Ok, let me repeat that you don't need limiters and let's forget about loudspeaker protection.
Instead we should take a look at your gain structure: 1,35v is more or less equal to +4dbu ( a little bit less than 5dbu but let's say it is 4 and you already havd 1db headroom).
So if we take the 85db(a) calibration when you play a -20dbfs signal you should have an analog output voltage from your dac at approx: 0.1228v ( -16dbu).
From this we now need efficiency/impedance of your loudspeakers, voltage gain of your amp and the listening distance of listening spot to loudspeakers.
From there we will see if your system is able to withstand this spl at first.
Then we will talk about limiting ( and i will restate that as long as you don't do live acts there is no needs for limiters
).
Dbfs is a standard defining the max digital out level. Don't mix it with analog voltage: they are linked yes but there is no direct relation and it makes things messy to understand.
What you determined in finding your 1,35v is the max input voltage of your amp before clipping ( the level at which you need 105dbspl at listening point). Your reference level is so -16dbu with 20db allowed for dynamic.
If all is good ( efficiency, listening distance and amp power) to meet that target you have more than enough dynamic range for the less compressed signal you could find and your loudspeakers are already protected as you have 1db headroom (maybe more but it'll depend of your amp and loudspeakers).
Instead we should take a look at your gain structure: 1,35v is more or less equal to +4dbu ( a little bit less than 5dbu but let's say it is 4 and you already havd 1db headroom).
So if we take the 85db(a) calibration when you play a -20dbfs signal you should have an analog output voltage from your dac at approx: 0.1228v ( -16dbu).
From this we now need efficiency/impedance of your loudspeakers, voltage gain of your amp and the listening distance of listening spot to loudspeakers.
From there we will see if your system is able to withstand this spl at first.
Then we will talk about limiting ( and i will restate that as long as you don't do live acts there is no needs for limiters
).Dbfs is a standard defining the max digital out level. Don't mix it with analog voltage: they are linked yes but there is no direct relation and it makes things messy to understand.
What you determined in finding your 1,35v is the max input voltage of your amp before clipping ( the level at which you need 105dbspl at listening point). Your reference level is so -16dbu with 20db allowed for dynamic.
If all is good ( efficiency, listening distance and amp power) to meet that target you have more than enough dynamic range for the less compressed signal you could find and your loudspeakers are already protected as you have 1db headroom (maybe more but it'll depend of your amp and loudspeakers).
Dear krivium,
I want to make it very clear that I do not have a processor, amplifiers or speakers. So let me rephrase the question for you.
Question:
Assume that I have a hypothetical multi-way sound system that needs: (a) peak limiting and (b) RMS limiting.
Now if you're the sound engineer, what attack/release settings would you suggest for (a) and (b), given the crossover frequencies?
Please note that this is not about me requiring help to setup my home music system that does not need protection (I already know that). And, if you have an issue with the term 'dBFS', let us simply say 'dB' for now, and not dBu.
0dB level clips my hypothetical amplifier. Let 0dB return any voltage, power or SPL in any of the popular units, but that's the maximum unclipped voltage my hypothetical amplifier can take.
Regards.
I want to make it very clear that I do not have a processor, amplifiers or speakers. So let me rephrase the question for you.
Question:
Assume that I have a hypothetical multi-way sound system that needs: (a) peak limiting and (b) RMS limiting.
Now if you're the sound engineer, what attack/release settings would you suggest for (a) and (b), given the crossover frequencies?
Please note that this is not about me requiring help to setup my home music system that does not need protection (I already know that). And, if you have an issue with the term 'dBFS', let us simply say 'dB' for now, and not dBu.
0dB level clips my hypothetical amplifier. Let 0dB return any voltage, power or SPL in any of the popular units, but that's the maximum unclipped voltage my hypothetical amplifier can take.
Regards.
There is no proper there is only your desired outcome...
It depends on which attribute you are focusing on....there are other specifications that have been left out, in particular, the knee...if there is one...and the ratio...limiting is usually like 10:1 or higher...rms reads averages peak reads....peaks...the threshold is point of action unless there is a knee...if there is a knee, depending on the shape the limiting may actuate before reaching the threshold... Per frequency is a tricky question...the trend I've seen suggest faster times with higher frequency.
This is the default setting on one of my linear phase limiters...
Before you read into that...consider what it is you are saying...If you are wanting speaker protection...fastest attack time is the goal no matter frequency, release times adjusted to taste by ear but judging by the default setting (and experience) on the limiter above higher frequencies are more sensitive to longer release than lower ones...complete transparency only leans away from protection OR turning the volume down, since you shouldn't be clipping if aiming for accuracy. Going into more detail has nothing to do with loudspeakers per say and more to do with sound design/signal design.
Limiting isnt something you want to do...its something you have to do.....Your music has already been dynamically tailored, our/my goal (in this part of the forum) is accuracy not colorization...If you are hitting the limiter and don't want to...buy bigger amp/speakers...If you are in a PA situation, the limiter keeps you from destroying your speakers when **** gets out of hand, but at the sacrifice of sound quality. If you are a sound engineer you are using a limiter to create certain effects or controlling dynamics etc etc from a creative stand point or technical...and you are not using an home amplifier with built in limiter to achieve it...unless thats the only way to get said desired creative vision expressed lol! Better said, maybe that is the only way to achieve desired sound...The limiters we use in the studio have way more options but there are limiters built with tubes and transistors etc etc that provide a colorization that digital is just on the vertgeof being able to replicate 100%...then bye bye analog hardware in the studio...except for that one crowd. I have my learned ways of creating transparent limiting....but it has nothing to do with protecting a loudspeaker and everything to do with signal design.
Attachments
Last edited:
Then we will talk about limiting ( and i will restate that as long as you don't do live acts there is no needs for limiters
Hi krivium, i'm in sync with your comments that are trying to focus on gain structure rather than limiting...and in general agreement that home audio will seldom need limiters.
But i can think of a number of cases in my past, with home gear that really wasn't all that loud, where i have lost drivers when a limiter would have saved them.
Home gear tends towards less efficiency, and if an amp is sized for headroom, it can be pretty easy to overdrive a speaker when trying to reach satisfying SPL.
Particularly on the two ends of the spectrum, subs and tweeters.
Heck, i'll bet half the reason folks are getting away with Linkwitz transforms on their subs without thermal or excursion issues, is simply because they are under amped for sufficient headroom.
So i guess i'm saying i think it's a bold overstatement to say it takes live acts to need a limiter.
A lot of the home stuff is lack luster, I wonder if thats changing, minimal effort involved in the programing might be an understatement but I also have never built these devices just a user...on the computer this is so 1980....non the less, with some tools you can be pretty effective if the amp has a limiter...specify a desired output level, hook your electrical monitoring equipment to output, play music, adjust threshold of limiter until output is limited to desired output expectations...it can be more convenient if the amp is monitoring output real time...most aren't in my experience.
Hi Camplo, I'll respond to both of your posts in this one.
It super duper rocks imo, that you posted "There is no proper there is only your desired outcome.."
That's the big picture, and big picture is most often missed on our DIY forums, imho.
DIY generally likes to debate about stuff -60dB down Lol....
Limiters and compressors are quite related, but different tools. Compressors are more for input channels and sonic sculpting.
Good live sound and studio guys know how to use them, I don't. Like you were showing, knee, ratio, ... add key frequency...add all kinds of things to compressors.
None of those normally exist with simpler limiters, used on output channels, for either peak or RMS. I do feel I'm technically good at limiters, but acknowledge it's still an art.
Frequency clearly has a role in such limiters, as it effects their attack and release times.
We do not want the same attack time across the frequency spectrum, even for peak limiting. Not at all, and even more so for RMS.
Ok, moving on to your second post....
I've never seen anyone try to set limiters with music....i doubt it can even be done.
Takes a more steady state signal. Sine waves are the standard.
And yep, having amps that display real time output is nice, but most folks want the security of a known DVM, reading the amp voltage to the speakers.
(Voltage rules, ime nobody who knows what they are doing thinks about limiters in terms of wattage...)
More info than most home audio cares about i guess.....
It super duper rocks imo, that you posted "There is no proper there is only your desired outcome.."
That's the big picture, and big picture is most often missed on our DIY forums, imho.
DIY generally likes to debate about stuff -60dB down Lol....
Limiters and compressors are quite related, but different tools. Compressors are more for input channels and sonic sculpting.
Good live sound and studio guys know how to use them, I don't. Like you were showing, knee, ratio, ... add key frequency...add all kinds of things to compressors.
None of those normally exist with simpler limiters, used on output channels, for either peak or RMS. I do feel I'm technically good at limiters, but acknowledge it's still an art.
Frequency clearly has a role in such limiters, as it effects their attack and release times.
We do not want the same attack time across the frequency spectrum, even for peak limiting. Not at all, and even more so for RMS.
Ok, moving on to your second post....
I've never seen anyone try to set limiters with music....i doubt it can even be done.
Takes a more steady state signal. Sine waves are the standard.
And yep, having amps that display real time output is nice, but most folks want the security of a known DVM, reading the amp voltage to the speakers.
(Voltage rules, ime nobody who knows what they are doing thinks about limiters in terms of wattage...)
More info than most home audio cares about i guess.....
Thank you camplo, for sharing your views and settings. I am guessing that attack/release are in milliseconds, please mention if otherwise.
Just like you, many of us have noticed the time-constants change with frequency, but never understood much about any direct connection between them. I put up the question mostly for the purpose of knowledge gathering, as I don't have any proper hardware yet.
Just like you, many of us have noticed the time-constants change with frequency, but never understood much about any direct connection between them. I put up the question mostly for the purpose of knowledge gathering, as I don't have any proper hardware yet.
Last edited:
WE don't have to argue too much into detail...I can tell from your perspective you are not a studio guy. Thats not to say you aren't educated on the topic just that you don't talk like a person whos spent years designing signal ....The limiter commonly is the first tool on a track...no ones (in regards to sound design) thinking too much about input/output in this fashion as compressor vs limiter. The signal (music, vocals etc) comes in...we design it...then it goes out lol....
"Limiters and compressors are quite related, but different tools" - the major difference between a limiter and compressor is the ratio of the dynamic attenuation ie... if I take a compressor and turn the ratio up to 10:1 or higher....its is now a limiter...So limiting has a special place and became a separate tool. Most limiting is going to be wanted with a fast attack so you don't always see attack settings on a limiter its stuck in whatever the fastest times it could do because limiting is usally meant o be as transparent as possible....limiters once tended to be broadband, addressing the whole signal not just a specific band unless a filter has been placed before the limiter...Multiband compressors/limiters eventually came and then we can be more specific about how the limiters reacts to frequency to some extent.
The multiband limiter has its own sacrifices and thus will not make the wideband limiter obsolete anytime soon.... splitting and reassembly of the signal has consequences long story short...a multiband limiter is a conjunction of filters placed before multiple limiters in one tool....
"We do not want the same attack time across the frequency spectrum" - well if you don't have a broad band limiter you ain't got a choice do you...
"Frequency clearly has a role in such limiters, as it effects their attack and release times" - a simple limiter reacts to signal as it crosses its threshold with no regards to frequency
"I've never seen anyone try to set limiters with music" _ Once again you seemed removed from the actual act of sound design with this comment...I literally set limiters with music any time I mix/master a song....as they say in the studio world...use your ears...In my example for home audio limiter setting all I said was to play music and monitor the signal...I never specified voltage or wattage.
In the playback world, if interested in scientific sound quality...a limiter/compressor will never touch your signal...
"More info than most home audio cares about i guess....." - the biggest point....a limiter to the home audio guy is just a way to protect his drivers.
"Limiters and compressors are quite related, but different tools" - the major difference between a limiter and compressor is the ratio of the dynamic attenuation ie... if I take a compressor and turn the ratio up to 10:1 or higher....its is now a limiter...So limiting has a special place and became a separate tool. Most limiting is going to be wanted with a fast attack so you don't always see attack settings on a limiter its stuck in whatever the fastest times it could do because limiting is usally meant o be as transparent as possible....limiters once tended to be broadband, addressing the whole signal not just a specific band unless a filter has been placed before the limiter...Multiband compressors/limiters eventually came and then we can be more specific about how the limiters reacts to frequency to some extent.
The multiband limiter has its own sacrifices and thus will not make the wideband limiter obsolete anytime soon.... splitting and reassembly of the signal has consequences long story short...a multiband limiter is a conjunction of filters placed before multiple limiters in one tool....
"We do not want the same attack time across the frequency spectrum" - well if you don't have a broad band limiter you ain't got a choice do you...
"Frequency clearly has a role in such limiters, as it effects their attack and release times" - a simple limiter reacts to signal as it crosses its threshold with no regards to frequency
"I've never seen anyone try to set limiters with music" _ Once again you seemed removed from the actual act of sound design with this comment...I literally set limiters with music any time I mix/master a song....as they say in the studio world...use your ears...In my example for home audio limiter setting all I said was to play music and monitor the signal...I never specified voltage or wattage.
In the playback world, if interested in scientific sound quality...a limiter/compressor will never touch your signal...
"More info than most home audio cares about i guess....." - the biggest point....a limiter to the home audio guy is just a way to protect his drivers.
Last edited:
The connection is your ear....if you had such tools you could play with the settings and it will make a lot more sense...Get equalizer Apo and download a multiband limiter vst and go to town then come back. I could try to put it into words but hearing works so much quicker. Multi band limiting allows one to apply limiting more selectively and thus end up with a more tailored result or more transparent result...
I repeat...from a user end viewpoint of music...you don't want limiting to actually happen in your playback...for this position in the food chain it should only be used for driver protection...the only other possible use is in trying to increase the loudness potential of the content at the sacrifice of sound quality...If you are normally listening with music under constant dynamic attenuation that is unfortunate. They are already over compressing main stream music as it is...ever heard of the loudness war?
To give clarity to why a multiband limiter is a good thing for the listener vs a normal limiter....OK, the bass is usually the highest in the signal level...maybe midbass at times...whatever the case maybe... lets say the multiband limiter has 4 channels separating 20hz-20khz...the peaks in the signal are crossing the threshold.... the signal is only going to be attenuated on the channel that the peaks exist on....as opposed to what the normal wideband limiter does, which is, attenuate the whole signal... no matter where the peaks are coming from when they cross the threshold...So basically when you turn up your music to the point that its hitting the limiter...it can protect your speakers and maintain some more of the sound quality by only attenuating on the channels that have been triggered....
All the home stereo stuff remains primitive...I have a 16 channel linear phase multiband limiter, thats old news for vst software (approximately 2011)....I barely use it unless a signal is causing a need for limiting with more transparency of said act, or tailorization of the signal, in this part of the art there aren't always "rules"...signal design can said to be an extension of creativity, adding to the music (content) itself. Which is why we don't do it as the listener unless wanting to change the creative outcome.
Last edited:
Correction, If you don't have a multiband limiter! And once again, the multiband limiter/compressor is not without fault...the filters placed before each separate limiter causes distortion...a linear phase multiband limiter causes delay and pre ringing etc etc though the technology only gets better with time. In the studio a regular old full band limiter is one the most used tools along with eq and compression and all the other familiars included (reverb, saturation, delays, what have you)
Last edited:
Hehe.. hi Camplo!
Well i'am happy not to have explained all this. I disagree on some points ( i'm definitely from an other genration about some productions techniques you now all use) but this is why i didn't answered straight about settings to Newvirus.
Things are not as simple as they may seems and as Camplo and Mark i think a clear target as to be defined for the use of each tools put into the signal pass.
Newvirus i hope you didn't thoughts i was trying to patronize on you about gain structure as this wasn't the case nor my intention. The fact is this is related and how i would approach things as a sound engineer ( in a technician way).
But to do i need more infos to explain my thougths with example one can follow as each case needs a specific or taylored approach.
Eg: what is the source? Some media which has already be mastered or live input with widely varying dynamic contents? For how long the system will be in use? Some hours a day as in most domestics rooms or this is a freeparty where the system will be pounded full blast 24h a day for 3 days non stop?
Who will use the system? Family members or a drunk punk? All this have to be taken into account.
About frequency and time constants relationships: keep it plain simple.
For a whole cycle of a sinus at a given frequency we have a formula which will help to put things in perspective: 1/freq= 1 cycle duration in s.
A 20hz cycle have a duration of 50ms
A 50hz cycle is 20ms
A 250hz cycle is 4ms
A 500hz is 2ms, 1khz 1ms,etc,etc,...
With a limiter we try to 'grab' the signal as soon as possible so it make sense to define attack to the highest freq of interest within a band. For release it'll depend of band and what you want to achieve: idiot proof protection ( which will be highly audible and not pretty at all - the situation described by Waxx about Dj which understand what headroom is or not) or some subtle protection?
Sure a long release time will be more transparent but during this time if something happen...it'll happen!
Well i'am happy not to have explained all this. I disagree on some points ( i'm definitely from an other genration about some productions techniques you now all use) but this is why i didn't answered straight about settings to Newvirus.
Things are not as simple as they may seems and as Camplo and Mark i think a clear target as to be defined for the use of each tools put into the signal pass.
Newvirus i hope you didn't thoughts i was trying to patronize on you about gain structure as this wasn't the case nor my intention. The fact is this is related and how i would approach things as a sound engineer ( in a technician way).
But to do i need more infos to explain my thougths with example one can follow as each case needs a specific or taylored approach.
Eg: what is the source? Some media which has already be mastered or live input with widely varying dynamic contents? For how long the system will be in use? Some hours a day as in most domestics rooms or this is a freeparty where the system will be pounded full blast 24h a day for 3 days non stop?
Who will use the system? Family members or a drunk punk? All this have to be taken into account.
About frequency and time constants relationships: keep it plain simple.
For a whole cycle of a sinus at a given frequency we have a formula which will help to put things in perspective: 1/freq= 1 cycle duration in s.
A 20hz cycle have a duration of 50ms
A 50hz cycle is 20ms
A 250hz cycle is 4ms
A 500hz is 2ms, 1khz 1ms,etc,etc,...
With a limiter we try to 'grab' the signal as soon as possible so it make sense to define attack to the highest freq of interest within a band. For release it'll depend of band and what you want to achieve: idiot proof protection ( which will be highly audible and not pretty at all - the situation described by Waxx about Dj which understand what headroom is or not) or some subtle protection?
Sure a long release time will be more transparent but during this time if something happen...it'll happen!
I'm at least kinda well rounded, I would imagine that you come from the time before the loudness war, and a limiter was the last tool to be reached for in the studio? its hard to generalize with the transition to electronic instruments, removing recording engineers out of the picture on some level.
There is a group that are somewhat elitist or purist and no dynamic attenuation should take place at any point in the signal...true to life sort of speak...sounds like someone who only wants to listen to live instrument recordings with talented recording engineers in the loop lol...might as well throw in talented musicians who can create said content without the need for a limiter (unwanted clicks and pops, deliberate and desirable dynamic character upon use of instrument during recording etc etc)....
From a certain perspective limiting/compression is a form of distortion. The same one we tend to design against with high efficiency systems lacking....dynamic compression of the original content
There is a group that are somewhat elitist or purist and no dynamic attenuation should take place at any point in the signal...true to life sort of speak...sounds like someone who only wants to listen to live instrument recordings with talented recording engineers in the loop lol...might as well throw in talented musicians who can create said content without the need for a limiter (unwanted clicks and pops, deliberate and desirable dynamic character upon use of instrument during recording etc etc)....
From a certain perspective limiting/compression is a form of distortion. The same one we tend to design against with high efficiency systems lacking....dynamic compression of the original content
Last edited:
No, i was well into loudness war i'm 44yo...was even part of it at my own level.
I think it comes more from where and with who i learned. And the kind of musics i worked on.
I've got nothing against dynamic processors ( even designed one varimu for my own use). In fact after using a 8900 for the first time i somewhat became obsessed. Same thing with Ear 660...
I think it comes more from where and with who i learned. And the kind of musics i worked on.
I've got nothing against dynamic processors ( even designed one varimu for my own use). In fact after using a 8900 for the first time i somewhat became obsessed. Same thing with Ear 660...
The short answer was done by krivium, there is no standard.
If the goal is the absolute protection, just set your digital limiter into look ahead mode (brick wall) and forget it. You can call it negative attack time. If you can choose release time, set it Auto, and it should be the most versatile setting. Absolute sound quality? Don't use any dynamic processor.
You can choose anything between those 2 choices based on your goal and preference.
If the goal is the absolute protection, just set your digital limiter into look ahead mode (brick wall) and forget it. You can call it negative attack time. If you can choose release time, set it Auto, and it should be the most versatile setting. Absolute sound quality? Don't use any dynamic processor.
You can choose anything between those 2 choices based on your goal and preference.
krivium said:
Thank you for that.
krivium said:
How about subtle first, then idiot-proof and then the differences between both? That should be useful knowledge to many.
You may assume the following:
* Source is a 0dB normalised studio master. no extra processing, paraEQ etc. for now.
* Max. 6 hrs/day and disciplined audience with sound hearing (no drunkards/drug addicts)
* Subtle peak protection [against unexpected issues / incorrect settings etc.]
* 1dB headroom, i.e. 0dBFS from processor will not clip the amplifier (happy ?).
* 3-way speaker with subwoofer (+10dB): 80Hz, 300Hz, 1kHz crossover frequencies.
Last edited:
Ok this is better but... it miss the most important parts ( loudspeakers type, reference, amplifiers used)!
No issue as you gave hints with xover points and as this is hypothetical i'll use a 'virtual' soundsystem ( and it'll be easier to put some points in highlight).
So from 1/2/5 you'll have a typical domestic use but with a horn loaded system ( at least for the highs with a 1" cd) and probably home theater oriented ( someone will have to explain me why the sub is supposed to be able to output 10db more spl than the rest of the system in ht? Never seen that implemented in control rooms dedicated to movie mixing and multichanel...).
So we can expect medium to high efficiency 'pro' pa drivers ( i hope so). Let's say you are around 98dbspl ( if the loudspeaker was passive because it doesn't mean anything with a dsp filtered multiamp system... that said as it is hypothetical we will use hypothetical drivers which are all 8r/ 98dbspl 1w/1m whatever the way).
Let's say we keep a target at 85dbspl with 20db allowed for peaks at listening point and 3m distance from loudspeakers ( we keep on the ht theme too).
So we will have -9db loss from distance and we need the loudspeakers to be able to produce 116dbspl at 1meter ( max peak out, average (rms) it'll be 96dbspl ).
From 98 to 116 is 18db difference so we'll need 64w (effectively used) to reach our target level. We could be ok with a tda7293 based amplifier speced for 64w into 8r (but in reality no as we will see), so we will use a good old pa amplifier of 256w/8r ( a small one) with a voltage gain of 35db and with a max input sensitivity of 0dbu ( 0,775v rms will output 256w from the amp).
Your processor will output 1,35v at 0dbfs which means it's max out is at +5dbu. And let's say your processor use the same spec than your current dac.
So we can now define our gain structure (which we will keep the same for each ways, in reality...this'll be different) :
processor/dac to amp.
Simple enough.
But there is already some level matching to be performed: with 0,775v at input the amp will output 256w so if we send 1,35v it'll already clip like hell. And we only need 64w to be used effectively so we will have to get rid of some more db too ( 6 to be accurate).
So we will attenuate dsp dac output 11db ( 5dbu to 0 dbu for processor dac to amp and then 6db to output 64w rather than 256w):
This will leave us with a maximum output allowed from processors DAC. of -6dbu (0.3882v).
Either you do this 'digitally' by specifying -11db as your max out within your dac or you use an Lpad or Hpad between dac analog out and amp input ( i favour this solution as it is safer cause attenuation is 'fixed' and canot be modified by users/ if you use digital attenuator in your dac nothing tells you someone isn't going to put it back at 0dbfs rather than -11dbfs - this is where drunk punk and family members can converge
).Ok so we have defined our gain structure, we have headroom from the DAC but this is no problem as in 4 you said : '0dbfs from processor 'will not' clip the amp' and i have wasted 192w by choosing too big an amplifier.
And so we will downsize to 64w and only -5db attenuation (rather than the -11 needed for 256w amp).. as we are cheapstake ( isn't it the point to diy? Doing things on the cheap... not my pov but let's say yes for the moment.) and because headroom is not something we take into account.
So all is well and we can now setup our limiters.
Except the processor dac could still clip the amp in this situation. Why? Because of a nasty effects dacs can have called ISP ( Inter Sample Peaks).
(For further info: Inter-Sample Peaks Tutorial )
So what db value we could expect from ISP? Something with max 3db seems well enough headroom ( so a 128w amp and -8db attenuation). And as we already faced unexpected issues running the system without headroom allowed why not add +3db more too, just to be sure...
So we are now back to -11db attenuation and 256w amplifier. Now we can expect things to be ok.
And from my experience with the system's gain structure run this way there is little chance anything will ever happen to clip ( if no one have access to the 'spare' 11db in use -iow no one will play anything higher than -11db on processor out volume- and that you'll attenuate the level of your source material to match 85dbspl average at listening spot if not wide dynamic source material - for typical 'pop' material produced before second half of 90's- expect 14db rather than 20, so you'll have to attenuate 6db more on your dac volume at -17db, 12db for 'broadcasted' material (radio, tv,..) so -19db on your processor level. Of course if you implemented an analog pad substract the 11db attenuation for your dsp volume settings...).
So this is why i told you there is no needs for protections if your gain structure is well thought in advance: with headroom allowed if the dac produce an ISP the amplifier won't clip, the drivers won't too and all is safe... ( even more so if you implemented analogue pad in my view, even one with 3 settings: one for high dynamic contents (k-20, 20db), K--14, 14db and k-12 with 12db dynamic allowed).
But you want to introduce another stage of treatment so here we go:
In your situation ( with mastered material) the only risk you'll run into is from peaks from dacs. So subtle it could be:
For the time constants we already talked about so take 0,1ms attack ( 10khz) for the high band, 1ms for the 300/1k band.
I would leave the sub and low way alone but if i had to put one more limiter it would be on the 80/300hz and i would use 3.3ms attack.
The sub don't need anything in my view: if it clip you won't hear it and it shouldn't be hurted if it happen. And i suggest you to try limit sub and bass on a track to see by yourself how nice it sounds ( hint sub and low are... slow to react, this is the nature of the beasts).
For release i would go from 4x to 16x attack time of band in question but it'll have to be defined by the kind of music you'll play ( breakcore at 190bpm and ambient music won't require same settings to be 'transparent' sounding).
So now the fun part: where to put the thresholds? As we are performing subtle protection it seems to make the limiters starts there action when the 'ceiling' of normal playback is reached ( once you need more than 64w).
The interesting thing is as we have 6db headroom from the amp i suggest to choose to setup thresholds at 0dbfs or at max -0,1/-0.3dbfs.
Protection will happen only when a peak reach digital ceiling... You may hear it or not, it'll depend from the source. But this way you can be sure you won't use the headroom ( which is here to not have to use limiter hence i keep on telling it is not needed).
If you setup threshold lower expect the limiters to be trigged constantly on 'loudness war' approved material and treatment will not be subtle anymore ( you'll hear it).
So here it is for subtle from my part...
I'm sure others will have different pov. Let's see what will be proposed.
Idiotproof will be for later... but i'm sure you won't like it ( i won't).
Last edited:
Another article about limiter setup from a reputable brand:
https://www.google.com/url?sa=t&sou...FjAAegQIBhAC&usg=AOvVaw3qJEu3XtHcAGMUwDZ5vwNj
There is common points with the one Mark posted.
https://www.google.com/url?sa=t&sou...FjAAegQIBhAC&usg=AOvVaw3qJEu3XtHcAGMUwDZ5vwNj
There is common points with the one Mark posted.
Thank you for the release timing calculations. However, regarding thresholds, I would only like to say what I told Mark100 in #17, that the same limiter threshold (-0.1 dBFS) could be valid for any amplifier, speaker, SPL and listening distance or simply put, even without any of those, as shown below:
Digital source:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ 0dBFS (-6dB, quarter power)
D/A+ISP:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ +3dBFS (-3dB, half power)
D/A+ISP+safety:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ +6dBFS (0dB, max power)
Limiter threshold:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ -0.1dBFS (-6.1dB, onset of limiting)
The unit dBu was brought in for no reason, with absolutely no consequence!
Now, I feel that a threshold of +0.1dBFS would be alright as well. The source on its own, without any processing (or mistakes) would then never be caught by the limiter.
Well, there're few things in common. However, a key difference is that Mark was indicating different time-constants for peak and RMS limiting, while the tables on pgs. 1&7 of the above article seem to suggest exactly the same timings for both limiter types. Anyway, thanks for sharing that.
Digital source:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ 0dBFS (-6dB, quarter power)
D/A+ISP:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ +3dBFS (-3dB, half power)
D/A+ISP+safety:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ +6dBFS (0dB, max power)
Limiter threshold:
❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚❚ -0.1dBFS (-6.1dB, onset of limiting)
The unit dBu was brought in for no reason, with absolutely no consequence!
Now, I feel that a threshold of +0.1dBFS would be alright as well. The source on its own, without any processing (or mistakes) would then never be caught by the limiter.
krivium said:
Well, there're few things in common. However, a key difference is that Mark was indicating different time-constants for peak and RMS limiting, while the tables on pgs. 1&7 of the above article seem to suggest exactly the same timings for both limiter types. Anyway, thanks for sharing that.
Hi Newvirus,
About dbfs it is a defined value, as dbspl, dbu, dbv, etc,etc,.. so you can't use them to define anything else than the standard it is associated with.
Db ( without anything related) are relative scale only giving relative value difference ( hence why it is often misleading) so better keep the one defined ( standard) for what they are and don't mix them with relative scale.
You can't go above 0dbfs. It is a digital ceiling.
This is a defined norm accepted everywhere in the world and defined by an AES/EBU norm:
dBFS - Wikipedia
In analog the ceiling is located the other way around and defined by noisefloor and the other end of scale is open ended ( as long as the receiver is able to cope with voltage nothing stop you to send 100v rms for example). Digital is closed end both end of scale.
So you can decide to use it as you do but it doesn't mean anything, and from pro you'll still have same answer: it doesn't exist or mean anything.
There is relation to dbu. In fact there is standard defined in pro world: + 4dbu is the standard pro line level. As you could guess the max output of analog gear is often +24dbu for studio world ( with minor variations thought: +21dbu exist as well but it come from defining the usable dynamic range from 20db to 17db).
There is some relation between dbu and dbfs: in pro world it is often defined that 0vu ( analog) is defined at -18dbfs ( it is the case within my Yamaha 02r desk as well as most digital gear i met in the last 25years). It's not 20db as i stated but close enough ( and we learned about ISP around 2005 in the field i was involved as it is in part responsible of the 'cold digital' sound associated with first gen digital gear. In practice i've used the first meters including this ISP indicators around 2010).
So anything above dbfs doesn't exist, and your way to display it is just wrong.
You could rephrase it as once past 0dbfs you 'eat' headroom allowed in analog.
But yeah you are right if your gain structure is well thought there is no need for a limiter ( this is the whole purpose of headroom, protection against this kind of things. Youcould call it 'passive' approach in rgards to limiters which coud be called 'active' solution)
About rms limiting i didn't mentioned it for a good reason ( i only talked about peaks protection): there is no need to protect against thermal compression in your case.
The only risk you'll see is from peaks. If you had to run full blast for longer period of time or with variable dynamic input signal the question will be different ( the free party example, live example or the dj Waxx talked about).
In that case yes an rms compressor would or could be needed.
But the way to define the thresholds are differents ( based on AES drivers spec mainly). You focus on time constants but for me this is a no brainer ( as Plasnu stated if there is something automated just use it ), where to put this protections into action is different.
Like Camplo i suggest you to find some compressor/limiter vst plug ins and play a bit with them to get what it is about with already processed materials ( mastered: where we put the last limiters and compressors on the 2bus ( final stereo track). You'll soon understand why i'm cautious about this.
There is difference of response not only from time constant but ratio ( rms will not use high ratio, limiting will) and threshold are as important.
About dbfs it is a defined value, as dbspl, dbu, dbv, etc,etc,.. so you can't use them to define anything else than the standard it is associated with.
Db ( without anything related) are relative scale only giving relative value difference ( hence why it is often misleading) so better keep the one defined ( standard) for what they are and don't mix them with relative scale.
You can't go above 0dbfs. It is a digital ceiling.
This is a defined norm accepted everywhere in the world and defined by an AES/EBU norm:
dBFS - Wikipedia
In analog the ceiling is located the other way around and defined by noisefloor and the other end of scale is open ended ( as long as the receiver is able to cope with voltage nothing stop you to send 100v rms for example). Digital is closed end both end of scale.
So you can decide to use it as you do but it doesn't mean anything, and from pro you'll still have same answer: it doesn't exist or mean anything.
There is relation to dbu. In fact there is standard defined in pro world: + 4dbu is the standard pro line level. As you could guess the max output of analog gear is often +24dbu for studio world ( with minor variations thought: +21dbu exist as well but it come from defining the usable dynamic range from 20db to 17db).
There is some relation between dbu and dbfs: in pro world it is often defined that 0vu ( analog) is defined at -18dbfs ( it is the case within my Yamaha 02r desk as well as most digital gear i met in the last 25years). It's not 20db as i stated but close enough ( and we learned about ISP around 2005 in the field i was involved as it is in part responsible of the 'cold digital' sound associated with first gen digital gear. In practice i've used the first meters including this ISP indicators around 2010).
So anything above dbfs doesn't exist, and your way to display it is just wrong.
You could rephrase it as once past 0dbfs you 'eat' headroom allowed in analog.
But yeah you are right if your gain structure is well thought there is no need for a limiter ( this is the whole purpose of headroom, protection against this kind of things. Youcould call it 'passive' approach in rgards to limiters which coud be called 'active' solution)
About rms limiting i didn't mentioned it for a good reason ( i only talked about peaks protection): there is no need to protect against thermal compression in your case.
The only risk you'll see is from peaks. If you had to run full blast for longer period of time or with variable dynamic input signal the question will be different ( the free party example, live example or the dj Waxx talked about).
In that case yes an rms compressor would or could be needed.
But the way to define the thresholds are differents ( based on AES drivers spec mainly). You focus on time constants but for me this is a no brainer ( as Plasnu stated if there is something automated just use it ), where to put this protections into action is different.
Like Camplo i suggest you to find some compressor/limiter vst plug ins and play a bit with them to get what it is about with already processed materials ( mastered: where we put the last limiters and compressors on the 2bus ( final stereo track). You'll soon understand why i'm cautious about this.
There is difference of response not only from time constant but ratio ( rms will not use high ratio, limiting will) and threshold are as important.
Last edited:
No it's not.
Since the beginning i talk about cycles allowed to clip. There is a reason: except if you use 'predictive' limiting ( lookahead, with audio delayed wrt sidechain signal driving your limiter) there is no brickwall limiters. Well it is not true as you could use a diode but... even if it has it's place in the producer's tools palette it does have a sound as it rise harmonic distortion ( Neve 33609 is a studio example, guitarists use the same principle and it is called a stompbox distortion.
). A bit counterproductive you'll agree i'm sure as this is what we try to avoid...What would happen if we didn't have oversized the amps and allowed for one cycle delay before action?
With 64w amp you will clip even with limiters on duty.
With 256w amp you will never clip ( even with limiters off)... the passive approach.
And why gain structure is important. Active solution is another step of protection but it is dedicated to specific case ( and doesn't discard the passive approach to be applied).
Last edited: