Get an exclusive sneak peek into our new project. >>> Click Here

Multi sub optimizing tips

1671 Views 24 Replies 9 Participants Last post by The Bogg, Oct 11, 2019

The Bogg

Discussion Starter · Oct 1, 2019

I decided to start this thread after coming to the realization that a simple error in setup that went unnoticed for over a year had a profound effect on the bass in my dedicated room. Like many of you, I have multiple subs which I bought more for evening out the response across multiple seats than for raw output. Each JL F113 was placed on the quarter wavelength spot on the sidewalls, with 2 per side. One of the subs was slightly away from the exact spot the others occupied because of the location of the door. I had expected fairly uniform bass throughout the theatre space based on this configuration but it wasn't as uniform as I had expected and I assumed it was because one of the subs was about a foot off the desired spot.

I finally got around to recalibrating the system using Dirac Live on my Datasat RS20. The previous calibration a year ago was rushed a bit and was decent so I hadn't redone it. As I look back at the graphs it seems so obvious that something was going on, lol, but I never questioned it until now as I was considering using bass management to optimize the in-room response. I will cut to the chase: at some point some little fingers had flipped the polarity switch on my left front sub. So 3 subs were set with normal polarity and 1 with reversed polarity.

I will post the graphs when I get home. Previously you could walk around the theater with sub test tones playing and hear the peaks and valleys in different locations. I just figured since it wasn't an optimally sized or shaped room that this was "normal". After flipping the polarity switch the difference is mindboggling - the tones sound the same as you walk around the room and the measured response at each of the 7 seats is IDENTICAL. I guess the theory and the reality do actually match, lol.

See less See more

Reactions: 2

1 - 20 of 25 Posts

Steve Bruzonsky

· #2 · Oct 1, 2019

Now that is simply mind"Bogg"ling!!!

Glad you found the error. Its so easy with our large complex home theatre systems for something so "tiny" like a polarity switch on one of multiple subwoofers to screw stuff up. And with all the good things we hear it can take months or more, or never, to find the error of our ways. I've done this in more ways than one.

My four subs will be going in the same spots as yours!!!!

Reactions: 1

The Bogg

Discussion Starter · #3 · Oct 2, 2019

heh, "bogg"ling indeed.

One might ask, how could I not notice this mistake by listening? Here's a graph of the uncorrected vs corrected (by Dirac) only at the main listening position, which is where I spend almost all of my time:

See less See more

The Bogg

Discussion Starter · #4 · Oct 2, 2019

As you see from the previous post and picture, the measured response at the main listening position is not so bad so one wouldn't think there was much wrong. But here's a multipoint measurement with 7 measurements corresponding to the 2 rows of seats - 3 in front and 4 in back. This one is with no dirac:

See less See more

The Bogg

Discussion Starter · #5 · Oct 2, 2019

This one is with dirac turned on:

See less See more

The Bogg

Discussion Starter · #6 · Oct 2, 2019

In both of the above cases the response variation between seats is terrible! That's why it's important to fight for the money seat.

Now that the polarity switch has been put in the right position, here's a sweep from the hotseat with Dirac turned off with an overlay of the same position with the wrong polarity:

See less See more

The Bogg

Discussion Starter · #7 · Oct 2, 2019

A couple of noteworthy findings on the graph above - all of a sudden you are seeing response NOT taper off rapidly above 60hz (note that the crossover on the subs is bypassed and I didn't set any crossover in the Datasat. I double-checked this, lol). Secondly, the response below 20hz doesn't taper off quite as quickly. Here's where things get really interesting...take a look at the multipoint measurements (at all 7 seats). Forget the fact that it isn't flat, these were taken with Dirac off, but check out the seat to seat variation now:

See less See more

torii

· #8 · Oct 2, 2019

I went a year with using audyssey w/out turning off limiters on rythmik subs when doing room corrections....limiters off when running audyssey then after turn limiters on....having limiters on when measuring wasnt optimum. so many little things can throw off the best result. my front L/R positions werent optimum either, inch here/there...it all matters. need to be ocd.

Reactions: 1

The Bogg

Discussion Starter · #9 · Oct 2, 2019

In case it wasn't clear from the last graph, those were 7 different measurements at the seven different seats! I couldn't believe my eyes (and still wonder if I did something wrong, lol). This was the point in having multiple subs "ideally" placed in a room. The response isn't measurably flat yet but the consistency is noteworthy.

I have each sub hooked up to a discrete output on the Datasat so I can adjust the distances independently. I didn't know if this was worth fiddling with but since my 2 rear subs are a little closer to my main seat than the front pair I thought I'd experiment with delaying the rear subs (they have the same polarity as the front ones - some people advocate flipping the polarity in the rear pair but I didn't find it useful in my room). This graph compares the standard setting vs adding a 2ms delay to the 2 rear subs:

See less See more

Reactions: 1

The Bogg

Discussion Starter · #10 · Oct 2, 2019

The noteworthy thing is that you get a "free" 5db boost at 20hz without messing anything else up! Dirac had already flattened the response at 20hz (not shown here) but now there will be more headroom if needed. Here's a multipoint measurement comparing the 2ms rear sub delay effect at all 7 positions:

See less See more

The Bogg

Discussion Starter · #11 · Oct 2, 2019

That last graph has 14 (!) measurements on it and shows that the 5db boost at 20hz carries over to all positions without any side effects.

To summarize, the effects of having one sub with the wrong polarity are:
1. huge seat-to-seat and row-to-row variation in measured and audible response
2. premature response taper above 60hz (with no crossover on) due to cancellation effects
3. a bit more rapid taper below 20hz

Check your settings! Next I will redo Dirac and get the hump at the upper end of the bass range sorted out.

appelz

· #12 · Oct 4, 2019

The Bogg said:
In case it wasn't clear from the last graph, those were 7 different measurements at the seven different seats! I couldn't believe my eyes (and still wonder if I did something wrong, lol). This was the point in having multiple subs "ideally" placed in a room. The response isn't measurably flat yet but the consistency is noteworthy.

I have each sub hooked up to a discrete output on the Datasat so I can adjust the distances independently. I didn't know if this was worth fiddling with but since my 2 rear subs are a little closer to my main seat than the front pair I thought I'd experiment with delaying the rear subs (they have the same polarity as the front ones - some people advocate flipping the polarity in the rear pair but I didn't find it useful in my room). This graph compares the standard setting vs adding a 2ms delay to the 2 rear subs:

Math and Physics are great!

Also, they should have the same polarity.

Reactions: 1

Nyal Mellor

· #13 · Oct 4, 2019

appelz said:
Math and Physics are great!

Also, they should have the same polarity.

You can run rears out of phase, they need delay as well. Check out CABS / DBA techniques.

Sent from my iPhone using Tapatalk

appelz

· #14 · Oct 4, 2019 (Edited)

Nyal Mellor said:
You can run rears out of phase, they need delay as well. Check out CABS / DBA techniques.

Sent from my iPhone using Tapatalk

I was specifically referring to the Welti/Devantier method, but yes, there are other techniques that specify flipping polarity.

Let's make sure for the benefit of others that we don't confuse phase and polarity. Phase should link with a shift in time/delay while polarity is a reversal of the signal by means of DSP or swapping negative/positive terminals.

Bigus

· #15 · Oct 5, 2019

Worth noting that most common topologies of analog all pass filters used in audio equipment do not produce a phase shift linearly related to frequency (which is what a true time delay does), and therefore are only approximately the same as a time delay over a small bandwidth centered on the target frequency, typically the xo point.

However, it is possible to construct analog phase shifting circuits that are linear with respect to frequency over a wide bandwidth. Though not common in audio, these would be functionally the same as a time delay.

Polarity flips are a 180 degree phase shift irrespective of frequency and as noted are therefore different from time delays and common analog phase shifting all pass circuits. However, again, while not common in audio it is possible to construct an appropriately cascade of all pass filters and arrive at a constant phase shift irrespective of frequency. Set to 180, this is functionally equivalent to a polarity flip.

Reactions: 1

Mark Seaton

· #16 · Oct 5, 2019

Bigus said:
Worth noting that most common topologies of analog all pass filters used in audio equipment do not produce a phase shift linearly related to frequency (which is what a true time delay does), and therefore are only approximately the same as a time delay over a small bandwidth centered on the target frequency, typically the xo point.

However, it is possible to construct analog phase shifting circuits that are linear with respect to frequency over a wide bandwidth. Though not common in audio, these would be functionally the same as a time delay.

Polarity flips are a 180 degree phase shift irrespective of frequency and as noted are therefore different from time delays and common analog phase shifting all pass circuits. However, again, while not common in audio it is possible to construct an appropriately cascade of all pass filters and arrive at a constant phase shift irrespective of frequency. Set to 180, this is functionally equivalent to a polarity flip.

I'm not sure if it just wasn't stated clearly, or if there was confusion, but a constant delay creates a phase shift which directly decreases as frequency decreases. If a circuit actually provides a constant phase shift, that results in group delay that inversely increases as frequency decreases since the period of the wavelengths increase as frequency lowers, so the same fractional delay has to as well.

For system optimization I prefer delays, but both can be used. Phase adjustments were easier to implement in earlier subwoofers with a focus on the subwoofer-speaker crossover range. Using more than a symmetrical front pair of subwoofers back when these came into popularity was also a rarity. Once making adjustments to better blend multiple subwoofers rather than just the overlap range of a sub-speaker crossover, IMO delay is much more useful, as it is much less likely to result in big cancellation at lower frequencies in the course of better aligning the upper frequency range. A 4-6ms delay can be quite significant in the upper 1/2 of the subwoofer range, but is less than 1/8th of a wavelength (45 deg) at 20Hz, and 1/2 that at 10Hz. this still allows subs to combine constructively at the lowest frequencies, where a broad band phase shift to fix an issue above 40Hz can leave the lowest frequencies of 2 subwoofers fighting each other.

See less See more

Reactions: 1

Bigus

· #17 · Oct 5, 2019

We aren't saying anything different, and I tried to make it clear that although such circuits exist they are not common in audio use.

The Bogg

Discussion Starter · #18 · Oct 5, 2019

Ah, that's very interesting Mark...I didn't realize that the phase shift was less at lower frequencies. Purely academic for me, lol. My subs (JL F113s) have a polarity switch as well as continuously variable phase adjustment. I left the phase adjustments at zero. I recall when Adam was here he temporarily played with the phase adjustment on the rear pair, looked at his graph, and promptly put it back to zero. But then again at that time the front subs were placed asymmetrically in order to counteract a 50hz cancellation at the hotseat for 2 channel listening.

Something to consider...if you make use of the sub controls it will likely have an effect on the total delay in the sub channel and that needs to be rechecked at the final adjustment stages. For example, when I had my Anthem D2 I manually entered the physical distances of each speaker and sub and it wasn't until a while later when I discovered REW that I learned/realized that the subs with dsp have a different effective distance because of the dsp delay which needs to be taken into consideration otherwise crossover overlap will be off.

See less See more

Reactions: 1

The Bogg

Discussion Starter · #19 · Oct 5, 2019

I was beginning to doubt the validity of my multiposition measurements so I pulled out my old uncalibrated Emotiva mic with the generic calibration file I previously used and it does confirm the seat-to-seat measurements are very close. I did the measurements while my wife and kids were upstairs and you can see on those sweeps that the footfalls mess up the measurements when they occur so it confirms that the measurements are "real-time". Despite foam underpadding below the hardwood, a massive kitchen island above the theatre, loads of insulation between joists, double 5/8inch drywall with green glue on kinetics hangers (the cheap ones!), and more insulation beneath fabric my 5foot zero 107lb wife still sounds like a hippo stomping around, lol.

mmiles

· #20 · Oct 6, 2019

^^^

Might I make a suggestion?

Edit your last post and remove *hippo* in the event your 107lb wife ever figures out your AVS sign in password.

Reactions: 1

1 - 20 of 25 Posts

This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.