M-Audio Audiophile Inputs

[GDoxta]

Okay, I know the topic has been convered a million times, but I have
one question which hasn't been asked directly. I notice that a lot of
audio interfaces have "+4dBu/-10dBV" input capabilities. I have also
noticed that the +4dBu seems to partner up with balanced inputs and
the -10dBV goes with the unbalanced inputs. My question is this:
since the M-Audio Audiophile 2496 PCI only has -10dBV inputs (yes it
also has "consumer input" but that isn't +4dBu), does that limit my
hottest signal to a "-10dB" level on input before digital clipping
occurs?

steve

[P Stamler]

> I notice that a lot of
>audio interfaces have "+4dBu/-10dBV" input capabilities. I have also
>noticed that the +4dBu seems to partner up with balanced inputs and
>the -10dBV goes with the unbalanced inputs. My question is this:
>since the M-Audio Audiophile 2496 PCI only has -10dBV inputs (yes it
>also has "consumer input" but that isn't +4dBu), does that limit my
>hottest signal to a "-10dB" level on input before digital clipping
>occurs?

No. Fasten your seat belt. We have to distinguish between dB used to express a
voltage level (as in dBV, which denotes decibels relative to a 1V level) and dB
used to express modulation level (as in dBFS, which denotes decibels relative
to full scale, full scale being the level right before digital clipping).

We also have to look at the idea of "nominal level". This dates back to the
days when we all used slow-responding VU meters and analog tape, and it's
basically the level that would show 0 VU if you were using a VU meter. An input
rated at -10dBV means that -10dBV (0.316 V RMS, or the peak-to-peak equivalent
-- let's not start *that* discussion) is the nominal level. Digital devices
don't record nominal level as full-scale; instead, they record it as a lower
level, allowing a certain amount of headroom.

How much lower varies with the manufacturer. Many use somewhere around -15dBFS
-- that is, a nominal input level (say -10dBV when the input sensitivity is set
to that nominal level) yields a digital signal of -15dB, relative to clipping.
So you'd have 15dB of headroom over nominal level before digital clipping
began, and you could accept up to +5dBV from your preamp/board before you
clipped.

When I tested the M-Audio Delta 66 card, I found that nominal level
corresponded to -10dBFS -- that is, you got only 10dB of headroom over nominal
level before you got to digital clipping, which is a little tight. I don't know
if the Audiophile specs are similar.

In practice, this means you need to watch the meters on your computer screen to
make sure you're not clipping. I do, and it works fine. But if you go by meters
on your mixer, you're likely to get into trouble unless you do some careful
calibrating.

Peace,
Paul

[David Satz]

steve (GDoxta) wrote:

> since the M-Audio Audiophile 2496 PCI only has -10dBV inputs (yes it
> also has "consumer input" but that isn't +4dBu), does that limit my
> hottest signal to a "-10dB" level on input before digital clipping
> occurs?

Everything that Paul Stamler wrote is true. Restating the main points:

[1] When an input level is quoted, it simply means that a steady tone of
that level _can_ drive the unit to whatever "0 VU" would have been, back
in the day when everybody used VU. (That's what's meant by a "nominal"
level.) But the VU standard understood that signals vary, and that you
could always go somewhat above 0 VU briefly without pain. So right away,
that fact tells you that 0 VU could never equal digital full scale.

How far above 0 VU you could go, for how long and on what type of signal,
was part of what an engineer would have to learn about each piece of
recording gear he was working with (and back then, it was almost always
"he" rather than "he or she").

[2] Most devices have input gain adjustments to let them accept signals
higher or lower than the specified level. There's no fixed definition of
how such an input gain control must work if present, so it could allow
you to use only weaker signals or only stronger signals, or it could be
designed to allow either one. It could have a narrow range or a wide
range. It could let you fade all the way down to zero or it could have
some fixed gain as its lowest setting. It could be that if you use the
control near its extreme settings, the circuit clips or is noisy or
sounds bad in some other way.

All such behavior would have to be discovered on a case by case basis.
Knowing--not simply assuming--the exact behavior of the controls on all
your equipment is still part of an engineer's responsibilities.

[GDoxta]

Thanks! It actually makes sense.

steve

[Mike Rivers]

In article > writes:

> since the M-Audio Audiophile 2496 PCI only has -10dBV inputs (yes it
> also has "consumer input" but that isn't +4dBu), does that limit my
> hottest signal to a "-10dB" level on input before digital clipping
> occurs?

No. -10 dBV corresponds to some level below clipping that M-Audio has
decided is safe for you. If you connect it to something with a nominal
-10 dBV output, you will have difficulty clipping, and you may even
have difficulty reaching full scale digital level without SOMETHING
ELSE in the chain before it clipping.

On the other hand, if you feed it from something with a +4 dBu nominal
output level, you will find it quite easy to reach clipping at the
sound card without the mic preamp, mixer, etc. clipping.

It's all about headroom and gain management. Twiddle the knobs until
it doesn't clip and you'll be OK.


--
I'm really Mike Rivers - )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me here: double-m-eleven-double-zero at yahoo

[Peter Larsen]

David Satz wrote:

> How far above 0 VU you could go, for how long and on what type of signal,
> was part of what an engineer would have to learn about each piece of
> recording gear he was working with (and back then, it was almost always
> "he" rather than "he or she").

And about what hardware, 10 dB was/is a good lead for recording on
analog tape and it will only rarely be exceeded as "non-displayed
transients", but usually it was recommended to use CCIR peak metering if
you wanted to run with a lead of only 10 dB, the general suggestion
would be to use 15 dB lead. With modern tapes of that (and this) day a
10 dB lead equals the 1 kHz headroom with 0 VU adjusted to 320 nW.
Household machines with VU meters defaulted to 0 VU corresponding 185
nW, and "high performance" machines 250 nanowebers. As long as you know
how the actual tape recorder was adjusted then you could run them
eqaully hard by using varying amounts of red ....

If it was about something that would do hard clipping - as opposed to
the soft clipping of analog tape - then 15 to 18 dB vU meter lead would
be the recommended setting. Which is why DAT's tend to have a marker on
their meters at -18 dB. With 0 VU typically being at + 4 dBm it should
begin to get obvious why most hardware is expected to handle a peak
level of +22 dBm.

My old Audio & Design Recording "Gemini" compressor is designed for
having a limiter threshold at +14 dBm out, i. e. only 10 dB above
nomimal "0 VU" level, it makes perfect sense because real world
compressor/limiters of its kind DO overshoot.

> All such behavior would have to be discovered on a case by case basis.
> Knowing--not simply assuming--the exact behavior of the controls on all
> your equipment is still part of an engineer's responsibilities.

Indeed. Something to embroider and hang on the wall ....


Kind regards

Peter Larsen

--
************************************************** ***********
* My site is at: http://www.muyiovatki.dk *
************************************************** ***********