It used to be “0” to “100”, now the numbers go “0” to “106”.
We were wondering when people might notice The ability to go to 11 (well 10.6) was added a few software versions ago. It only goes to 106 when the 20dB attenuator is engaged. Paul wanted to help a few customers that had systems that were sensitive enough that the full level was too high but the 20dB attenuator was too low.
Ted and Paul, you have been found out once again!
Nice. And pretty funny too.
I was afraid someone might notice that
The few that really needed it were warned that in extreme cases there might be a little distortion at 106 - but probably not in anything people would listen to. On the other hand if we let everyone on the planet use 106 I’m sure that some user or more likely someone doing measurements would notice. There are SACDs that temporarily go to +3dB FS. Add that to the 3dB of 106 and we’ve reached all 1’s or all 0’s. If that situation persists very long (which it shouldn’t with SACDs) there would be noticeable distortion.
It’s a compromise for the few that don’t use preamps and have systems that are too loud without the 20dB attenuator, but too quiet with it.
I still find bizarre that some SACDs have levels above FS. Weird in concept. And why would one do this?
Tape, vinyl, etc. all allow stuff above full scale. It’s just PCM that doesn’t let you soft clip. It’s an art deciding how to set the level when recording to get just a little material above 0dBFS, but the ability to record with a little headroom is wonderful.
In essence SACD specs +3dB of headroom for brief moments - typically above 3dB the distortion rises fast.
I think of it differently.
With PCM, there is a hard stop at full scale. It is the loudest you can go as the maximum representable value has been reached. For example, if your system goes to 1,000 - 1,000 is full scale, there is nothing bigger. But there is also no distortion.
With analog, we have arbitrarily assigned levels, such as +4dBu established by convention. Part of this, is that a properly calibrated analog system will not distort at +4dBu and provides some headroom. There is not a true analog full scale, as above certain level for each format distortion begins and increases with loudness.
When recording records to DSD, I always make sure to use the extra headroom, and the result is amazing. I actually always assumed that this is one of the reasons that DSD sounds so much more like analog than PCM…
I just figured the 106 was part of the beta Firmware, so I didn’t say anything. I’m using it all the time for further burning in of the DAC when I’m away from home for work.
wim said When recording records to DSD, I always make sure to use the extra headroom, and the result is amazing.There is no difference recording to digital with peaks set, for example, to -6dBFS or -12dBFS or -20dBFS as far as sound quality.
There is only the theortetical possibility of losing some dynamic resolution at very low levels. As a practical matter however, with 24-bit depth, one has 144dB+ to work with and you can easily throw out 40dB and more with no impact.
On the other hand, it is unwise to record with peaks approaching 0dBFS as any overs result in nasty distortion.
The rules are different for DSD - 20 bits is a little less to work with and some engineers don’t want to be changing levels after recording.
But remember that we don’t really have 24 bit accurate A/Ds (without spending gobs of money): 20 bits is easy these days, 24 not so much. I’d also argue that there is certainly an impact if you throw away 40dB even with if you had a full 24 bits. The difference between 16 bits and more bits is often noticeable (but not always.)
Very valid points. I admit to exaggerating.
The issue is that those of us familiar with recording with analog are accustomed to slamming the meters as the S/N is limited with analog you need to take advantage of every dB you can get, especially if you will be mixing multiple tracks down to stereo. Noise is additive. Of course, many also like the sound of tape compression/saturation as an added benefit.
With digital you are best to leave at least 12dB of headroom over peaks. This avoids intersample overs and places most equipment in a range where it sounds its best. Gain structure is a mysterious thing. This also provides a solid, real world 17-bit to 18-bit depth which will easily capture any music (also keep in mind the quietest concert halls have a background noise level of 20dB).
One indeed needs to be deliberate with DSD and to set the levels you want for the final product. DSD is, in some way, even less forgiving the tape. Ironic.