0dbfs Sine Wave measurement in DP shows +6dB in mixer??

[ptfigg]

Can someone elaborate on this for me please ...

I have file which is a precise 0dBfs sine wave. I play it in Peak, Soundtrack Pro and any other audio app. and the meters reflect 0dB.

I bring the file into DP, lay it in a track, and play it back. The meter on the channel registers +6dB [above zero center].

Channel slider is at zero.

I am not quite sure I understand what this is all about. Can someone please clear this up for me?

-ptfigg.

[newrigel]

What actually are you doing this sine wave thing for?
If your that discerning there's a calibration utility in DP just for this type of thing... but you make a good point! I mean what's the use if your DAW is shredding the inputs to the desk! Look on page 748 and check out this calibration plug. Are you using any plug-in on the track in DP? What I/O are you using and does it have any type of trim? Also, DP has a way better way of creating headroom internally than the other applications you mentioned. There's allot of things that can go on with a DAW that the other apps just don't do so this hotter signal is due to that increased level. Also DP has a exponential buss structure ( just like a real console ) so you can have tons of tracks and not loose too much headroom in the process while multiplying the tracks. I usually listen and don't rely too heavy on metering anyway (unless the meters always in the red of course)....

[ptfigg]

HVK,

Hi. First let me point out I have an 896HD on a Dual Mac G5.

I have lots of analog gear, mostly Aphex and dbx which I use for broadcasting purposes. I record in RT using DP.

I just picked up an Aphex Dominator II Limiter, and its being used at the last stage of the chain. The idea is to record in RT yielding a recording with proper RMS values and equalization, etc. This saves me lots of time -- that is, not having to perform these tasks in post and such. it works well.

The issue with the sine wav is this:

The limiter and also my Aphex Compellor -- need a 1khz sine wav for proper input calibration. Needless to say I dont have a test tone generator. I have DP set up for direct hardware playthrough.

I tried sending the sine wav through the 896 into DP which feeds the chain, and ultimatly the limiter.

The channel sliders need to be maxed in order to match the measured level within Inspector for example.

I just want to be clear with why DP works this way. The chain has a constant +4 operating level throughout, which no additional increase in gain on the 896 input.

-ptfigg.

[Timeline]

Some levels for reference.

If you set a sinewave to o on the DP fader you will be approximately at '0' level =+4 on the OP of your IO. The fader position does not mater but the green level is reasonabley acurate to '0'. The fader position is dependant on the recorded level.

A 1K sinewave recorded at full level to the last bit will put the fader at just under -14 to produce a '0' level. I use this meathod to figure out DP's fader position as most +4 sources should be at about -14 to -15 area in digital.

Not very scientific but that's how DP designed the faders.

[Christian]

So you are calibrating to a sine wave playing at 0dbFS? Or are you calibrating so that 0 dbVU= -20dbFS (standard US film/TV calibration)?

I say this because I've never heard of calibrating to 0dbFS... you would have no headroom if you did!

As for DP's 'green meters', just like Digi's meters, Logics meters, etc... they are innaccurate and arbitrary. You'll note that if meters actually went in the 'red' on a digital system you would get digital clipping... for the most part, this doesn't happen in DP. I would caution against using these meters for any sense of measurement. If you need to have a full television mix with peaks at -10dbFS and RMS levels in a certain range (I find -28 to -22 dbFS works for me on a full mix with dialog), then strap a metering plug across the master fader and work from there. Same theory for stem mixes.

Not sure if this answered your question?

[ptfigg]

Chris,

For example, The Aphex Compellor has 3 selectable operating levels.

+4 ... +8 ... -10

The piece needs to be 'normalized' or I should say the input calibration meter needs a 0dbfs tone in oder to determine which operating level to select. I was sending the sine wav from the 896HD, and out into the chain [compellor, limiter, etc].

This is how I relaized the sliders in DP need to be maxed in order to output the signal at is measured level.

-ptfigg.

[croyal]

DP meters are analog style- "0" on the meter is similar 0VU (as on an analog deck)- which is why it goes up to +6 before clipping. If it were digital 0, you wouldn't (couldn't) have anything above 0, right? It may be callibrated so that +6db (displayed)=0dbfs (digital), but not necessarily. The metering is inaccurate, so you need to callibrate to something else along the way.

The "Trim plug" meters (for one example) are capable of showing 0dbfs- so insert that plug on the track and view it to see the actual level of the signal in digital terms.


Chris

[Kubi]

0dB in digital simply means: full level, no headroom left. DP decided (as was mentioned before) to fashion it's mixing board in a pseudo-analog way, and they decided to have 0dB be 6.02 dB below full level. Hence, your full level sinewave shows up as +6dB.

Either value is meaningless in the analog world, since it entirely depends on your interface's setting what voltage corresponds to full level digital.

DP's setting with a mere 6dB of headroom is actually too claustrophobic, IMO. 0dB (analog) in a +4 environment should equal 1.28 Volts, and that in turn should be calibrated in digital to correspond to about -18 to -12dB.

In reality, since 99% of what I do is digital now anyway, this all means usually not much anymore. I simply let DP peak (not average) around 0dB, which gives me another 6dB of headroom if needed. If I'm tracking, I let the input average around -12dB, maybe lower if it's a very peak-y signal (i.e. percussion)

In 24bits, that's plenty of resolution. In the very end, during mastering, I (or whomever is mastering) finalize the levels, while EQing, compressing, limiting etc., in a single step, to peak at -0.1dB or so (that is, 0.1dB under full digital level.)

[magicd]

Nothing arbitrary here.

The numbers on the audio track faders in the DP mixer represent what the fader is doing. If the fader is set to 0, there is no gain or attenuation happening from that fader. The track faders have six db of additional available gain. If you push the track fader up to maximum you will be adding 6db of gain to the signal.

The VU meter for the track fader does not have a visual db reference. Your normalized sine wave should show full VU all the way to the top of the meter when the fader itself is set to 0. In other words, the VU meter in the track goes up to full scale but does not show additional gain past that point. If you want a numerical VU reference of the signal, put a Trim plug-in on the track and use the meter in the Trim plug. The Trim plug-in VU meter will also show up to 20db above full scale.

If the track fader is sent through a master fader or bus/aux track, going above full scale will not cause clipping, unless the next gain stage also allows the signal to go above full scale. What that means is that the peak indicator for the track fader could be lit, but as long as the final output to the interface is attenuated back into the 24 bit sample scale, you are not actually clipping output.

Because DP uses 32 bit floating point precision in it's data path, there is something close to 1500db of dynamic range within the mix path. Because the output of DP is 24 bit integer (no decimal point), the output of DP must be scaled back to within the 24 bit sample range or else you will get clipping on the audio interface or bounced file.

If the track fader is sent directly to the audio interface, and as long as no other signals are also being sent to the interface (which would introduce summing gain), the only way you could clip the output of the interface would be if you added gain by pushing the track fader above 0 or adding a plug-in which created more gain. This is because the sample itself can not exceed full scale. The only way a full scale sample can get louder is if gain is added at some point later in the signal chain.


The line level outputs of all current MOTU audio interfaces are calibrated with 14 db of headroom as referenced to the analog 0 +4db standard. This means that if your normalized sine wave is sent to the interface you will see a full scale VU readout on the interface itself and if that signal is then sent to an analog device with a properly calibrated VU, you will see a total of 18db of signal (14db above the analog 0 +4db reference). Conversely, if you attenuate the sine wave by 14db in DP, the interface output VU will show 14db below the output ceiling and that signal will then drive an analog VU meter to exactly the 0db reference.

Gain matching with analog gear is critical to the gear working to spec. However, there doesn't need to be any guessing here. The outputs of the interface can not be any louder than 14db above the analog 0 +4db reference.
Just knowing that a piece of gear is referenced to +4db is not enough information. You also need to know how much headroom above that reference the device offers. For example, if the analog gear is referenced to +4db but has headroom of 12 additional db, the full scale signal from the interface would clip the analog input by 2db. So for proper calibration, you really need to know the available headroom of the device, not just that it has +4db reference.

Let me know if you need any clarification.

Magic Dave

[ptfigg]

M.Dave,

Excellent information, and good to hear from everyone as well.

M.Dave, Here is the chain -->

RE-20, SM7B, and various source components on a Mackie Onyx 1620->

Aphex 109 inserted into a mic channel -->

Main outs on the Mackie feed the Compellor inputs-->

Compellor outs to Aphex Dominator II inputs -->

Dominator outs feed to a pair of inputs on the 896HD.

All components using +4 operating level, as well as the 896HD inputs.

What do I need to do here you calibrate headroom and such properly?

Thanks again ...

-ptfigg.

[Timeline]

Nothing arbitrary here.


Gain matching with analog gear is critical to the gear working to spec. However, there doesn't need to be any guessing here. The outputs of the interface can not be any louder than 14db above the analog 0 +4db reference.
Just knowing that a piece of gear is referenced to +4db is not enough information. You also need to know how much headroom above that reference the device offers. For example, if the analog gear is referenced to +4db but has headroom of 12 additional db, the full scale signal from the interface would clip the analog input by 2db. So for proper calibration, you really need to know the available headroom of the device, not just that it has +4db reference.

Let me know if you need any clarification.

Magic Dave

Has someone finally admitted here that +18 db is the clip point in MOTU IO's

[magicd]

Wow. A Compellor and a Dominator. Old school. I used to own a Compellor. I used it for analog mastering and liked it a lot. I tried out a Dominator for awhile but didn't like it.

You don't say what the outs of the 896HD are going to. If you're doing your mixdown inside the computer, the 896HD outs are being used for monitoring only and calibration is less of an issue. Everything else you describe is on the front end.

The mics hit the pre-amps on the Mackie. The Mackie pres will have a peak indicator to tell you if the mic signal is overloading. Make the loudest signal that is going to hit the mic and turn up the Mackie pre amp until the clip LED lights. Then back off the pre a hair. Experience tells me that musicians are always louder in the song than in sound check. Keep an eye on those clip indicators. If you overload the input to the pre you will get distortion and you won't be able to fix that in the mix.

The Aphex 109 is in the insert, which means it is post pre amp (post pre?) The Mackie doesn't give you a separate insert gain control and you need to set the pre amp gain to match the mic signal. Therefore you'll need to calibrate the input and the output of the 109. Once the pre amp is set, watch the input of the 109. The 109 does not have VU meters. It only has a clip LED. Start off with the +4db calibration on the 109. I'm not sure of the reference of the Mackie inserts and I do know that they are not balanced. If you can't get enough signal into the 109 with the +4db setting, that's when you switch it to -10db reference. Set the EQ controls to flat and switch between bypass and non-bypass. Assuming that the gain control of the 109 only works when the EQ is switched in, set the gain control so that bypass and un-bypass are the same level. Any meter in the signal path after the mic input can be used to check this. Your ears work too.

Put the channel fader on the Mackie to unity gain. Bring up the master fader on the Mackie and watch the Mackie VU until you are close to unity gain output. It's ok to have a little headroom here.

Set the Compellor and Dominator to +4db reference. Use the clip lights on each device to determine input level. If you use a lot of compression, you may have to raise the gain on output to compensate. Use balanced cables. Set the inputs of the 896HD to +4db fixed. Watch the inputs of the 896HD and attenuate (or raise the gain of) the Dominator until you are close to the top of the 896HD VU.

Most engineers will tell you to leave yourself a little headroom on the input to the recording interface. The 896HD has 111db of dynamic range, which is more than most people need. Also, because you are heavily compressing the input signal, you have already reduced the dynamic range of that signal. If you clip the input of the 896HD, you'll get distortion. You can add as much gain as you like in DP, so try to keep the input VU in the top third of the 896HD input VU, while taking care not to hit the top LED of the meter.

Personally, I have stopped using compressors or limiters on the front end. You can't get rid of compression. You have a ton of dynamic range on the interface. In my experience, the only signals that generally have unpredictable spikes are voice and bass guitar. When I record bass and vocals, I still don't put compression on the front end. I just make sure I've got enough headroom to allow for any incoming peaks.

The Dominator was originally advertized as a mastering limiter. The Compellor is a more gentle leveler. I used the Compellor on my final stereo mixes going out to a Toshiba VCR with a PCM converter (14 bits. This was back in the late 1980s). I found that the Dominator did it's job, but was very obvious and I didn't like the sound I got. By contrast, the Compellor did a beautiful job of bringing out detail in the mix.
What I now do is use the MW Compressor in place of the Compellor, and the MW Limiter in place of the Dominator.

My signal chain goes like this:
mic straight in to the 896HD

In mixdown, I use a compressor and possibly a de-esser (on vocal tracks) on the track itself. If I use subgroup, such as for a drum mix, I may put a compressor on the subgroup. On my master fader I have the MW Compressor going to the MW EQ and then to the MW Limiter.
The heaviest amount of compression is applied on the track itself. By the time I get to the master fader, my personal rule is that I don't ever want to see more than 4db of gain reduction happening in the MW Compressor or MW Limiter.

Dave

[magicd]

Has someone finally admitted here that +18 db is the clip point in MOTU IO's

Admit what? That's the spec of the unit (actually, to be precise, 18dbv is not the clip point. 18dbv is the ceiling. Anything above 18dbv will clip the input). It's also the same spec for the Traveler, 828mkII, 2408mk3, 24i/o and HD192 interfaces. It's very easy to check on input and output. All you need is a test tone generator and an accurate analog VU meter.


Dave

[Kubi]

Thanks, Dave! Great explanation, and it's funny how after 15 years lil ol me still can get things mixed up, even though I knew that, sort of.

Anyway, I guess the source of the confusion is that the VU meter next to each fader shows the "real level" of the track in question, while the scale between the fader and the readout shows the gain increase/reduction by the fader gain stage. Obviously, the numbers refer to what the fader is doing, while the VU meter shows how hot the signal is on the scale from -infinity to 0dB. Hence my (erroneous) comment on the "arbitrarity", where in reality the numbers simply refer to the fader and not the VU meter.

[ptfigg]

Awesome Help Dave.

Thanks to Dave and everyone else here ...

Bye the way the signal is sent out via firewire to a Mac running DP.

-paul.
-nyc, ny