Plugin Clipping Internally: Bad?

[DarkEcho]

I am running Battery 3, and when I load up a kit and start playing- Battery's internal master level meter usually goes into the red. I have to decrease it by 8dBFS usually, to get it to what I would consider to be a nice, healthy level. (All green, sometimes dipping into the yellow).

The main question I have though, is, because these plugins are generally floating-point.. Does it really matter if it's in the red like mad internally as long as I bring down the DAW fader and avoid actual clipping?

What do you do with floating plugins that are internally in the red?

[theblue1]

I'm pretty sure they only put the clip light on to look retro.


J/K.

[narco]

the answer is

what do you think?

use your ears.

narco (not intended to be rude, although it may come across that way)

[DarkEcho]

I have very under trained ears so I don't trust them.

But when I slowly lowered the volume of Battery's fader and increased the level of my monitors- I couldnt hear anything significantly different.

But again- I don't know what I'm listening for- so I do not count on my ears.

Instead- I count on the wisdom of this forum full of people who know better than I =)

[jikkyboy11]

i don't let my drums hit the red either. I have mine hitting the yellow, but not the red. I process the drums more with 3rd party vst compressors, eq, etc.

And up too high will clip it within battery. just open battery standalone and use the compressor and eq inside it and max it out. U'll hear the clip. Try to keep it in the yellow is a safe bet.

[zakco]

Some plugins can clip and be fairly unnoticable, others will produce audible distortion almost immediately. With Battery it may depend on which samples are being triggered when the clipping occurs...but as a general rule with digital audio: RED=BAD.

Chances are though, that regardless whether or not the clipping is audible, you should practice good gain staging in your daw and that means running with at least 4-6 db of headroom everywhere in the mixer. So...bring 'er down!

-Z-

[Bitfiend]

I know in Battery 2 if you hit the red it is clipping (one of the very, very few VST's that will actually clip, as mentioned most have a floating point). It was supposed to be fixed in B3.
I don't have B3 so i couldn't tell you. But in B2 you have to back it down.
you could check the wave form as well.

[wakestyle]

are you using it in a host sequencer? If so I would trust that it doesn't go over -0.0.

I think battery 2 is way better btw.

[peeder]

It's up to the programmer to decide if they want to clip or not. Generally in a floating point system, programmers with IQ < 100 will have it clip, and those with IQ > 100 won't. Those with IQ > 120 provide saturation models where they "soft clip" i.e. get some pleasant distortion approaching or going over 0dbR. There is no standard for when a plugin should saturate although 0dbR is probably most common.

Programmers trying to achieve compatibility with the old fixed point system in PTHD get a pass on having it clip at 0dbR...but digi should have 86'd that architecture years ago now...

[theblue1]

Originally Posted by zakco

Some plugins can clip and be fairly unnoticable, others will produce audible distortion almost immediately. With Battery it may depend on which samples are being triggered when the clipping occurs...but as a general rule with digital audio: RED=BAD.

Chances are though, that regardless whether or not the clipping is audible, you should practice good gain staging in your daw and that means running with at least 4-6 db of headroom everywhere in the mixer. So...bring 'er down!

-Z-

This is the most sensible practice, seems to me. While outright clipping might not be a problem with many, modern, well-designed plugs -- many of our most important plugs will presumably have an optimal range: compressor/limiter plugs. (Granted, there's likely no reason a compressor plug couldn't be designed to operate on any relative level range, but take a look at the compressor plugs you've got and see if they aren't, by and large, oriented around 0 dBfs.)

[DarkEcho]

So are you saying that, while Battery3 itself going into the red might not produce audible clipping, when putting a plugin that has an artificial saturation curve, you might get undesirable behavior from other plugins down that have behavior dependant on level?

[DarkEcho]

Say Battery 3 DID produce audible distortion when it goes into the red (I guess it woudln't be floating point in that case?)

If I lowered the fader in my DAW, the signal would still be clipped- just quieter, right?

So while the distortion might not be super obvious- I should probably just be safe and keep battery NOT in the red...

HOWEVER- if a plugin is floating point- is there any advantage/disadvantage of running it hotter if you can?

[zakco]

Basically, my point is this:

Regardless of floating point headroom, the digital mix bus in your daw cannot output a finished file with levels above 0dbfs. this means that if you are running at excessive levels at one stage of the signal path, it is inevitable that you will need to attenuate the signal at some point down the line. Why allow excessive levels at one stage and then need to bring it down at another?

I like to keep all points in the chain operating at optimal levels. This begins with tracking at moderate levels (peaking between -10 and -6 db for most sources), bringing down unruly plugins (like battery) that generate too much level. If you do this you will never need to be concerned with internal clipping of plugins, nor will you need to lower your master in order to output a clean 2mix.

Simple really.

-Z-

[peeder]

Originally Posted by zakco

Basically, my point is this:

Regardless of floating point headroom, the digital mix bus in your daw cannot output a finished file with levels above 0dbfs. this means that if you are running at excessive levels at one stage of the signal path, it is inevitable that you will need to attenuate the signal at some point down the line. Why allow excessive levels at one stage and then need to bring it down at another?

Same reason you do so in analog: to achieve saturation.

Not all plugins offer saturation models, and as I already wrote, if a plugin just up and decides to go square at 0dbR, without offering any saturation model...the programmer is just laying booby traps for you.

There are people that hate digital purely because of these asinine programmers who have done this. In analog, you saturate before going square, and there is "give" and a standard nominal level. Digital should work like this, and the best digital does work like this (some plugins even block going square to make things more "foolproof") but some programmers have it out for their customers and insist on placing sharp objects in the playground.

[zakco]

Originally Posted by peeder

Same reason you do so in analog: to achieve saturation.

Not all plugins offer saturation models, and as I already wrote, if a plugin just up and decides to go square at 0dbR, without offering any saturation model...the programmer is just laying booby traps for you.

There are people that hate digital purely because of these asinine programmers who have done this. In analog, you saturate before going square, and there is "give" and a standard nominal level. Digital should work like this, and the best digital does work like this (some plugins even block going square to make things more "foolproof") but some programmers have it out for their customers and insist on placing sharp objects in the playground.

I know there are exceptions ( I do some truly silly gain staging with the EMI TG plugins) but these cases are the exception I think. Sure you can intentionally run things hot for effect, but that doesn't change the fact that good gain distribution is a wise practice in your DAW...

-Z-

[DarkEcho]

Ok- I have found that there is a difficulty in testing for tonal changes in a system without the effects of volume interfering with the results.

Say I try to turn down the master fader while I turn up battery 3 to listen for any changes (other than volume). I am not going to get it perfect obviously-

What is the best method of pushing a plugins levels to see if it affects the quality of the sound without having to mickey-mouse the levels?

It would be awesome of there was a plugin that automated the channels fader to maintain a specific level of volume...

so say you set it to -4dBFS, then start turning up battery- it will compensate by turning down the fader to maintain -4dBFS so that you can focus on listening for the tonal/character changes instead of matching volume...


any ideas!?

I was thinking maybe a compressor- but that imparts a whole other bias.

[peeder]

Import a 500Hz or 1KHz sine wave tone into battery, and play it through at a variety of levels, into an RTA (you don't say which DAW...oh actually you do..just use the Logic channel EQ analyzer set to its max resolution...and there are plenty of RTA's available...e.g. the free apEQ demo, Ozone demo). Look for the harmonics to come up when you reach clipping...and you can either get a full square harmonic ladder or you can get 2nd/3rd etc. with a saturation model.

You could also listen, just pull your master fader down so you aren't clipping the DAC. A volume-correcter thing would be a nice gadget but it's not necessary.

But regardless, you can just not clip battery and trim up the signal if you want downstream. I would hope battery is floating point output and maintains dynamic range...that would actually be a more interesting test than the headroom test. I.e. lowering battery's output control and then gaining up after that to see if any details are lost. Sonalksis FreeG gives you up to 54db of gain in one plug, and I'm always looking for a plug that can give more (anyone know of one?).

A 32bit float system has 900db of headroom over 0dbR, and 151 db of dynamic range.

And math is math...there is no intrinsic difference to a floating point signal anywhere across its >1500db of level range.

[DarkEcho]

Peeder, thanks for the suggestions- I did your test with the master volume maxed out in Battery 3 with a sinewave at 1000hz playing. Looking at the EQ analyzer, im not sure what I should be looking for- I just see a smooth pencil like spire sticking up. I assume that I am looking for smaller obelisks to pop up in random places other than 1000hz or to see artifacts in the 1000hz tower?

[peeder]

Originally Posted by DarkEcho

Peeder, thanks for the suggestions- I did your test with the master volume maxed out in Battery 3 with a sinewave at 1000hz playing. Looking at the EQ analyzer, im not sure what I should be looking for- I just see a smooth pencil like spire sticking up. I assume that I am looking for smaller obelisks to pop up in random places other than 1000hz or to see artifacts in the 1000hz tower?

You will see spikes to the right of the main spike as harmonic distortion shows up.

Try some of the distortion plugins fed from the tone generator thing and you will see the distortion. Square wave looks like an even series of spikes...the tone generator will let you see those.

Welcome to answering your own questions!

[DarkEcho]

Ok, so if I want to check a Plugins behavior in response to level, I can run a sine wave through it and push its internal fader while watching through a real time analyzer and I should only see one spike.

If distortion is taking place, it will look like this:


with spikes occurring elsewhere.

So distortion = more spikes than the original 1000hz tone, but nothing will happen to the shape of the 1000hz tone, it should stay roughly the same as other spikes start to appear.

[peeder]

Originally Posted by DarkEcho

Ok, so if I want to check a Plugins behavior in response to level, I can run a sine wave through it and push its internal fader while watching through a real time analyzer and I should only see one spike.

If distortion is taking place, it will look like this:


with spikes occurring elsewhere.

So distortion = more spikes than the original 1000hz tone, but nothing will happen to the shape of the 1000hz tone, it should stay roughly the same as other spikes start to appear.

Yup good work.

That distortion pattern in your picture is fairly "euphonic" and will actually be fairly hard to hear. That's a "good distortion" pattern as opposed to square (where the spikes would have equal heights). Note that since it is harmonic distortion, the spikes occur at multiples of the fundamental frequency (for 1000, 2000 is 2nd harmonic, 3000 3rd, 4000 4th, etc.). These harmonics correspond to musical intervals...2nd harmonic is an octave, 3rd harmonic is a perfect fifth...

Note that you can drag down the scale labels on that analyzer to see the noise floor (no idear why they don't let you just resize the whole plugin window). To see very low noise floors, add gain plugins before the analyzer.

[Charlie-O]

Awsome work you guys!