+16dBu @ 0dBFS outputs?

[kelldammit]

why would someone use a spec like that to describe analog outs?
what does it really mean when interfacing to standard +4 analog gear?
i thought +16dBu was clip (or close to it) even in analog? this being the case, i can see the logic, but not the sense.

i ran into this when i hooked up my saffire to mackie hr624 monitors.
their input trims are usually set to "normal" (i.e. full blast...around 8:00), which is supposed to be able to handle +4. with the monitor out knob on the saffire barely up, those speakers were SCREAMING. i had to back the input trims to one notch above "off" (around 3:00) on the speakers to be able to get listening levels other than "totally off" and "too loud".
i'm using balanced trs out of the saffire to xlr line ins on the mackies...

anyone else dealt with this sort of thing?

kell

[qo]

I imagine it's the Saffire that has the spec you cite. This makes sense to me since 0dBFS is the highest level achievable with digital. So, they are correlating that with their output level (which will be the highest level achievable with that box). This level (dBu) isn't "loudness." So, "screaming" (which I'm assuming is a subjective loadness term ) isn't really applicable.

dBu is a measure of voltage irrespective of load (u = unloaded). 0dBu = .775 volts, always. You can equate dBu to dBm only if the load is 600 ohms (dBm is dependent on load). That is, 0 dBm equals .775 volts only into a 600 ohm load.

i thought +16dBu was clip (or close to it) even in analog?

dBu is only relevant to analog (inapplicable to digital). +16dBu is 5 volts. Now, it could be true (indeed is likely) that a digital signal has +/- 5 volt swings, but that's only the analog reference for the signal, and not a measure of how many bits are 1 vs 0 in a given sample (which is dBFS afaik).

which is supposed to be able to handle +4

This is probably +4dBm. And, this is a always a source of confusion. dBm and dBu only have the same 0 reference into 600 ohms. So, you have to look at the input impedence of the Mackies to be able to equate these.

[kelldammit]

firstly, thanks for the response!
on most other line level output devices, i've only had to back the input attenuation down to -10 (about high noon) on the speakers, if at all (and only IF i was feeling like being nice to my neighbors). at this setting i had a nice range of volumes via the interface's monitor out control. but with the monitor output attenuation nearly maxed on the saffire, and still being nearly deafened, it made me wonder what the hell was going on, so i went digging for the spec...hence my irritation at the one given. all i wanted was a practical nominal line out level assuming a typical load. but i did some more research anyway...

dBu is a measure of voltage irrespective of load (u = unloaded). 0dBu = .775 volts, always. <snip> dBu is only relevant to analog (inapplicable to digital). +16dBu is 5 volts. Now, it could be true (indeed is likely) that a digital signal has +/- 5 volt swings, but that's only the analog reference for the signal, and not a measure of how many bits are 1 vs 0 in a given sample (which is dBFS afaik).

decibels full scale (dBFS) IS an analog reference to digital. it's equal to the maximum RMS voltage going into an a/d that will produce a digital signal of all 1's (but not induce clipping). it is ALSO the RMS voltage output from a d/a when an all 1's (but no overs) 997hz digital sine is sent to it. so from one unit to another, it could vary a bit due to design, part specs, etc.
apparently there are other variants as well, insofar as whether square or sine waves are used for the incoming/outbound signals.

we often use dBFS to refer to levels within the digital system itself (i.e. the software), but since dBFS relates to voltage, it's apparently incorrect. hell, at 32bit float, >0 doesn't necessarily mean clipping...so afaik, nothing but a bit meter will tell you what bits are actually in use at any given time. now i wonder what the meters in DP are really showing, and when we move a slider down 2dB, it's 2dB relative to what? what's the reference?
always new stuff to learn!

kell

[qo]

Ah, ok, thanks for clarifying dBFS. But this just leads to another question. If dBFS is referenced to an incoming analog level resulting in "all ones" output, then it might mean different things depending on what bit depth, or AD, is being used (since different bit depths represent different dynamic ranges and different AD's might have different input stages). So, does not dBFS have an analog ('scuz the pun) to e.g. dBm which is referenced to 1 milliwatt, or dBu (referenced to .775 volts)? That is, I would think that part of dBFS' definition would include some absolute reference?

it's 2dB relative to what?

Yes, good question since dB is all about relations between two values. To say 4dB is to say nothing (although, to say 4dBu does say something since dBu does have a zero reference i.e. .775 volts). To say that the ratio of 5 watts to 3 watts, in dB, is (10 x log(5/3)) = 2.21 dBW, describes the relative power levels in a useful way (i.e. 5 watts is 2.21 (dBW) more power than 2 watts); dBW referenced to 1 watt.

While we're on the subject, in case anyone is interested, the equations for acoustic power i.e. the "screaming" part, (and for current and voltage) are similar to the above, just substitute 20 for 10 as the multiplier e.g., what's the difference, in dBSPL (sound pressure level, measured in dynes/cm^2) between 120 and 100 db?

20 x log(120/100) = 1.58

If follows that twice the (acoustic) power is 6dB e.g.:

20 x log(100/50) = 6.02

whereas twice the electrical power (dBm) is 3 dB e.g.:

10 x log(100/50) = 3.01

For those that always wanted to use the log key on OSX's calculator in the course of your everyday work, look no further, heh.