Topic: limiter in general

[BWolf]

so if you run the limiter, will it makes soft parts loud and loud parts soft  (ie even everything out)....

is it better to run levels in general.  i am helping a friend take a deck for a board patch and he knows nothing about it.  so should i get the limiter all set up before hand?  do you need to set anything on the R4 besides just putting the limiter on and letting him plug it into the board?  do you need to set levels?

[Nate]

I have very limited experience with the limiter, but from what I understand, if  you think you'll clip the levels (like if you are taping Kiss and the pyrotechnics are blowing the levels) or if you are in a very quiet situation and the clapping might make you clip....that is when you I would go with the limiter.

I know a few guys that used to use the limiter on their HHB's...but other than that...I've only used mine a few times.

I still don't know how to set levels with a limiter though?   

thanks

[Brian Skalinder]

Generally speaking, all limiters are compressors, but not all compressors are limiters.  Compressors reduce dynamic range, i.e. the difference between the loudest part and the softest part.  So in that sense, yes, a limiter / compressor evens things out.  Starter link on compression:  http://en.wikipedia.org/wiki/Audio_level_compression

The R-4's hard-switch "limiter" is actually a compressor.  It has a threshold of -10 dB and a ratio of 3:1.  When engaged, it applies to all 4 channels and operates in the analog realm.  The R-4 also includes a limiter / compressor function through the Effects menu, but it operates in the digital realm and so isn't very useful - if you're going to limit / compress in the digital realm, might as well do it in post-production on the computer where you have more control.

After engaging the R-4's "limiter", one still must set levels, just as one would do without the limiter engaged.  The difference with the limiter engaged:  after reaching - 10dB, the levels won't rise as fast as with the limiter off.

I don't think recording off the board requires the limiter.  But I don't record from the board very often.  I sometimes use it on-stage to protect against unexpected spikes in levels, and maybe if I expect the audience to roar above the levels of the music, but that's about it.

[bobbygeeWOW]

Maybe I'l jump on this.. a limiter limits the level that a signal can reach to some set threshold, often 0db to prevent clipping. Another way to look at it is that its a compressor with an infinite ratio once the threshold is met.  So there's no gradual gain reduction, its kinda immediate instead, but uses some rounding algorithm to make it sound nice rather than chopping it off squarely.

In digital post, software can "look ahead" to the next coupla samples of your file, determine the signal is going to go over based on the gain being applied, and limit the signal to the chosen threshold. Interestingly this can't function in an ultimate sense within the analog domain due to the real time factor - in order to tell that a signal is going to go over say 0db, the signal has to go over 0db first, and that hasn't happened yet!  I think most analog limiters allow a brief transient through before clamping down on the signal.

Sony has the new PCM-D1 recorder out that digitally buffers the incoming signal at -20db gain, then analyses that level and applies limiting before writing the track at the full gain level. This way it achieves real clip protection providing you don't have spikes over 20db. The use of wildly excessive limiting of course allows the full-tilt in-yer-face, grossly dynamic-free productions that are so popular nowadays..

but its been on my mind a little lately because the stuff I'm taping often has audiences louder than the music, which has me setting levels to the audience reaction instead of optimizing gain levels for the music.. anyway, hope that helps!

[NJFunk]

Generally speaking, all limiters are compressors, but not all compressors are limiters.  Compressors reduce dynamic range, i.e. the difference between the loudest part and the softest part.  So in that sense, yes, a limiter / compressor evens things out.  Starter link on compression:  http://en.wikipedia.org/wiki/Audio_level_compression

The R-4's hard-switch "limiter" is actually a compressor.  It has a threshold of -10 dB and a ratio of 3:1.  When engaged, it applies to all 4 channels and operates in the analog realm.  The R-4 also includes a limiter / compressor function through the Effects menu, but it operates in the digital realm and so isn't very useful - if you're going to limit / compress in the digital realm, might as well do it in post-production on the computer where you have more control.

After engaging the R-4's "limiter", one still must set levels, just as one would do without the limiter engaged.  The difference with the limiter engaged:  after reaching - 10dB, the levels won't rise as fast as with the limiter off.

I don't think recording off the board requires the limiter.  But I don't record from the board very often.  I sometimes use it on-stage to protect against unexpected spikes in levels, and maybe if I expect the audience to roar above the levels of the music, but that's about it.

Actually, I think using the limiter on a board feed can be useful, especially in small venues.  Often, board feeds are not balanced with the vocals super high in the mix and certin instruments low (partucularly bass and guitar).  If applied correctly, the limiter can raise the levels of the undermic'ed instruments in the mix relative to the vocals in the analog realm before digitizing the music, which IMHO, is usually better than trying to do it digitally.  Of course, this is a venue & situation dependent argument.  If you have larger room with a perfectly mixed board feed, then there is not reason to use the limiter unless you have some reason to fear clipping.

[NJFunk]

Maybe I'l jump on this.. a limiter limits the level that a signal can reach to some set threshold, often 0db to prevent clipping. Another way to look at it is that its a compressor with an infinite ratio once the threshold is met.  So there's no gradual gain reduction, its kinda immediate instead, but uses some rounding algorithm to make it sound nice rather than chopping it off squarely.

In digital post, software can "look ahead" to the next coupla samples of your file, determine the signal is going to go over based on the gain being applied, and limit the signal to the chosen threshold. Interestingly this can't function in an ultimate sense within the analog domain due to the real time factor - in order to tell that a signal is going to go over say 0db, the signal has to go over 0db first, and that hasn't happened yet!  I think most analog limiters allow a brief transient through before clamping down on the signal.

Sony has the new PCM-D1 recorder out that digitally buffers the incoming signal at -20db gain, then analyses that level and applies limiting before writing the track at the full gain level. This way it achieves real clip protection providing you don't have spikes over 20db. The use of wildly excessive limiting of course allows the full-tilt in-yer-face, grossly dynamic-free productions that are so popular nowadays..

but its been on my mind a little lately because the stuff I'm taping often has audiences louder than the music, which has me setting levels to the audience reaction instead of optimizing gain levels for the music.. anyway, hope that helps!

This is not the way that the R4's limiter switch works.  It applies compression at a 3:1 ratio starting at -10db.  Therefore, it gives you an additional 20db of headroom before you clip (at 3:1, you can "fit" 30db of levels in the space of 10).  But, if you really crank it, you will clip even with the limiter on.

[ghellquist]

In digital post, software can "look ahead" to the next coupla samples of your file, determine the signal is going to go over based on the gain being applied, and limit the signal to the chosen threshold. Interestingly this can't function in an ultimate sense within the analog domain due to the real time factor - in order to tell that a signal is going to go over say 0db, the signal has to go over 0db first, and that hasn't happened yet!  I think most analog limiters allow a brief transient through before clamping down on the signal.

Actually, some equipment used to do it in the analog domain as well. I have a Sony MXP42 mixer, a four channel battery powered film mixer. When you engage the limiter an analog delay line is entered into the signal path. It delays the signal about 50 microseconds, enough to let the circuits work. I am not too fond of the limiter stuff, but the mic pres are first class with transformers on the input. There is one more for sale where mine came from, see here:

http://korgen.svt.se/form/begagnad.nsf/SVTFS?OpenFrameSet&Frame=main&Src=%2Fform%2Fbegagnad.nsf%2Fwutrustning_ext%3FOpenPage%26Count%3D1000%26AutoFramed

Last I looked, price was set at 2.500 SEK, with current exchange rate (6,83 SEK per USD) that would be around 370 USD. It is sort of a reversed auction, this is used equipment from the Swedish state controlled television and radio, and they lower the price until the stuff is sold. If anyone is seriously interested, you may contact me, I might be able to arrange the details (I work about two miles away from the TV house).

In general though, with 24 bit equipment, the best way to go is to set levels low enough for the limiter to never engage. Then in post processing using a better look-ahead software limiter to raise the level to whatever crushed level you preferr. In this way it is possible to try different levels until things sounds at best. A good starting point software might be Magix Audio Studio deLuxe, it goes for about 80 USD.
http://site.magix.net/english-us/home/music/music-studio-11-deluxe/

(Warning: it should be music-studio, not the other program).

Anyway, just some musings.

Gunnar

[bobbygeeWOW]

Hm, that's all interesting stuff. Does anyone know how the 7xx's do limiting? is it a time based buffer as well?
One solution I've been wondering about is the reality of recording with lower levels at 24 bit, then raising in post, using limiting or (preferably) the volume envelope tool to keep the crowd down - does the noise floor rise at a similar rate as in 16 bit land? Is the noise floor actually functionally in real life lower at 24 bit so this won't become an issue?

(...not referring to the audience noise floor here) 

[Nate]

I run the limiter quite frequently with the 722, in acoustic situations. It helps to even out the "signal-to-noise ratio" (music to applause ratio). It helps to find a more happy medium with higher music mastering levels, and compressing the applause. Keeps things closer to 6dB difference, as opposed to a much more out of balance s-n-r.

Wow, I actually agree with Moke's ideas!    I run the limiter on my R-4 for the exact same reasons and in the same situations.   Better to be safe than sorry is my thoughts in that situation, better to compress the clapping than the music.   

So, if you are going to send your novice buddy out to tape a show with your gear, how do you advise him?   Set the levels at 50% forget about it?    From reading the discussion it does vary depending on the d vs a realm.

We are talking about the guy that can set up the gear...but ends up having his girlfriend break it down.   I know...I know...don't send it out....then again I come from the school of thought that I'd rather have my gear get used by a novice than sit home collect dust during most D & D games like many of you fockers out there.

+5 Constitution points

I'm a wizard.   
~Harry Pothead and the Taper

[ghellquist]

Does anyone know how the 7xx's do limiting? is it a time based buffer as well?
One solution I've been wondering about is the reality of recording with lower levels at 24 bit, then raising in post, using limiting or (preferably) the volume envelope tool to keep the crowd down - does the noise floor rise at a similar rate as in 16 bit land? Is the noise floor actually functionally in real life lower at 24 bit so this won't become an issue?

No buffer as far as I know. It is a normal analog circuit kicking in at about -6dB. The purpose is to allow you to get at least something even if you set the volumes all wrong. As limiter goes, it is decent, but all limiters does modify the sound.

The 722 is meant to be operated with peaks at about -12dB maximum. I tend to get the best results with the peaks a bit into the yellow band but avoiiding red leds totally. The first yellow lead is at -20dB, the first red at -12dB.

The noise floor will of course come up in post processing. But the availabe range on the 722 in 24 bit mode is about 110dB measured. To compare, a 16 bit unit can have a theoretical maximum of about 96 dB but many units measure well below that, some gives about 70dB. So the idea when using the 722 is to "waste" about 12dB of range in the recording in order to be sure to never even engage the limiter. This way you come home with a totally clean recording, no limiting, no overs, far above noise floor. At home you post process the recording to raise levels in the controlled environment where you can actually hear what is happening and can try different tools until you reach the result you are after. This allows you to get much better end results, at the expense of a small bit of work at home. Once happy you can burn CD-s or translate into mp3-s or whatever format you preferr. If you save the original format files as backup, you could come back at a later time and redo the post-processing, perhaps you have gained experience on the way or have new, better tools.
 
Gunnar

[poorlyconditioned]

I generally record 24bit about -12dB average level.  Sometimes it goes a bit higher, but not much.  No limiter.  This is with the R4.

I find compression/limiting useful occasionally in post though.  Suppose the music peaks at 12dB and claps come through a lot higher, like -3dB.  Then I just turn on compression above 12dB to bring things down to the same level as the music.  Then I might try to fade in/out between songs to reduce the clapping.

Best thing is to just get the mics in the right place to begin with.  I've learned that lesson now.

  Richard