Dirac Live Overdriving Subwoofer

[LocutusEstBorg]

I have an 80 Hz crossover enabled in my Sound Blaster Zx with speakers set to "Small" so measurement is done with the crossover applied. As you can see there are massive nulls and resonances in my room. The LFE channel also has a massive dip of -16 dB at 50Hz. I have a Wharfedale SPC-12 and I can only listen to my system at something like -12.0 dB before the subwoofer severely distorts. The PC is connected via analog to my receiver so there is no DSP in the receiver.

 

Is it possible that the crossover is behaving unpredictably when receiving filtered audio from each channel which has massive cuts and boosts? Could the total boosts in each channel be clipping once they are crossed over?

 

As far as I understand, unless Dirac does its own crossover before filtering there will always be problems. The speakers and sub have to be measured without crossover. Then based on the findings, Dirac should apply its own crossover and mix the LFE channel before applying any filtering. Applying a digital crossover at a later stage will always result in some kind of clipping or overdrive.

 

Basically measurement should be done without crossover and filtering should be done after crossover. It can be done with advanced routing but that has problems of its own and defeats the simplicity of the Dirac playback device.

[blueixus]

Looks to me like you have some fairly typical room issues. i see a similar pattern with my own system although not quite as pronounced. I have used Dirac to correct these nulls & peaks, but i run it through the system with the crossovers intact.

[Kal Rubinson]

EQ should always come after bass management. It is meant to EQ the speaker/room and not the source channel.

[wgscott]

Turn up the gain more on the sub, remeasure and re-correct, then most of it will be attenuation.

[dallasjustice]

What is the slope of your crossover on your main R/L speakers? It looks like the speakers and the sub are fighting each and you are getting the huge null at 50hz due to an integration issue between sub and speakers? Try flipping the phase 180 degrees on the sub just to be sure and remeasure. You have to get the sub integration reasonably correct before you try to use DIRAC to correct. Of course, you will have problems apply your filter because DIRAC must come BEFORE the crossover as Kal said. If you use Jriver, try opening 2 instances of Jriver and then select Jriver ASIO as output in DAP. You can then set your crossover/delay in Jriver loopback zone.

 

If you don't have Jriver, give up. It's not gonna happen.

 

I have an 80 Hz crossover enabled in my Sound Blaster Zx with speakers set to "Small" so measurement is done with the crossover applied. As you can see there are massive nulls and resonances in my room. The LFE channel also has a massive dip of -16 dB at 50Hz. I have a Wharfedale SPC-12 and I can only listen to my system at something like -12.0 dB before the subwoofer severely distorts. The PC is connected via analog to my receiver so there is no DSP in the receiver.

 

Is it possible that the crossover is behaving unpredictably when receiving filtered audio from each channel which has massive cuts and boosts? Could the total boosts in each channel be clipping once they are crossed over?

 

As far as I understand, unless Dirac does its own crossover before filtering there will always be problems. The speakers and sub have to be measured without crossover. Then based on the findings, Dirac should apply its own crossover and mix the LFE channel before applying any filtering. Applying a digital crossover at a later stage will always result in some kind of clipping or overdrive.

 

Basically measurement should be done without crossover and filtering should be done after crossover. It can be done with advanced routing but that has problems of its own and defeats the simplicity of the Dirac playback device.

[LocutusEstBorg]

I have decided not to buy Dirac since their customer support is bad for a $850 product.

 

What is the slope of your crossover on your main R/L speakers? It looks like the speakers and the sub are fighting each and you are getting the huge null at 50hz due to an integration issue between sub and speakers? Try flipping the phase 180 degrees on the sub just to be sure and remeasure. You have to get the sub integration reasonably correct before you try to use DIRAC to correct. Of course, you will have problems apply your filter because DIRAC must come BEFORE the crossover as Kal said. If you use Jriver, try opening 2 instances of Jriver and then select Jriver ASIO as output in DAP. You can then set your crossover/delay in Jriver loopback zone.

 

If you don't have Jriver, give up. It's not gonna happen.

It's just a simple Crossover checkbox with a frequency slider that gets applied to all output at the final stage before output. You mean Dirac filtering should come after Crossover - which means ideally Dirac has to implement it.

 

EQ should always come after bass management. It is meant to EQ the speaker/room and not the source channel.

Yes. This seems impossible when using Dirac as the default playback device which should work globally for all apps and games.

 

Turn up the gain more on the sub, remeasure and re-correct, then most of it will be attenuation.

This makes the other channels too soft and I can hear the noise floor of my PC when I turn up the volume.

[dallasjustice]

No. I mean DIRAC comes before the crossover. DIRAC doesn't have a crossover. You need to do your own crossover somewhere else in the chain. By the looks of your magnitude response, you should spend a little time integrating the sub first. There's no easy button for subwoofer integration. It takes a little work.

 

I have decided not to buy Dirac since their customer support is bad for a $850 product.

 

 

It's just a simple Crossover checkbox with a frequency slider that gets applied to all output at the final stage before output. You mean Dirac filtering should come after Crossover - which means ideally Dirac has to implement it.

 

 

Yes. This seems impossible when using Dirac as the default playback device which should work globally for all apps and games.

 

 

This makes the other channels too soft and I can hear the noise floor of my PC when I turn up the volume.

[Kal Rubinson]

Yes. This seems impossible when using Dirac as the default playback device which should work globally for all apps and games.

You can implement bass management in JRMC and, probably, in other players prior to feeding the signals to Dirac.

[dallasjustice]

He's using DIRAC with 2CH. Bass management must come after DAP. Otherwise the bass channels will not be filtered.

 

There's another way to do it with a DLCT MCH 2.1 setup described by retro somewhere in this forum.

 

You can implement bass management in JRMC and, probably, in other players prior to feeding the signals to Dirac.

[Kal Rubinson]

He's using DIRAC with 2CH. Bass management must come after DAP. Otherwise the bass channels will not be filtered.

With stereo DAP only and the BM in place for the calibration, yes. I tend to think in terms of mch and the more general solution.:-)

[wgscott]

Turn up the gain more on the sub, remeasure and re-correct, then most of it will be attenuation.

 

Yeah, I wasn't thinking. This presupposes you have your sub hooked up at speaker-level like I do. Then Dirac just sees it as part of the main 2.0 system.

[LocutusEstBorg]

No. I mean DIRAC comes before the crossover. DIRAC doesn't have a crossover. You need to do your own crossover somewhere else in the chain. By the looks of your magnitude response, you should spend a little time integrating the sub first. There's no easy button for subwoofer integration. It takes a little work.

 

That's how I have it setup now. During measurement my sound card crossover affects the test tones. During playback the Dirac Processor comes first, then my sound card applies crossovers. The problem I see with this is the filtered audio is being modified again by my sound card crossover. What is the guarantee that the LFE channel won't clip once it's crossed over, given the massive boost Dirac will have to apply in my case?

[LocutusEstBorg]

You can implement bass management in JRMC and, probably, in other players prior to feeding the signals to Dirac.

I need it to work globally for all audio in games as well. There is no question of applying any additional processing before the Dirac Processor.

 

He's using DIRAC with 2CH. Bass management must come after DAP. Otherwise the bass channels will not be filtered.

 

There's another way to do it with a DLCT MCH 2.1 setup described by retro somewhere in this forum.

I am using 5.1 with a dedicated subwoofer output. The Dirac Processor playback device also shows up as a 5.1 device with dedicated LFE channel input. This goes to my sound card for crossover and to a 5.1 receiver for power amplification (sub is pre-amp).

[LocutusEstBorg]

The jRiver MC20 WDM driver means I can now perform bass management first and then pass the output channels to Dirac and to my power amp.

 

However won't this result in the crossover point being slightly imperfect?

[LocutusEstBorg]

Turn up the gain more on the sub, remeasure and re-correct, then most of it will be attenuation.

When I do this, the measured LFE level doesn't change. The crossedover frequencies are boosted relative to the higher frequencies and this shows up in the Front Left / Right graphs, but the LFE graph is at exactly the same height as before!

[wgscott]

....

 

Yeah, I wasn't thinking. This presupposes you have your sub hooked up at speaker-level like I do. Then Dirac just sees it as part of the main 2.0 system.

[flak]

Some additional info for users of the multichannel version of Dirac Live where two options are available...

 

In the standard 5.1 and 7.1 configurations our filter design will assume that the LFE channel needs an additional 10 dB boost from the receiver so the gain calibration will make the LFE +10dB louder than the other channels. If the receiver already has +10dB on the LFE, it will be kept that way.

In Custom mode, the gain calibration will make all channels equally loud. So using custom, if the receiver has +10dB on the LFE, this will be undone.

 

Ciao, Flavio