question about tracking too hot......

[miroslav]

My point on the OP's question (assuming it's the same as what you just posted) would be that if his front end's self noise is (for the sake of argument) -70 while tracking at -12dBFS, you can bet that it'll be -60 (or higher) at -2dBFS (plus distortion, minus clarity, etc.).

.............

I've had dozens - DOZENS of projects wrecked from the word "go" because of tracking too hot ...

OK...I understand what you are saying here...but wouldn't that mainly apply when there was an unusual amount of make-up gain being employed to "artificially" raise the input/output level of the front end in order to get a "hot" track in the DAW?
IOW...what if your source was loud enough so that the front end was getting and putting out a nice, hot signal even when its controls were set to nominal gain...
....would you then turn down the gain just to bring the level in at -12dBFS or would you leave it as-is if it was peaking at say...-6dBFS...?

In any case - I'd rather "need to" add 20dB of digital gain than have to reduce it 2dB.

Now this I'm not getting.
Once it’s in the DAW...what difference is there raising or lowering the digital level?
Isn't it the same math?

 

[Massive Master]

"It depends" -- If it's a remarkably good preamp, I might leave it alone. If it's questionable, I'd pad it. If it was a very "smooth & steady" signal - A heavy guitar, synth pads, etc., I'd almost definitely pad it. High low-frequency transient (a typical drum) maybe leave it alone. High high frequency transient (triangle, finger cymbals, tambourine), mega-pad it.

The other thing -- What I meant was that I'd rather a signal be tracked 20dB "too low" into the DAW (requiring 20dB of additional digital gain) than have it come in too hot. Not because of the math - Because of the "purity" of the original analog signal.

And if anyone wasn't clear on the whole point here (I've had a PM or two), this is all about the analog signal and capturing it as "ideally" as possible - Once it's captured and digitized, I couldn't give a rat's rear end about levels as long as they're not clipping.

 

[dastrick]

Oh God, I don't even know anymore...

My point on the OP's question (assuming it's the same as what you just posted) would be that if his front end's self noise is (for the sake of argument) -70 while tracking at -12dBFS, you can bet that it'll be -60 (or higher) at -2dBFS (plus distortion, minus clarity, etc.).

In any case - I'd rather "need to" add 20dB of digital gain than have to reduce it 2dB.

I think it was Ronan - and I think it was in this thread - who makes the same case -- I've had dozens - DOZENS of projects wrecked from the word "go" because of tracking too hot (I'm working on one at his very moment). I've never - NEVER had one come in that seemed compromised in any way by tracking "too quietly."

On a personal side-note, one of the coolest sounding recordings I've ever made peaked at maybe -30dBFS and didn't have one individual track that peaked above -40dBFS. That wasn't by design - That was a "one-shot / take your best guess because you have no second chance to hit RECORD again" recording. There was no noticable 'noise' on it, there were no artifacts of 'not getting enough gain' etc., etc. No doubt - If I'd have had the time, everything would've been peaking at more typical levels (somewhere between -20 and -15dBFS is where I like to be peaking). But it wasn't in the cards and I couldn't change a thing after the recording started.

But I wasn't worried about it either... Not for a second.

My question has been answered (and then some). And the answer is "no, it is not the same". Once my signal is in my DAW, I can pretty much raise the gain as much as I want (not that I would) and it wouldn't matter unless it clips. What I want to be careful of is overloading my mics, preamps, and the signal going into the converters. I get it. Thanks to everyone for taking the time.

 

[Tim Gillett]

Oh God, I don't even know anymore...

My point on the OP's question (assuming it's the same as what you just posted) would be that if his front end's self noise is (for the sake of argument) -70 while tracking at -12dBFS, you can bet that it'll be -60 (or higher) at -2dBFS (plus distortion, minus clarity, etc.).

In any case - I'd rather "need to" add 20dB of digital gain than have to reduce it 2dB.

John,
You're right. In practice the snr from a mic/pre combination will not be as good as a reasonable A/D. My bad.

Cheers Tim

 

[Ethan Winer]

But in the analog world, running your preamps at +20db won't sound the same as running them at +4db.

Yes, gain staging matters with analog, but not as much as many seem to believe. I would never run anything at +20 because it's not necessary and just wastes electricity and makes gear run hot aging the components faster. I think a good approach is to set your mixer so outputting around 0 to +5 or so hits digital zero at the converter inputs. But anything within +/- 10 dB of that is fine, especially when there are no transformers in the path.

Understand that analog gear is linear. If a preamp sounds noticeably different at +10 versus -30, then something is wrong with the preamp.

--Ethan

 

[Farview]

Yes, gain staging matters with analog, but not as much as many seem to believe. I would never run anything at +20 because it's not necessary and just wastes electricity and makes gear run hot aging the components faster. I think a good approach is to set your mixer so outputting around 0 to +5 or so hits digital zero at the converter inputs. But anything within +/- 10 dB of that is fine, especially when there are no transformers in the path.

This assumes that anyone has the ability to calibrate their converters. A lot of people on this board don't. A good number of them are using the preamps built into the converters...

Understand that analog gear is linear. If a preamp sounds noticeably different at +10 versus -30, then something is wrong with the preamp.

--Ethan

That depends on the preamp. Neve preamps sound significantly different when you push them a little. And that is a good thing, if that's what you are going for. Of course Behringer toob preamps do too, but not in a good way.

I agree, a well designed clean preamp probably won't sound different. But I get the idea that most people around here are not using really well designed preamps, most are using cheap preamps that come on their interface or on a $200 8 channel mixer. These do tend to get pinched sounding when pushed (as would anything with a wallwart power supply).

The people around here who are lucky enough to get some great preamps are getting the Neve's and API's that do have transformers that will make them sound different at different output levels.


It's obviously very easy to optimise a specific signal chain if you know what you are doing and you know the limitations of all of the gear being used.

If you record to tape and dump to digital, there is one answer.

If you use very clean, very linear preamps into converters that you can calibrate, there is another answer.

If you use an all-in-one interface, that's another answer, etc...

I'm not sure I get why people get so bent out of shape about 'safe' advice. Is it really that bad to run everything at line level? Is that really bad advice worthy of debunking?


The actual answer to the OP's question is: It depends.

IF you are recording soft syths or anything else that never existed in the real world, no it doesn't make any difference whether it was recorded at -12 and normalized or just recorded at -2.

If you are recording an analog source, it might make a difference depending on the equipment being used in the analog domain.

Most of the rest of this thread is just people arguing over the minutest of details. Even then most of those arguments have more to do with the type of equipment used by and workflows of the people involved in the arguments. Everyone is right, in the scenario that they generally work in.

 

[miroslav]

Everyone is right, in the scenario that they generally work in.

 

[noisewreck]

I almost feel like I am in the Cave!

OK, here's a twist, a monkey wrench a coconut and a hubcap or two into this whole debate/mess...

Since many of us are using audio interfaces that combine preamps and converters into one box, and since many of these contraptions only use peak metering using digital meters that by nature are fast enough to follow every nook and cranny of the incoming audio, which basically means there are no RMS metering facilities, how does one figure when they're around -18 or -15 dBFS RMS? It's not so difficult with sustained sounds such as synth pads and violins, however it's far more murky when dealing with material with lots of transients such as drums, other percussion instruments, plucked instruments, etc.

Then what? Do you then say: "I am just gonna track so things don't peak past -12dBFS so that I don't cook my preamps"?

 

[Farview]

I almost feel like I am in the Cave!

OK, here's a twist, a monkey wrench a coconut and a hubcap or two into this whole debate/mess...

Since many of us are using audio interfaces that combine preamps and converters into one box, and since many of these contraptions only use peak metering using digital meters that by nature are fast enough to follow every nook and cranny of the incoming audio, which basically means there are no RMS metering facilities, how does one figure when they're around -18 or -15 dBFS RMS? It's not so difficult with sustained sounds such as synth pads and violins, however it's far more murky when dealing with material with lots of transients such as drums, other percussion instruments, plucked instruments, etc.

Then what? Do you then say: "I am just gonna track so things don't peak past -12dBFS so that I don't cook my preamps"?

With drums and other things with giant transients, you can't get them to average at -18dbfs without clipping. In that instance, just get your peaks around -6dbfs or so and hit record. The main reason I say -6dbfs is just in case the drummer gets a little extra happy...

For the most part, the transients don't matter as long as you don't clip. It's the sustained level that will adversly affect the signal chain.

 

[SouthSIDE Glen]

Since many of us are using audio interfaces that combine preamps and converters into one box, and since many of these contraptions only use peak metering using digital meters that by nature are fast enough to follow every nook and cranny of the incoming audio, which basically means there are no RMS metering facilities, how does one figure when they're around -18 or -15 dBFS RMS?

One can get close enough pretty easily and pretty cheaply, if they just get a 1kHz test tone off the net for free, and then go out and spend ten bucks on a Radio Shack multi-tester.

Assuming first (but not last, stay with me here) one is running into their integrated interface at a designed +4dBu line level, one can run that test tone line in at 1.23v (checking that level with the multi-tester), as this is what +4dBu represents. Send that direct into the computer with all input levels on the computer (driver and DAW) set to unity gain. The read the meters onthe DAW and that will tell you the converted digital signal level when sending a +4dBu (0VU) signal into the converter.

With mics, the knob setting will vary, but again, the key for measurement is not what's going into the preamp, nor just how much the preamp has to work, but rather the amplified voltage being fed into the converter. This means that yo can still use the same meter reading to determine that voltage going into the converter. the conversion factor will remain the same either way. The trick there is that depending upon the matchup between mic and pre, you may not actually wnat to be riding around 0VU/+4dbu. i fthe mic has a weak output and the preamp does not provide enough gain, you might have to crank things to far to get totat level. there is where you have to use your ears to decide if a lower level will be better than the "nominal" level or not.

---

As far as the whole "how critical is gain staging" debate, I grant that it's criticality increases with the number of links in the chain and most home reckers don't have that many links. BUT, there's another way to look at it - one that I didn't really consider until I started hanging on these boards:

For these guys a few dB here or there may not make that great of a difference because a simpler chain can be more forgiving. But that's not the common problem around here. Rather, the common problem is an ignorance or complete disregard for the whole idea altogether; with false myths like "record has hot as you can" and "use all the bits to maximize resolution" floating around everywhere like Internet viruses for accurate information, not to mention a very sad lack of proper metering in today's econo-chains, it's not a question of gain structure varying by a handful of dB this way or that, but rather with gain levels swinging wildly all over the place. I don't care what anybody says, there is nothing good about that situation, which 99 times out of 100 times causes degradation of either tracking quality or mix quality or both.

The *knowledge* is the real controlling factor here, the real power. The exact numbers shouldn't need to be argued over at that point by anybody except the most die-hard of gear sluts. Just knowing that one should be somehwere near a *designed and intended* sweet spot, and where that spot is, can reduce a huge chunk of the problems that many newbs have with their efforts.

G.

 

[dastrick]

And just for the record I am using an all in one interface. I was using Presonus Firepods, but I have a Presonus Firestudio being shipped right now. My DAW is Cubase. So I guess the only way that I have to monitor my levels is after it's been through the AD conversion.

 

[Farview]

And just for the record I am using an all in one interface. I was using Presonus Firepods, but I have a Presonus Firestudio being shipped right now. My DAW is Cubase. So I guess the only way that I have to monitor my levels is after it's been through the AD conversion.

Yup, you are correct. Just aim for sustained notes to sit around half way up the meter and make sure percussive stuff doesn't clip.

 

[Ethan Winer]

Neve preamps sound significantly different when you push them a little. And that is a good thing, if that's what you are going for. Of course Behringer toob preamps do too, but not in a good way.

I never consider "color" type gear because it goes against everything I believe in.

I agree, a well designed clean preamp probably won't sound different. But I get the idea that most people around here are not using really well designed preamps, most are using cheap preamps that come on their interface or on a $200 8 channel mixer. These do tend to get pinched sounding when pushed (as would anything with a wallwart power supply).

I have experience with only a few brands. But if we can believe the specs, I can't see why even a cheap preamp would change quality at higher levels below hard clipping. I've seen wall warts that can put out 5 amps or more! "Pinched" might be another way of saying distorted, and of course that can be easily measured. If a preamp is spec'd to some very small value of distortion up to the point of hard clipping, then it should sound fine at all levels below clipping. My aging Mackie 1202 is neither high end nor budget, and the direct outs I use have very low distortion to well beyond the levels I ever ask of it.

One of the points I make in my new AES Audio Myths Workshop video is that our ears have a "sweet spot" where things sound better or worse. Most gear is linear and doesn't change nearly as much with different levels. So it seems to me that varying the preamp gain and output level changes our perception more than it actually changes the sound.

The people around here who are lucky enough to get some great preamps are getting the Neve's and API's that do have transformers that will make them sound different at different output levels.

I agree, but to my ears those changes are not desirable or useful. If you record clean you can dirty it up later when mixing, and change your mind about how much dirt you want. If you record with dirt (or EQ or compression) there's little you can do later when the final mix takes shape and you can better hear what fits and what doesn't. Of course this is just opinion, not science fact.

Most of the rest of this thread is just people arguing over the minutest of details.

I surely agree with that.

--Ethan

 

[Ethan Winer]

many of these contraptions only use peak metering using digital meters that by nature are fast enough to follow every nook and cranny of the incoming audio, which basically means there are no RMS metering facilities, how does one figure when they're around -18 or -15 dBFS RMS?

RMS is irrelevant in this context. All that matters when setting levels in a DAW is that you don't exceed digital zero. Of course, you don't want to record at -50 or whatever either. The main use for knowing RMS levels is when you want to assess how loud something sounds. I haven't looked at an RMS type meter in at least ten years.

Do you then say: "I am just gonna track so things don't peak past -12dBFS so that I don't cook my preamps"?

Even that's not needed. Especially if you use a converter with built-in preamps. Simply adjust the preamp level so the digital audio never clips. It's that simple.

--Ethan

 

[Farview]

I never consider "color" type gear because it goes against everything I believe in.

So, everything you said only really pertains to the equipment you chose to work with and the way you choose to work.

That's fair, but it does ignore some of the most sought after preamps and the reason why they are so popular. It also completely ignores most of the budget stuff and the starved plate toob stuff that a lot of people on this board are using.

So in reality, only some analog is linear, it just happens that all the stuff you decided to use for the purpose you use it for is.

I have experience with only a few brands. But if we can believe the specs, I can't see why even a cheap preamp would change quality at higher levels below hard clipping. I've seen wall warts that can put out 5 amps or more! "Pinched" might be another way of saying distorted, and of course that can be easily measured. If a preamp is spec'd to some very small value of distortion up to the point of hard clipping, then it should sound fine at all levels below clipping. My aging Mackie 1202 is neither high end nor budget, and the direct outs I use have very low distortion to well beyond the levels I ever ask of it.

I have no experience with the 1202, but I know for a fact that the Mackie 8 buss mixers from the mid 90's could not output a really hot signal. In fact, 0dbVU wasn't even +4 or -10, it was somewhere in the middle. So if you were trying to push the level of something like a synth or heavily distorted guitar up to around 0dbfs, the direct outs would fall apart well before you got there. That's one of the reasons why that mixer got the reputation for being a little thin and screachy sounding, because people were pushing the levels into +4 recorders and trying to get as close to 0dbfs as possible. They were running a +22db signal out of something that had a line level somewhere around -2db.

I agree, but to my ears those changes are not desirable or useful. If you record clean you can dirty it up later when mixing, and change your mind about how much dirt you want. If you record with dirt (or EQ or compression) there's little you can do later when the final mix takes shape and you can better hear what fits and what doesn't. Of course this is just opinion, not science fact.--Ethan

In the genres of music you decide to record, they may not be. But in some forms of music, it's almost expected.

 

[SouthSIDE Glen]

Oh my GOD. I simply cannot believe the bullpucky direction ths thread has taken. I'm trying to keep my composure here, but that's hard to do when I'm busy picking my incredulous jaw off the floor.

Ignoring preamps "of color" when talking about this subject is like talking about life in a large, American metropolitan city and ignoring people "of color"; You wind up ignoring the vast majority of the actual facts. This is not Connecticut.

If someone can't hear the character difference between a Mackie mixer preamp, a Tascam interface preamp, an mAudio Fastrack preamp and a Presonus firebox preamp, they probably have no business participating in this thread other than to ask questions. And that's not even talking about the real difference between the character of Neve, SSL or Digidesign preamps on the pro level where most commercial recordings pass through.

And before Ethan comes back with a faulty test that A/Bs examples of each, I'm not saying that everyone can or should be able to *recognize* which is which out of context. I'm saying that differences in performance character are audibly identifiable as "better", "worse" or "inconsequential" for any given situational requirement, and such relevant decisions are made in studios every day.

And folks need to get their head out of spec sheets ind into the real, physical world. This came up recently in another thread, and while there are those intelligent folks here who I often agree with, those that claim they can tell the personality of any circuit by looking at a spec sheet are involved in a degree of self-deception equal to those that say they can predict what it will be like outside in four hours by checking the current temperature and humidity. Specs have their use and place, and can give some helpful general indications and classings, but they fall far short of telling the whole real story.

Additionally, one simply cannot accurately talk about the quality of "digital levels" while ignoring the entire analog chain going into the converter. No more than one can describe modern physics by talking about space and ignoring time.

This is an analog world we're recording (at least as long as we stay above the quantum level ), and the recording chain is almost entirely analog as well, regardless of the medium we are recording to. In a "for the want of a nail" fashion, the final intrinsic digital recording quality is determined by what's coming out of the converter (and so should the signal level), the converted level is determined by the analog signal going into it, the quality of the converted signal is determined a little by the quality of the converter and a LOT by the quality of the analog signal fed into it, and that quality is determined by the personality of the analog circuitry and the skill of the person piloting them. (As important as it is overall, for the sake of this argument, we'll ignore the quality of the performance, since that's not what's at issue here).

G.

 

[Bristol Posse]

This is an analog world we're recording (at least as long as we stay above the quantum level )"]This is an analog world we're recording (at least as long as we stay above the quantum level)

I prefer the quantum level. at that level the possibilities that I have both recorded and sold a million albums and not recorded anything at all exist simulataneously. So long as I don't open my eyes and look at my computer the probability wave will never resolve into an actually

also I have found this cat, the name tag say Schroedinger...... any takers

 

[noisewreck]

Even that's not needed. Especially if you use a converter with built-in preamps. Simply adjust the preamp level so the digital audio never clips. It's that simple.

--Ethan

Ethan, while I have a lot of respect for you, I am going to have to disagree with this statement, simply because in my experience with my equipment tracking close to 0dBFS definitely caused issues when all parts were recorded and I was starting to mix. While things didn't really sound distorted, it was always a struggle to get things fit come mixing time. Things have improved considerably since I have started to take it easy when tracking.

My recording chain (when dealing with external devices) is extremely simple, a couple of synths, and in some rare occasions a mic (SM7B). In fact, about 80% of the time, the only thing that ends up inside the DAW from the ouside world is my Kurzweil K2600XS, which does have rather hot analog outputs (+21dBu Max on separate outputs, using balanced cables, and +27dBu Max on the Mix Outputs, again when using balanced cables). As long as I ensure that the gain staging is setup so that I don't digitally clip the signal inside the K2600, the analog outputs simply don't overload and the signal quality does not change even when pushing close to it's limits.

The second component is my Aardvark Pro Q-10. While it is a pretty decent sounding interface, it is also somewhat noisy, with a self noise floor of around -72dB or so, which is certainly rather high for 24bit recording.

That's about it for the background.

Before, I used to record so that I'd pretty much hit -1dbFS, and like I said, mixing was always difficult. I then started experimenting with lower recording levels, however, recording around -18dBFS (peaking) would cause noise buildup from the various tracks, so that was also an issue. The Q-10 allows me to lower the preamp level to -8dB (from unity gain), and when recording from Kurzweil with this setting I peak around -6dB or so (sometimes lower maybe down to -10dBFS), and that seems to give me the best overall balance.

So, from my experience, with the above equipment, there was certainly an issue when tracking close to 0dBFS.

Glen, a couple of questions regarding your post that I need some clarification on.

1kHz test tone. I can easily generate a 1kHz sine from the K2600. Will this work or do I need a specific test equipment for generating this?

I also have a multimeter, but I am not sure where I'd be measuring the voltages. Are you suggesting I simply check the voltages coming out of the Audio Outs of my Kurzweil (for example) while having it play a 1kHz sine wave and calibrate it's output so I read 1.23V and then connect that output to the audio in of my Aardvark Q-10 and adjust it's levels so that I read -18dBFS on the meters? I am somewhat confused about this process so if you don't mind dumbing down the process for me I'd appreciate.

 

[Paulie Jay]

I prefer the quantum level. at that level the possibilities that I have both recorded and sold a million albums and not recorded anything at all exist simulataneously. So long as I don't open my eyes and look at my computer the probability wave will never resolve into an actually

also I have found this cat, the name tag say Schroedinger...... any takers

I've got him right here in this cardboard box. Or do I...?

 

[miroslav]

Before, I used to record so that I'd pretty much hit -1dbFS, and like I said, mixing was always difficult.

What was difficult about it?

I'm assuming even though you tracked real hot, the tracks were OK quality-wise...so when you started mixing, why did they give you problems?
I mean....once you had them in the DAW, the actual levels would be irrelevant. You could raise or lower them without any degradation.