Comparative volumes

[volltreffer]

I assume that this may have two reasons:

a) former digital mixers might have had not too much headroom. Staying a factor of two below the limits seems reasonable. (And if two peaks add, they're still within the limit).

b) the meters often act a little slower, so reaching -4 may be 0 fast peaks...

Just ideas..

aXel

 

[masteringhouse]

I assume that this may have two reasons:

a) former digital mixers might have had not too much headroom. Staying a factor of two below the limits seems reasonable. (And if two peaks add, they're still within the limit).

b) the meters often act a little slower, so reaching -4 may be 0 fast peaks...

Just ideas..

aXel

This is true for an anlog or VU meter, but not for a digital or peak meter. They should be instantaneous since there aren't any ballistics involved. As far as headroom, anything below 0dBFS is fine in the digital world.

 

[Farview]

This is true for an anlog or VU meter, but not for a digital or peak meter. They should be instantaneous since there aren't any ballistics involved. As far as headroom, anything below 0dBFS is fine in the digital world.

The peak meters do have ballistics (of sorts) involved. Technically, you can't go over 0db so there is no way to measure an 'over' other than by counting consecutive full scale samples. Some meters will turn the red light on after 4 (for example) consecutive full scale samples, some after 8, or 10. (These number are just examples, I don't know what the exact numbers are for any meters in particular)

The other headroom comment was refering to digital mixers that have a fixed bit depth mix bus. If you have 2 signals that peak at -6 at the same time, they add together and you get 0. As you can see, 24 tracks that peak at -6 is going to run you out of headroom quickly. This isn't really a problem with 32 bit floating point mix busses in a daw.

 

[masteringhouse]

The peak meters do have ballistics (of sorts) involved.

Last I looked a digital meter had no moving parts, hence no "ballistics". You can program a digital meter react like a VU but there still are no ballistics per se.

Technically, you can't go over 0db so there is no way to measure an 'over' other than by counting consecutive full scale samples. Some meters will turn the red light on after 4 (for example) consecutive full scale samples, some after 8, or 10. (These number are just examples, I don't know what the exact numbers are for any meters in particular).

If your meter is reading 8 consecutive samples as an over you better throw it away let alone 10. 3 is what I've seen as a standard in most high end gear.

The other headroom comment was refering to digital mixers that have a fixed bit depth mix bus. If you have 2 signals that peak at -6 at the same time, they add together and you get 0. As you can see, 24 tracks that peak at -6 is going to run you out of headroom quickly. This isn't really a problem with 32 bit floating point mix busses in a daw.

We're talking about recording a final mix at -4dBFS or less. And yes it's still a problem with 32 bit floating point too. Floating point doesn't allow you to record an infinite amount of gain, so there is still a limit. Also very likely most folks here will be going into plugins that are 24 bit. What happens with a floating point input when you exceed the the range of 24 bit with a 24 bit plugin?

Clipping ...

 

[kylen]

... Also very likely most folks here will be going into plugins that are 24 bit...

I'm using Audition and Sonar apps which have 32bit float internal busses, my plugs are mainly Voxengo which use 64bit float internal processing. So I guess the LCD is 32bit float in my case...plus I use 32bit float wave files in Audition, Sonar will do that later this month in SOnar4 (I think!).

 

[Farview]

If your meter is reading 8 consecutive samples as an over you better throw it away let alone 10. 3 is what I've seen as a standard in most high end gear.

You did read what I said in parenthesis
(These number are just examples, I don't know what the exact numbers are for any meters in particular)

We're talking about recording a final mix at -4dBFS or less. And yes it's still a problem with 32 bit floating point too. Floating point doesn't allow you to record an infinite amount of gain, so there is still a limit.

I was confused, Massive was talking about it being bad for individual tracks and that combined with volltreffers comment about older digital boards not having headroom, it got me thinking in that direction. And no, floating point doesn't let you have an infinite amount of gain, but it doesn't run out of headroom instantly like a fixed point system would.

Also very likely most folks here will be going into plugins that are 24 bit. What happens with a floating point input when you exceed the the range of 24 bit with a 24 bit plugin?

I don't think I have any 24bit plugins. But you are correct, it would clip. If you are using a 24 bit fixed point plugin across your master bus, you deserve what ever you get.

 

[masteringhouse]

I don't think I have any 24bit plugins. But you are correct, it would clip. If you are using a 24 bit fixed point plugin across your master bus, you deserve what ever you get.

Cool Farview, (sorry if I seemed a bit terse in my response a long day at the studio today).

One thing to consider though is if you are routing your bus to external digital hardware (like I do). In these cases you still have to be concerned about keeping levels within a 24 bit range. In cases like this dithering to 24 bit before hitting the device is important. Pro Tools TDM systems have a dithered mixer to help with this issue.

 

[Farview]

That sort of thing never comes up for me. I either mix completely in the box, or I run everything back through the mixer and use outboard effects. I just never run into that scenario.