question about tracking too hot......

[miroslav]

...and someone who makes audio furniture and wall fittings for a living who disagrees with them...

WOAH there!
I'm not trying to take sides...but you are WAY out of line here, probably unknowingly, but still, you might have done some checking before making this kind of statement.

I've known Ethan for a long time through forums and also having met him in person at AES.
Not sure it he still has that historic info on his website...but Ethan was involved with pro audio recording, and if I'm not mistaken, actually owned and ran a pro studio, when most of the guys on these forums were still in diapers or shooting hoops in their school playgrounds.

He stepped away from that and turned his attention to bass traps...first offering FREE designs to people that he had researched and built, which took a LOT of acoustics and audio knowledge...and then most currently running a ready-made bass trap design and building business that does well and is highly respected in the audio industry.

Referring to him as nothing more than a builder of furniture and wall fittings is a gross misconception.

I’ve disagreed with Ethan myself at times in the past…but I do know that he is NOT some kind of audio dolt. I know he would never blow his own horn as he isn’t that type of person…
...so I want to set that straight for the record.

 

[miroslav]

Yes, the whole mess is because Ethan is looking at it from the point where the signal is already digital. And everyone else is screaming that they are NOT talking about that.

Actually...the initial post was purely talking about the digital side if I recall...and I believe there were a few other posts initially focusing more on the digital side (or failing to clearly mention the analog side)...until the thread blew open and many different angles were presented.

When I proposed a question to John (Massive Master)...what if the front end was putting out a hot signal that pushed the digital meters high, even IF the analog front end was operating within its "sweet spot"...how would he handle that at the digital end...?
He kinda gave me several answers that seemed more of personal preference choices than anything else (which is fine) about turning DOWN the analog front end for certain things in order to give the converters less signal…
...but that didn't really prove that a hot signal on the DAW side was necessarily a bad thing if that same signal was hot-but-fine on the analog side.

There was a certain amount of suggestion being made that you should target the lower range rather than higher range on the digital side…but that’s not absolute if you disconnect from what the front end is doing in a given situation.
I mean...if...IF....you have a front end that is running comfortably within its operating range, and you are intentionally running it on the hot side because you like what it does to the signal...who cares if the DAW meter is showing -2dBFS or -20dBFS...???

That is the only question/point I've wanted to see clarified...but people keep introducing other scenarios or saying that it's "safer" to keep your sweet spot narrower on most occasions.
Well fine...that may be an OK approach, but it starts to smell of a "preset" kind of perspective...which may be OK for total newbies to follow...but I don't think this thread was only aimed at them.
So I think there is room for more expanded points of view rather than to just sell it all as a recording 101 SOP....yes?
Heck...I think all the newbies ducked and ran once Glen started tossing out the technical arguments (I mean that in a nice way Glen).

 

[Farview]

But it seems it's easier for Glen to bow out than answer that direct specific question. That's okay. Again, I'm not here to piss anyone off. I do have an agenda! But it's not making people hate me.

The thing that changes is the amount of distortion in the analog chain. Of course there are exceptions to this, but the vast majority of preamps do distort or compress as they get close to the limits.

Once inside the daw, even though you really can't clip a 32 float, some plugins do clip at a certain point. This is especially true of plugins that emulate analog gear. Part of what the plugin would emulate would be the distortion caused by putting a really hot signal through it.

Another annoying thing would be compressor plugins with 'soft knee', because you could easily get into a situation where the threshold is set to 0dbfs and you are still farther into the reduction than you want to be.

Maybe, maybe not. Define "soiling itself."

Seriously, this is exactly the sort of precision (in wording) I'm trying to enforce by being such a hard-ass. The OP's question asked about tracking at -12 versus -2, and the correct answer is there's no difference assuming common sense is used when setting preamp and converter levels.

--Ethan

The whole thing is a big maybe. Assuming that you were able to cleanly get the signal to the converters at both -12 and -2, there would be no difference. The reality is that you can't assume assume that.

 

[Farview]

I'm not taking the stance that Ethan is taking in that I don't believe that -18dB can be a sweet spot, but I really haven't heard any technical explanation as to why it would be. I have no agenda here, other than to find out why.

There are a couple of assumtions about the -18dbfs being the sweetspot.

1. That the preamp/analog chain was designed to be in it's sweet spot at line level

2. that the converters were calibrated so that line level = -18dbfs.


If you know that your converters are line level = -15dbfs, then -15 would be the sweet spot. I'm not sure how, but -18 became the generic number thrown around in these discussions. I assumed that it was because it was right about in the middle of where manufacturers tend to calibrate their 24 bit converters. (I've seen everything from -15 to -22 being line level)

Since sometimes it's hard to find out exactly how your converters are calibrated AND 2 or 3db either way isn't a big deal, -18dbfs = line level is a safe bet when used as an rms value.

The point of the whole thing is to run the entire analog chain at line level and you won't have to worry about the digital levels. (as long as they don't clip)

 

[miroslav]

But I think that's the crux of the debate in refering to a single point as the "sweet spot" (-18 or -15 or -12 dBFS...etc)

Much of the counterpoint to that is saying that the "swet spot" CAN BE much wider on the digital side as long as you are happy what your analog front end is doing.

 

[Farview]

But I think that's the crux of the debate in refering to a single point as the "sweet spot" (-18 or -15 or -12 dBFS...etc)

Much of the counterpoint to that is saying that the "swet spot" CAN BE much wider on the digital side as long as you are happy what your analog front end is doing.

Absolutely, the analog sweet spot is what is being refered to. The digital sweet spot is so wide that it's hard to miss without really trying.

 

[SouthSIDE Glen]

But I think that's the crux of the debate in refering to a single point as the "sweet spot" (-18 or -15 or -12 dBFS...etc)

Much of the counterpoint to that is saying that the "swet spot" CAN BE much wider on the digital side as long as you are happy what your analog front end is doing.

I for one never referred to the "sweet spot" as a single point. In fact it could be reasonably argued that the "sweet spot" - or maybe "sweet range" might be more accurate to say - for any given analog device can vary in size and location on the overall voltage scale. I seriously doubt there'd be that much argument from any side of the debate on that one (at least, God, I pray not! )

That said, though, the vast majority of non-consumer-level analog gear *is* designed to include 0VU as residing well within that sweet spot, no matter how narrow or wide that "spot" may be. Some gear may be designed or built better or worse in that regard, and some engineers may disagree on whether the "sweet spot" actually sounds better than overdrive or underdrive or not, but more often than not, one can take this as a generally correct general principle.

On the digital canvas of, let's say, 32-bit size (~140dB), by traditional definitions the "sweet spot" is 140dB wide, since one bit register is as good as another. This is where Ethan is coming from when he says that digital recording level does not matter as long as you don't clip. And he's of course right as far as that goes. No argument there. And the fact that the typical analog signal coming into the digital world only has some 60-70 dB of dynamic range, give or take a handful of dB amongst us friends, there's typically twice as much digital room to fit the analog signal as is technically needed.

But did you ever wonder why A/D converters are never calibrated to convert 0VU to -50dBFS or lower? Even on a high-dynamic, high-S/N analog signal of 70dB range, converting 0VU to -50dB should still leave plenty of room on the bottom and a shitload of summing headroom on the upper end. It would, on the surface, seem ideal by the myopic "it doesn't matter" view of the engineering universe. But you never see anything even close to that.

Instead, what the Engineers Who Decide These Things have done is cluster their conversion factors in the extremely narrow mid-teens to low twenties negative dbFS range, with the mid teens tending to date back mostly to 16bit assumptions and the range dropping towards the twenties as 32bit or better became/becomes more standardized in the market. This tight grouping is not by accident. I have explained several times, and no longer wish to here, why this is. For those who do not grasp this explanation or who disagree with it, I invite their alternative explanations.

G.

 

[noisewreck]

Actually...the initial post was purely talking about the digital side if I recall...and I believe there were a few other posts initially focusing more on the digital side (or failing to clearly mention the analog side)...until the thread blew open and many different angles were presented.

The initial question was

I've tracked to -12dbFS, then in my DAW I increase the gain by 10db (not moving the fader, but processing the track to increase the gain).

Is that the same as tracking at -2dbFS?

The question maybe worded in such a way as to infer digital only point of view, but I think the OP himself was ignoring the fact that the signal HAS to go through the analog stage BEFORE getting into digital. Because he's specifically talking about TRACKING. And unless you're using a digital synth and connect it to your interface via digital (S/PDIF, AES/EBU, ADAT, etc) method, your signal is going to be in analog up to the point of hitting your AD converter. And BECAUSE of this, many of the fine gentlemen here have been giving it seeminlgy futile attempts to explain that one should not ignore the characteristics of the analog path from source all the way to the AD converter inputs.

So, given that, the answer to the OP's question is "depends".

First Point:
If your analog front end can handle tracking at input levels that will translate to -2dBFS in your DAW (while leaving everything at unity gain in the DAW) w/o distorting or otherwise coloring the sound, then most likely there is no difference, between OP's two scenarios. However, if your analog chain will distort or otherwise color the sound at that high a level, then the answer is, there is a difference.

Second Point:
Another twist. What if you have some noisy device in the chain? If you track so that you're hitting -12dBFS and then raise the signal to -2dBFS in the daw, you're effectively raising the level of the noise by 10dB as well. Now, if you can structure your gain stages so that you're feeding louder signal into that noisy device, w/o affecting it's signal tone/quality much, thus giving yourself better S/N ratio, which then means that the signal you're feeding into your converter translates to -2dBFS in your DAW, then effectively you've given yourself a better S/N.

Most people here though will say that with most modern gear, you're not really going to have to worry much about noise floors, and you should worry more about not distorting your analog chain, and thus will put the emphasis on the First Point.

Point being, OPs question cannot be answered by taking the digital/DAW domain in isolation, unless you're doing everything ITB (i.e. work exclusively with virtual instruments and do not record any actual hardware), OR you interface your external devices only digitally, so there is no analog circuit involved anywhere, which pretty much means using digital synths and samplers with digital outputs, maybe in conjunction with digital mixers and other digital processors, all communicating with each other digitally.

 

[SouthSIDE Glen]

The initial question was
...
First Point:
...
Second Point:
...
Point being, OPs question cannot be answered by taking the digital/DAW domain in isolation, unless you're doing everything ITB (i.e. work exclusively with virtual instruments and do not record any actual hardware)

"You must spread some Reputation around before giving it to noisewreck again."

Damn BBS points rules .

G.

 

[bozmillar]

I guess what it comes down to is why does this sweet range exist at all? it's not because the engineers design it in there, it's because there must be some physical limitation on the amps used. When you get farther away from this ideal range, what happens to the signal?

 

[Bristol Posse]

I guess what it comes down to is why does this sweet range exist at all? it's not because the engineers design it in there, it's because there must be some physical limitation on the amps used. When you get farther away from this ideal range, what happens to the signal?

The only answer to that is try it and see with your gear

Since this came up on that other thread you mentioned I have spent several days with a set of coleman VU meters, tone generators and multimeters calibrating everything. I've tested all my preamps and mics at different levels into their headroom with various sources to see what happens. I've deliberately clipped my converters with different sources to see what happens. Tried different preamp/mic/converters combinations on various sources at Line level RMS and deeper into the headroom while not clipping the converters to see what happens. I've done more number crunching on this little hobby of mine in the last couple of days than I would expect to do on a seven figure deal, but it keeps me out of trouble.

The upshot of all of it is I now really understand what is going on in my chain in various combinations and what is likely to happen when I push those variouse combinations. I'm not going to try and graph anything or do null tests it was fairly obvious to my ears what was going on in the literaly hundreds of test I have done and unless you have my exact gear list in your studio it wouldn't make any difference anyway and quite frankly 4 days was enough time spent on this. I will probably never bother with my DAW meters again except to verify I never go above 0dBFS when mixing

This doesn't qualify me at all to say what will happen with any one elses set up unless it happens to be identical to mine, because unless someone else is using that same chain, then all of that testing I have done is irrelevant to them. "Colorful" pre amps will distort by design above 0VU although how much, how the distortion sounds and how far I can push it before to my own sensibilities it sounds bad (again totally subjective) will vary from type to type, source to source and day to day (depending on how much the kids have been driving me up the wall cos they're stuck inside because of the terrible weather) . Clean pres *should* stay clean but again I can't say that is categorically the case for every possible amplifier available in the universe of audio recording. All I know is that mine did up to a point and then went "bad" at absolute maximum gain where the noise floor really came up, although it was at significantly above 0VU on all but very, very quiet sources that I had micd with a low output dynamic.

So if you really want to know what happens to the signal outside the sweetspot in the only place that really matters, YOUR GEAR, test your gear and see. I'm glad I did.

 

[MCI2424]

The only answer to that is try it and see with your gear...................
The upshot of all of it is I now really understand what is going on in my chain in various combinations and what is likely to happen when I push those variouse combinations and see. ....."QUOTE]

Congrats. You win. Best post ever. This is the actual answer.

 

[miroslav]

The initial question was


The question maybe worded in such a way as to infer digital only point of view, but I think the OP himself was ignoring the fact that the signal HAS to go through the analog stage BEFORE getting into digital. Because he's specifically talking about TRACKING. And unless you're using a digital synth and connect it to your interface via digital (S/PDIF, AES/EBU, ADAT, etc) method, your signal is going to be in analog up to the point of hitting your AD converter. And BECAUSE of this, many of the fine gentlemen here have been giving it seeminlgy futile attempts to explain that one should not ignore the characteristics of the analog path from source all the way to the AD converter inputs.

So, given that, the answer to the OP's question is "depends".

First Point:
If your analog front end can handle tracking at input levels that will translate to -2dBFS in your DAW (while leaving everything at unity gain in the DAW) w/o distorting or otherwise coloring the sound, then most likely there is no difference, between OP's two scenarios. However, if your analog chain will distort or otherwise color the sound at that high a level, then the answer is, there is a difference.

Second Point:
Another twist. What if you have some noisy device in the chain? If you track so that you're hitting -12dBFS and then raise the signal to -2dBFS in the daw, you're effectively raising the level of the noise by 10dB as well. Now, if you can structure your gain stages so that you're feeding louder signal into that noisy device, w/o affecting it's signal tone/quality much, thus giving yourself better S/N ratio, which then means that the signal you're feeding into your converter translates to -2dBFS in your DAW, then effectively you've given yourself a better S/N.

Most people here though will say that with most modern gear, you're not really going to have to worry much about noise floors, and you should worry more about not distorting your analog chain, and thus will put the emphasis on the First Point.

Point being, OPs question cannot be answered by taking the digital/DAW domain in isolation, unless you're doing everything ITB (i.e. work exclusively with virtual instruments and do not record any actual hardware), OR you interface your external devices only digitally, so there is no analog circuit involved anywhere, which pretty much means using digital synths and samplers with digital outputs, maybe in conjunction with digital mixers and other digital processors, all communicating with each other digitally.

All these points are fine, and have been already made in one way or another...and even agreed to in one way or another by almost everyone.
However...there was still some inference throughout this thread by some that you should avoid tracking at levels that hit your digital converters too hot...and that point is very vague with only a lot of "what if" scenarios to support it and only under certain conditions. I don’t think everyone is always trying to get only pristine clean signals when it comes to using analog gear.

I've already said in my own words that IF you are happy with your front end sound...WHO CARES what the converter is registering on its meters as long as it's not clipping.
Like Glen was pointing out...yeah....some people like to use their particular analog gear on the outer edges of the typical analog "sweet spot/range" where 0VU is the typical "center point".

AFA as the OP's original question....yes, he is indirectly/unknowingly talking about the analog side of things...but I think his question WAS actually focused on the DIGITAL operating range. These days it seems that's what everyone references even though there is always an analog front end involved.
So...the answer to his questions AFA the digital side is concerned is…NO...it matters not.
Get your front end set where you like it…and ignore the digital meters as long as they are not clipping.
I see no benefit in turning down the analog front end JUST to have a lower digital signal…???...especially since you can pretty much do what you like with those signals once they are digital under most cases.
If your cheap plug-ins are crapping out…that’s a different issue. If you are not adjusting/balancing properly when you sum out to stereo…that also is a different issue....etc...etc...
IOW…I don’t see that taking all these concerns and making them solely a burden that your analog front end has to bear AND to correct…is the only/better way to go.

 

[SouthSIDE Glen]

The problem with that is that sometimes the correct answer to a question is not the kind of answer the OP expects or asks for. The question is, do we give them the kind of answer they want to hear, even if it winds up being a misleading answer? Or do we give them the corect answer?

Way back on page one of this disaster the OP was given the answer that one cannot consider the digital without considering the analog as well, and then was explained how the two examples in his question did indeed differ in that light.

And you know what? Way back on page one the OP got it, understood it, and was thankful for it, even spreading rep around for it. And hasn't had any followup questions since. That should have been case closed at that point.

But instead here folks are 6 pages and some 140 or so posts later and folks are still freakin' arguing abut what they even think the OP was even asking. Jesus freaking Christ. people.

George, did you ever wonder why we got stuck with "the wrong half" in here?

So...the answer to his questions AFA the digital side is concerned is…NO...it matters not.

Sure it does.

G.

 

[miroslav]

Well...now you too are spurring on the debate with that....


I think Fairview had the best comment back on page 2, which I quoted then and again now:

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

There is too much effort to impose personal SOPs on others and refer to them as the better or more correct approach.

 

[SouthSIDE Glen]

Well...now you too are spurring on the debate with that....

The "debate" has been nothing but a merry-go-round going nowhere for seven pages. Not because of lack of effort to explain on one side, but because of a lack of effort to understand on the other.

What I find interesting is that people have kept asking questions of one side, and several of us have repeatedly tried answering them. I have gotten tired of repeating myself. Yet, with one half-hearted exception by someone who themselves admitted that they didn't understand the point, I have gotten NO responses to the questions I have asked in a couple of different posts in the last couple of pages. And here's the headline to that: those answers bear directly on the resolution to this whole conflict.

Instead, we got a bunch of inane bickering over what the OP was actually asking 7 pages ago. The fact is that anyone who doesn't understand the OP question probably should not be trying to provide an answer to it.

The same is true of the questions I asked, which is, I assume, why there have been no real answers provided yet.

G.

 

[miroslav]

Well...I also asked the same simple question several times...and I got some "maybe" "depends" "sometimes" type of replies.
Basically….
If your front end is doing what you want...what difference does it make if the digital meters are hitting at -2 or -20 dBFS?

If it doesn't make a difference...then that's one answer that ALSO answers in one way the OP's first question.
All the other scenarios that people have introduced in order to debate the question from their angle and/or from positions/SOPs they prefer...that's something altogether different. Not necessarily right or wrong…just different…but at the heart of the thread, there HAS been a theme about sticking to well-known sweet spots in order to be “safe”…which may also be a correct answer/approach, but that doesn’t necessarily disprove the initial point of digital levels NOT mattering (just not clipping) IF you are perfectly happy with your front end signal…which is what some of the counterpoints have simply argued.
There has been no real proof been given why you WOULD/SHOULD turn down your front end JUST to give the converter a lower signal…or reason why, if your front end was behaving as it should and more importantly as you wanted it to behave in a given situation.

 

[SouthSIDE Glen]

If your front end is doing what you want...what difference does it make if the digital meters are hitting at -2 or -20 dBFS?

As has been explained without equivocation several times already, because such levels make the job of mixing easier for everyone, leave less room for newb mixing mistakes, and result much more naturally/easily in mix levels preferred by mastering engineers because they wind up polishing gold instead of turds.

Now, I'd like to hear your (or anybody's for that matter) response to post #147 first. Then responses with explanations to post #132. So far I have gotten ZERO responses to post 147, and only one response with no explanation and an admission of non-understanding to post 132. Time to ante up, the pot is light.

These questions and the responses bear directly upon your question, miro.

G.

 

[NYMorningstar]

And you know what? Way back on page one the OP got it, understood it, and was thankful for it, even spreading rep around for it. And hasn't had any followup questions since. That should have been case closed at that point.

G.

The good side of it though is it's gives a lot of different angles on the subject and everyone reading it is learning a little here, a little there.

I see alot of different engineers work from different angles. Some play the desk like an instrument and others set the gain structure an leave it be, go figure...

Thanks for posting again on it. On a side note, you know how old people ramble on about subjects... I was wondering what we're going to do when we get old? Wait a minute, we are old! How'd that happen?

 

[bozmillar]

But did you ever wonder why A/D converters are never calibrated to convert 0VU to -50dBFS or lower? Even on a high-dynamic, high-S/N analog signal of 70dB range, converting 0VU to -50dB should still leave plenty of room on the bottom and a shitload of summing headroom on the upper end. It would, on the surface, seem ideal by the myopic "it doesn't matter" view of the engineering universe. But you never see anything even close to that.

Instead, what the Engineers Who Decide These Things have done is cluster their conversion factors in the extremely narrow mid-teens to low twenties negative dbFS range, with the mid teens tending to date back mostly to 16bit assumptions and the range dropping towards the twenties as 32bit or better became/becomes more standardized in the market. This tight grouping is not by accident. I have explained several times, and no longer wish to here, why this is. For those who do not grasp this explanation or who disagree with it, I invite their alternative explanations.

G.

there are plenty of reasons not to map 0dBVU to -50dBFS. This is mostly because the A/D converter does not have a dynamic range of 144 dB even if it is 24 bit (there are no 32 bit converters). The dynamic range is typically around 100dB for the good converters. mapping 0dBU to -50dBFS would limit the dynamic range in the nominal range to about 50dB, which would make it less than the analog front end.