Peter,
Your last posts seem to be asking more about SNR than baseline noise levels. In classic audio engineering, since we are dealing with log relationships, the SNR(dB) = dB(signal) - dB(noise).

Without going into the math (which involves some ambiguities), the digital dB(signal) reference is often -20dBFS in the US, or -18dBFS by the BBC, using a 1000Hz pure sine wave. This is a little different than asking for an analog dBV (RMS voltage) reference, but the references in bold correspond roughly to +4dBu (0VU) and 0dBu (PPM4), respectively (where 0dBu = .775 VRMS).

Thus, if Rocky is seeing -101.5dB noise "floor" on a 0dBFS meter (such as in Vegas or Sound Forge), he is reporting -81.5 SNR in the US.

Although some shady equipment manufacturers use 0dBFS as a reference (this raising their advertised SNR), the practical approach is to leave some headroom for linear gain as you have noted.
* EBU R68 is used in most European countries, specifying +18 dBu at 0 dBFS
(Snagged from Wikipedia)
I can send you a pristine BBC test tone if you wish. It used to be a free download.
HTH

I've been using E-MU's 1212m sound card for years. great value !

Has decent converters for the two analog in/outs, but also has several digital inputs as well (SPDIF optical and coax as well as ADAT and AES capability).

You can add a word clock daughter card ( I use word clock to keep all my digital audio gear synced) to it, or when you are ready to expand to 5.1/7.1 a micro dock.

Check it out on E-MU's web site. For the price (around $169 street) the quality is amazing!

Used fantastic M-Audio for years in XP.

Bought a new 64 bit system with 64bit Windows 7 Pro.
No drivers to be had from M-Audio.

Dumped M-Audio and went (with much fear and trepidation from previous experiences) to Creative Titanium Pro X-fi.

No appreciable noise, great performance, much surprise here and I am very happy with it.

Presonus Audiobox USB works well for me

musicvid

Perhaps we should not bore others by prolonging this. I was just seeking clarification what your reference level was. If it is full scale then what you are measuring is the dynamic range, but this gives you no indication of signal to noise ratio for a given microphone, which is what I was describing. Both things are important.

If you are recording a rock concert at 100+ dB sound level, then noise may not be much of an issue, but with an orchestra, say, you may need both wide dynamic range and low noise.

I was looking for comparative objective data because most comments so far in this thread have been subjective. You can't properly compare different products assessed by different people that way.

My Presonus Firebox with 24 bit sampling has good dynamic range but the noise expressed as a voltage at the mic input (dB re 1 mV or whatever) is worse than that of my cheap Realistic mic preamp - rather disappointing! :(

Peter, no offense meant.

I agree with you completely that noisy pre's are totally unwanted in an audio interface. As DSE pointed out years back, that's hardly the only consideration.

If your personal frame of reference revolves around analog scales and measurements, that is entirely understandable. I spent a good portion of my post-college years in a tape studio.

However, I think this discussion is about the digital side, which is a top-down world, and one in which the only contextual relationships to SPLs, RMS, voltages, power, and the like (analog terms) are relative at best, confusing, and sometimes untranslatable.

I tried above to establish the keystone relationships, although somewhat ambiguous among regions, for informational reasons. Several indicators in your last response show that was not helpful, so I agree that there is no value in pursuing it further.

OK, enough is enough. Thanks anyway. I'll just point out that the noise that I am talking about originates in the analog circuitry (mic preamp) and that microphones are analog, so there is no escaping analog concepts when considering the noise performance of audio input devices.

I was staying out of this but I have to reiterate.

The preamp before the conversion hardware, if it has reasonable gain, in a reasonably quiet system, will completely determine the signal to noise in the system. What comes after the preamp doesn't matter if the 2 assumptions I stated are true. This is because the noise contribution to SNR in succeeding stages is reduced by the gain in the preceding stages.

This is true whether it is an audio system or an RF receiver. The noise figure is set by the first stage of gain and that stage's interaction with the signal source.

rocky

"I was staying out of this but I have to reiterate."

With my apologies Rocky, I see now that I totally missed your response exactly eleven posts up. If I had seen it, I think I would have stayed totally out of the rest of the discussion. It is entirely correct and to the point.

The basic question of quantifying noise levels in A->D environments could have easily ended there, without any additional input.

Here's that BBC Test Tone (-18dBFS) if it will be of use:
http://shell.dim.com/~musicvid/18dbfsstereotone.wav
(Right-click and "Save Link As")

WOW - this turned into a interesting conversation!

As I have mentioned in several debates with users in forums at Cakewalk, Cubase etc... technically speaking there is little difference between a SB Blaster and a higher priced 'Pro Audio' interface. Open one up and have a look.... common converters are produced by Cirrus Logic and or Asahi Kasai - AD & DA converters have been around for a long time now... there are few that would outshine others. In fact Mackie's new Firewire interfaces use the same Cirrus Logic converters that the Sound Blast XiFi use.

Where these "pro" interfaces put themselves into a different product range is the use of their Pre Amps... which again if you were to open a few different interfaces up you will find that many use a similar Integrated Circuit Op Amps - some of slightly better than others but not very far off in terms of quality. But for me I am using my Joe Meek SixQ mic pre amp which uses the highly acclaimed Burr Brown Op Amp with an onboard digital converter (sending the digital audio out to the SB).

The one other area is the DSP of each interface.... I guess here some processors on some cards may be better than others (i.e MOTU's DSP with CueMix) but often the results are in par with each other. And the since the release of the Sound Blaster Live! card - Creative Labs has employed the EMU 10K1 chip (now using the EMU20K1 chip in the XiFi cards) which has never been a slouch in that dept. (note that is was designed EMU).

And drivers... well yeah this one dept that can make or break an interface. PreSonus drivers have always giving me trouble - esp. in the clock area. My MOTU - no driver issues... going back in time... my Echo Audio Mona - no issues, my Tascam US-428 and US-422 no problems - my Steinberg VSL-2020 huge problems... my M-Audio interfaces (sorry guys) huge problems (granted the last M-Audio interface I used was back in 2000, things could have improved since)... my Sound Blaster cards - no driver issues... very solid (I have had over 20 diff. interfaces some good in the driver dept. some bad).

My point is that the Sound Blaster is often looked down upon by so called 'Audio Professionals' - yet if you are to really look at their guts and what is put into them - you will find very little diff. between a Sound Blaster and a Pro Audio interface. It's marketing.... Creative is aware that there is 'Pro Audio' market - but why worry about it when they have the 'gaming/consumer' market all wrapped up. Same with the "Pro" interfaces - they are never going to market them as a 'Consumer' grade devices... no they want people to think they are 'Pro' devices so the 'pro' users will buy them. It's all how the device is marketed.

As for the discussions on Noise Level.... never go directly with what the vendor states in it's specs. If anything that would be the noise level of the converter used and not the noise from the Op Amps and any other gain stage (which adds far more noise) before it reaches the converters, and the specs are typically from the converter's manufacturers - so they would have no way of knowing what other factors of the design would create noise.

If you want to test the quality of your interfaces converters - then start by using a line level - far less gain being applied these inputs and yes - record the input with nothing connected - this will be your hardware's noise floor. Do another recording with the pre amp input... but again remember different gains stages will create different noise floors - as to does the impedance and mic that is connected to them.

There is also the discussion about the referenced level of the converters. Often +4dbu in the analog world will equal -18DBFS at the converter level. This follows the traditional views of recording... giving you head room for the peaks that may clip at 0DBFS (around +22dbu). You want your nominal levels to be a +4dbu (often marked as 0 on most outboard pre amps - to get this headroom. Now not every manufacture will use the same reference level - there is a so called standard for the broadcast industry but not always followed. My MOTU ref level is -12DBFS. Also with my Joe Meek Mic Pre, the reference level is -12DBFS when using the digital output of the mic pre.

But advertised SNR will of course use 0DBFS as the reference - thus giving the illusion of a 12 or 18 db improvement on paper - but really most should be recording with nominal levels at -12DBFS or -18DBFS. And as somebody mentioned, if you are recording quiet passages (voice overs, soft music etc) then yes the noise floor is important - if recording some heavy rock or pop music - not so important.

Food for thought....

Cheers,
Chris

Creative's pro audio cards are sold under the E-MU brand and are actually very, very good.

http://www.emu.com/products/welcome.asp?category=505&