Subwoofer Crossover Advice

[watchnerd]

For the first time I'm experimenting with adding a subwoofer to my mains.

Mains: Dynaudio Contour 20

Manufacturer Specs: https://www.dynaudio.com/home-audio/contour/contour-20#tech-specs

Stereophile Measurements: https://www.stereophile.com/content/dynaudio-contour-20-loudspeaker-measurements

Subwoofer: M&K MX-700 (15+ years old, had it sitting in the garage)

2 x 8" drivers, push pull

Setup:

I'm currently doing all the crossover duties inside the Devialet Expert 400 using its built-in DSP.

Subwoofer is set to bypass, gain is set to Reference / 0 dB.

High Pass for Mains: 70 Hz, 2nd order

Low Pass for Subs: 70 Hz, 2nd order

I set this rather arbitrarily at one seemed to be a common sense starting point and it seems okay....not quite seamless, but not boomy or overpowering.

Questions:

1. What slope is best to use for both high and low pass?

2. Should I use a shallower rolloff on the mains high pass than the subs low pass?

3. Is there a better way of determining the crossover point than my seat of the pants guesstimate?

4. What's the best way to set the level? (I more or less just did it by ear)

 

[amirm]

3. Is there a better way of determining the crossover point than my seat of the pants guesstimate?

I will take this one. The answer is almost always 80 Hz. Frequencies below are not locatable so allow the sub to go anywhere in the room to optimize its performance. Setting it higher risks audibility of their location. Setting too low loses the advantage of placement.

 

[amirm]

4. What's the best way to set the level? (I more or less just did it by ear)

You have to measure and EQ the subs. Without it, one track will have too much bass, and another too little depending on whether they hit the room modes or not.

Minimum should be one or two parametric EQs to dial down the peaks.

Once there you can set by ear.

 

[andreasmaaan]

Do you have a measurement mic?

1. What slope is best to use for both high and low pass?

I would try with 4th order slopes, less can go wrong than with 2nd order slopes, especially if you're unable to measure. In theory, with 2nd order filters, the units should be connected in opposite polarity (but it all depends on what slope you're actually ending up with acoustically and where the unis are located with respect to each other).

2. Should I use a shallower rolloff on the mains high pass than the subs low pass?

Those Dynaudios seem to begin rolling off at about 60Hz. You'll probably get a better result with a shallower slope, but again the best thing would be to measure if possible.

Agree with @amirm that 80Hz would probably be a better xo frequency.

 

[watchnerd]

Do you have a measurement mic?



I would try with 4th order slopes, less can go wrong than with 2nd order slopes, especially if you're unable to measure. In theory, with 2nd order filters, the units should be connected in opposite polarity (but it all depends on what slope you're actually ending up with acoustically and where the unis are located with respect to each other).



Those Dynaudios seem to begin rolling off at about 60Hz. You'll probably get a better result with a shallower slope, but again the best thing would be to measure if possible.

Agree with @amirm that 80Hz would probably be a better xo frequency.

Yes, I have an UMIK I can use to measure.

Re: slope, I'm a bit confused as you say to try 4th order, but then talk about a shallower slope.

Do you mean 4th order on the low pass and 1st order on the high pass?

 

[watchnerd]

I will take this one. The answer is almost always 80 Hz. Frequencies below are not locatable so allow the sub to go anywhere in the room to optimize its performance. Setting it higher risks audibility of their location. Setting too low loses the advantage of placement.

I tried 80 and it sounded muddy in my room, but that was just with one track.

Sounds like I need to go measure.

 

[andreasmaaan]

Yes, I have an UMIK I can use to measure.

Re: slope, I'm a bit confused as you say to try 4th order, but then talk about a shallower slope.

Do you mean 4th order on the low pass and 1st order on the high pass?

Sorry, what I meant was that 4th order acoustic slopes are likely to be a good starting point in terms of ease of integration. Acoustic 2nd order slopes will be difficult because the drivers will need to behave linearly over a wider bandwidth in the first place, and because one of the units will need its polarity reverses to integrate properly.

However, to achieve acoustic 4th order slopes, you may need to use an electrical 2nd order slope for example on the speakers. You’ll need to determine what’s best with measurement and trial and error.

If your sub is near to the speakers, 100Hz may also be a better xo frequency IMHO.

 

[watchnerd]

Sorry, what I meant was that 4th order acoustic slopes are likely to be a good starting point in terms of ease of integration. Acoustic 2nd order slopes will be difficult because the drivers will need to behave linearly over a wider bandwidth in the first place, and because one of the units will need its polarity reverses to integrate properly.

However, to achieve acoustic 4th order slopes, you may need to use an electrical 2nd order slope for example on the speakers. You’ll need to determine what’s best with measurement and trial and error.

If your sub is near to the speakers, 100Hz may also be a better xo frequency IMHO.

Got it: acoustical vs electrical; now it makes sense.

BTW, my old school M&K MX-700 is good for experimenting, but I don't think I'll be using it in the long run because it lacks the fancy wireless management and EQ of more modern subs.

The new ELAC series looks interesting:

https://www.elac.com/product/sub3010/?r=us

https://www.elac.com/forums/topic/elac-subwoofer-frequently-asked-questions-by-andrew-jones/

Q5. How does the Auto EQ feature on the S10EQ and S12EQ subwoofer work?

A5. The Auto EQ works in a way that is subtly different from conventional EQ methods. In those, a calibrated microphone is required, and the response measured at the listening location is adjusted to match some “ideal” response determined by the software engineer. This is not the response that was designed by the speaker engineer themselves! Ideally, when the subwoofer engineer designs the subwoofer they have an idea of what performance they want the speaker to achieve. The limitation is that the room will modify this, hence the need for EQ. But the ideal is that the EQ gives at the listening location the response desired by the speaker engineer. This is the approach utilized in ELACs’ process. It allows for a clever change in the requirement of the microphone. We no longer require a calibrated microphone. In fact we utilize the microphone built into the smartphone itself!

The process is to hold the phone very close to the subwoofer, the so-called “nearfield”. At this location, the measured response is devoid of almost any influence from the room, and is measuring in fact the designed response of the subwoofer. Once this response is captured, the smartphone is relocated to the listening location and a second measurement made. Next a set of filters is adjusted automatically to make the response measured at the listening location match that made in the nearfield of the subwoofer. The actual performance of the microphone is now of no importance: it is the same for both measurements and we are simply adjusting to minimize the difference! The sound heard at the listening location is now the one that the designer wanted you to have. It’s a very cool technique.

Q6. Is the microphone in my cell phone or tablet good enough for setting up the S10EQ or S12EQ?

A6. The requirement for the microphone is simply that it has a response that has some capability down to 20Hz with useable signal to noise ratio. The exact response is of no importance since that function is eliminated form influencing our process. We have found that the majority of modern phones and tablets meet this requirement. For those few that do not, it is possible to purchase a simple low cost calibrated mic that plugs into the mic input of the phone or tablet to give better performance.

 

[Soniclife]

I started with 80hz 2nd order crossover, but I also worked out the delay from positioning and added a guess for the sub's DSP delay, and then used the configurator to delay the mains so they should be in phase, as my sub is behind the mains. I set the level by ear over several days, my sub has a remote so it's easy and fun to fiddle, then when I had settled on what sounded correct I checked with REW, I was within the acceptable curve range on frequency response so left that alone. I think I had to fiddle slightly with the delay setting after measuring, but didn't hear a difference.

I still use SAM on the mains, whilst crossed over to the sub, I get slightly better phase measurements like that.

Set the Devialet remote up so you can switch between the mains full range and no sub, or crossed over to the sub, I found this enlightening switching back and forth, and gives a reference point.

 

[Soniclife]

BTW, my old school M&K MX-700 is good for experimenting, but I don't think I'll be using it in the long run because it lacks the fancy wireless management and EQ of more modern subs.

The new ELAC series looks interesting:

Also look at Martin Logan and Paradigm, they have something similar built in. SVS do as well I think.

 

[Soniclife]

Acoustic 2nd order slopes will be difficult because the drivers will need to behave linearly over a wider bandwidth in the first place, and because one of the units will need its polarity reverses to integrate properly.

Can you explain the polarity reversal bit please.

 

[MZKM]

Excellent video that mostly focuses on bass and subwoofer integration:

____

Crossover point, two things to take into account:

1) Room modes, you want to find the best position in the room for your sub, you need to do measurements (or the sub crawl for those without a measurements mic), but if one location is awesome for 20Hz-60Hz but cruddy for 70Hz-100Hz, and if you need an 80Hz crossover, then you may not want that spot.

2) Speaker limitations, if you do a 2nd order filter LR at 70Hz on the speakers, that means they are 12dB down at 35Hz, which your speakers may not handle well; that also means your subwoofer is 12dB down at 140Hz, which it also may not like. Dynamic compression is also an issue, how linear is the performance at high output. I would go with a 100dB target (if your system can handle that) and send full range to your speakers, and then measure 20Hz to maybe 150Hz on the subwoofer (after finding its best placement by using the RTA and signal generator in REW).

If you just want your subwoofer to pick up where your speakers leave off (assuming room modes are not an issue), find either the -12dB or -24dB point (2nd vs 4th order) of your speakers and double that frequency; and if 2nd order, see if your subwoofer can handle 2x that XO frequency (4x the -12dB or -24dB point).

____

https://www.audiocheck.net/audiotests_subwooferkicktest.php

You can use that test to check level matching for those that don’t have a measurement mic nor SPL meter.

Keep in mind your subwoofer and speakers should not measure flat, it should be tilted, just like your speakers.

____

BTW, my old school M&K MX-700 is good for experimenting, but I don't think I'll be using it in the long run because it lacks the fancy wireless management and EQ of more modern subs.

I would use a MiniDSP SHD as your DSP/preamp. That way you can better integrate the sub as well as DSP both sub and speakers. Rythmik, HSU, PSA, and JTR all make awesome subs for the money (SVS is good bang for buck if looking at performance only until you get to their higher end with the digital display, auto-EQ, etc.). In terms of full range and no size constraints, B&W, MartinLogan, REL, Paradigm, Vandersteen, etc. don’t hold a candle to the brands mentioned above with regards to bang for buck (REL’s flagship $7500 model doesn’t dig as deep as Rythmik’s cheapest ~$510 sub). I haven’t seen measurements of new models from HSU and PSA, but mid to high tier Rythmik and JTR subs are very “musical” as well, low distortion, good group delay, etc. , even from their ported models.

You can get a wireless subwoofer adaptor for like $80, don’t limit your subwoofer choose with just that.

 

[MZKM]

Can you explain the polarity reversal bit please.

https://www.diyaudio.com/forums/multi-way/141806-phase-shift-filter.html#post1794713v

The reason the woofer's polarity stays the same and the tweeter's polarity switches is because the high-pass filter has a 90 degree phase-lead at the crossover point, and the low-pass filter has a 90 degree phase-lag at the crossover point. So the phase-lead with the phase-lag is what causes the 180 degree phase shift.

By wiring the tweeter in reverse your woofer and tweeter are in phase at the crossover point relative to one another but still 90 degrees out of phase relative to the input signal.

Of course, that’s with a perfectly time aligned system, which is why having a variable phase knob on a sub is helpful, or having fine distance adjustment/delay for the sub from your receiver/MiniDSP/etc.

 

[Soniclife]

https://www.diyaudio.com/forums/multi-way/141806-phase-shift-filter.html#post1794713v



Of course, that’s with a perfectly time aligned system, which is why having a variable phase knob on a sub is helpful, or having fine distance adjustment/delay for the sub from your receiver/MiniDSP/etc.

Does that not only apply to analogue crossovers? In this case the crossover is being doing digitally and does not change phase.

 

[Willem]

I happily use a B&W PV1d subwoofer and use it the way REL recommends: main speakers at full range, and the sub with a low pass filter only, and set at a pretty low crossover frequency.
My main system consists of Quad 2805 electrostatics and the cleanest crossover frequency was 33 Hz, with 4th order slope and 0 degree phase to preserve as much as possible of the clean output of the stats.
I have also experimented using the Harbeth P3ESR speakers from my secundary system in this same large room and with the same subwoofer. The setting recommended by B&W for these closed speakers and their specific -6 dB point turned out to be dead right: crossover at 45 Hz, 2nd order slope and 180 degree phase. Again, integration was audibly seamless.
In both systems I used the Antimode 8033 dsp room eq unit to tame room modes, and with good effect. For even better results I would recommend using two small subs rather than one large sub.

 

[MZKM]

Does that not only apply to analogue crossovers? In this case the crossover is being doing digitally and does not change phase.

Remember that your subwoofer has some delay on its own native amp processing/DSP. This is why if you ever use auto-EQ from a home theater receiver, the detected distance for subs is never physically correct, my sub is about 13ft from my MLP, but Audyssey on my Denon set it to about 16.2ft (which is pretty bang on when I checked with REW).

 

[andreasmaaan]

Can you explain the polarity reversal bit please.

Yes sure, sorry I’m on the phone today and yesterday so my responses are not very detailed and are full of typos.

A 2nd order minimum-phase xover results in 180 degrees of phase shift. As a result of this, reversing the polarity of one of the drivers is necessary to ensure correct summation at the crossover point. Leave both drivers with the same polarity and the 180 degree phase shift will cause a null at the crossover point.

(Of course, this all assumes the two drivers are coincident in space, which is never the case in practice, and particularly not when the sub(s) are placed a significant distance from the speakers. So in reality, how the drivers sum will depend on not only driver polarity but also relative distance to the listener, and of course all the other room-related factors that normally affect any speaker to sub crossover, etc.)

Does that not only apply to analogue crossovers? In this case the crossover is being doing digitally and does not change phase.

A digital linear-phase FIR filter will of course not affect the phase of the output. But to my knowledge, the Devialet uses minimum-phase IIR crossover filters, which have the same effect on phase as passive/analogue filters.

 

[Soniclife]

A digital linear-phase FIR filter will of course not affect the phase of the output. But to my knowledge, the Devialet uses minimum-phase IIR crossover filters, which have the same effect on phase as passive/analogue filters.

Have a look at the following short thread.
https://devialetchat.com/Thread-PREOUT-and-SPEAKERS-filters-type-solved-BESSEL-filters

 

[watchnerd]

I still use SAM on the mains, whilst crossed over to the sub, I get slightly better phase measurements like that.

Set the Devialet remote up so you can switch between the mains full range and no sub, or crossed over to the sub, I found this enlightening switching back and forth, and gives a reference point.

Were you running the mains full range?

I thought SAM wasn't supposed to be used or was disabled if using the mains with a high pass.

 

[Soniclife]

Were you running the mains full range?

I thought SAM wasn't supposed to be used or was disabled if using the mains with a high pass.

When I run the mains full range with no sub I used SAM.

When I use a sub using the Devialet to cross over, so HP to mains, and mono LP to sub I continue to use SAM on the mains. Theoretically this should make the roll off much more predicable for the mains.

You should not use SAM if you are using the high level into the sub from the speaker sockets, like REL recommend, as the SAM correction will also go to the sub, which will be all kinds of wrong.