TOPIC:

Team,
I thought I read somewhere that you can limit the increase Dirac applied to dips. I.e. adjust a setting to limit any increase to 4dB let’s say. Have I imagined this? My DDRC24 and Dirac software doesn’t seem to provide any such “advanced” options. Am I missing something very obvious?
Thanks

asx77 wrote:

Team,
I thought I read somewhere that you can limit the increase Dirac applied to dips. I.e. adjust a setting to limit any increase to 4dB let’s say. Have I imagined this? My DDRC24 and Dirac software doesn’t seem to provide any such “advanced” options. Am I missing something very obvious?
Thanks

I think this was possible on Dirac v1 but isn't on v2 or v3.

You can stop dips being boosted by making the target curve follow them.

Seems a shame they removed that option. Putting your own dip on the target curve sounds like the solution but it introduces some guesswork in terms of how wide or narrow to make it. Still I’ll probably not add my own dips. Just wanted confirmation I wasn’t missing something obvious! So thanks.

You may know but I believe Dirac itself limits the max boost to 10 dB.

Not only does Dirac limit the max boost. From analyzing the sweeps, it knows which parts of the response are non-minimum phase and therefore won’t respond to normal PEQ boosts (most dips in the bass). Together with advanced FIR filtering, it can alleviate some dips, but never much. So leave the target curve be.

So the reason it’s not possible to adjust in v2 or v3 is because it doesn’t need to be.

9radua1 wrote:

Not only does Dirac limit the max boost. From analyzing the sweeps, it knows which parts of the response are non-minimum phase and therefore won’t respond to normal PEQ boosts (most dips in the bass). Together with advanced FIR filtering, it can alleviate some dips, but never much. So leave the target curve be.

So the reason it’s not possible to adjust in v2 or v3 is because it doesn’t need to be.

You've misunderstood the problem I'm afraid. Boosts are still applied in v2 and v3, and actually even just phase changes can result in the peak signal levels increasing. If this is done on a signal at or very close to the maximum (0 dB) digital level then this can cause digital clipping and hence distortion. The way to prevent this is to make sure a minimum level of digital attenuation is always applied, with 10-15 dB being appropriate.

Sure, I dont disagree with any of that. Just that the compensate attenuation will automatically be applied by Dirac to stop clipping, so why would you follow a dip with the target curve?

9radua1 wrote:

Sure, I dont disagree with any of that. Just that the compensate attenuation will automatically be applied by Dirac to stop clipping, so why would you follow a dip with the target curve?

Unfortunately Dirac doesn't automatically compensate to prevent clipping though.

Edit: I mentioned modifying the target curve because it was asked how the boost applied could be limited and I believe this is the only way to do so.

Ok. I thought the volume calibration was there for this exact reason. To cap the master output at -12 or -32 dB to allow ample headroom for the filters.

If one wants to limit the boost in a specific region, changing the target curve may be the only way. But that presumes Dirac will in fact try to boost large non-minimum phase dips to reach a target curve that’s beyond the calibrated headroom. I just don’t know that that’s true.

9radua1 wrote:

Ok. I thought the volume calibration was there for this exact reason. To cap the master output at -12 or -32 dB to allow ample headroom for the filters.

No, the volume calibration is about getting a good signal-to-noise ratio for the measurements without being so loud as to potentially damage the speakers. It has no impact on the EQ applied.

9radua1 wrote:

If one wants to limit the boost in a specific region, changing the target curve may be the only way. But that presumes Dirac will in fact try to boost large non-minimum phase dips to reach a target curve that’s beyond the calibrated headroom. I just don’t know that that’s true.

Thinking about minimum phase is a distraction here. The fact is that Dirac applies boosts, and even if it didn't then phase changes alone could still cause clipping.

If you want to see what boosts are being applied by your own filters you can do so by making two measurements with REW: one with Dirac on and one with Dirac off, without moving the microphone in between. If you then divide the Dirac on measurement by the Dirac off measurement you will see what the applied filter looks like.

This definitely is a real issue and we do occasionally see people posting here with distortion problems because they haven't applied attenuation to prevent digital clipping.

Ultrasonic wrote:

It has no impact on the EQ applied.

So how does the system secure headroom for processing? In my understanding, gain staging is about SNR, as you say, but also about the input-output ratio of the component. If processing on a channel is causing clipping, reduce the channel volume or the plug-in volume just like on any DAW. In the old v1 you had a volume control on the plug-in to attenuate the output if the filters were overloaded (I don’t know if it’s still there in the new plug-in version).

How does Dirac secure headroom if not by volume calibration?

9radua1 wrote:

How does Dirac secure headroom if not by volume calibration?

It doesn't, hence the advice in this thread. I'm not saying this is smart but it's the truth.

As you don't seem to be believing me, here are some other threads on this topic, starting with one with advice along similar lines from miniDSP's devteam:

www.minidsp.com/forum/dirac-series-suppo...-preventing-clipping

www.minidsp.com/forum/shd-series/16778-s...stortion-2-ch-purity

www.minidsp.com/forum/dirac-series-suppo...nd-internal-headroom

The whole “at what level should I set my master volume to avoid clipping?” Can get very in-depth.

After reading lots of threads including those just shared by Ultrasonic I decided to stop worrying so much!!

Without my downstream integrated amp on I did a 0db (-3db) sweep with REW. I made it nice and slow and did sweeps in small chunks e.g. 20hz-200hz, 200hz-400hz... While doing the sweep I watched the output dB values on the DDRC24.

The max I got was +6dB so I set my Master volume at -8dB.

Although playing with the ddrc is fun, I’m hoping this is good enough so I can get on and enjoy listening to music rather than listening to sine sweeps !!

asx77 wrote:

The whole “at what level should I set my master volume to avoid clipping?” Can get very in-depth.

After reading lots of threads including those just shared by Ultrasonic I decided to stop worrying so much!!

Without my downstream integrated amp on I did a 0db (-3db) sweep with REW. I made it nice and slow and did sweeps in small chunks e.g. 20hz-200hz, 200hz-400hz... While doing the sweep I watched the output dB values on the DDRC24.

The max I got was +6dB so I set my Master volume at -8dB.

Although playing with the ddrc is fun, I’m hoping this is good enough so I can get on and enjoy listening to music rather than listening to sine sweeps !!

I still think my method of taking one measurement with Dirac on and dividing it by one with Dirac off is quicker, simpler and gives results that are at least as useful (and I'd argue better).

Hi, when you say "divide" do you mean just subtract one from the other?

Without dirac shouldn't the reading you get just be a flat 0db?