Noise Floor (how low is low)

[Mixxit12]

We have 2 workstations in our studio. Mac running a Pro Tools TDM rig. The other is a PC running various apps (including Pro Tools M-Powered). The PC is the box in question. Recently I've noticed more noise is the system. When I hit record on the PC I'm seeing the meters jump up to about -66db. Sometimes as low as -72db. It used to be a lot lower than this. However, the noise isn't really audible. The DAT machine shows about -60db. The TDM rig shows nothing on the meters.

What is average for you guys? Do I need to go hunting for the gremlin (presumably AC) or would you consider that an acceptable level? I tend to believe it's way high, but I want to be sure before we start looking for the needle in the heystack.

Thanks for the input!!

 

[mshilarious]

If we are talking acoustic noise floor, and I think we are, then you can't really express it in terms of dBFS, you just need to know the dBSPL. That's because the noise level in your software is dependent upon what you are recording. Drum overheads will be much louder than acoustic guitar, therefore the acoustic signal-to-noise ratio will be much higher, and the noise floor in dBFS correspondingly lower.

Anyway, a quiet room is 20dBSPL or less. That's going to be nearly impossible to achieve with 2 PCs in a room, probably you are closer to 40dBSPL, maybe worse. Still, you can work around that. Again, it's all about signal to noise ratio. It won't be an issue for loud sources. For quiet sources, you need to try to quiet down or isolate the PCs, move tracking away from them, or use directional mics pointed away from that.

I would suspect that your noise level is largely fan noise, not AC interference. If you have an FFT analyzer, look at the noise spectrum. If it's AC interference, it's easy to spot, because you can see the 60Hz peak, as well as all of its harmonics through the entire audio range as distinct peaks. It's actually the harmonics that are the problem, because the ear is very insensitive to noise at 60Hz.

The ear is much more sensitive to noise in the 1kHz to 4kHz range, which if it's fan noise, is relatively easy to block with portable absorbers.

If you do see AC interference, it's probably not ground loops as most people expect. Common noise sources are flourescent lights, dimmer switches, electric heaters, refrigerators. Eliminate those sources and see if your problem improves. If not, then you need to look into grounding of your gear (using a single circuit, checking your wiring, etc.), balancing cabling, and those sorts of solutions, which I would try first before resorting to power conditioning, which often doesn't help.

 

[SouthSIDE Glen]

Recently I've noticed more noise is the system.

I'm goiung to try this from a different angle and assume you're talking system noise and not ambent acoustic noise. In which case the responding question I would have is, "what do you mean by 'system'? i.e. what is the signal path defining the "sysem" you're talking about?

- 66dBFS of system noise is not absolutely god-awful, but it's certainly nothing to brag about and worth looking at improving. Assuming that noise os on input and actually winds up on the recordings (have you double-checked that?) that floor will be on every track you record and will build up with the number of tracks you have.

MSH brngs up some good suggestions as far as analyzing the noise, especially if it's AC RF bleeding into the signal chain somewhere from neon or a fridge condensor or something like that.

If external sources seem to be eliminated by the system noise is still there, I'd add another basic troubleshooting method; staring with just the PC alone and working you way back in the signal chain by adding one link at a time (including intermediate cables), check your noise level at each length of the chain and see if you can fine if there is a weak link like a noisy piece of gear or a particular cable which may be the major culprit adding the noise.

While doing this, make sure that you have the gain structure in each point in your chain set up properly, and that you don't accidentally have (for example) the gain on a mic pre or converter accidentally cranked higher than normal. If the gain structure looks normal, check the wiring quality and positioning around the suspect gear, as well as the connection jacks on the suspected gear itself.

EDIT: Here's something I thought of after posting:
Assuming you're using a halfway decent FW/USB/PCI interface and not a stock Soundblaster-class PC soundcard, make sure in your editor software setup that the soundcard has not accidentally gotten re-activated for input along with your main interface. The sudden jump in noise floor in your system could be cause by a sudden re-arming your your PC soundcard in your software.

HTH,

G.

 

[Nick98338]

Well, I just checked. Opened a track I recorded with an AT 3035. Gain set at the proper level to record a vocal track at about -15dbfs. The silence at the beginning of the track is about -60dbfs. The absolute bottom of the meter is -90dbfs.

 

[Mixxit12]

Sorry, I should have been more clear about the "noise floor" term. I am indeed talking about noise in the system, not acoustic noise in my room. Actually our live room is really quiet (assuming the studio next door isn't recording hip-hop).

I have tried some of the stuff that Glen suggested (thanks for the detailed info by the way). If I unplug the signal to the PC and drag the box away from everything, the floor drops to about -80. If I shut off our board but leave the snake plugged into the interface, I get about -74. So it appears that it's a mixture of stuff, the majority of it coming from our board. Unfortunately it's a 64 input board with a sizable patch bay. I really don't want to try to isloate each and every piece of gear we have in there.

The noise is in the input chain, but until we recorded a tune with 51 tracks this weekend, I didn't notice the hum. That said, it shouldn't be a huge deal but I would love some more input regarding lowering system noise. I'm definitely going to follow some of the suggestions from Mshilarious and Glen. Much thanks to both of you.

Any one else have any ideas?

 

[SouthSIDE Glen]

If I unplug the signal to the PC and drag the box away from everything, the floor drops to about -80. If I shut off our board but leave the snake plugged into the interface, I get about -74.

First, glad to hear that you're making some progress in troubleshooting the source(s) of the noise. It sounds like you're on the right track. And yeah, t-shooting the entire signal chain when it branches out to scores of paths is unreasonable .

I'm a bit intrigued by the above quote. A few qualifying questions I'd like to toss out there.

First, could you describe the board-to-PC interface with the snake? What actually are you using between the board and the PC (hardware, cable type, interface type, etc.)?

Second, in that first drop to -80dBFS you mention, does the drop happen when you unplug or when you move?

Third, did you move the position of anything relative to each other at or around the time the noise level increased? You might be able to attack the problem by simple physical arrangement change. For example, is your computer monitor position and/or the snake position relative to the monitor (or anything else with a possibly noisy RF field) possibly the issue?

Fourth, make sure as I said earlier that your standard PC sound card is not armed in your recording software along with whatever driver runs your i/f to the board. Even if you are getting external noise, the addition of unnecessary and unwanted noise from the additional sound card could be adding to what your getting from the rest to the point where either one of them could be tolerable, but the two together is pushing it up to the -60s.

For example (just a hypothetical), if your PC card were providing that -80dBFS background level and the rest of your chain the extra 24dB of noise, pushing you up to -66dBFS. if you got rid of the extraneous -90dBFS level from the PC card, then your system would only be pushing your floor 24dB up from the theoretical floor, which would be very acceptable.

G.

 

[Mixxit12]

Hey Glen, thanks for the quick reply and suggestions!

The noise falls off when I unplug the snake, the physical location has little to do with it, sadly. Just moving the box would have been a great solution.

The PC has an 8 channel snake running from the patch bay to the Delta 1010 interface (stop laughing). I would be inclined to blame the relatively cheap interface except it's never been a problem before. Obviously our TDM system's interface is a great deal better and it's not seeing the noise. It's also on the other side of the control room. Thoughts on that?

I will definitely check the PC's standard card. I'm not at the studio now so I'll have to check it tomorrow.

Thanks again Glen for all of your insight!

 

[mshilarious]

Hey Glen, thanks for the quick reply and suggestions!

The noise falls off when I unplug the snake

It's not unusual for an unterminated open cable to have that kind of noise. It's also not terribly important, unless you like to record open channels for no particular reason.

Once you plug a mic into a channel, of course then you have the acoustic noise floor of the room, and the self-noise of the mic to deal with, but underneath that, the self-noise of your interface will drop dramatically. It's just really hard to measure with a microphone attached. If you were so inclined, you could terminate a mic lead with a dummy 150 ohm load and see what that does to your noise.

 

[Mixxit12]

Thanks for the info mshilarious! Regardless of what input (whether it's one or all) is assigned to the PC we get basically the same reading.