What is Dirac/DLBC doing to my sound ?

[pollock0424]

Hello,
I have been tweaking my new AVR (Integra DRX 8.4) setup and it has Dirac and DLBC options also. Firstly, I optimized my placement the best I could without any room correction and then I used room correction to see if gets any better. I would like to understand why Dirac is behaving the way it is. I am using a subwoofer and I am sitting 7.9ft from the tweeters and the speakers are 7.7ft apart from each other. I took REW measurements with and without Dirac and I ensured that the mic location didn't change at all. I made sure that physically everything was same for the two cases.

Here's my speaker arrangement:

As you can see, Dirac introduced a major dip at 60hz. I am neither a musician nor a trained listener, so take my subjective evaluations with a pinch of salt! I listened to my favourite tracks "A/B ing" with and without Dirac and I could tell that the "weight" in the low end was gone! Although, midrange and stringed instruments sounded more natural with Dirac ON. With Dirac OFF, the midrange was a tiny bit shouty and stringed instruments sounded just a bit overblown! Let's look at these plots with "psy" smoothing, at the end of the day what I am hearing matters the most to me.

Now, superimposing the L+R plot with and without Dirac. It is clear from this plot that loss of "weight" in the bass and clarity of mid-range and stringed instruments ?

I wanted to see what is happening so dug into RT60, GD etc. You can clearly see that RT60 has gone up for the bass/sub bass and GD also is looking worse now with Dirac. Earlier, I tweaked the placement and phase knob on the subwoofer to get the best possible excess GD. The subwoofer was crossed at 55hz and was placed next to the left speaker and I also used the provided one band PEQ on the subwoofer to tame the nasty mode at 55hz, which you can see in the Room sim also. Without subwoofer, this mode almost made it unbearable... Also, I made sure that the crossover was set to 55hz in DLBC also. Integra has a mobile app wherein one can adjust the crossover, boost/cuts etc.

Now, looking at the RT60 plots! Again, Dirac has somehow managed to mess up the bass region.

Here are IR plots, with Dirac ON, IR seems to go significantly into the negative direction... what's going on here ? I can see a strong reflection at 5ms.

I like how the system sounds without Dirac, but I also like Dirac on some tracks. With much pain and suffering I have achieved what I have, and I would like your help in understanding what is going on ? Also, here is a link to the REW project, if you're interested.

 

[voodooless]

Subwoofer was crossed at 55hz and was placed next to the left speaker and I also used the provided one band PEQ on the subwoofer to tame the nasty mode at 55hz, which you can see in the Room sim also. Without subwoofer, this mode almost made it unbearable... Also, I made sure that the crossover was set to 55hz in DLBC also

That is not what the manual says that you should do. You’re supposed to set the subwoofers crossover as high as possible, and leave it all to the AVR to figure out. Since the crossover is relatively close to your new dip, this may be part of the problem.

 

[pollock0424]

@voodooless : I have tried that and the crossover was automatically set to 70hz. Because I have only one subwoofer, I was clearly able to tell when the subwoofer was playing. Integration was not well done. So, here's something that I will try. I will set the crossover point to the maximum on the subwoofer and vary the crossover on the Integra AVR phone app. I don't think it is a good idea for the crossover to be higher if there is only one subwoofer.

 

[voodooless]

I don't think it is a good idea for the crossover to be higher if there is only one subwoofer.

Generally, even with one sub, 70 Hz should not be locatable. Maybe it was due to the same setup issue?

 

[pollock0424]

Also, the PEQ cut was turned on when I ran Dirac. Do you think that could've played a role in the dip?

I tried using LFE mode on the subwoofer which basically means that the AVR does everything such as xover, gain/cut in the bass region. All the controls on the subwoofer are defeated in LFE mode. It almost sounded like the sub was playing well into 150hz and the bass was awful ..

 

[voodooless]

It almost sounded like the sub was playing well into 150hz and the bass was awful ..

That’s strange.. something may not be correctly setup somewhere. Maybe someone with experience with Integra/Onkyo AVRs has a better grasp on this.

 

[Flak]

As you can see, Dirac introduced a major dip at 60hz. I am neither a musician nor a trained listener, so take my subjective evaluations with a pinch of salt! I listened to my favourite tracks "A/B ing" with and without Dirac and I could tell that the "weight" in the low end was gone!

With reference to the dip in the crossover region, this post might be relevant (click "See more"):

https://www.avsforum.com/threads/dirac-live-bass-control-and-crossover-frequencies.3177010/

The "weight" in the low end is determined by the target curve (I'd suggest using the same for all speaker groups, including the subwoofer) so it should be possible to get as much weight as desired.
b.t.w. Not in this case, but when multiple subwoofers are used one should take the attached document into account.

In any case, Happy New Year!!

 

[pollock0424]

@Flak
Thanks for your input. I will refer to your link. However, here's the house curve that I applied for the 2.1 channels. You can see the crossover bar set at 55hz in DLBC software and the curtains also.

What's unusual is that I already gave a decent sounding system to Dirac. I did most of the work and yet dirac messed up. So, the opinion that do the best and then use room correction doesn't hold !

I will try experimenting again. If nothing works out then I will reach out to Dirac.

 

[ban25]

I don't have experience with the mobile app, but I'm suspicious as to whether or not DLBC is actually enabled, if it's letting you set the crossover in there. When you open Integra Control Pro on your phone and look at the different Dirac slots (OFF, 1, 2, 3) do you see "Bass Control" under those slots like this:

It'd be helpful to know what speakers/subs you are using. 55 Hz crossover sounds a bit low. I have mine at 70 Hz and I have also had good success with 90 Hz. If possible, try downloading the Dirac Live software on your PC/Mac from dirac.com and run the calibration from there. If you do that, please post pics of your Dirac calibration so we can see what it's trying to do.

 

[pollock0424]

@ban25
Yes, I have the bass control enabled and it exactly looks like your screenshot. I also have a PC version as I also have a miniDSP SHD with Dirac full range license that I used for over a year. The issue with PC based Dirac is the delay that I takes exporting filters... is it still 10 minutes? I tried a month ago and gave up on the PC version because of that..

I am using Thiel CS3.6 speakers, -3db@24hz anechoic. My subwoofer is a Rythmik E15HP. Without Dirac, I am almost there.. just a couple of PEQ filters and I should be done but I want to use Dirac/DLBC and make it work because I paid quite a bit of money.

 

[ban25]

@ban25
Yes, I have the bass control enabled and it exactly looks like your screenshot. I also have a PC version as I also have a miniDSP SHD with Dirac full range license that I used for over a year. The issue with PC based Dirac is the delay that I takes exporting filters... is it still 10 minutes? I tried a month ago and gave up on the PC version because of that..

I use a Mac and the filter export takes maybe 2 minutes, no big deal. This is with version 3.8.2.

 

[pollock0424]

Okay, folks! I tried two more times and this time Dirac did a decent to good job. I reset the subwoofer phase, crossover to defaults and I also defeated the PEQ on the sub and let Dirac do its thing. The second time, soundstage was skewed to the left. In my experience with Dirac, this soundstage skewing happens time to time, and it probably is related to human error or some reflection. I tried again for the third time and everything went well, I used the "tightly focused imaging" option though. My viewing/listening position has only 3 seats, and mostly it's just me using it. Here are some plots and charts. I made my conclusions..

As you can see, Dirac cleaned out the excess SPL from 1.5khz to and 150hz to 200hz. Vocals and strings seemed natural now. DLBC set the xover to 70hz and you can see that it introduced a peak around 70hz, I will tweak this one out just by playing with the curve around there. However, I lost more SPL in the 100hz region and the dip is wider too! Now, coming to the excess GD plots.

The above plot was obtained by tweaking with the phase knob to get the best possible excess GD, I didn't tweak the delay settings in the AVR manually. Now, let's see how
did DLBC do when phase/PEQ konbs were defeated on the subwoofer.

It is clear to me during my listening, bass is cleaner with DLBC. Without DLBC excess delay in sub bass was around 20ms, and it decreased to 10ms with DLBC. All this came at a cost of losing SPL around 100hz. I will keep trying to how far I can take this optimization.

 

[ban25]

Okay, folks! I tried two more times and this time Dirac did a decent to good job. I reset the subwoofer phase, crossover to defaults and I also defeated the PEQ on the sub and let Dirac do its thing. The second time, soundstage was skewed to the left. In my experience with Dirac, this soundstage skewing happens time to time, and it probably is related to human error or some reflection. I tried again for the third time and everything went well, I used the "tightly focused imaging" option though. My viewing/listening position has only 3 seats, and mostly it's just me using it. Here are some plots and charts. I made my conclusions..

As you can see, Dirac cleaned out the excess SPL from 1.5khz to and 150hz to 200hz. Vocals and strings seemed natural now. DLBC set the xover to 70hz and you can see that it introduced a peak around 70hz, I will tweak this one out just by playing with the curve around there. However, I lost more SPL in the 100hz region and the dip is wider too! Now, coming to the excess GD plots.


The above plot was obtained by tweaking with the phase knob to get the best possible excess GD, I didn't tweak the delay settings in the AVR manually. Now, let's see how
did DLBC do when phase/PEQ konbs were defeated on the subwoofer.


It is clear to me during my listening, bass is cleaner with DLBC. Without DLBC excess delay in sub bass was around 20ms, and it decreased to 10ms with DLBC. All this came at a cost of losing SPL around 100hz. I will keep trying to how far I can take this optimization.

Ideally, you only want to use tight imaging for a situation where a single listener's head will not move, i.e. desktop/near-field use or single seat lounge chair.

Some housekeeping items:

1. Are you using a boom microphone tripod to position your measurement microphone? Something like this:

Amazon.com

2. If you haven't done so already, it's worth following these tips on how to maximize your microphone gain during calibration:

Configure microphone and setting speaker levels for measurements in Dirac Live v2 & v3

3. Maybe try a DLBC crossover of 90 Hz. The good thing about Dirac/DLBC is once you have your measurements dialed in, you can play around with curves and crossovers and upload to different filter slots for A/B comparisons to your heart's content!

 

[pollock0424]

@ban25
Yes sir! I am using a tripod with boom for mic. I use it in my day job

This time, I ensured that all the levels were approximately the same. Fortunately, I had to tweak only a couple of channels to achieve this equal levels. Yes, I was thinking about 90hz crossover, I will give it a try. I will be getting another sub soon.

This time, I used my PC to do thr calibration. I takes about 5 minutes for export.

 

[OCA]

I lost more SPL in the 100hz region and the dip is wider too!

That dip (112.76Hz for the left speaker in your mdat) is about 5ft (4.9898294ft) extra distance travelled from your woofer driver to the LP. I don't know your exact placement but seems like front wall reflection if your speakers are about 2.5ft from the front wall. If you can move them 0.3ft further away from that wall and crossover the sub at 100Hz, you can get rid of it.

 

[pollock0424]

That dip (112.76Hz for the left speaker in your mdat) is about 5ft (4.9898294ft) extra distance travelled from your woofer driver to the LP. I don't know your exact placement but seems like front wall reflection if your speakers are about 2.5ft from the front wall. If you can move them 0.3ft further away from that wall and crossover the sub at 100Hz, you can get rid of it.

I'm trying to understand that dip too! I have a picture of my setup attached in the first post of this thread. Woofer is 3.4 ft into the room from the front wall. I will do some REW sweeps today playing with moving things around.

 

[OCA]

I'm trying to understand that dip too! I have a picture of my setup attached in the first post of this thread. Woofer is 3.4 ft into the room from the front wall. I will do some REW sweeps today playing with moving things around.

More parameters are required to calculate the first reflection when it's not from the front wall. I checked your dimensions in my XL file from what I could extract from your Room sim data. I don't have the exact numbers for the woofer height and LP to woofer distance (it's a bit longer than LP to tweeter and it matters) but looks like it's near side wall reflection or ceiling reflection or both (most probable given the large dip area):

First reflection delays should not be similar for that very reason. Focals come with a formula to minimize this in their manuals. It might help you:

Here's the XL file for you to play with (sorry it's in meters). Remember the 1st peak frequency is the actual reflection wave and the distance should be adjusted in the formula (C26 to C31) according to the driver which produces that frequency. In your case (crossover @ 500Hz) only floor reflection is generated by the midrange, the rest are all sourced at the woofer.

First Wall Reflections.xlsx

Actual Fixed dimensions,Room Length,3.5052 Room Width,4.7244 Room Height,2.5908 Woofer from floor,0.7874 Midrange from floor,0.8200 LP from floor,0.9906 Variables,Woofer from front wall,1.0160 Woofer from side wall,1.0414 LP from woofer,2.3700 LP from midrange,2.3622 Calculated dimensions,LP fr...

docs.google.com

 

[pollock0424]

@OCA
that is good info, I think my speakers are in a bad spot for the 100hz dip. I just inspected my earlier measurements when the speakers were pushed closer to the front wall, dips around 100hz didn't seem that bad.
Woofer to floor: 1.26ft
Woofer to front wall: 3.5ft
Woofer to side wall: 4.3ft
Woofer to ceiling:7.785ft
Woofer to LP: 7.6ft
12 Inch woofer plays up to 500hz.

midrange from floor:2.85ft
midrange from side wall:4.324
midrange from front wall:3.379
midrange to LP: 7.75ft
Speakers are Thiel CS3.6 and they have a sloping baffle.

Double dip at 100hz most likely because of the similar distances from front wall and side wall. Earlier, I used to have the speakers pushed further back closer to the front wall, I just checked those plots and the dip at 100hz is not that bad.. I was still seeing 75db in the lowest point inside the dip.

I entered the appropriate measurements after I have re-measured. Here's what the first reflections look like

 

[OCA]

Rear wall reflection now also comes from the mid driver and since your speakers are tilted, you need that "mid from rear wall" dimension. We also don't have a "LP from woofer" number.

Here's an SI metric version of the Excel, I think you should double check all the fixed dimensions above.

First Wall Reflections_feet.xlsx

Actual Fixed,Room Length,11.5000 Room Width,15.5000 Room Height,9.0450 Woofer from floor,1.2600 Midrange from floor,2.8500 LP from floor,3.2500 Variable,Woofer from front wall,3.5000 Woofer from side wall,4.3000 LP from woofer,7.9000 LP from midrange,7.9000 Calculated,Midrange from rear wall,8....

docs.google.com

A wall reflection causes a 180 deg phase shift at the first dip and floor reflection for instance can usually be easily spotted in the phase graph as it's the very first wall reflection. This is from my left speaker:

Unfortunately, following reflections all mix up with each other and get chaotic very quickly and floor reflection is the only one we don't need to bother. It's been there with every sound we hear in nature and is efficiently ignored in our brains. Still, it might help you accurately obtain some of the dimensions in your room. Acoustic centers of drivers can be tricky to spot.

 

[pollock0424]

@OCA
How do you simulate this speaker boundary reflections ? I mean the equations ? And here's some encouraging news, I was able to lift the dip by moving the speakers closer to the front wall. See the below chart:

I think that currently, ceiling and front wall reflections are dominating for the left speaker. In the measurements below, there is a sharp dip at 115Hz and 350hz. Not sure why this is NOT happening with the right speaker with the same magnitude, although they're symmetrical when considering ceiling and front wall. My subwoofer is a 20"x20"x20" and is right next to the left speaker as shown in the room sim. Nothing is to the right of the Right speaker as I have a door with 6 inch foam with air gaps stuck on it (mostly for thermal insulation and if it helps sound then that's a win!). Perhaps the subwoofer is contributing to this Speaker Boundary cancellations ?

Looking at Filtered IR@80Hz, no problem zone.

Now, let's look at Filtered IR@120Hz, problem zone! This FIR dies down very quickly (probably because the wavelength is smaller ?)