I've always balanced and eq'ed my pa systems by ear, with an occasional use of swept sine waves. I've gotten pretty decent results, but now that I'm building custom subs, I want to learn how to do this right.
This weekend, for the first time, I used RTA software. (Real Audio's TrueRTA, which I'm not sure was the right purchase, but good enough for now.)
Man, I thought it would be easy! Famous last words.
My first attempt was with pink noise, and a 1/6th octave read out. I played around with processor setting while watching the RTA display, then checked with actual song audio when I thought I was close, or when I had to choose between two different settings. I must say, it sound better than when I did it by ear, but that bloody RTA display refuses to get flat. - And when I do get CLOSE to flat, down to maybe 35 hz, the song audio sounds overly LF-heavy, to the point that I'm losing "punch" from the subs.
Uggh. I'm totally in the weeds here.
I don't expect easy answers on a forum, but can someone point me to a good tutorial that explains the typical steps used to achieve this goal?
--------------------------------------
Also: I intended to use TrueRTA for both PA setups and room acoustic tuning. (The latter is something I did years ago, as an extra business designing recording studios.) . I'm disappointed that the software I bought has no warble tones and no waterfall display, which were critical components of my old hardware-based system.
Can anyone recommend a really good software system that can do all of this?
thx.
This weekend, for the first time, I used RTA software. (Real Audio's TrueRTA, which I'm not sure was the right purchase, but good enough for now.)
Man, I thought it would be easy! Famous last words.
My first attempt was with pink noise, and a 1/6th octave read out. I played around with processor setting while watching the RTA display, then checked with actual song audio when I thought I was close, or when I had to choose between two different settings. I must say, it sound better than when I did it by ear, but that bloody RTA display refuses to get flat. - And when I do get CLOSE to flat, down to maybe 35 hz, the song audio sounds overly LF-heavy, to the point that I'm losing "punch" from the subs.
Uggh. I'm totally in the weeds here.
I don't expect easy answers on a forum, but can someone point me to a good tutorial that explains the typical steps used to achieve this goal?
--------------------------------------
Also: I intended to use TrueRTA for both PA setups and room acoustic tuning. (The latter is something I did years ago, as an extra business designing recording studios.) . I'm disappointed that the software I bought has no warble tones and no waterfall display, which were critical components of my old hardware-based system.
Can anyone recommend a really good software system that can do all of this?
thx.
First question is where are you doing this work... indoors or outside?
The only way you will know you are only looking at speaker response(especially with subs) is to measure outdoor and well away from buildings and structures. Indoors the room size and shape and the subs position within the room totally dominates response. tops are easier to do indoors but you have to close mic the cabs and gate the response.
You do want to know what your speakers raw response looks like so you're not trying to make them do something they can't.. particularly relevent with subs. But you also have to know that you can't fix all response anomalies with EQ alone, some of it will come from phase and time correction so you will need software that can display that.
My observations with a live RTA are similar to yours... there is just too much going on with too much variation to make many conclusions. What it will show you is how much the speaker response varies when the listening position(mic) is moved around. That info will come in handy later on when you attempt to do in-room measurements and you come to the realization you have to take a series of measurements across the space and average them.. throwing out any outliers.
I like to use sine sweeps and then apply 1/12th smoothing to what is captured, it keeps the average response shape without taking away too much data. You wont ever remove every single peak or though but you can produce a relatively linear response... although I don't go for ruler flat I like to target something that looks like this Harman response curve with a lowend boost and gently rolloff at the topend. That dropoff above 10khz is also OK, if you can do a little better that is fine but don't wreck the system trying to make it go all the way out to 20khz as not many can hear it.
The only way you will know you are only looking at speaker response(especially with subs) is to measure outdoor and well away from buildings and structures. Indoors the room size and shape and the subs position within the room totally dominates response. tops are easier to do indoors but you have to close mic the cabs and gate the response.
You do want to know what your speakers raw response looks like so you're not trying to make them do something they can't.. particularly relevent with subs. But you also have to know that you can't fix all response anomalies with EQ alone, some of it will come from phase and time correction so you will need software that can display that.
My observations with a live RTA are similar to yours... there is just too much going on with too much variation to make many conclusions. What it will show you is how much the speaker response varies when the listening position(mic) is moved around. That info will come in handy later on when you attempt to do in-room measurements and you come to the realization you have to take a series of measurements across the space and average them.. throwing out any outliers.
I like to use sine sweeps and then apply 1/12th smoothing to what is captured, it keeps the average response shape without taking away too much data. You wont ever remove every single peak or though but you can produce a relatively linear response... although I don't go for ruler flat I like to target something that looks like this Harman response curve with a lowend boost and gently rolloff at the topend. That dropoff above 10khz is also OK, if you can do a little better that is fine but don't wreck the system trying to make it go all the way out to 20khz as not many can hear it.
I'm indoors and yea I understand the problems with that, but I have to get things close so that later I spend minimal time outside. The neighbors would have me strung up. lol ....
I actually made a make-shift acoustic chamber for the test, using 4" of 4 lb/cubic inch mineral wool and then 1/2" of soundboard, (I used to build recordind studios for a living) so that drastically minimizes all nodes down to maybe 80 Hz. After that, I dunno. I suppose I could stack up a bunch of cinder blocks, if I HAD a bunch of cinder blocks, lol.... But then I'd be getting reinforcement from THAT, so still no accuracy.
I'm also keeping the volume low, to at least minimize any resonances.
Once I take it outside, I shouldn't have to use any tones above 120 Hz since my basement test-chamber is clean above that. That should help a lot with the neighbors. When they hear THAT, they'll probably think it's just trucks going by on the highway.
I actually made a make-shift acoustic chamber for the test, using 4" of 4 lb/cubic inch mineral wool and then 1/2" of soundboard, (I used to build recordind studios for a living) so that drastically minimizes all nodes down to maybe 80 Hz. After that, I dunno. I suppose I could stack up a bunch of cinder blocks, if I HAD a bunch of cinder blocks, lol.... But then I'd be getting reinforcement from THAT, so still no accuracy.
I'm also keeping the volume low, to at least minimize any resonances.
Once I take it outside, I shouldn't have to use any tones above 120 Hz since my basement test-chamber is clean above that. That should help a lot with the neighbors. When they hear THAT, they'll probably think it's just trucks going by on the highway.
Last edited:
QUOTE: "I like to use sine sweeps and then apply 1/12th smoothing to what is captured, it keeps the average response shape without taking away too much data"
-----------------------------------------
What software are you using for this?
FWIW, after a full day of tweaking, I realized that my software can do a sort of "real-time averaging" of the pink noise. It also can display the bands as bars, which is much easier to read. Once I got this happening, things went a little better.
Like you, the manual recommends NOT using pink noise for speaker setups, but rather something they call a "quick sweep." It's not working on my system for some reason, but the description sound as you describe: an automatic, low-distortion sine wave sweep. It doesn't mention smoothing, but must do so, as a regular sine wave is way too thin and will show massive peaks and valleys all over the place.
AND YET: I like the idea of using averaged pink noise first, as long as it is continuously updated. This is because I can then make changes to my processor settings, and see them in "real time." (Only a few seconds of lag each time, as I start & stop the analyzer.) I figure this can get me close, a lot faster than constantly running sweep tests. - I'm thinking that I would then run the sweep test once my pink noise tweaks get me in the ballpark. Maybe save a few setups that are all close, then sweep those for selection and fine tuning.
Does that make sense, or can sine sweeps be done really fast?
------------------
My last thought for today:
The interraction between filters is such a bloody challenge. SO much back & forth tweaking......
I'm wondering if the newer EAW processor might be easier to work with, as it has FIR filters (I think mine has IIR) and so probably much less phase shift. - That thing is EXPENSIVE, though. Ughh...
-----------------------------------------
What software are you using for this?
FWIW, after a full day of tweaking, I realized that my software can do a sort of "real-time averaging" of the pink noise. It also can display the bands as bars, which is much easier to read. Once I got this happening, things went a little better.
Like you, the manual recommends NOT using pink noise for speaker setups, but rather something they call a "quick sweep." It's not working on my system for some reason, but the description sound as you describe: an automatic, low-distortion sine wave sweep. It doesn't mention smoothing, but must do so, as a regular sine wave is way too thin and will show massive peaks and valleys all over the place.
AND YET: I like the idea of using averaged pink noise first, as long as it is continuously updated. This is because I can then make changes to my processor settings, and see them in "real time." (Only a few seconds of lag each time, as I start & stop the analyzer.) I figure this can get me close, a lot faster than constantly running sweep tests. - I'm thinking that I would then run the sweep test once my pink noise tweaks get me in the ballpark. Maybe save a few setups that are all close, then sweep those for selection and fine tuning.
Does that make sense, or can sine sweeps be done really fast?
------------------
My last thought for today:
The interraction between filters is such a bloody challenge. SO much back & forth tweaking......
I'm wondering if the newer EAW processor might be easier to work with, as it has FIR filters (I think mine has IIR) and so probably much less phase shift. - That thing is EXPENSIVE, though. Ughh...
I'm far from an expert on this but I personally feel that trying to tune subs for a flat response is not the best idea. Just choose a design from the outset that is close to what you want, and leave it flat. The reason being is that 1) say if you are boosting 3db at 35hz to bring it up then unless you have at least 3db of overhead (twice a much power as you need) you are losing 3db of overall level. That's quite a lot. Perhaps not in tracks that dont contain that frequency but once you play one that does then you will possibly have a noticeable drop in level (e.g. punch), particularly if you are already running close to the limiters.
2) Your sub may be flat but then as soon as its placed in a room the response will likely go off anyway because of the room nodes, which then change whether the room is full or empty. Maybe for a permanent install or if you notice in a given room there is a particularly wild node then an EQ tweak could smooth out a bit (EQ cut rather than boost though!). Boosting to bring up a sucked out frequency due to either sub response or room node, you may need an awful lot of extra power to do this effectively.
2) Your sub may be flat but then as soon as its placed in a room the response will likely go off anyway because of the room nodes, which then change whether the room is full or empty. Maybe for a permanent install or if you notice in a given room there is a particularly wild node then an EQ tweak could smooth out a bit (EQ cut rather than boost though!). Boosting to bring up a sucked out frequency due to either sub response or room node, you may need an awful lot of extra power to do this effectively.
Hey, Damos.
Your points re the problems are valid, but I think it's always best to start form "neutral." Or maybe neutral plus a little extra on the bottom.
Also, I keep my subs clustered in the middle, so therefore I get a fairly clean "power alley" (that's an actual term 🙂 ) in the dance area. This can then caused even larger peaks & dips at the sides and back of the room, but you can never really deal with that, anyway.
I suppose it would be wise to do an RTA test in the power alley, at each venue, then do a quick compensation for the worst non-linearities, but personally I would never have time. I'd LOVE to have some kind of automatic eq that could do that instantly and at low volume.
Hmmm... I wonder if that exists?
Your points re the problems are valid, but I think it's always best to start form "neutral." Or maybe neutral plus a little extra on the bottom.
Also, I keep my subs clustered in the middle, so therefore I get a fairly clean "power alley" (that's an actual term 🙂 ) in the dance area. This can then caused even larger peaks & dips at the sides and back of the room, but you can never really deal with that, anyway.
I suppose it would be wise to do an RTA test in the power alley, at each venue, then do a quick compensation for the worst non-linearities, but personally I would never have time. I'd LOVE to have some kind of automatic eq that could do that instantly and at low volume.
Hmmm... I wonder if that exists?
Room nodes cannot be corrected with EQ at all, the ONLY solution is moving the subs or adding more in different locations. This is one of the main reasons that system measurement in temporary locations is of little value, we are often very limited in speaker location options anyway so there isn't much that can be done even if you know exactly what the problem looks like. That's not to say you shouldn't try.. sometimes separated subs can work better(provide more even response) than a cluster indoors.. depends on the size and shape of the room, but in many cases you just have to live with what you get.
Some room modes can be corrected with EQ, but not all of them.
For instance, my listening/living room has a strong 40Hz resonance, which makes small speakers sound big, and big speakers sound ugly. Using EQ to reduce the SPL in that area gives me a LF response that's flat at the listening position.
My room is quite helpful. Others are less so. Dips in the frequency response should be approached with caution: +3dB = 2x power, so large EQ boosts can be very problematic.
Chris
For instance, my listening/living room has a strong 40Hz resonance, which makes small speakers sound big, and big speakers sound ugly. Using EQ to reduce the SPL in that area gives me a LF response that's flat at the listening position.
My room is quite helpful. Others are less so. Dips in the frequency response should be approached with caution: +3dB = 2x power, so large EQ boosts can be very problematic.
Chris
That is the important take away, yes you can reduce the 40hz intensity at that position but that doesn't change the number and location of nodes in the room as a whole, and there will now be positions in the room with a 40hz hole. This is what I referring to with the "can't fix it with EQ" comment.
Measuring subs should be done >2m ground plane outdoors. https://data-bass.com/#/articles/5cc0bc36a75a260004255c88?_k=r7n1f7 once you have the subs correctly processed to flat in this enviroment you then array them either outdoors or in a room and conduct further measurments to correct the resulting array response (moving mic, multiple mic positions etc.). I will then integrate the array with the main speakers (delay and gain and crossover). After this full system EQ. There may however be complications with gain staging or internal DSP headroom that cause some of the corrections to be ommited (E.G delibratly running subs hot to get closer to the overall response target). There are many approaches though, it would take a whole book to cover tuning speakers, it gets especialy difficult if all the work has to be done in situe without outdoor measurments.
And to furthere chis comment about room modes some just don't respond to EQ, you put more power in but as your seeing an acoustic cancelation the SPL dosen't increase at that frequency. You have to be careful with any EQ boost (or effective EQ boost if your performing wide band cuts) in PA systems due to them eating headroom.
And to furthere chis comment about room modes some just don't respond to EQ, you put more power in but as your seeing an acoustic cancelation the SPL dosen't increase at that frequency. You have to be careful with any EQ boost (or effective EQ boost if your performing wide band cuts) in PA systems due to them eating headroom.
That's why I said to EQ for the power alley, where it matters the most and the only place you can really achieve anything workable, anyway.
FWIW, some guys like to spread their subs (using poles to hold up the tops, etc.) This will even out the room a little, creating even more comb filtering, but with less overall peaks and dips. For a listening environment, that's helpful, but for pumping the dancefloor, it's a (very common) mistake. Been there, done that. 🙂
Worry about the dancefloor, and forget how it sounds out where people are just sitting and talking. If the club owner wants to make THAT better, then bring in a waterfall display and a construction crew. 😀
Kipman, that's something I didn't know. (Will read that article later)
But wouldn't it make more sense to have the subs on the ground, to re-create the floor boundary effect when used in a club?
I can understand 2+m if the subs will be flown, but what mobile DJ / band does that?
The ">2m" part refers to the distance from mic to speaker, not speaker to ground. They are indeed measured on the ground, as specified by the following two words.
There isnt much point because as mentioned above the response can change hugely depending on how full the venue is. Generally speaking a full venue sounds so much better as all the people help to absorb reverb and standing waves. Basically, never tune a system in an empty room.
True, but not TOO much below about 300 Hz, which is the frequencies we'd be most worried about. Better to do something rather than nothing.
- And since volume doesn't matter too much for just RTA tuning, and the "automatic" eq I ask about would be able to adjust incredibly fast, you probably COULD get away with doing in in a crowded room. -Especially if you restrict it to 300 Hz and below. Set it to as many frequency bands as possible (but testing with warble or a wide sine sweep) then limit all boost / cut to maybe 3-5 dB.
Does anyone know what the pros do, at major indoor concerts?
Dual channel FFT allows you to get the transfer function out of any signal, so you can monitor the house in real time.
I finally had a chance to read this and think about it. It raises even more questions / thoughts.
It seems like that dB article is describing testing a sub BY ITSELF, for the purpose of dialing-in corrective EQ. SO yeah, having the mic closer than any boundary is going to be important. (And good to keep in mind.) And of course you would do this first, if building custom subs.
However, this thread is concerned with balancing an entire front-end, and when you think about it, the dB method is probably not a good idea for that. No matter what venue you work at, unless you fly your subs, you will always be dealing with ground plane reinforcement, so why not set your test-system up with the subs on the ground?
- You will still have to tweak for every room, if you have the time and equipment to do so, but at least you will be starting from a point that is closer to reality. Otherwise, you will have to compensate for "too much LF," at every single gig.
Even if you like your subs "purposely too loud" (as most people do, I think) you wouldn't want just that one frequency too loud. You'd want the whole sub too loud. So ......
Is my thinking somehow wrong on this?
Sorry, I'm not sure I follow your query.
None of the DataBass article, Kipman's or my posts in this thread, have said anything other than that the subs should be on the ground.