TOPIC:

Hi all,

After moving my listening position I did a new full setup with my SHD with Dirac and PEQ. Although Dirac took care of most of the problematic areas till up to 500hz, I have a massive dip between roughly 45 and 60hz. I verified this with measurements with REW. I decided to try to compensate this with rather aggressive PEQ settings:

Filter type: SUB_EQ
Frequency: 50hz
Gain: +10db
Q: 3

Although this made an audible improvement, I started to notice some clipping/distortion from time to time.

I have other eq bands in use with less gain, max. 4db, and filtering.  

Based on my limited understanding/expertise on this matter, I suspect that this PEQ settings cause clipping  in the SHD’s DAC. Or should I consider other causes?

My main question is, how can I objectively determine if the SHD’s DAC is clipping? And how can I avoid this clipping without sacrificing my PEQ settings?

As far as I understand, the gain sliders on SHD’s outputs tab in the plugin come after the DAC in the signal chain and therefore wouldn’t effect any clipping in the DAC. Or is this a wrong assumption?

Hoping to learn from your insights.

Here is a measurement after applying this aggressive PEQ settings: [attachment=4481]

I can think of two causes of dips that Dirac doesn't compensate for.

1. Dips larger than 10dB.
Dirac can only provide 10dB boost so any dips  will only be reduced by 10dB, if they are any larger a dip will remain.
Some people solve this by setting the Dirac reference level along the bottom of the response dips, rather than the average. Dirac then only has to use cuts, which are not limited, and not any boosts. This is better than using PEQ. The reasons Dirac has the 10dB limit is to try to avoid clipping, with PEQ you are just using a crude bypass of its limit and running into the exact problem it tries to avoid.

2. Dips due to Speaker Boundary Inteference Responses (SBIR). Because these are due to an almost perfect reflection from walls (and floors and ceilings) causing almost prefect cancellation no amount of boost can solve these. You put 10dB more out and 10dB more reflects back and still perfectly cancels out. The same for any amount of boost. Neither Dirac nor PEQ can solve this problem, absorber room treatment works well (I did this) but costs rather a lot, speaker positioning may be able to help a little.

As for avoiding clipping, turning the volume control down should work. If your amp has a volume control turn that up a bit more.

Thank you for your reply. Indeed Dirac corrected this dip up to 10db already.

You mention changing the Dirac reference level to the dip. I couldn’t find anything about this is the Dirac documentation. Could you help me understand how to do this? For example, if I use the default target curve, how can I lower it as a whole? Or do I need to adjust all data points manually one by one?

I confess its so long since I've used Dirac I can't actually remember how to move the target.

But I rather don’t mess with anything above 600hz, I want to leave my speakers untouched. When lowering the target curve I have no choice other than correcting the full bandwidth. 

I'm still on Dirac 2 but it has sliding lines at each end of the frequency range that limit the frequency range Dirac operates on.

entripy wrote:

I'm still on Dirac 2 but it has sliding lines at each end of the frequency range that limit the frequency range Dirac operates on.

Yes, the same applies to Dirac 3. But if I lower the target curve with 6db, I can’t leave any bandwidth untouched. 

Dirac ignores the target outside the frequency range selected by the lines. For example if you reduce the high frequency limit to 600Hz, Dirac provides no compensation above 600Hz so the target above 600Hz is irrelevant.

matthijsdebooij wrote:

Hi all,

After moving my listening position I did a new full setup with my SHD with Dirac and PEQ. Although Dirac took care of most of the problematic areas till up to 500hz, I have a massive dip between roughly 45 and 60hz. I verified this with measurements with REW. I decided to try to compensate this with rather aggressive PEQ settings:

Filter type: SUB_EQ
Frequency: 50hz
Gain: +10db
Q: 3

Although this made an audible improvement, I started to notice some clipping/distortion from time to time.

I have other eq bands in use with less gain, max. 4db, and filtering.  

Based on my limited understanding/expertise on this matter, I suspect that this PEQ settings cause clipping  in the SHD’s DAC. Or should I consider other causes?

My main question is, how can I objectively determine if the SHD’s DAC is clipping? And how can I avoid this clipping without sacrificing my PEQ settings?

As far as I understand, the gain sliders on SHD’s outputs tab in the plugin come after the DAC in the signal chain and therefore wouldn’t effect any clipping in the DAC. Or is this a wrong assumption?
 

There are two things I recommend to determine if you are seeing clipping in the SHD.

1) Look at the output meter levels, as long as these are less than 0 dB you are not clipping the SHD.
2) I like to run a frequency sweep measurement in REW with the SHD as an input / output device to see how much net boost I have applied. Once you understand net boost as long as you offset that boost with the SHD volume control (i.e. if you have +12 dB boost you will not clip as long as the long level is -12 dB or lower). It doesn't necessarily mean you will clip at higher volume levels as it depends on how the track was mastered but it gives a boundary to understand when it is possible.

Michael

entripy wrote:

Dirac ignores the target outside the frequency range selected by the lines. For example if you reduce the high frequency limit to 600Hz, Dirac provides no compensation above 600Hz so the target above 600Hz is irrelevant.
 

Then this would happen:

 

1) Look at the output meter levels, as long as these are less than 0 dB you are not clipping the SHD.

The output meters in the “outputs” tab stay way below 0 since normal listening levels for me mean the SHD volume is set to -44 db. So the ouput meters stay below -44db. So I doubt you mean these, right?

matthijsdebooij wrote:

1) Look at the output meter levels, as long as these are less than 0 dB you are not clipping the SHD.

The output meters in the “outputs” tab stay way below 0 since normal listening levels for me mean the SHD volume is set to -44 db. So the ouput meters stay below -44db. So I doubt you mean these, right?

I 100% mean those, if they are not near 0 dB you are not clipping in the SHD. Are you saying you were hearing distortion at a -44 dB volume control setting?

Michael

100% mean those, if they are not near 0 dB you are not clipping in the SHD. Are you saying you were hearing distortion at a -44 dB volume control setting?

Michael

The SHD is directly connected to a power amp (700 watts at 4 ohm). As I mentioned in my initial post, I’m not worried about clipping the power amp, but clipping the SHD’s DAC. I believe (=assume) these output meters come after the DAC in the signal chain. And therefore don’t provide any meaningful information about internal clipping within the SHD. But I might be wrong. 

matthijsdebooij wrote:

100% mean those, if they are not near 0 dB you are not clipping in the SHD. Are you saying you were hearing distortion at a -44 dB volume control setting?

Michael

The SHD is directly connected to a power amp (700 watts at 4 ohm). As I mentioned in my initial post, I’m not worried about clipping the power amp, but clipping the SHD’s DAC. I believe (=assume) these output meters come after the DAC in the signal chain. And therefore don’t provide any meaningful information about internal clipping within the SHD. But I might be wrong. 
 

Those output meters come right BEFORE the DAC. If they are below 0 dB you are not clipping the DAC.

Michael

My main thought reading this is that I wouldn't want to be using a 10 dB boost in addition to Dirac Live. In fact I wouldn't use any manually entered PEQ in addition to Dirac.

If you want to get an idea of what your total DSP is doing you can use REW to make one measueremnt with Dirac and your PEQ enabled, followed by anothter with both disabled without moving the microphone. Then in REW divide the first measurement by the second and look at the result. Personally I'd be reluctant to use a 10 dB boost unless I knew I had a lot of headroom, and you may have up to 20 dB with what you've done I think.

Note that large boosts can significantly increase distortion produced by the speakers themselves, rather than digital clipping being the only concern.