TOPIC:

Some questions.

1. When measuring with DIRAC do PEQ filters loaded modify the sweep signal. In other words are the PEQ filters upstream or downstream from the DIRAC audio test signal generation.

2. What is the maximum correction possible for DIRAC 3 - I have seen +/- 10dB and +/- 12.5 dB, but presumably it is only + because surely cutting is less limited - safer and easier to implement.

Reason being is I am trying to correct a problematic room, and using REW generated PEQ filters to make major corrections so that DIRAC only has to do minor ones, but if the PEQ is upstream from DIRAC this only screws up the DIRAC result.

Thanks.

PEQ's, delays, XOVER all influence the dirac sweep.
So make sure they are set correctly before the dirac measurement and use exaclty these settings when using a calibration filter generated with this measurement. (It seems you know this already )

As far as I know all signals paths are floating point, so this would mean there is no (Internal) clipping by combined boosts set by dirac and PEQs. I did some testing with 0dbFs input signals and could not get any clipping when setting max boost on places where dirac was also boosting. Still the output is limited to 0dBFS so make sure that max volume is not going past (0dBfs - max boost)

This is from the manual

So does DIRAC internally provide headroom - meaning can one run the DIRAC volume at 0 dB and not experience clipping?

I adjust volume using my downstream DAC which has a hybrid gain control, but I am uncertain how much headroom to allow for, particularly when I use PEQ filters which have plus 5 or 10 dB gain.

Presumably I need to adjust the SHD DIRAC digital volume down to minus 10 dB for example.

If you boost valleys in the measured curve 10dB in my experience this can cause clipping. It's better to lower the target curve and not boost more than 3dB. Alternatively you can make the target curve follow the deepest valleys a bit. I do a little of both.

Thanks jaapina.

What I was asking was where does DIRAC apply it's headroom?

How much headroom does DIRAC apply?

And if one PEQ filters AND also DIRAC on the SHD, for a nominal gain (let's say 3 dB) where should the SHD digital volume be set??

Where are the internal software / hardware headroom allowances, and how much is allowed for?

In my testing (but not accepted by everybody) Dirac needs 10dB headroom which the SHD kindly supplies. 0dB SHD gain can often be applied without clipping because it depends on what the input signal level is. You have to remember that the SHD has a gain control, not a level control.

The reason I think it is 10dB is because Dirac 1 used to allow the user to change it but it defaulted to 10dB. With Dirac 2 the user could no longer change it but the overall SHD gain seemed to be the same to me which suggested to me that it is still 10dB. It can also be seem that Dirac can boost deep valleys no more than 10dB, presumably because it knows it only has that much headroom safely.

So assuming Dirac/SHD cooperate to give headroom all you can say is with xdB PEQ (or any other) gain, the SHD gain control may need to be set xdB lower than it was was before.

Because the SHD uses floating point DSP there isn't really any clipping until it reaches the output (either a DAC or a digital output). Since any gain structure works best with attenuation as late as possible in the chain the main SHD gain control is the best place for user level adjustment. The SHD plugin also allows you to apply separate input/output gains. Typically its advisable to set the output gains such that you can run the SHD gain at or near 0dB without your speakers exploding. This is the safest approach which avoids issues if some Volumio plugin suddenly requests 0dB SHD gain, which has been known to annoy people who haven't set things up safely like that.

The final point is that the SHD has a generous dynamic range which means it quite often works really well with non-optimal gain settings

entripy wrote:

You have to remember that the SHD has a gain control, not a level control.

What do you mean by this?

Thanks
Wayne

I mean that the control on the SHD controls how much the signal is multiplied by. You can get the same level out with 0dB setting and a quiet input as with -30dB setting and a loud input. As such there's almost no point wondering what setting avoids clipping as almost any setting can cause clipping with input signals of certain levels. Thinking of it as a level control tends to cause confusion because of this disconnection with clipping and output levels. Both gain and level can be measured in dB which doesn't help.

I think the confusion can be understood since if people only use digital inputs with music from a certain era the input signal level will tend to be a constant maximum level and the gain control will then approximate a level control. For people who use analogue inputs or multiple digital inputs with different gains the approximation rapidly starts to disappear.

entripy wrote:

I mean that the control on the SHD controls how much the signal is multiplied by. You can get the same level out with 0dB setting and a quiet input as with -30dB setting and a loud input. As such there's almost no point wondering what setting avoids clipping as almost any setting can cause clipping with input signals of certain levels. Thinking of it as a level control tends to cause confusion because of this disconnection with clipping and output levels. Both gain and level can be measured in dB which doesn't help.

I think the confusion can be understood since if people only use digital inputs with music from a certain era the input signal level will tend to be a constant maximum level and the gain control will then approximate a level control. For people who use analogue inputs or multiple digital inputs with different gains the approximation rapidly starts to disappear.

Thank you for your clarification . This is how all “level” controls work in consumer electronics. Same as gain.