nlc201's explanation of proper gain straging deserves a seperate thread

[C.Lambrechts]

I've taken the liberty copying this post by nlc201 and making a seperate thread out of it.

The original thread was : 'Do I need a Dangerous 2 Buss, and a Benchmark DAC-1?' by eskay.


http://gearslutz.com/board/showthrea...5&pagenumber=1


I felt like this deserves the extra reading and a seperate thread :




Here's nlc201's excellent post :


-------------------

The main argument for recording lower is really one of proper gain straging like Bassmac said. In a nutshell (I'm sure fellow slutz will extrapolate), digital gear was originally, and still is , designed to interface with professional analog gear operating at +4 dBu. Most converters range from -14 dbfs to -20 dbfs = +4 dBu = 0 VU. That extra 16 dB or so is intended as headroom and was/is thought necessary to prevent those nasty digital overs. Most pro analog gear is designed to work at these "low" levels (+4 dBu nominal). Now, if you record compressed audio at close to 0 dBfs, you're gonna be runing your gear at around +20 dBu! That's starting to get close to the limit for a lot of gear. Granted, some gear sounds good when it's run that hot. But a lot of gear is much more efficient at the levels intended. A hot running pre/EQ is going to be severely lacking in headroom which can very negatively impact the signal. If you've got on;ly 4-5 dB of realistic headroom to EQ with, where can you really go with it?

In digital terms, recording hot runs the risk of overs. And don't think that because the meter says you aren't clipping doesen't necessarily make it so. Most digital meters are calibrated to at least 4 samples of full scale digital audio in a row are needed to set off the clip light. Some are even more. Bottom line, you may be clipping and not even know it. People say, "what's the big deal if I don't hear it?". Well, many have found really hot audio that doesen't have the more obvious artifacts of clipping can still be degraded in general. Light digital distortion can really add up and ruin your day.

Now add to this fact that many people who record digitally like to add plugins or other dsp. An additive EQ inserted on a screamingly hot track has absolutley no headroom with which to work. A little 2-3 dB boost can clip the audio with ease. It reminds me of working on really shitty analog consoles with no headroom where things started to distort at the first sign of EQ. The solution for most is to turn down the input gain as to give yourself the headroom needed to operate. But, as Bob pointed out, this is another step in the process. Another mathematical calculation that is really not needed at all. And, like DAWs, plugins can clip and not tell you about it. The chances of plugs clipping on digital audio at -3 dBfs is quite high.

Many people like to claim they like to get "those extra bits". But you really have to look at where those bits are coming from. First of all, the average 24-bit ADC has a dynamic range of really not much more than 110 dB. Simply put, since every bit is the equivalent of 6 dB of dynamic range, most audio recordings done at 24-bit are in actuality closer to 20-bit at best. But who cares? Look at the specs of the average boutique mic pre. I can tell you for certain that my old Neves ain't got that kind of dynamic range! A more realistic number would be in the 90's or less. So, it is true that by recording hotter, you get a better S/N on the converter. However, you're really just increasing the level of detail of which the noise of the mic/mic pre/compressor/etc. is rendered. Add to that fact the dither that will evetually be added at the 16-bit level (assuming your target format is CD). So, that 12 dB (or 2 bits) you sacrifice is essentially going to be sacrificed anyway. Well, why record at 24-bit then? I think it's because the converters simply sound better. Also, the signals add up with a better S/N ratio bringing out the low level detail. Remember that the ultimate S/N of 16-bit is 96 dB, but there are very few, if any, 16-bit converters capable of that.

So, by recording at lower level, you gain the following advantages........

1. Running your analog gear at nominal levels. Not just going in either! Mixing out of a hot running DAW into console eats your mix headroom like crazy. True the nice console can typically handle it, but why burn your headroom if you don't have to? There's always the line amps on the console but there's yet another unnecessary gain stage.

2. Much lower risk of nasty digital overs and lower level digital distortion that may not evern be read on digital meters. Applies to both ADCs and DACS!

3. Internal DSP of the DAW and or plugins doesen't require an extra gain change just to be able to function in a proper gain realm. Less policing of pugins to make sure they're not clipping internally. Because let's face it, you're not gonna stare at the meters of a whole bunch of plugs throughout the entire mix. Wouldn't it be more efficient to just mix knowing that, barring any extreme DSP process, there's very little chance of clipping?

4. Much easier to interface with other outboard mixdown gear. You think a 1/2" machine wants to see +20 dBu? How about a compressor? In the old "hot" days, I remember times when hitting a mix buss compressor guaranteed at least 2-3 dB of attenuation with the threshold at maximum. Boy, that made it easy to use. I also loved it when my mix buss EQ distorted on certain very loud parts. That was just grand. Anyway, I digress......

5. No real loss in S/N ratio unless you find that rendering preamp noise accurately will make the lead vocal sound better.





OK, enough preaching. This is how I see it and I invite everyone to poke holes in my theory and correct the mistakes I made (I'm sure there are at least a few!).

Bottom line for me though.....Once I started recording at nominal levels instead of trying to get every extra bit, things sounded better. Very noticably better......

Flame away!

---------------

[NathanEldred]

I agree, 24 bit digital can get away with plenty of headroom before full scale. It isn't necessary to obsess about hitting digital zero (or really extremely close). I would rather have the headroom for the sake of digital and analog than the extra "bits".

[AnthonyV]

What about as far as summing goes? I find that even when keeping my recording levels modest (PT faders usually in the mid yellow or so), I end up having to turn down all my faders just so that I don't end up clipping the master. I know that turning down the master fader will do the same thing but its a mental thing for me to keep that at 0. So should I be recording at even lower levels to avoid this?

So I guess my question is: What exactly would be considered optimal record level in PT/DAW realm?

Thanks for the great info so far!

[Ckevperry]

Originally posted by AnthonyV
What about as far as summing goes? I find that even when keeping my recording levels modest (PT faders usually in the mid yellow or so), I end up having to turn down all my faders just so that I don't end up clipping the master. I know that turning down the master fader will do the same thing but its a mental thing for me to keep that at 0. So should I be recording at even lower levels to avoid this?

So I guess my question is: What exactly would be considered optimal record level in PT/DAW realm?

Thanks for the great info so far!

The mixer has more than enough headroom to handle whatever levels you want to use. (Provided you like what your analog side is doing at those levels.)

You just have to decide yourself what kind of headroom you need for plugin processing down the road.

[CV7]

Thanks C.L. Great info! So then, where you guys hanging around?
-10? -14? whats the range that you think is optimal?

[bassmac]

Originally posted by CV7
So then, where you guys hanging around? -10? -14? whats the range that you think is optimal?

The optimal number would simply be whatever your converters are calibrated to. As nlc201 mentioned; most converters range from -14 dbfs to -20 dbfs.

In my case, my PT|192 I/O is calibrated to -18 dbfs, therefore [-18 dbfs = +4 dBu = 0 VU]

[CV7]

Thanx Bassmac.

[alphajerk]

the optimal range is the difference between the peak level of the signal and the noise floor of the room/gear. you dont LOSE bits by recording lower levels. the RESOLUTION is the SAME. working at 16 or 24 bit is only a difference of resolution and noise floor.

there is no reason except to purposely distort a signal to get anywhere near 0dbfs on a convertor.

[Bob Olhsson]

I think 24 bit recording is a revolution for both remote recording and for obtaining usable first takes. One of the few areas where it leaves analog in the dust.