0dbVU and digital converters

[Jimbo]

I recently read a post on a different forum regarding optimal tracking levels.

This poster claims that converters are tested and spec'd at 0dbVU (-20 dbFs American). His conclusion is that for optimal converter performance, recording levels should hover around -20 dbFs.

Any comments on this?

Thanks,

- Jim

[Farview]

It's not really the performance of the converters, it's the performance of the analog chain before the A/D and after the D/A.

[Jimbo]

I was referring to the whole conversion circuit (not just the chips), but I'm glad you explained this further. thumbsup

So it sounds like signals hotter than -20 dBFs will lower the performance of the input buffer to the chip. Thus, a degraded signal will undergo A/D conversion.

Time to lower the gain!

Thanks,

- Jim

[Jimbo]

Let's take a situation where the incoming signal's range is roughly 10dB.

I should be setting the level to run between -25 and -15 dBFs, right?

This assumes that signal degradation is not linear as it moves away from -20dBFs.

[tony666]

Isn't he talking about this :

Most VU meters are average signal indicators. On an analogue mixer a signal can be 0Vu (=+4dBu)but it's peak level can be higher (Vu doesn't detect those peaks).

When you match the 0VU to -20 dBFS with a sine wave, you give the converter 20 dB of headroom for capturing signal tops. The sine waves are used for calibration, real audio signal have higher peaks and all type of forms. The difference between peak and average(Vu) levels is in most cases not above 16 dB. So with 0VU = -20dBFS , maximum peak levels will be at -4dBFs.

[Jimbo]

I'm really asking two things (at this point):

1. Are A/D converters optimized to sound best at -20dBFs (0 dBVU)?

2. If so, is the signal degradation linear as it moves away from -20dBFS?

Thanks,

- Jim

[tony666]

Originally Posted by Jimbo

1. Are A/D converters optimized to sound best at -20dBFs (0 dBVU)?

On page 12 of this ******//www.lavryengineering.com/whit...296MKIIIc1.pdf Lavry manual you can read the harmonic distortion and noise levels at different levels. Didn't know this myself. It's the lowest at -40dBFS, but you probably loose dynamic range at that point?

[Jay Kahrs]

Originally Posted by Jimbo

I recently read a post on a different forum regarding optimal tracking levels.

This poster claims that converters are tested and spec'd at 0dbVU (-20 dbFs American). His conclusion is that for optimal converter performance, recording levels should hover around -20 dbFs.

Any comments on this?

A/D converters are optimized to sound 'best' depending on how you calibarte them.

In our shop, most things (that can BE cal'd anyway) are set to -18dBfs when there's a 0dbVU signal at +4.

There's no law saying that you can't run at -12dBfs to 0VU...but IMO cramming all that signal into digital-land is just bad juju...eats up all your headroom.

IMO -20dBfs is too low for most things...I normally print guitars and 'steady state' things around -15 or so...maybe -10 depending how how they match up to whatever other instruments are already there.

My best advice?

Don't print too low & don't print too hot!

[Jimbo]

Originally Posted by Jay Kahrs

Don't print too low & don't print too hot!

[/Quote]

But, that begs the question that I'm asking: what is the mid-point. If my facts are straight, it seems to be around -20 dBFs. Also, as one person pointed out, you don't want to sacrifice too much resolution either.

Resolution is a whole 'nuther factor -- especially when you consider how much more resolution a low gain 24 bit signal has than the high-gain 16 bit signal that's going to come off of a CD.

Anyhoo,,,,

[Farview]

Originally Posted by Jimbo

I'm really asking two things (at this point):

1. Are A/D converters optimized to sound best at -20dBFs (0 dBVU)?

It depends on where they are calibrated. The converter chip doesn't really care, the analog path to the chip is what will degrade the signal. So the answer to your question would depend on the equipment used.

Originally Posted by Jimbo

2. If so, is the signal degradation linear as it moves away from -20dBFS?

It could be, it still depends on the equipment involved. There really is no standard calibration.

If you record too hot, you are using up the headroom in the analog signal path. What your analog signal path is will determine how much it will degrade the signal. Once digitized, you will end up having to turn the signal down in the DAW so you don't run it out of headroom. Your software will determine how much that will degrade the signal. Once the signal gets converted back to analog, if it's too hot, you will be using up all the headroom in that analog path as well. The equipment being used will determine how much it will degrade the signal.

All this stuff about not hitting the converters too hard isn't about protecting the converters. It's about good gain staging. All of your signals should be running right around line level because that is where everything in your signal chain is designed to run. If your converters are calibrated so that 0dbVU = -20dbFS, you should be running at that level. My old DAT machine is calibrated so that 0dbVU = -12dbFS. So it's all relative.

[NathanEldred]

Originally Posted by Jimbo

But, that begs the question that I'm asking: what is the mid-point.

-16dbFS is probably the mid point when running a 1khz sine tone @ +4 into the converter. I'd say if you are staying in the box to mix, err on the lower side of the 'middle' (-16/-18 or -20 for extreme dynamics) and maybe a little hotter of the 'middle' when integrating with analog gear (-16/-14)....headroom is good, but with analog you are dealing with signal to noise issues, and most importantly the sweet spot that the analog outboard likes to see.

[Jimbo]

Thanks for the replies, folks!

[RobR]

I have an excellent suggestion.

Don't clip.