• 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). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Need advice: how to time-align single subwoofer with pair of speakers?

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,190
Likes
2,289
Location
Zg, Cro
#1
I'm considering buying a SW to enhance LF resposne of my Castle Harlech S2 speakers. Unfortunately, in terms of avialable space and WAF compatibility I can afford to put only one of them (I'm thinking of Kef Kube12b), so my question is how to time-align a single subwoofer with pair of speakers as distance between SW and speakers would differ considerably. Harlechs are sitting in a triangle formation in front of LP (app 4m away) while SW will sit close to the left side wall which means distance toward left speaker will be app 0.7m while distance toward right speaker will be app 4.3m. That means that SW will be also app 3.8m from the LP as it will sit near the left spear.

I will be doing integration via filters designed with rePhase and BruteFIR convolver plugin for Volumio. I plan to implement high-pass filters to releave Harlechs of duty to play LF below 80Hz. I plan to feed SW via stereo line input from MiniDSP U-DAC8 DAC.

So my question is: how do I time align a single SW with 2 speakers that are not at the same distance from it?
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,190
Likes
2,289
Location
Zg, Cro
#2
I just discovered that internal filter can be turned off by turning the crossover knob to „LFE“ position, which seems as an optimal solution to me as I intend to do the filtering with BruteFIR.


LFE switch.JPG



Regarding phase – should I time align SW to my left main speaker and hope for the best sound or there is a better strategy? :)
 

Ron Texas

Major Contributor
Joined
Jun 10, 2018
Messages
1,820
Likes
1,281
Location
Equidistant From Everywhere
#3
My experience is phase is about avoiding cancellation at certain frequencies in the crossover range. Measure with REW using both settings and choose the one which avoids cancellation. You might want to try placing the Sub midway between the mains and against the front wall and see how it sounds. That's my current setup and it works better than having the sub off in a corner. However, this is highly room dependent.
 

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
3,185
Likes
3,921
Location
Monument, CO
#4
There are a myriad of ways; here is a simple'ish version. The most important place to align the sub to the mains is at the crossover frequency. Above and below that frequency, the contributions from the "other" source(s) diminish(es), so alignment becomes less critical. That is, below crossover the mains are fading away, and above the sub is fading away, so each contributes less to the overall sound in the "other" speaker's band of interest. Obviously, with DSP you can set up filters to get everything nice and lined up across the spectrum by checking the step response, frequency response, etc. But for a quick first pass, get it right at the crossover frequency. That will happen when the signals from main and subs are aligned (in phase), so the easiest thing to do is to play a tone at the crossover frequency and adjust the phase of the mains and/or sub(s) for the loudest signal at the main listening position (MLP).

Note crossover and speaker design (sealed vs. ported, how drivers are wired, etc.) as well as filter latency can make the delay different than the physical distance. Also, if the sub is closer to the MLP than the mains, you may have to delay the sub's signal, but if the mains are closer then you may have to delay the signal to the mains. Etc.

The more complex way with DSP is to run frequency sweeps and adjust phase/delay per-frequency point by tweaking the filters for the flattest frequency response. Or adjust for cleanest step/pulse response. Etc.

Note SBIR, room modes, etc. can corrupt readings no matter the method chosen. Measurement programs often allow you to "gate" the measurement window to use direct sound, but sometimes that is impossible or impractical (e.g. gating can cause frequency bands to be neglected or falsely suppressed or emphasized).

HTH - Don
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,190
Likes
2,289
Location
Zg, Cro
#5
There are a myriad of ways; here is a simple'ish version. The most important place to align the sub to the mains is at the crossover frequency. Above and below that frequency, the contributions from the "other" source(s) diminish(es), so alignment becomes less critical. That is, below crossover the mains are fading away, and above the sub is fading away, so each contributes less to the overall sound in the "other" speaker's band of interest. Obviously, with DSP you can set up filters to get everything nice and lined up across the spectrum by checking the step response, frequency response, etc. But for a quick first pass, get it right at the crossover frequency. That will happen when the signals from main and subs are aligned (in phase), so the easiest thing to do is to play a tone at the crossover frequency and adjust the phase of the mains and/or sub(s) for the loudest signal at the main listening position (MLP).

Note crossover and speaker design (sealed vs. ported, how drivers are wired, etc.) as well as filter latency can make the delay different than the physical distance. Also, if the sub is closer to the MLP than the mains, you may have to delay the sub's signal, but if the mains are closer then you may have to delay the signal to the mains. Etc.

The more complex way with DSP is to run frequency sweeps and adjust phase/delay per-frequency point by tweaking the filters for the flattest frequency response. Or adjust for cleanest step/pulse response. Etc.

Note SBIR, room modes, etc. can corrupt readings no matter the method chosen. Measurement programs often allow you to "gate" the measurement window to use direct sound, but sometimes that is impossible or impractical (e.g. gating can cause frequency bands to be neglected or falsely suppressed or emphasized).

HTH - Don
Thank you for yout thoughts!

You pretty much answered my main dilemma: as I will be sending both channels to the 1 sub I can only time align it relative to the mains at one point, and that is MLP. If I would have 2 subs (one on each side wall) I would probably be sending left channel to left sub and right channel to right sub but I would again have to time align both subs to MLP, correct?

One more thing: what do you suggest for XO, Linkwitz-Riley or Butterworth? Should rollof rate be 24db/octave or something else?
 
Last edited:

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
3,185
Likes
3,921
Location
Monument, CO
#6
Thank you for yout thoughts!

You pretty much answered my main dilemma: as I will be sending both channels to the 1 sub I can only time align it relative to the mains at one point, and that is MLP. If I would have 2 subs (one on each side wall) I would probably be sending left channel to left sub and right channel to right sub but I would again have to time align both subs to MLP, correct?

One more thing: what do you suggest for XO, Linkwitz-Riley or Butterworth? Should rollof rate be 24db/octave or something else?

Every sub needs to be aligned to the mains for the signal at the MLP, stereo or mono. I ran stereo subs for along time, vacillated back and forth, and now run mono. So little content is actually stereo in the sub region, I decided I couldn't hear it anyway, and stereo just adds one more layer of complexity to the task.

Technically the L-R crossover is a cascade of second-order Butterworth stages IIRC but in any event I would use the standard 24 dB/oct L-R. That provides maximally-flat response; better for frequency response, less so for pulse response, but in filter design it is hard to get both. Frequency response has been shown to be what we are most sensitive to in listening tests (e.g. see Dr. Toole's book etc.) You can always go down the filter rabbit hole later. If it helps, last time I set up a bi-amped system, I used a L-R crossover and it worked well for me.

IME/IMO/FWIWFM/etc. - Don
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,190
Likes
2,289
Location
Zg, Cro
#7
Every sub needs to be aligned to the mains for the signal at the MLP, stereo or mono. I ran stereo subs for along time, vacillated back and forth, and now run mono. So little content is actually stereo in the sub region, I decided I couldn't hear it anyway, and stereo just adds one more layer of complexity to the task.
Are you saying you have two subs and you feed them both with mono signal? How did you sum up stereo signal into mono?

Technically the L-R crossover is a cascade of second-order Butterworth stages IIRC but in any event I would use the standard 24 dB/oct L-R. That provides maximally-flat response; better for frequency response, less so for pulse response, but in filter design it is hard to get both. Frequency response has been shown to be what we are most sensitive to in listening tests (e.g. see Dr. Toole's book etc.) You can always go down the filter rabbit hole later. If it helps, last time I set up a bi-amped system, I used a L-R crossover and it worked well for me.
I see, thank you!
 

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
3,185
Likes
3,921
Location
Monument, CO
#8
Are you saying you have two subs and you feed them both with mono signal? How did you sum up stereo signal into mono?
I have four subs and my processor has two sub outputs that can be mono, dual-mono (same signal but delay and amplitude handled separately for each), or stereo. I use them in dual-mono to feed front and back pairs or subs. Most mixes sum the bass to mono anyway. In the past when I needed to sum I just used a pair of 10k resistors (100k for my tube preamp) to sum the L/R preamp outputs. Think of a "Y" connection with the legs being the resistors and the point they combine the mono sub output, e.g.

L -> 10k ->mono sub
R -> 10k ->mono sub

HTH - Don
 
Top Bottom