TOPIC:

I have notticed that I there is bass distortions when I applied RoomFeel_EQ_Target_Curve.
I have configured DDRC-24 to use 0dB output gain 0dB master volume. I also use 100% volume in my audio source to achieve full 0 dBFs for maximum dynamic range.
The RoomFeel_EQ_Target_Curve has 6.6dB gain at 31Hz. It seem there is not enough headroom for such gain.
I corrected my curves to not exceed 2.5dB gain, but I would like to know, what maximum gain I can use for creating target curves and be sure that there will be no clipping at all.
Maybe devteam could answer to this?

@plus5volt

yes, I think that it's a bit of combination here:
a) using digital signal = very little headroom
b) Having a larger boost in lower end with a target curve maybe above the speaker response.
c) having the master volume control at 100% (controlling the input gain).

Have a trial at making your measurement with your master gain not as high within Dirac live. If level is too low, make sure that you UMIK-1 recording level is high enough or just catch up the volume at the amplifier (i.e. increase level there).

This way you don't have to be at 100% and run into potential internal gain structure issues.

Devteam

Draw the shape of target keeping all of it underneath the “all before” response normalized graph, will not clip and will sound great, might have to bump the input of the downstream device by 3 or 6db if like to listen really loudly ,
.02c

This was default RoomFeel EQ target, with slight dip around 3khz. Some songs audibly suffer from bass distortion.


Then I created this, 2.5 db peak. Distortion much reduced. But I notticed that there is sill sometimes.


My current target has peak at 0db.


I am satisfied with the result, but I wonder if there is any headroom I waste with this target, or why default target has 6.6db peak. I would like to be sure I use full available dbFS without clipping.
DDRC-24 has no spdif output, so I am not sure how I can detect clipping during measurements with microphone in REW, I will try during the weekend.

I did few more tests, and I can say, that ANY target curve I tested so far causes distortion.
Including my 'Auto target' curve which has peak at 2.5dB, or my 'flat' curves that have peaks at 0db,

I tested it at 24b and 16b 48kHz, both toslink and usb interface of my DDRC-24. Both have identical distortion.
My audio source is windows pc, and I play spotify, player and all system playback volumes set to 100%.

For test I used is randomly found song on Spotify "Losing (Remix) by Vintage Culture, Kalozy, JORD",
beat sound that is at around 0:43s is ugly distorted.
I checked FFT of this song, it has no deep bass line, it does not go below 45Hz at this distorted moments.
Spotify app has normalisation disabled and high quality enabled.

This is not speakers or amplifier issue. Because when I set 90% (-1,5dB) on my audio source master volume on PC,
then distortion dissapears, and I can set my amplifier volume level as loud as I want, and even very loud playback has no distortion.

Something is wrong with the signal processing. Can anybody try to reproduce this as well?

I did even more experiments and found out that I just need correct output volume level at the DDRC to -1.5-2dB, and I suppose it is done after dirac correction.
When I use dirac target curve with higher peak, then I have to set lower output volume to get rid of distortion.
After this experiments, I started to think if this is maybe clipping analog end or D/A section of the DDRC-24 unit, unfortunatelly i have no scope to verify this.

This is an issue that I raised in here a while back - the fundamental problem is that the user gets zero feedback from the calibration tools as to what level of boost is applied by the Dirac filters. If you are running the master volume at 0dB, you essentially have no headroom to play with - by the time the (internal) signal reaches the DAC, you had better have attenuated it to the point where there are no peaks in the signal that exceed 0dB. Not an easy thing to do if the only thing you have to go on is the rather idealized/smoothed before & after curves that are shown in the calibration tool. It would be entirely possible for the calibration tool to report what the max boost is in a given set of filters; equally, it would be entirely possible for the miniDSP hardware to report a "clipping" condition. So far, neither Dirac nor miniDSP seem to have any answer to this other than "trial and error", which I find disappointing.

Am I the only one running DDRC at -35dB with 50% Windows volume?
I like to control my volume digitally (more convenient). Just put speaker amp dial to middle
With 24-bits I think 0dB digital output is not worth that much...I doubt DAC inside DDRC24 is audiophile grade anyways.

50% volume means that you no longer use 24bit, but signal delivered to ddrc has up to 12bit resolution in best possible scenario, then after all ddrc processing you reduce signal details further -35dB means you rescale ddrc output to 10% possible dynamic range. Why it still sounds decent, because In windows volume conversion done acurately, because is applied to signal converted to 32bit floats (between your audio player and system mixer).
12bit should provide decent sound, because good quality magnetic audio tape has quality compared to 10, excellent tape comparable to 12 bits, but 12bit you use has very low noise, that is why you still hear decent quality.
If you cannot hear the difference, dont bother, but try imagine it is possible that some people say they can hear difference in hi-res audio...

and yes, DAC insde the DDRC is audiophile grade.
but of course I should ask about your definition of audiophile, some audiophiles find vinyls better sounding than cd

Looks like DDRC-24 does 32-bit internal processing...I'll try 100% @ 24-bit in Windows and control volume with DDRC24...
-60dB master volume now LOL...dirac input levels in DDRC24 seem to be around -7 to -16dBFs without clipping though.

Maybe it is 32 bit, but only fixed point, not floating point. DAC it uses is AK4626, it accepts up to 24bit. After Dirac processing, conversion 32 to 24 must be done, and master volume is probably executed in AK4626 chip on 24 bit data.
Do you ever use 100% windows and 100% ddrc ? You should consider to lower analog amplification to the maximum level you usually use, and use more amplification in the amplifier when you want to wake up your neighbours

I am still trying to understand my distortion issue, why decreasing master volume output helps to reduce the issue. There should be enough analog headroom, because now when I have curves with peak at max 0db, overall volume is much higher when I disable filters and no distortion.
@devteam is not very helpful here to explain my case.

3 analog volume controls for me, unfortunately.
Left, Right active studio speakers (with a notch at 12 O'clock) and active subwoofer with its own volume at some level set for flat response with mains...digital volume control is the only way for this setup :/

devteam wrote:

@plus5volt

Have a trial at making your measurement with your master gain not as high within Dirac live. If level is too low, make sure that you UMIK-1 recording level is high enough or just catch up the volume at the amplifier (i.e. increase level there).

Devteam

@DevTeam - the Dirac calibration software (DDRC-24) allows the UMIK-1 level to be increased beyond 0dB - I have always assumed that to do so would cause problems. Am I correct, or is it OK to use the absolute max level on the mic?