Prevent DSP caused digital clipping best practices

[czt]

I never hesitate to put advanced limiter after convolver (with a max response: 0 dB rePhase filter) in fb2k.

 

[mitchco]

@levimax have a look at: https://hydrogenaud.io/index.php/topic,43195.msg378879.html#msg378879

HLC uses a similar approach, but there are gain sliders for each filterbank and a master trim adjust for the overall gain. HLC also has peak meters so one can easily identify if there is any clipping.

 

[levimax]

I never hesitate to put advanced limiter after convolver (with a max response: 0 dB rePhase filter) in fb2k.

@levimax have a look at: https://hydrogenaud.io/index.php/topic,43195.msg378879.html#msg378879

HLC uses a similar approach, but there are gain sliders for each filterbank and a master trim adjust for the overall gain. HLC also has peak meters so one can easily identify if there is any clipping.

Thank you for the information. Since my DAC stopped working I got new one (SMSL D100) and thought maybe it would react differently to "digital clipping" but it doesn't. I spent some time playing around with this and here are my observations and questions.

I have a DIY 4 ways system (3-way plus 2 subs) with analog active crossovers all LR4. I did the best I could setting levels between the drivers with the crossovers and then created some FIR filters using REW and Rephase for final speaker / room EQ. The filters are a little aggressive (I allowed 9 dB max boost) as I have a lot of power with 8 channels of amplification and I am happy with how they sound.

I knew I would have to allow for the filters and make some compensation to prevent digital clipping so I created a 0 dB sweep in REW and played it with Foobar 2000. As expected without backing off on the level adjustment in the convolver plug in digital clipping was clear both looking at the "meters" in Foobar and listening. If you never tried it digital clipping on a sweep is VERY audible (at least in the lower FR range where most of my correction are)... you don't need golden ears... when digital clipping happens it gets louder and the tone spikes up higher. Here is what surprised me. If I used the "automatic level adjust" on the plug in I could still hear digital clipping. I then tried the "Foobar Limiter" and while it completely eliminated clipping according to the "meter" it did NOTHING to actually stop the audible clipping. Next I ran the sweep and manually adjusted the level in the convolver plug in 1 dB at a time until the digital clipping stopped. This happened at -8 dB which is a lot. At this point the meters stayed well below clipping and the sweep sounded smooth from 20 Hz on up.

Here is my question. Is listening to a 0 dB sweep a legitimate way to test for digital clipping? If it is then I would say the "automatic" clipping eliminators and clipping"meters" are inadequate and misleading. If a 0 dB sweep is "too hard" of a test then what would be the best way to test for digital clipping (assuming you don't trust the automatic limiter programs or meters).