-70db noise floor - decent?

[wearashirt]

I have -70db white noise, noise floor from my physical mixing console. Question is - is that bad? I only realized it today, even though I've already had a track mastered from my analog process and I didn't seem to hear anything distracting from the result. I use a 1990 Soundcraft Spirit Live and an aggregated Behringer 1820 + Scarlett 6i6.

To compare, I was recently without my console (had a blown IC replaced) and had to mix ITB*. I suddenly realized that an ITB mix with my speakers cranked, there was no such white noise.

I'm suddenly paralyzed that I may be over my head with this analog thing...do I accept the -70db white noise? I notice it only comes on when I turn the channels on (i.e., bus them out into the L-R stereo master bus). If they're off, the inputs into my audio interface read a tad above -90db.

Can someone educate me on the mixing noisefloor of other vintage studio mixing consoles? (The Spirit LIVE has a counterpart called Spirit STUDIO).

There really is something truly different about mixing analog. Not necessarily better than ITB, but when I summed my mix* into the mixer, I felt I had to redo a bunch of EQ profiles, compression, and reverb. Some were drastic. I realized that this physical mixing console presents audio in a weirdly different way. Like, some elements can be made louder without necessarily getting in the way of another element. But then again, maybe it's just my imagination.

Are we better today with zero noisefloor vs the old days of tape hiss or suboptimal mixing consoles of B-list studios?

Would love to show you guys but how do I attach a photo?

[James Perrett]

Have you matched the desk output level to your audio interface's input? You'll end up with the best noise performance if 0dBFS is set to around 1dB below your desk's clipping point.

According to the specs in the manual, the RMS output noise with all channels routed and line inputs at unity gain terminated by a 150 ohm resistor is -81dBu. The maximum output level is +21dBu which will give you a signal to noise ratio of over 100dB.

However, it appears that your Behringer interface can't handle the full output of your desk but the Scarlett can if you turn the gain control to minimum so probably best to discard the Behringer and just stick with the Scarlett. Learning about matching levels and calibrating them is absolutely vital in the analogue world.

[wearashirt]

Have you matched the desk output level to your audio interface's input? You'll end up with the best noise performance if 0dBFS is set to around 1dB below your desk's clipping point.

This sounds something that was intuitive to me from the start - my signals need to be loud enough to beat the noisefloor. Is that right?

I've done two official mixes on the board. One mix peaks at -6, and another peaks at -12. Does this mean my mixes need to get up to 0dB on my interface input, just shy of 1db of the clipping point of my desk? I've thought of this, but I worried about my inputs..

Alas, the reason my IC blew was because I from inside the DAW, I bussed together two different kick sounds that add together during the chorus of the song -- going into a mono channel on the mixer. So I also have to worry about being too hot.

I will research on level matching. Thanks for that, James and for always replying in my various questions.

[James Perrett]

After looking at the specs for the various devices, the simple thing to do is to turn the gain right down on the Scarlett and go from the line output on the desk to line inputs 1 and 2 on the Scarlett. That will get you within a dB or so of the correct level.

The Behringer simply isn't designed to be used with devices that output professional levels and the Scarlett inputs 3 and 4 will also distort well before the desk runs out of steam.

NB - I'm assuming you have a 2nd Gen 6i6. I can't find the specs for the first generation.

[wearashirt]

The Behringer simply isn't designed to be used with devices that output professional levels and the Scarlett inputs 3 and 4 will also distort well before the desk runs out of steam.

Is this gleaned from the max input gain / impedance indicated?

[Mike Stranks]

Once you've got your gain-staging sorted you may grab a few more dB... but -70 from a mixer of that type and vintage is nothing to sweat over.

Back when I started in this caper back in the early 70s, in the serious amateur/semi-pro world we were ecstatic if we could get an end to end noise floor lower than -45dB...

[James Perrett]

The Behringer simply isn't designed to be used with devices that output professional levels and the Scarlett inputs 3 and 4 will also distort well before the desk runs out of steam.

Is this gleaned from the max input gain / impedance indicated?

Yes, that's from the max input spec for the Behringer which is +11dBu and inputs 3 and 4 on the Scarlett which can handle up to +16dBu. Inputs 1 and 2 on the Scarlett can handle +22dBu. The max output for the Soundcraft desk is +21dBu and the distortion figures are quoted for an output of +20dBu which means that the Soundcraft is happy to work at high levels (though you would normally aim for an average level of +4dBu to give plenty of headroom for peaks).

[Hugh Robjohns]

Is a noise floor of -70dBFS decent?

As Mike says, its adequate for many musical genre, but its not what I'd call decent.

You should be able to achieve an analogue noise floor of well over -95dBFS with the correct gain structure in the interface (and console).

James has made some good recommendations which will help a lot with the mixer-interface gain structure, and with that corrected you'll be better placed to optimise the console gain structure too.

[wearashirt]

Yes, that's from the max input spec for the Behringer which is +11dBu and inputs 3 and 4 on the Scarlett which can handle up to +16dBu. Inputs 1 and 2 on the Scarlett can handle +22dBu. The max output for the Soundcraft desk is +21dBu and the distortion figures are quoted for an output of +20dBu which means that the Soundcraft is happy to work at high levels (though you would normally aim for an average level of +4dBu to give plenty of headroom for peaks).

To add, of course these interfaces have pad attentuation options. Will that help me out, or is that proper to use in this case?

It's not in the manual, but some guy said the pad on the Behringer umc 1820 is -15db. INST is +18db. Do they add together for a total of +33db of max. input?

For the scarlett, I'm not sure whether the max input levels indicated take into account pad attentuation, since enabling the pad is done inside their Mixcontrol software. If so, then the Scarlett with Pad enabled is up to +32db of max input.

[wearashirt]

You should be able to achieve an analogue noise floor of well over -95dBFS with the correct gain structure in the interface (and console).

Sigh, well I guess that's well beyond my league. Are professional consoles able to achieve that? -95dbFS noise floor while the mix soars comfortably at -10 or even 0db?

[Arpangel]

Back in the 90’s I was using a Soundcraft RacPac, and an M-Audio Audiophile.
I never noticed any noise intrusion, to hear noise I literally had to turn the gain right up on my monitoring.
Hugh is right, you should be getting better performance than this, gain structure is the key, even with all faders raised on my mixer noise was never an issue in normal circumstances, that is, most of the time, and our music was very dynamic, if there was any major noise it would have shown up immediately.
One caveat, I now have a Behringer 1820, and it’s not the quietest of interfaces, but even so, it should be possible to get a decent performance from it.

[Hugh Robjohns]

Are professional consoles able to achieve that? -95dbFS noise floor while the mix soars comfortably at -10 or even 0db?

Yes.

Check out this comparison:

The section on the left shows the SMPTE's recommended analogue-digital alignment. It is based on the idea that the highest peak signal level a professional analogue is intended to handle before clipping is around +24dBu. The analogue console noise floor is going to be somewhere between -90 and -100dBu, depending on the console quality and the number of routed channels etc.

Let's take worst case and go for -90dBu which means the console's dynamic range is 114dB (90+24). Pretty much all modern budget A-D converters can manage 114dB dynamic range and most will be around 116dB or so. The very best are up in the low 120dB dynamic range area these days.

So recording from an analogue console with a correctly aligned converter will show the console's analogue noise floor at around -114dBFS or maybe a little lower. If you peak your music to -10dBFS the noise floor is still over 100dB below the peak music level!

In contrast, the section on the right of the diagram above shows the equivalent with your set up. I've lowered the maximum analogue level to +21dBu, the noise floor up to -81dBu, and the converter's peak level to +11dBu, all as per James' figures above from the specs he found.

As you can see, the dynamic range of the Spirit console is lower than the SMPTE expectations at 102dB (81+21) rather than 114dB -- but that is expected given the modest cost of the console. Nevertheless, 102dB is well within the dynamic range capability of the most basic converter.

Sadly, though, because the Behringer interface's converter's peak level is only +11dBu -- ten decibels below the peak level of the console -- you're inherently throwing away 10dB of dynamic range and raising the apparent noise floor level just through that misalignment. So you're actually only capturing, at best 92dB of dynamic range in the digital domain.

In reality, though, that's still pretty good, and way more than most reel-reel tapes could ever manage!

In theory, by the numbers, the captured noise floor from your console should end up around -92dBu when using the Behringer interface, but that assumes line level inputs to the console and unity gain throughout... Clearly, that's not the case, and as you're running mic preamps with high gain the console's noise floor will be higher. And if you're peaking your music to -10dBFS it will appear 10dB higher still...

So a -70dBFS noise floor from the console is not an unreasonable figure under the circumstances. Optimising the gain structure into the interface (or using a different, better-matched, interface) would improve matters considerably. And you may find further improvements by optimising the gain structure through the console itself.

Hope that helps to clarify the situation, and highlights the importance of correct analogue-digital alignment in a converter/interface. It is called audio engineering for a reason...

[wearashirt]

I would love to ask more questions, but of course I have to be respectful of everyone's time as well as my own reputation around here..

I have an idea to improve my set up, if you folks can also chime in. Basically, I have to maximize my line levels to beat the noise.

1. Maximize the input level coming into my mixing console (max input of 10db mic input, 30db for line input), which is lower than max outputs of my interfaces (16.5 dbu for behringer, 11 dbu for scarlett)*

*How do I know what dbU I'm at, based on the output meters inside my DAW?

2. Build up a mix.
3. Grab more clean gain from the tops of my channel strips.

This should yield a final mix that hits the orange at 0 to +6db. (The red LEDs come up to 15db, all seen printed on the mixing consoles chassis). Hopefully, it meets the max input gain of any of my interfaces.

[Hugh Robjohns]

*How do I know I'm at 16.5 dbu, based on the output meters inside my DAW?

The manuals for your interfaces will state somewhere in the specs what analogue level is expected (input) or delivered (output) for a 0dBFS digital signal.

Once you know that you can work out what any digital value in the DAW equates to in the analogue world from each interface.

For example, the Behringer UMC 1820 has a maximum (0dBFS) output level at the main and line outputs of +16dBu.

So if you wanted a signal level of +4dBu, you'd need to be seeing -12dBFS on the DAW meters (16-12 = 4)

[cashhewn]

In order to thoroughly flog this late horse:

So that I am perfectly clear on the concept, if I want to “measure the noise floor” of my analog console (just because, ok?), do I meter the output of the mix bus with the master fader at unity or at minus infinity?

Without any channels engaged (i.e. nothing is fed to the mix bus) and the master fader at minus infinity (all the way down) I read around -96dBfs via RME’s DIGICheck at the UFXIII’s AES input via the ADI-2FS converter that is converting the analog signal from the console mixbus. Conversely, all things equal except the console master fader at unity, I get around -70dBfs.

At the end of the day, I’m making records on this thing and about to release one, and I think the results are fine. Can you hear some analog hiss or noise floor in the milliseconds before a track begins? Yes you can. Has it ever presented itself as a problem to me artistically? No it has not. Heck me and the mastering engineer mastered this record I’m about to release to a 1/2” Mara machines deck for some tape benefit (really seemed to roll off and soften the highs in a fantastic yet subtle way), so this is just to say there are “degradations” all over the place and I am happy with the results. I love early seventies music.

In any case, I’m just curious about me ol’ console.

[Hugh Robjohns]

So that I am perfectly clear on the concept, if I want to “measure the noise floor” of my analog console (just because, ok?), do I meter the output of the mix bus with the master fader at unity or at minus infinity?

If you close the master fader you'll only be measuring the noise from the mixer's output driver. With the master fader at unity you'll be measuring the noise from the output driver, plus the mix-bus amplifiers, plus the contribution from however many channels are routed to the mix bus.

Which scenario best reflects normal use?

Without any channels engaged (i.e. nothing is fed to the mix bus) and the master fader at minus infinity (all the way down) I read around -96dBfs... Conversely, all things equal except the console master fader at unity, I get around -70dBfs. ..

That sounds quite high to me... but remember there is no weighting on that measurement and its a sample-peak reading, not an average. Critically, to be meaningful at all we also need to know the analogue-digital alignment — what analogue signal level corresponds to 0dBFS.

A decent, modern, small analogue desk (>16 channels) should have an averaged noise floor below -100dBu, or thereabouts, and a bigg'un (say, 32 channels) about -86dBu. (Those figures come from the SSL Big Six and Audient ASP2024, respectively).

To make the numbers easy, if it was -106dBu with all channels routed, and with a clipping level of +24dBu, that gives a total usable dynamic range of about 130dB for a small desk, and maybe 110dB for a big desk (measured from line inputs).

So, if analogue clipping is aligned to 0dBFS, you should see a noise-floor around -130 up to about -110dBFS, and even lower if the main fader is closed — based on averaged, not peak noise but your measurements are based on peak meters

White noise crest factor (the difference between average and peak level) is about 12dB... so if you're measuring with sample peak meters expect numbers about 12dB worse than my theoretical ones above.

Which means real world electronic noise floor readings between -118 and -98dBFS sample peaks.

Even if you knock off 20dB for a small budget desk's higher noise-floor then you should still see -110dBFS average or -98dBFS peak... so -70dBFS still seems particularly poor for electronic console noise.

So, first, check your desk/interface alignment — if clipping levels aren't aligned the figures will be worse... and second check the channel input conditions. Obviously, if the preamps are max gain the noisefloor will be higher, and if the inputs are unterminated, higher still. Same with the number of channels routed to the output.

...and if you're recording with mics the ambient nousefloor may well still establish the dominant noise floor.

...so this is just to say there are “degradations” all over the place and I am happy with the results. I love early seventies music.

An era of hiss-a-go-go thanks to multitrack recording, but records still sold!

If you're happy with your setup why mess about with pointless measurements?

[James Perrett]

What is your working level?

I've just been using my A&H console so I've done a very quick test...

0dB on the desk meters is -12dBFS (possibly a bit high but I tend to use the computer meters rather than the desk meters)

Noise Levels -
With one channel routed and fader up - set to line input -84dBFS
With one channel routed and fader up - with mic active (studio noise floor) -64dBFS
As above but fader down -85dBFS
With output fader down -89dBFS

These are all peak levels

[Hugh Robjohns]

I've just been using my A&H console so I've done a very quick test...

0dB on the desk meters is -12dBFS

Noise Levels -
With one channel routed and fader up - set to line input -84dBFS

I have a feeling James' desk is a Sabre, which specs a one channel to output noise of -77dBu and a clipping level of +27dBu, giving a dynamic range of 104dB. Knock off 12dB for the crest factor gives a nominal sample peak noise-floor around -92dBFS.

0VU is +4dBu on that console, so knock off another 11dB for the clipping level alignment mismatch and I'd expect a peak measurement around -81dBFS — so a real world measurement of -84 is good.

[ef37a]

My head hurts!
But cashewn said "Alas, the reason my IC blew was because I from inside the DAW, I bussed together two different kick sounds that add together during the chorus of the song -- going into a mono channel on the mixer. So I also have to worry about being too hot."

Now I am pretty sure the chip blowing was a complete coincidence. However it might have been the result of a static charge or an earth leakage current. "Modern" op amps are not damaged even running clipped into a short (but not to be advised at home folks!).

I have read a few forum reports of the Spirit mixer series having a rather feeble power supply?

Dave.

[Hugh Robjohns]

My head hurts!

Sums can do that!

But cashewn said "Alas, the reason my IC blew...

The OP, wearashirt did... four years ago.

[cashhewn]

Thank you for your help Hugh.

That sounds quite high to me... but remember there is no weighting on that measurement and it’s a sample-peak reading, not an average. Critically, to be meaningful at all we also need to know the analogue-digital alignment — what analogue signal level corresponds to 0dBFS.

I’m using an RME ADI-2FS to convert the output of the analogue console and that device is set to +19dBu, so I’m assuming that means I’m aligned for input to the DAW to be +19dBu = 0dBfs…

A decent, modern, small analogue desk (>16 channels) should have an averaged noise floor below -100dBu, or thereabouts, and a bigg'un (say, 32 channels) about -86dBu. (Those figures come from the SSL Big Six and Audient ASP2024, respectively).

I’m using (albeit refurbished and a little modified) Soundcraft 600 Series 16 channel 8 group console, and here are the original specs from the manual regarding noise (for the 24 channel model):
NOISE: (measurement bandwidth 20kHz, ref +4dBu)
Mic equivalent input noise, 150Ohm source res= -128dBv
Group mix noise 24 channels routed= -85dB
Group mix noise nothing routed= -98dB
Aux noise (master gain = 0dB)= -83dB

How do I measure this empirically?

And for maximum balanced output at master bus= +26dBu

To make the numbers easy, if it was -106dBu with all channels routed, and with a clipping level of +24dBu, that gives a total usable dynamic range of about 130dB for a small desk, and maybe 110dB for a big desk (measured from line inputs).

So, if analogue clipping is aligned to 0dBFS, you should see a noise-floor around -130 up to about -110dBFS, and even lower if the main fader is closed — based on averaged, not peak noise but your measurements are based on peak meters

So what do I make of this? Using the group, all channels routed figure, I reckon I have 111dB dynamic range (-85dB noise floor and +26 dBu desk clipping point).

However due to the ADI-2FS clipping at +19dBu, I am not aligned and I’m losing 7dB there I reckon.

White noise crest factor (the difference between average and peak level) is about 12dB... so if you're measuring with sample peak meters expect numbers about 12dB worse than my theoretical ones above.

DIGICheck has both peak and RMS figures and, yes, I think in the OP I was giving peak.

Which means real world electronic noise floor readings between -118 and -98dBFS sample peaks.

I’ll have to look again but yeah I think what I’m finding is well worse than this. In the order of around 20dB worse.

Even if you knock off 20dB for a small budget desk's higher noise-floor then you should still see -110dBFS average or -98dBFS peak... so -70dBFS still seems particularly poor for electronic console noise.

Hmm and if you knock another 12dB for RMS we’re getting close to what I’m seeing.

So, first, check your desk/interface alignment — if clipping levels aren't aligned the figures will be worse... and second check the channel input conditions. Obviously, if the preamps are max gain the noisefloor will be higher, and if the inputs are unterminated, higher still. Same with the number of channels routed to the output.

Right, we know the alignment is 6dB off. Preamps are dialed all the way down.
Inputs unterminated: I’m forced to show my ignorance now. The inputs of the desk are in line mode and are normalled from the tape returns which are normalled via patch bay from the UFXIII and a Focusrite Clarett…is this terminated or not? I do not understand termination properly from a circuit point of view sadly.

…and if you're recording with mics the ambient nousefloor may well still establish the dominant noise floor.

A bit of a red herring but an excellent point!!

If you're happy with your setup why mess about with pointless measurements?

Because I’m an admitted fool who wants to understand if my console is in technical decent shape or not after spending 4K to acquire and refurbish it possibly.

[cashhewn]

What is your working level?

I've just been using my A&H console so I've done a very quick test...

0dB on the desk meters is -12dBFS (possibly a bit high but I tend to use the computer meters rather than the desk meters)

Noise Levels -
With one channel routed and fader up - set to line input -84dBFS
With one channel routed and fader up - with mic active (studio noise floor) -64dBFS
As above but fader down -85dBFS
With output fader down -89dBFS

These are all peak levels

James, many thanks, had to step away from the studio but I’ll have a look at this when I’m back. Thanks again for the comparison!

[cashhewn]

My head hurts!
But cashewn said "Alas, the reason my IC blew was because I from inside the DAW, I bussed together two different kick sounds that add together during the chorus of the song -- going into a mono channel on the mixer. So I also have to worry about being too hot."

Now I am pretty sure the chip blowing was a complete coincidence. However it might have been the result of a static charge or an earth leakage current. "Modern" op amps are not damaged even running clipped into a short (but not to be advised at home folks!).

I have read a few forum reports of the Spirit mixer series having a rather feeble power supply?

Dave.

Forgive me, but I’ve never blown an IC and I don’t remember saying any of this, and I don’t have a Spirit, so I’m pretty sure I’m being misquoted here?
EDIT: that quote is the OP, not I. But this is all my fault for flogging that horse I mentioned

[ef37a]

My head hurts!
But cashewn said "Alas, the reason my IC blew was because I from inside the DAW, I bussed together two different kick sounds that add together during the chorus of the song -- going into a mono channel on the mixer. So I also have to worry about being too hot."

Now I am pretty sure the chip blowing was a complete coincidence. However it might have been the result of a static charge or an earth leakage current. "Modern" op amps are not damaged even running clipped into a short (but not to be advised at home folks!).

I have read a few forum reports of the Spirit mixer series having a rather feeble power supply?

Dave.

Forgive me, but I’ve never blown an IC and I don’t remember saying any of this, and I don’t have a Spirit, so I’m pretty sure I’m being misquoted here?
EDIT: that quote is the OP, not I. But this is all my fault for flogging that horse I mentioned

Oops! Please excuse this daft old bugger!

Dave.

[Hugh Robjohns]

I’m using an RME ADI-2FS to convert the output of the analogue console and that device is set to +19dBu, so I’m assuming that means I’m aligned for input to the DAW to be +19dBu = 0dBfs…

Yep. So, with the desk's clipping level of +26dBu, you're (theoretically) losing 7dB of headroom and raising the noise-floor by the same amount.

However due to the ADI-2FS clipping at +19dBu, I am not aligned and I’m losing 7dB there I reckon.

Yes, and around another 12dB for the crest factor because you're measuring peak rather than average level.

So, -70 -12 -7 = 89dBFS ... which is a much more reasonable noise-floor figure!

DIGICheck has both peak and RMS figures and, yes, I think in the OP I was giving peak.

Using the RMS option would make life a lot easier.

I’ll have to look again but yeah I think what I’m finding is well worse than this. In the order of around 20dB worse.

Yeah. I'd really expect the RMS noise-floor to be at least -90dBFS or better — assuming aligned analogue and digital clipping.... but I think we've found reasons why you're not there.

The inputs of the desk are in line mode and are normalled from the tape returns which are normalled via patch bay from the UFXIII and a Focusrite Clarett…is this terminated or not?

Yes...

...I’m an admitted fool who wants to understand if my console is in technical decent shape or not after spending 4K to acquire and refurbish it possibly.

The person to ask is the person who refurbished it as they should have the right gear to confirm it is meeting its original published specs.