Secrets of those dB / VU levels
 

Several recent threads have prompted me to sort out something that has always been vague for me--the fundamental meaning of dBFS and peak meter indications, so I thought I'd share it.

Audio level matching sometimes seems like a hall of mirrors. Not only are we trying to match outputs with inputs, but we're expressing the problem in terms of a non-unit that is in itself ratio of actual unit quantities we usually don't specify, and the ratio has different names (dB, dBV, dBu, dBm, dBFS, VU) depending on what the unnamed reference quantity is. Decibels save us from having to manipulate very large and very small numbers, but sometimes it's nice to know what the anchor in terms of physical quantities is. The quest leads down a bit of a winding path in the case of VU and dBFS, but it ends at a simple place. Here's my take. Correct me if I'm wrong, but:

In the electronic realm, 0VU started out being defined as 0.775 volts (which is also 0 dBu and the voltage associate with 0 dBm power) and that's how VU meters were calibrated. But engineers found that the resistance that had to be added to increase the meter's impedance to an accpetable level in a standard measurement configuration reduced the reading by 4 dBu. They decided to live with it, and let +4 dBu ( 1.228 volts) correspond to 0VU. So although VU is a sliding scale, it is most often set to correspond to an output of +4 dBu, or 1.228 volts in professional equipment.

Since 0VU is converted to -20 dBFS in an ATSC compliant device, and since 20 dB is a factor of ten in the voltage world, 0dBFS corresponds to 12.28 volts.

16 bit audio breaks that 12.28 volts into (2^16)-1 steps, each assigned an integer number from 0 to (2^16)-1. 24 bit audio breaks 12.28 volts into (2^24)-1 steps, and so on.

So when any LED peak meter in any ATSC compliant device (or program) indicates for example -20dB, it's saying "I have just assigned (or read) a number that will someday cause a digital to analog converter to produce 1.228 volts at some identifiable output.

Reference levels for completeness:
0 dBV = 1 volt (aka 0 dB)
0 dBu = 0.775 volt
0 dBm = 0.001 watt
0 VU corresponds (usually) to 1.228 volts @ 1 Khz*
0 dBFS corresponds to 12.28 volts @ 1 Khz*

And finally, although it wasn't the purpose of this thread to rehash this, it would be remiss not to mention that the relative level of any particular voltage in terms of any of the above is 20 times the log of of the quotient given by voltage / reference. For dBm its 10 times the power quotient.

* dBV, dBu and dBm are based upon truly absolute physical units, and do not vary depending on context. By contrast, VU and dBFS are adjustable scales whose zero points can be set where we want them. In addition, their readings are "effective" weighted values of complicated waveforms, so instruments that measure them are standardly calibrated with 1000 Hz sine wave voltages.

[Edited to correct for Steve House's observations below. Thanks, Steve]

My references (Jay Rose's books) differ from some of your numbers. With the pro standards, 0dBu corresponds to 0.775v but 0VU on the meters is +4dBu or 1.228v. With most consumer gear, OTOH, 0dBV corresponds to 1v and 0VU on the meters correspnds to -10dBv or 0.316v. Just to keep in really interesting, my Mackie mixer's meter is calibrated to read 0VU when the level at the main output is 0dBu or 0.775v!

So now my question gets to the ATSC standard: Is it the 0dBu, 0.775v, or the 0VU, +4dBu, 1.228v, level that is supposed to be aligned with -20dBFS point?

Just answered my own question - according to an article I just located on the Broadcast Engineering website, 0VU=+4dBu=-20dBFS

Fred,
This I'm not following. From previous posts, I know you have more of an engineering and math orientation than I, but, isn't analog line level equal to roughly 1v in all worlds?

If I understand where you're going with this (and I'm not sure I do), if I send 0db tone from my mixer, and set a reference of -20db on my digital recording device, are you then saying that it will take about 7.75 volts of signal out of my mixer to peg my digital recorder at 0db? 'Cause I don't think my mixer will send 7.75 volts under any circumstances, but even with 0 analog referenced to -20 digital, I think a substansial peak (+7 analog?) might peak the recorder to 0...

Really, I'm just trying to reconcile the math with my general experience, I might be misinterpreting either. It's hard to imagine any device providing 7.75 volts... but we see 0dbFS all the time.

My understanding is:

Analog:
0 VU: defined as the point where distortion begins to kick in.
+4 dBu and dBm: standard professional reference level; the point where distortion begins to kick in.
-10dBV: standard consumer reference level; the point where distortion begins to kick in. Associated with consumer equipment.

The difference between +4 dBu and -10dBV is ~11.78dB.

Digital:
0dBFS: defined as the point where digital clipping kicks in.
-20dBFS: standard reference level for professional formats. i.e. tone at -20dBFS should correspond to 0VU.
-12dBFS: standard reference level for consumer formats.

Analog distortion versus digital clipping:
Digital clipping sounds absolute terrible, while analog distortion doesn't. Some analog distortion is ok, and intentional distortion may even be desireable. With digital, levels should be way below clipping. With analog, hitting a little over 0 VU is ok (especially when you have a little headroom above that).

More information:
http://www.prorec.com/prorec/article...25675400514576

I think they are relative units??? You need to know the context... i.e. professional or consumer levels... -10 dBV versus +4 dBu (or dBm), -12dBFS versus -20dBFS.

I hate it when that happens! You think you've put a nice dinner on the table, but find you forgot to cook the meat. But thanks, Steve--your comments came in time to correct my original post.

You can find a lot of sites that contradict each other on 0VU. I ultimately went with a VU meter manufacturer who said that the meter movements themselves were always calibrated to 0VU = 0 dBm. But that doesn't mean that they're deployed that way in equipment. Having read a bunch of the other stuff I tried, as you can see, to hedge my language. But ultimately, with my cat circling my legs looking to be fed, I made the mistake of assuming 0VU = 0 dBu --> -20 dBFS.

But for the record, "0VU=+4dBu=-20dBFS" is not a correct statement either.
"0dBu equals 0.775 volts" is true--it's a definition
"0VU is often set to correspond to +4dBu (1.228 volts)" is true.
"+4dBu has been adopted as -20dBFS" is apparently true (according to the Broadcast Engineering website, which is probably reliable. I still haven't found the ATSC or other authoritative specification that says so, but DSE says it's out there.)

Yes, if "roughly" means 0.3v to 1.228 volts.No, in that case it would take 10 volts to hit 0dBFS, since 0dB is 1 volt.Even a mic can put out 7 volts. An AT4073 listening to a jet takeoff could do it.

Regarding seeing 0dBFS all the time, even after calibrating your camcorder to -20dBFS = 1 volt, that doesn't stop you from hooking up an input device that is nominally outputting at hotter than 1 volt.

You are absolutely correct! -- while the convention is to calibrate the meters in pro gear so that 0dBVU corresponds to a signal level of +4dBu, in fact the VU scale is purely relative and one could pick any arbitrary signal level one likes to define as "0VU." As an example, as I mentioned before, my Mackie compact mixer's meters are calibrated so when the meter reads 0VU, it's sending 0.775v or 0dBu on the main output (and to Mackie's credit they clearly document the calibration in the manual, and even print a reminder note next to the meter on the mixer's panel). I think the best way to interpret the needle hovering around "0VU" is as meaning "This is the signal level that this piece of equipment is designed to normally operate at." That assumes the meters on the gear are intended to actually mean something and are not just for show (which is by no means a given with consumer gear).

Steve, that is about as succinct an explanation as I've ever read.
Spot on!