How would you test for s/n and hum / buzz in your studio?

[harry]

Hey Guys,

I just finished recording some ideas at home and noticed that i could hear some hum in the mix..... it turns out that this is an earth loop / unbalanced to balanced problem caused by my monitor mixer... the audio itself is cleaner...

But - on inspecting some DI'ed tracks i found the noise floor to be at -48db (according to the daws scale) This is pretty crap right?

So - I'm not sure if it is the mic pre/di, or something else in the patchbay that is causing this...

So - how would you all test the S/N ratio and Hum Buzz of all your equipment?
I have a 16 channels of AD/DA conversion, and about 16-24 channels of outboard, plus heaps of synths and instruments...

The room has 3 powerpoints which come off a splitter that goes to its own fuse on the meterbox

I was thinking of disconnecting everything from the patchbay and measuring the S/N using wavelab or something and then connecting bits of gear one by one to see which individual bits or combinations of gear adds hiss, buzzing or noise....

I have a heap of 1:1 audio transformers (lundahl, sowter, carnhill) sitting idle that i could put into service as D'I boxes / Galvanic isolation ect...

What would you do guys?

[heathen]

Make sure everything is running from 1 power point, very important. Anything that's not can be connected to a balancing transformer box, which should work just fine. Though if it's RF it could be the patchbay, would be a good idea to test 1 piece of gear at a time too.

[harry]

Thanks Heath,

yeah i must try and plug in everything into one point on the wall... (its on a 20 amp circuit breaker and its a 4 socket point)

Not really a popular subject eh?

I wonder how rick goes through his setup to sort out all the buzzes and hums...

[heathen]

Just the simplest way I know of, I just daisychain my power boards from 1 outlet. It really should make a difference.

[rick]

i do it the time consuming - and nutso/ expensive way

nothing simple or easy to report i am afraid

first you start by freezing your cables , then get back to me and if our still feeling sane
i will give you the next step :)

[heathen]

When you did your cables Rick was it just a quick dip in liquid nitrogen or did you leave them in there a while to stew in thier own cryogenic juices? Did it make a perceivable difference? I'm just gonna stick with the cheapo method but I am interested.

[rick]

they went for a swim in liquid nitrogen from friday night through to monday morning
slow drop down to - hypo low temp slow rise up

does it make a perceivable difference ?

i dunno
probably.. ..probably not !
its done now so i cannot a/b test it
so lets just say i did it - its done , its not worse , i hope its better

seemed like a good idea at the time though :)

if you noise floor is -48db something is not so great
get it to about -70db and you all good

on a quiet day ( night only actually ) we are at -104db
most of the time -97db and rising
depending what i plug in it could go up to -65db

all unweighted figures of coarse

[heathen]

I read a little this morning on a company who does cryogenic connectors and cables and farout they are bloody expensive, I added 1 aussie connector, a 2ft cable and shipping, grand total for 1 power cable was $255usd.

Anyway even if it is no better or subjectively better then it still was worth giving a go.

To be honest with noise floor mine used to sit at around -92db- -89db in my old house, though now in the new place I was amazed when I noticed I had a floor of -102db, I think after patching a few bits in it went to about -96db. Power in the old place was all wired in the 1950's. Anyway I'm pretty happy with the new houses power.

[harry]

As much as i'd like to Rick, I think i'm a few steps short of 'freezing cables' here at Abbey Road in my bedroom

I did some more testing and found that with my convertors normalised to my patchbay and everything turned off except convertors and computer - wavelabs peak monitoring shows noise floor at around -74...

turning everything on (20+ bits of gear) does not change the result much.... it comes down to about -72...

The real culprit for my shitty -48 noise figure was the 'third' gainstage in my chinaneve pres.... but thats another story...


So - im wondering if my convertors are set to -10 or +4.... does this affect the 'meaning' of what i am reading on the wavelab peak meter and is this the way to measure things?

i wonder how a unit with a supposed s/n ratio of 93db on its inputs is -74 in the realworld with nothing connected...

Harry Arvanitis

[rob]

harry, measuring noise floors is a tricky business. And it can be difficult to be confident of the results you get with anything other than a proper audio analyser.

However, the things that occur to me are :
some computers are noisy and depending on the interface's relationship to computer, re power and grounding this can cause issues. Note that laptops running off their charger are notoriously bad. I have analyser software and an M-Audio USB interface that is useless for noise floor measurements unless it is running off the battery.
Also, remeasure your system with all audio inputs and outputs physically disconnected from the interface.

then the question arises...what sort of noise. When folks say to me, "hey Rob, I've got this doo-hickey and it has gone noisy". My first question is what sort of noise?
White, buzzzz, hummmmm or a noise you can't hear but shows up on the metering. Clarifying this yields some helpful information. Saying something is noisy tells me jack shit.

I've just fired wavelab up on my PC and with a PCI Julia soundcard and with nothing connected I have a unweighted noise floor of -95dBu. If I record that and then normalise it I get a random pulse train, which shows that the noise is fairly white, and indeed playing it back after normalising I can hear a nice loud but uncoloured white noise.

Also bandwidth is important, sample at 44.1, thus limiting the audio bandwidth to ~ 20kHz and see if this changes the noise floor. My noise floor as measured above gets 10dB worse at a 192 sample rate.

[harry]

Thanks Rob!

So much valuable information in your post.... lots to take in.

Heres what i came up with:

- Physically disconnecting all i/o (apart from firewire cables) did not change S/N at all
- Turning every bit of gear on as opposed to everything off only worsens S/N by 1-3db
- Changing the 44.1-96 setting makes no change at all.
- Moving the power transformer further away it did not improve the S/N and actually made it a db worse at one point (probably its proximity to the computer..)

I then recorded the noise in wavelab and normalised

It sounds like a combination of white noise and a bass tone (50hz hum?)

I then analysed it in a transcription program that can analyse audio segments and then show you the power of each harmonic expressed in a graph above a keyboard...

- this showed a series of harmonics coming of a very low G (roughly 50 hz range id say) then octave above, fifth above that and 4th above that....

- analysis in wavelab showed these harmonics with the lowest one exibiting the biggest amplitude,
but the real surprise were peaks of a far greater amplitude in the 20hz range..... (would this be caused by the steep cutoff filters ... ie a process of the A/D converison process?)

So based on this info i suspect that perhaps:

- my external power supplys's transformer might be inducing mains hum into the convertors (poor regulation)
- i might get an improvement by disconecting the earth sheild in the firewire cables that connect the unit to the computer? - maybe they are picking up some audio junk from the computer (it does have a spiffy looking super regulated power supply...)
- the noise is essentially made up of:

mains hum,
low freq (subsonic) noise from the a/d's cutoff filter
and white noise

The good news i guess is that i dont seem to have any earth loop related issues and all other power packs dont seem to add noise to the system...

I got the convertor second hand and the power pack is not a factory one.. i know that sometimes for our market local distributors source a power pack and they are usually crap....

so i guess if i can be bothered:
look at isolating sheild in firewire cables and see if there is an improvement?
find another power pack or build a regulated power supply...

or stop caring and make some music....

and on it goes....

[rob]

first up and most importantly don't let this get in the way of making music. I had client once who spent two years getting his home studio "just right" before he could allow himself to start recording anything. I remember 6 months went by at one stage and having not heard from him assumed he'd got cracking at his music career only to recieve an urgent call out because he'd found that his Avalon 747 was 0.25dB out of balance from left to right and this was preventing him getting on with it!
Some of my happiest recording and creative musical ventures were done with no regard to the quality of the equipment...domestic sony tape recorders, recording cassette walkmans, crappy consoles, no brand microphones.

however, you should be getting a better noise floor out of your gear...but then again i'm not sure if you've said what the interface is? It certainly seems you do have some mains hum in there. The sub 50Hz stuff may just be an artifact of the FFT analysis of the 50Hz component.
Is this non-original plug pack the right one for the interface? Does it supply an AC or DC voltage to the interface. Does the plug pack have an earth pin? Is everything plugged into one power outlet?

Whilst there are rules for correct grounding / wiring. There are also the odd times when the best solution results from breaking the rules and there have been times when i've resolved an earth loop type hum with the most unlikely wiring scheme.

If you are handy with a sharp knife and some side cutters, then cutting the shield off the firewire cable at one end, near to the connector may be worth trying, but you may just bugger up a cable for no result, as, unless you have a 4pin firewire socket then a power ground is also present as well as the shielding ground.

happy hunting

[Linear]

Hi,

Firstly, you need to resolve your monitoring. You should never have anything that appears in your monitoring chain that isn't on the 2-buss. Way to test that is to have nothing connected to the console, all faders down, all tracks muted. Turn up your monitors - if you hear anything (and I mean ANYTHING) then that should be sorted first.

After that, unless you can hear the noise in the actual track while mixing, then I wouldn't sweat it too much. Note that this includes intros to songs where you might have just an acoustic/bass/vocal. If it's noticeable there, then it's a problem. That's why it's handy to run a 2-buss compressor (something I never used to do), it helps realise any problems before it hits mastering (always be super conservative though). For me noise depends on what's connected/gain structure/number of tracks, I wouldn't bother with testing individual stuff.

I've had situations where there is godaweful noise while the track isn't running, but it's masked when everything is going. I believe the noisiest things I have are the analog verbs (plate/spring) and a few quirky boxes like the ursa major, lexicon 200 and lexicon prime time, there's no making those things quieter other than gating (which is another can of worms). If you can automate mute on console then great, but I can't so I don't worry about it.

Note that the only time you need to be super careful is during stems - if you don't physically mute tracks that aren't being used (ie only mute things in your DAW) then you'll end up with noise x 10, or however many stems are required to make the original mix up again.

So if you're being super pedantic, it's good to have sets of stems - TV mix/vox only, Mix without bass/bass only, drums only/no drums etc.

But noise (in my opinion) should only be a problem when you can hear it, not something that you fart about and waste time measuring with scientific instruments. Believe me, there are so many other things you should be worry about in a recording/mix situation...

Chris

[harry]

Lots of useful info in those last few posts - thanks Rob and Chris!

btw - i'm not scared of noise...... most of my old school gear has a bit of hiss.... and the swooshing sound my tape echo creates makes me feel happy...

i have been mainly demoing ideas and creating rough drafts, but some of the takes are looking to be keepers...
i want to get this all sorted as i quit work for 2 days p/w to stay home and record music over the next 12 months...

i checked the i/o on the soundcard (seperate to the convertor box) and it shows a noise floor of about -87bd - normalized, it's all white noise....

where i'm worried about the convertors is if there is a strong 50hz hum at -70db (it's like a bass note) and it's present in say 16 tracks of audio the noise (mains hum) will compound...
ie: 1 track - 70, 2 tracks = -67, 4 tracks -64, 8 tracks -61, 16 tracks -58

and if i need to boost the levels on some tracks it will bring it out further.... this could be a problem...

actually i'm cursing that i wasn't aware of this earlier - i spent the last year archiving about 500 records (i'm a dj) in 24/96 taking pains to balance the tonearm and use fresh stylusses and a nice 'rane' phono preamp.... ect...
the rips sound very nice and the s/n on a vinyl is like 40-50db...

Rob, i checked the powersupply against whats stated in the manual - should be a 12v AC 1.5A supply - mine is a 12v AC 2.5 A supply..

I don't know much about AC supplies (ie where to get a good one).... i have a couple here but they are not powerful enough to test em out...

would a jaycar / altronics one be the go??

harry arvanitis

[rob]

since the plug pack is supplying AC to the converters it is has far less effect on quality of the final power supply. 50Hz noise suggests a ground loop problem, 100Hz noise suggesst a power supply problem....this is a rule of thumb and so like all rules can be wrong, however is usually applicable

i'm confused about your setup. So you have a soundcard in the PC but a converter box outside the PC? How do these two things link? Are you using a digital input on the internal card? If so if it has an optical input...use it, If it has an RCA style SPDIF, get an old RCA to RCA lead and butcher it so that the shield is disconnected near one end.

see if that changes anything.

[waitup]

Having a power supply that can deliver too much current isn't a problem

[Futureman]

where i'm worried about the convertors is if there is a strong 50hz hum at -70db (it's like a bass note) and it's present in say 16 tracks of audio the noise (mains hum) will compound...
ie: 1 track - 70, 2 tracks = -67, 4 tracks -64, 8 tracks -61, 16 tracks -58

One for the brains trust... but is that correct?

I thought, yes, the 'hum' would get louder, but so does the whole track as you add extra tracks.. so the net result is that the 'hum' stays at say -70db

Regards
Mike

[Futureman]

one for the brains trust... but is that correct?

I thought, yes, the 'hum' would get louder, but so does the whole track as you add extra tracks.. so the net result is that the 'hum' stays at say -70db

Knock knock?
Anyone home on that one?
lol.

[rob]

depends

if we assume that each track is recorded at a different time then the 'hum' will be uncorrelated and thus will add in the same way as uncorrelated noise or indeed a lot of individual music tracks. This yields a 3dB increase for every doubling of tracks

2 tracks up 3db, 4 tracks up 6dB, 8 tracks up 9dB, 16 tracks up 12dB. etc