): #1,009