|
Post by Johnkenn on Apr 1, 2020 16:28:05 GMT -6
I realize this might seem to be a noob topic, but it’s always good to return to. Hell - I’m hoping someone teaches me something i don’t know here. How do you guys gain stage from mic to pre input/output to compressor to whatever to AD? What is your peak into the AD? For instance - like my Upton going into the 73. The Upton (and lots of others) can be a little hot going in, so I’m usually pulling the fader down on the 73. Usually end up around 30-35 dB...but if the fader (or i guess it’s just attention) were wide open, I’d definitely be getting hairy when digging in. Would it be better to have like a -10 dB inline pad and keeping the fader all the way up? I honestly don’t really pay close attention to the meters - but obviously I try to peak just when I see the first orange bar. I think many would consider that too hot? I guess the proper way to gain stage the compressor is to (say on an LA2A style comp) pull the input almost all the way down and adjust the output to the same level as the mic and pre. Then as you pull up the input to add gain reduction, you reduce the output by the same amount. Am I close here? Of course, I don’t really ever do that - I just go to my “that works” position and adjust accordingly. But it would be nice to do it somewhat scientifically.
How hot are you recording into the AD? Often, I’m singing into a somewhat pre-mixed in-progress mix...so I probably record hotter than I should. I guess I should probably be pulling down my mixes a lot so I don’t have the vocal cranked to compete. Obviously, I can crank the headphones up and still record less hot - but then I often find myself making up gain with a second comp in the mix. I’m scattershooting here, but you get the general questions. .
|
|
|
Post by Guitar on Apr 1, 2020 16:45:47 GMT -6
With 24 bit digital you have 48 more decibels above silence than 16 bit. That is a ton of headroom (foot room). So as long as you are not clipping your ADC you are doing fine. Even a few peaks here and there might not be the end of the world. Being quite low will probably also be fine in the DAW, with clip gain.
I don't know too much about the analog gain staging, just to where everything sounds good I guess.
When I'm gain staging processors (usually plugins) yeah there's a lot of pushing and pulling levels until it's "right" but it always seems intuitive to me, rather than being able to describe a specific methodology that is used.
don't know if that helps or not but there it is
|
|
|
Post by chessparov on Apr 1, 2020 16:46:46 GMT -6
IMHO from what I understand, hitting the Pad will change the tone. IIRC it's generally suggested (24 bit) to stay within -18 to -10. Occasionally, you can go as high as -8 or so. I'm just starting to go to 24/44.1 (sometimes), and was compelled to look this up.
Your compressor method makes sense. FWIW I've tried keeping the compression Input "like usual", then reducing the Compression Output as needed. Seems to be fine then for me.
Great question. Thanks, Chris
|
|
|
Post by Johnkenn on Apr 1, 2020 16:55:58 GMT -6
IMHO from what I understand, hitting the Pad will change the tone. IIRC it's generally suggested (24 bit) to stay within -18 to -10. Occasionally, you can go as high as -8 or so. I'm just starting to go to 24/44.1 (sometimes), and was compelled to look this up. Your compressor method makes sense. FWIW I've tried keeping the compression Input "like usual", then reducing the Compression Output as needed. Seems to be fine then for me. Great question. Thanks, Chris Don’t know if it’s my method - think I’ve done that about two times lol. Usually just turn the input to get the correct GR and then turn the output to a reasonable level to go to the ad.
|
|
|
Post by chessparov on Apr 1, 2020 17:01:44 GMT -6
P.S.I think I was influenced a lot by Klaus, who recommends staying away from "The Pad", for vocals. One issue I suppose, could be punching in later and having more tonal variance than typically expected. Chris
|
|
|
Post by svart on Apr 1, 2020 17:09:57 GMT -6
I turn it up until it crackles, then turn it down a few db.
Start with everything down quite a ways, the do this with the preamp, then the next piece, then the next piece, etc.
Now you know where everything clips, and how far you can push things.
BTW, most pads are resistive PI pads, which have flat bandwidths into the GHz. Adding a pad should not change the tone unless it changes impedance matching into a transformer or something.
|
|
|
Post by Johnkenn on Apr 1, 2020 17:10:13 GMT -6
I was actually thinking an in-line pad on the cable.
|
|
|
Post by Johnkenn on Apr 1, 2020 17:12:58 GMT -6
I turn it up until it crackles, then turn it down a few db. Start with everything down quite a ways, the do this with the preamp, then the next piece, then the next piece, etc. Now you know where everything clips, and how far you can push things. I guess I’m kind’ve asking if there’s a more “scientific” approach. One issue I sometimes have being the singer and the engineer is that I’ll kind of adjust to what I think my loudest peak will be...then in actuality, I rarely get there. Maybe this is kind of a silly topic - I guess if it sounds good, it is good...but hey - we all have a little time to kill.
|
|
|
Post by svart on Apr 1, 2020 17:14:48 GMT -6
I turn it up until it crackles, then turn it down a few db. Start with everything down quite a ways, the do this with the preamp, then the next piece, then the next piece, etc. Now you know where everything clips, and how far you can push things. I guess I’m kind’ve asking if there’s a more “scientific” approach. One issue I sometimes have being the singer and the engineer is that I’ll kind of adjust to what I think my loudest peak will be...then in actuality, I rarely get there. Maybe this is kind of a silly topic - I guess if it sounds good, it is good...but hey - we all have a little time to kill. I guess there could be some science, but it's really dependant on the loudness of the signal. If you sing a little louder one day, or stand a little closer or further away, the staging won't be optimal.
|
|
|
Post by svart on Apr 1, 2020 17:20:27 GMT -6
I turn it up until it crackles, then turn it down a few db. Start with everything down quite a ways, the do this with the preamp, then the next piece, then the next piece, etc. Now you know where everything clips, and how far you can push things. I guess I’m kind’ve asking if there’s a more “scientific” approach. One issue I sometimes have being the singer and the engineer is that I’ll kind of adjust to what I think my loudest peak will be...then in actuality, I rarely get there. Maybe this is kind of a silly topic - I guess if it sounds good, it is good...but hey - we all have a little time to kill. I guess one more thing I do in sessions is have the musicians play some of their loudest parts and set the preamps for that, but then turn down another 6db or so. Musicians tend to play a little louder during takes than they do during getting levels.
|
|
|
Post by Tbone81 on Apr 1, 2020 17:39:04 GMT -6
If you have a preamp, like a 1073, with input trim and an output attenuator, its good practice to leave the output atten at unity (all the way up), set the input trim till you reach your desired level into the DAW, and use the output atten to dial or fine tune the level if you still need to. BUT, if you want to push the preamp into blishful distortion, you keep pushing the input trim up till you hear the preamp breakup in a pleasing way, and keep pulling the output trim down to compensate.
I usually keep my peaks between -12 and -6. At 24bit the only danger in recording too hot is clipping your converters, usually with fast transients (like drums) or really dynamic sources (like some vocalists). Even then, just because the red light hits intermittently doesn't necessarily mean you've yet clipped your converters.
|
|
|
Post by jampa on Apr 1, 2020 17:42:10 GMT -6
Musicians tend to play a little louder during takes than they do during getting levels. If I had a dollar every time this happened I'd be rich
|
|
|
Post by Johnkenn on Apr 1, 2020 19:07:34 GMT -6
Just for clarity...I understand how input and output works on a mic pre (I’ve been doing this for a little while)...maybe I should have asked how hot you’re going into the next stage...in order not to achieve distortion or clipping. I guess using my ears has served me well for several decades.
|
|
|
Post by popmann on Apr 1, 2020 19:13:06 GMT -6
I literally just took a break from recording a vocal--so I just pulled the peak stat on the last take: -14.2dbfs peak. And it's pretty limited analog. So despite being a dynamic vocal, the VU meter sits largely between -3dbfs and +1dbfs (with the VU calibrated to -18=0). If I didn't have the fast 2.5:1 VCA on that same vocal would probably peak at -6dbfs. It's so limited I COULD just ride the fader to sit it in the mix. I probably won't...but, just for perspective of how squeezed it is.
For digital reference, do you have the Waves VU meter? Insert it on the Cubase input channel**. Just keep that open while singing the tune. You can set it's calibration to whatever you use (the new Apollo is -24dbfs=0VU if you want to do textbook clean work)...but, for perspective, with the above peak level, mine is set to -18dbfs=0, which is also what the Burl is set to. I think people not using VUs, but instead fast digital meters adds to the confusion. Dourough meters and your DAWs are good for drum transients--where VUs are sort of useless.
**or I actually put it on the reverb send channel since that's always input monitor enabled. That's not a "better" solution...just a different one that works in all DAWs, since not every DAW has full time input channels you can put plugs on.
|
|
|
Post by chessparov on Apr 1, 2020 19:17:10 GMT -6
Just for clarity...I understand how input and output works on a mic pre (I’ve been doing this for a little while)...maybe I should have asked how hot you’re going into the next stage...in order not to achieve distortion or clipping. I guess using my ears has served me well for several decades. I figured you might know just a l-e-e-tle bit more than me. I just know enough to be mildly dangerous, in a Studio...Apartment! Chris
|
|
|
Post by Johnkenn on Apr 1, 2020 19:52:31 GMT -6
I literally just took a break from recording a vocal--so I just pulled the peak stat on the last take: -14.2dbfs peak. And it's pretty limited analog. So despite being a dynamic vocal, the VU meter sits largely between -3dbfs and +1dbfs (with the VU calibrated to -18=0). If I didn't have the fast 2.5:1 VCA on that same vocal would probably peak at -6dbfs. It's so limited I COULD just ride the fader to sit it in the mix. I probably won't...but, just for perspective of how squeezed it is. For digital reference, do you have the Waves VU meter? Insert it on the Cubase input channel**. Just keep that open while singing the tune. You can set it's calibration to whatever you use (the new Apollo is -24dbfs=0VU if you want to do textbook clean work)...but, for perspective, with the above peak level, mine is set to -18dbfs=0, which is also what the Burl is set to. I think people not using VUs, but instead fast digital meters adds to the confusion. Dourough meters and your DAWs are good for drum transients--where VUs are sort of useless. **or I actually put it on the reverb send channel since that's always input monitor enabled. That's not a "better" solution...just a different one that works in all DAWs, since not every DAW has full time input channels you can put plugs on. Yes! Thanks! I’ve definitely been coming in hotter than that.
|
|
|
Post by Johnkenn on Apr 1, 2020 19:54:49 GMT -6
Just for clarity...I understand how input and output works on a mic pre (I’ve been doing this for a little while)...maybe I should have asked how hot you’re going into the next stage...in order not to achieve distortion or clipping. I guess using my ears has served me well for several decades. I figured you might know just a l-e-e-tle bit more than me. I just know enough to be mildly dangerous, in a Studio...Apartment! Chris No - didn’t mean it to sound catty. I just was trying to explain what I was trying to ask. Jamie definitely had some good info there. Gonna do that in my next vocal.
|
|
|
Post by EmRR on Apr 1, 2020 20:17:25 GMT -6
Just for clarity...I understand how input and output works on a mic pre (I’ve been doing this for a little while)...maybe I should have asked how hot you’re going into the next stage...in order not to achieve distortion or clipping. I guess using my ears has served me well for several decades. It really depends what the next stage is. La2a has gain that had to be tossed to achieve unity. Other devices are unity throughput so you could set pre gain very differently for the two types. Almost all vari-mu will have inherent gain. With the la2a, if the gain is almost all the way down you are losing some air treble because of the pot position, since it’s a hi-Z pot with audio through it. The higher you can keep that pot and still achieve the gain reduction you want, the more air it’ll have, which could be an argument for lower preamp gain. It’s worth comparing opposing knob settings that yield same basic result to see if there’s other change. Most preamps have best signal to noise ratio at max gain, which seems counterintuitive. Ignoring that and using a ribbon in a high gain situation, I find I ‘think’ noise is better with gain slightly lower than max and more gain added from a later device, like comp makeup gain. And that’s definitely better noise performance than something like a cloudshitter up front. You may like the shitter for other reasons, but don’t buy the noise reason. It’s technically impossible and test gear has proven it. (off soapbox). Any unity gain comp (one with a makeup gain labeled +/-0 usually) or Eq you set the pre as if where the only thing in the path, unless it happens to be overdriving said device undesirably. The input stage of any following device can be an unknown, so first treat it like it’s unity and has appropriate headroom, then start adjusting the pre down if it’s not true.
|
|
|
Post by Ward on Apr 1, 2020 20:23:25 GMT -6
I was actually thinking an in-line pad on the cable. That is also my suggestion. In general, try to stay in the green for every stage of the chain... hitting 2-3 yellows on really big peaks. Nominal zero is -18dbFS. And that's where you want to be for the energetic parts of a performance, usually.
|
|
|
Post by nick8801 on Apr 1, 2020 20:29:38 GMT -6
I’m still a big fan of vu meters for gain staging. I like to keep the meter tapping zero on the peaks. As I add compression and eq I’ll push it a little past that. That leaves me ample headroom in the end to push my busses/mixes. You will find out pretty quickly if there is too much low end, or too much distortion mucking up your signal that way. When it comes to mastering I rely on other metering but I still like to see the meters just bouncing a bit at their ends.
|
|
|
Post by christopher on Apr 1, 2020 21:14:17 GMT -6
This is an interesting topic I’m not sure there is a perfect answer. Here’s a controversial explanation I came across in some thread or comment over 10 years ago that I’ve never seen since, but it was poking fun at “the idiots” who want to record near 0dB. They argued that for 24bit ADC the IC chips in the analog stage weren’t designed for that kind of dynamic range. So in order to look better on the S/N spec sheet, manufacturers design the front end to put noise floor lower, placing 0dBFS at the max the chip can handle.. In doing so, the optimum place for linear response is also pushed lower. They argued that the ideal place to aim for would be about -20dBFS, with peaks going higher some. this would supposedly result in less harshness, since the ICs start to be stressed and sound nasty the closer you get to 0dBFS. I’m not sure that has been my experience but I haven’t felt any negatives from peaking way lower and then normalizing.
|
|
|
Post by wiz on Apr 1, 2020 21:31:24 GMT -6
the thing is to optimise each stage, whilst getting the sound you want out of each stage, eg clean , fuzzy etc etc
24 bit, gives you an extremely big dynamic range.....
basically, get the sound you want out of each analog stage... and make sure you capture at a reasonable level at your AD , without clipping.
So assume U87 into 1073 into LA2A into AD.
You will find most of the time a 1073 will sit probably around the 35 mark, on my 1073 DMP i like the sound of the preamp at 50 and the output trim adjusted back a little, say 10 o'clock this would hit pretty much around -10 -12dBFS .. .this is with no compression.
Insert a LA2A and the LA2A will probably end up with its gain at around 20 on its dial... and reduction adjusted to taste.....
point is , get the sound at each stage you want... and then check the capture is not clipping, and you have some headroom..... -12dbfs will be plenty.... remember we are talking capture...
Cheers
Wiz
|
|
|
Post by Chad on Apr 1, 2020 21:43:40 GMT -6
I think about this stuff alot, myself. And, I often have these same questions, but then I don't ask them for some reason...
John, thanks for opening up this thread!
|
|
|
Post by Blackdawg on Apr 1, 2020 22:05:17 GMT -6
You have way more flexiblity in the analog world with gain staging than the digital, at least when it comes to pushing devices.
In analog, depending on the unit, you can over drive certain circuits/tubes/transformers/ect to get different colors and things. Which I think most who use analog gear are pretty familiar with, it's one of the best reasons to use analog gear.
In digital, if you hit 0 or go over, you're screwed.
For me, even though I record at 32bits in software(all my converters are only 24bit, not a lot of 32 bit converters out there) I push the levels as high as I can. I do a lot of classical recording, so getting the S/N ratio of the space and performance is everything. Changing the mic pre gain changes how I place the microphone and how the baseline sound is to start with. Which again for me is everything. I don't do tons of heavy EQ, certainly no compressor, rarely anything in post honestly. It's all about mic placement and gain into the recorder. So i push the gain, I try and peak my Grace micpre's at -3dBfs at the loudest part of the performance. I've learned Grace pre's very very well(m802 particularly, m108 since last year as well) and know how to set them usually after hearing a rehearsal once while also anticipating that it's going to be louder during the live performance because it always is.
I do this because in my case in post if you just "raise the gain" of the clip or even the volume too much the noise floor of the room becomes VERY obvious during the quiet sections. This isn't acceptable in the classical world it needs to be clean as possible and natural.
Now, most the music I think people here are working on do not have that as an issue I think it is safe to say. The noise floor is nothing most the time, you can hide things in a big production and filter the crap out of things.
That said, even when I record other stuff, I still shoot for peaking a -6dBFS or -8. How you get there with whatever analog chain you decide before you, doesn't really matter. You can get lots of cool color's by pushing and pulling the sound with gear. It's not like you aren't doing that in post anyways with plugins or if you use outboard.
That said, VU meters are great in the analog world to use.
But if you're tracking and don't have a true Peak meter showing you DBFS, that's a big gamble in my book. Wouldn't do that personally.
|
|
|
Post by johneppstein on Apr 1, 2020 22:44:57 GMT -6
I turn it up until it crackles, then turn it down a few db. Start with everything down quite a ways, the do this with the preamp, then the next piece, then the next piece, etc. Now you know where everything clips, and how far you can push things. I guess I’m kind’ve asking if there’s a more “scientific” approach. One issue I sometimes have being the singer and the engineer is that I’ll kind of adjust to what I think my loudest peak will be...then in actuality, I rarely get there. Maybe this is kind of a silly topic - I guess if it sounds good, it is good...but hey - we all have a little time to kill. Well, yeah, there IS a "more scientific" approach, the problem is that it varies with the source, the efficiency/nominal output level/headroom of the input transducer (if any), and the gain structure of the particular equipment chain on the channel. What you're talking about is the art and science of gain structure.
One of the great beauties of working on a full blown hardware console is that, excepting any outboard (which should be individually trimmed to fit the console gain structure, your gain structure is pretty consistent across all the channels of the mix. With the patchwork approach of many "modern" setups you lack that consistency, which makes it harder to visualize/understand what's going on, gain wise. And it's one of those cases of a simple subject that can become frustratingly complex if you don't have a good grasp of the optimal operating parameters of each piece of gear being used in each channel.
|
|