View Single Post
Old 10th September 2009
Originally Posted by Tweek Audio View Post
Is there any documentation that can display that plugins actually do behave like analog consoles and that running -20dBFS into a plugin is actaully better than running -0.1dBFS?
Actually, if you read the original thread, that's not really the issue being discussed. What's behind it is Paul Frindle's argument that various algorithsm create boosts in the signal that are not apparent. If you look at the standard DAW, the signal comes into the track or bus, goes through the plugs, and it's not until the *output* of the track or bus that you see the level.

But the level could have, for a number of reasons, actually gone over 0dBFS, even though the input to the track was below that and the output that you see on the track/bus meter is below that. It could be due to manual boosts of course, but more subtly due to certain types of processes that the plugs do.

So you could have the peak track level as stored be -5dB, then the first plug on that track is an EQ. You might then set a high pass on that EQ, which anyone would think would lower the actual level that comes out of the EQ, but it doesn't always. EQs often have substantial 'filter ring' around the point of the filter cutoff frequency. If the part you cut out constributed less than the part that's bosted by the filter ring, it can actually raise the level of the track by a substantial amount, some of them have really big ring and that's part of their sound.

So then the next plug after the EQ sees a signal that's now over 0dBFS, but because of what it does it brings down the signal before it comes out, and so what you see at the track fader might be -9dBFS. So you think all is well, but in fact one of both of those plugins might have clipped internally if they don't deal gracefully with signals over 0dBFS.

So the point was, if you aren't recording yourself, and therefore can track with reasonable levels (peaking say in the -12dB to -9dBFS range, though of course the occasional peak over that is no disaster), and therefore you get in tracks to mix that were recorded really hot, you may want to reduce them using a trim plug before the first plugin on the track, in order to avoid the above problem.

And the level wasn't -20dBFS *peak* it was RMS. This is a constant source of confusion in these discussions, where people aren't clear about whether the level referred to is peak or RMS. A 20dBFS RMS level will have peaks that are about in that same range I mentioned above when the signal contains peaky material, less so for less peaky material. So the level being suggested wasn't as low as some here might have felt was being indicated.

In theory, if you send a sine wave through an outboard piece of gear and adjust the output to 0dB *VU* on the outboard gear's meter, it'll come out about in that -20 to -18dBFS range in the digital world when you are set up for the usual +4 style pro levels, which most of us would be. So tracking at that level is what would make for about optimum signal to noise ratio in the outboard world, and it would give you plenty of room in the digital world to avoid the issues above. But if you didn't record it yourself, then a plug can be used to bring the level down to avoid the possible plugin clipping issues.

The problem may or may not occur. It depends on the plugin. But since you can't know for sure, the argument is that it's best to err on the side of caution.