need help with figuring out my ELC without reference grade gear

[Joe Bloggs]

This is a very vast topic I can't hope to answer in one go... but I'd just start by saying that I think that the idea of putting together your ELC (what does ELC stand for and where did this term emerge, btw? I think I had something to do with it but I'd be darned if I remember anything of it ) by listening to any number of steady pure tones (like 60 tones as you propose) is a bad idea.

To understand why just look at a typical loudspeaker FR measurement:



Ignore the overall hilly nature of the graph, because this wasn't measured via a measurement mic on a finally calibrated system in this case; what I want you to look at is the sheer number of jagged edges to the response, which gets crazier the higher in frequency you go. These are caused by the sheer number of reflected sound waves from different directions, at different distances, in your typical listening room. Each reflection causes its own comb filtering effect, and the combined result is a frequency response that looks like a million-teeth comb even after correction.

Thus any pure tone you listen to will inevitably fall on a point or a crack in the response and be several dBs off from the average value. Deriving ELC from speakers by listening is, well, hard.

Theoretically the best technical solution may be to replace pure tones with some synthetic signal that has a narrow but non-zero bandwidth (unlike a pure tone), sort of like a narrow-band noise signal but with steady amplitude. But I don't know if such a beast exists, and certainly don't know of any existing software that lets you sweep such a signal up and down in frequency in real time.

As an alternative, I suggest using your choice of signal generator to sweep the test frequency up and down more or less quickly. I find that at a certain speed, the spikes of the comb blend together so that on a calibrated system you can perceive the sweep as being more or less steady in amplitude after applying the ELC, whilst still being slow enough that you can pick out wider band peaks and dips that should be corrected in the speaker system or applied to the ELC, as the case may be...

Best regards,
Joe

 

[VNandor]

It came up like a month ago when I asked you something on your EQ tutorial thread.
 
Eww, that FR looks nasty, and looking at the responses here, I won't be able to get mine to be much better than something like this.
Thanks for the suggestion, Would it be good if I made a sine sweep from let's say 300Hz to 400Hz and made the sweep to go back and forth pretty fast? Something like this: https://drive.google.com/open?id=0B1RKepESfvrWdndqREpHWWpjMHc
And then using that particular tone for the sweeping.

 

[gregorio]

  What you describe however doesn't seem like an unsolvable problem with just EQ. I may not be able to boost certain frequencies but I could reduce the frequencies as I please? Because if that's the case, I could just turn up the volume to the point where I only have to use the EQ for reducing the frequncies.

 
Whether you boost the frequencies suffering from phase cancellation with tight (high Q) EQ or boost everything with your volume knob and use EQ to reduce everything else, makes no difference! Either way you are boosting equally both the level of the direct sound and of the reflections causing the cancellations. And, even if your suggestion would work in theory (which it wouldn't) you cannot use a PEQ to cut a wide range of frequencies with the accuracy which would be required.
 
In the case of peaks in response, phase summations (resonances) rather than phase cancellations, we have a slightly different outcome when attempting to correct with just EQ, though still NOT an ideal/acceptable solution. For example, a signal at -6dB relative to another signal will contain the same amount of total energy if it lasts twice a long. In other words, we could reduce a resonant frequency with a PEQ until the FR measures flat but unfortunately we are likely to hear this freq as sounding quieter because beyond very short durations our hearing can identify loudness/duration. This is where the use of a waterfall plot becomes essential.
 
Without stopping reflections from interacting with the direct sound, either by absorbing the reflections or re-directing them (in the case of diffusers), some of the issues you are going to face will respond well to treatment with EQ but others can only be marginally improved by the use of EQ or not improved at all. Ultimately, you're just going to have to accept a relatively high degree of inaccuracy. From your replies so far, I suspect that you're only going to appreciate the magnitude of the problem and the inefficacy of PEQ when you get a measurement mic/REW and discover for yourself.
 
G

 

[gregorio]

I hadn't seen Joe Bloggs' response before I posted my reply.
 
Quote:

 
Eww, that FR looks nasty, and looking at the responses here, I won't be able to get mine to be much better than something like this.

 
That FR does not look nasty, it looks pretty average. If anything, I'd generally expect one or two more very sizeable (>10dB) peaks and troughs between 100Hz and 600Hz. In that band, it looks pretty much like a top class recording studio would be hoping to achieve and that's with all the acoustic treatments at their disposal! You're right, you'd be doing well to equal or significantly improve upon the sort of FR in the posted example with EQ alone.
 
It's truly bizarre that some hardcore audiophiles are apparently sensitive to fractions of a dB in FR while simultaneously being utterly unaware/insensitive to the relatively massive FR differences/inaccuracies present in every listening environment. I'm not accusing you personally of being one of those audiofools, just sayin'
 
G

 

[Joe Bloggs]

Hmm, regarding the question of nulls in the frequency response: I'd just like to say that phase cancellations' already taken into account by the initial frequency response measurement, which is why you get like -30dB nulls. If you could boost the measured frequency by the stated deficiency amount, you do get your decibels back. Put another way, phase cancellation's a proportional phenomenon, but so are volume relationships. You get a frequency that is cancelled out 90%, now double the power. The new perceived power is 10% of double, which is still double that of the original.

Using my ears and mics to probe a loudspeaker environment and a PEQ capable of arbitrarily narrow boosts to arbitrary magnitudes, that's my experience in real life too. At least, up to a few hundred Hz--beyond that the peaks and nulls become so tightly packed and so location dependent that it doesn't make sense to try to correct them anymore.

 

[gregorio]

If you could boost the measured frequency by the stated deficiency amount, you do get your decibels back. Put another way, phase cancellation's a proportional phenomenon, but so are volume relationships.

Using my ears and mics to probe a loudspeaker environment and a PEQ capable of arbitrarily narrow boosts to arbitrary magnitudes, that's my experience in real life too.

 
That neither agrees with the theory nor with my personal experience. If a reflection arrives exactly 180deg out of phase and of exactly the same amplitude as the direct signal, you will get total phase cancellation, boosting the EQ at that freq, even by a million dB will still result in total phase cancellation at that listening location. In practice, we rarely see a reflection exactly 180deg out of phase AND of exactly the same amplitude, as there is usually some loss of energy in the reflection and therefore we typically get only partial cancellation, rarely more than about 60dB. Therefore EQ can have an effect, however I would expect a -30dB dip to usually require considerably more than 30dB of boost to level out because even with significant loss of amplitude in the reflection, boosting the direct sound by 30dB will almost always still increase the amplitude of the reflection by some amount, thereby increasing the amount of phase cancellation by that same amount. In my experience, adding say 30dB of boost to a -30dB dip would likely result in the dip being reduced to around -20dB or -10dB but still a significant dip. I've only played around with boosts of that size though, in practice it wouldn't be worth in effect loosing 30dB of system gain, even if 30dB of EQ would solve the problem and of course assuming you had a PEQ capable of 30dB of band gain. Additionally, as you mentioned, even if you did have a suitable EQ, it's use did completely solve the problem and you could afford to loose 30dB of system gain, it's still impractical above about 600Hz-800Hz anyway!
 
I don't know of any commercial studios who don't use some EQ for room correction but likewise I don't know of any able to get decent results by only using EQ correction. IME and from what I've been told by studio designers/acousticians, EQ can only be relied upon for roughly 10% of the necessary correction. The heavy lifting needs to be done by construction and physical treatments and generally the smaller the room, the more compromises you end up with. And by "decent results", we're still talking about FR variations which are massive compared to the minuscule differences many audiophiles seem to consider "night and day". I don't doubt or dispute that the OP can almost certainly improve his room with EQ alone, I'm just saying that even after that improvement it's still going to be pretty wildly inaccurate, probably by far more than he would believe until he tries and measures his room for himself.
 
G

 

[VNandor]

Fine. So it's quite obvious I will have to deal with rather big inaccuracies because of the room. So I assume even a very basic DIY treatment could improve the results from very bad to just simply bad. Is there anything I could do before starting to take the measurements, any universal rules? I have a matress-sized sponge around, carpets, rugs, and some wood waste. Or there isn't much I can do and have to take the measurements and place my sponges rugs and cobbled together planks around the wall and see how it changes the FR?

 
I could take out the speakers to the garden which is way more "open" than a room. Still, there are fences, the wall of the house, trees etc. but that way I would have to deal with less reflections. Obviously, I would have to deal with higher ambient noise that way... It's rather quiet here, if the wind starts to blow a bit there's not much else to hear. Should I do that or it really depends on how noisy the enviroment is (as "rather quiet" is not exactly a precise indicator of noise level)?
 
I'm really grateful for all the help I got by the way, I knew it would be a good idea to ask here how I should get started.

 

[gregorio]

  So I assume even a very basic DIY treatment could improve the results from very bad to just simply bad.

 
Actually, you can get an excellent bang for the buck with DIY treatments, obviously depending on how much time and effort you're willing to put in and how much you're prepared to alter the visual appearance of your room. Actual financial cost of the required materials is pretty low though, make up some wooden frames, fill with rockwool, cover with some nice looking material and hang in the right places, all for a couple of hundred bucks and you'd almost certainly get a far greater improvement than if you spent 10 or more times that amount on more expensive speakers! A couple of hundred bucks more and you can make your own diffusers, which would be better still. There are forums elsewhere dedicated to DIY acoustic treatment and adjectives such as "night and day" really are applicable here! If you haven't got the time, money or desire for that level of effort though, simple rugs hung against the wall at the primary reflection points will still make a noticeable difference. If you can hang them with a couple of inches gap between the rug and the wall, you will significantly enhance their absorption efficiency. Even with unlimited funds/treatments won't get you a flat FR though, it's just about making significant improvements.
 

  I could take out the speakers to the garden which is way more "open" than a room. Still, there are fences, the wall of the house, trees etc. but that way I would have to deal with less reflections.

 
Obviously that's only a solution for a one time test but yes, it would likely reduce the number of reflections and therefore the number of issues. The more in the middle of the garden, the further from fences or other reflective surfaces your speakers are, the less signal cancellation or summing you're going to get. It still won't be a flat FR but it should be significantly better than an untreated room in your house. Concrete is very acoustically reflective so putting your speakers on stands on say a lawn would be far better than on the patio and pointing towards a wooden fence better than pointing at your house.
 
G

 

[castleofargh]

I believe Joe was talking about attenuation from phase differences, not absolute cancellation. most of the time you can indeed boost or attenuate some frequencies, they may not respond with +1db for a 1db EQ, but they still may respond depending on the phase. 
how will the driver handle a +15db or +20db boost at that frequency that might be needed to get the FR we want? perhaps not too well, so the the problem doesn't just go away and it's still better to try and deal with the reflection itself when possible. that much is clear IMO. can I agree with both of you and not burn in hell? ^_^
 
 
 
@VNandor you have most likely between 2 and 5 db variations somewhere between left and right in most of your headphones, and almost nobody notices it unless it's large and placed on the frequencies where the voice is so it moves out of the center imaging position. also just taking the headphone off and putting it on again, you might just have created as much change in the signature because of the positioning variations. I'm certainly not saying that it's a reason to give up, there are a lot of things one can do to improve his audio experience thanks to objective data. just that when you start doing measurements and see how the unicorn utopia created by audiophiles is a giant joke, you stop being concerned that 1db will change your life.

 

[VNandor]

   
Actually, you can get an excellent bang for the buck with DIY treatments, obviously depending on how much time and effort you're willing to put in and how much you're prepared to alter the visual appearance of your room. Actual financial cost of the required materials is pretty low though, make up some wooden frames, fill with rockwool, cover with some nice looking material and hang in the right places, all for a couple of hundred bucks and you'd almost certainly get a far greater improvement than if you spent 10 or more times that amount on more expensive speakers! A couple of hundred bucks more and you can make your own diffusers, which would be better still. There are forums elsewhere dedicated to DIY acoustic treatment and adjectives such as "night and day" really are applicable here! If you haven't got the time, money or desire for that level of effort though, simple rugs hung against the wall at the primary reflection points will still make a noticeable difference. If you can hang them with a couple of inches gap between the rug and the wall, you will significantly enhance their absorption efficiency. Even with unlimited funds/treatments won't get you a flat FR though, it's just about making significant improvements.
 
 
Obviously that's only a solution for a one time test but yes, it would likely reduce the number of reflections and therefore the number of issues. The more in the middle of the garden, the further from fences or other reflective surfaces your speakers are, the less signal cancellation or summing you're going to get. It still won't be a flat FR but it should be significantly better than an untreated room in your house. Concrete is very acoustically reflective so putting your speakers on stands on say a lawn would be far better than on the patio and pointing towards a wooden fence better than pointing at your house.
 
G

Well, the original intention is to find the best one-time solution so I can get my ELC measured. Throwing the speakers out of the house is a more entincing idea than trying to treat the room;  it takes less time, and more importantly, less effort and hopefully I would get similar level of accuracy at the least.
 
Looking at the placement possibilities, I think I probably found the best possible place. I could place some chair on the top of the garage an place the speaker on that. The chair could be either padded or hard-surface don't know how much that counts. That would be about 3 meters above the street level which means there would be no fences to interfere. Our house's wall would be behind the speakers. The garage is 5m long so I can't get the measurements done farther than that but it's not like I had more place than this in the room. The street is like 8m wide and I can place the speaker in a way it points against the neighbour's lawn instead of their house so that's a lot of free place. That would definitely work better than a room? (I'm sure I can work it out with my neighbours.)
 
Is it better to use only one speaker than two? I really don't need anything stereo for this (I hope so) and the placement of the two speakers would just cause further complications.

 

[VNandor]

   
@VNandor you have most likely between 2 and 5 db variations somewhere between left and right in most of your headphones, and almost nobody notices it unless it's large and placed on the frequencies where the voice is so it moves out of the center imaging position. also just taking the headphone off and putting it on again, you might just have created as much change in the signature because of the positioning variations. I'm certainly not saying that it's a reason to give up, there are a lot of things one can do to improve his audio experience thanks to objective data. just that when you start doing measurements and see how the unicorn utopia created by audiophiles is a giant joke, you stop being concerned that 1db will change your life.

Hey, I can hear that 1dB when a sine sweep is played or when a signal is 1 dB louder than the other.

Probably couldn't notice an 1dB dip somwhere in a pink noise though. I'm not worrying of 1 dB hell even a live concert doesn't sound the same from all positions and I won't start to bitch about how the reflections (or maybe the lack of it) of the stage ruined my experience. Well, most of the times I don't. I don't think I'm chasing that last bit, I feel like every professional would run away from my idea of "measuring". But that's why I'm here, I'm hoping to get the best out of my possibilities even if the best is not all that great... and to learn a bit from the process.

 

[gregorio]

  I believe Joe was talking about attenuation from phase differences, not absolute cancellation. most of the time you can indeed boost or attenuate some frequencies, they may not respond with +1db for a 1db EQ, but they still may respond depending on the phase. 
how will the driver handle a +15db or +20db boost at that frequency that might be needed to get the FR we want?

 
 
Actually, if you read beyond the first sentence of my post I think you'll find I said pretty much the same thing.
 

  That would definitely work better than a room? (I'm sure I can work it out with my neighbours.)

 
Impossible to say. On the face of it, neither the wall behind the speakers nor the reflective surface of the garage roof are good acoustically, earth or earth + grass/plants would be much better. Measuring a significant distance from the speakers would also not help, you'd get a higher level of reflections relative to the direct sound. So within reason, the closer to the speakers the better, 3 or 4ft (1m or so) probably being about ideal and yes, definitely just use the one speaker.
 
G

 

[VNandor]

   
Impossible to say. On the face of it, neither the wall behind the speakers nor the reflective surface of the garage roof are good acoustically, earth or earth + grass/plants would be much better. Measuring a significant distance from the speakers would also not help, you'd get a higher level of reflections relative to the direct sound. So within reason, the closer to the speakers the better, 3 or 4ft (1m or so) probably being about ideal and yes, definitely just use the one speaker.
 
G

Point is, there wouldn't be anything around that could cause reflection if I placed it on the garage. Except the garage itself and the surface behind it. Actually I could apply a pretty neat "treatment" there. Throwing all the textiles I can find in the house on the garage and the wall behind doesn't take as much planning as trying to treat a room. That's just two surfaces and one is horizontal, which makes things a whole lot easier because there's no way i could apply space between the absorber and the surface.
 
If I placed the speaker in the lawn, there would be a 7m*7m square in front of it. From the side there's the garage made out of brick and the other side is a solid fence... definitely hard surface but I don't know its material. Slapping rugs on them wouldn't be too hard, hanging them down from the garage and simply just putting some on the fence. Nothing behind it which is a good thing I guess. And in front of there's a wooden fence made out of planks, and they are spaced, I could stick my hand between them.

 

[Joe Bloggs]

That neither agrees with the theory nor with my personal experience. If a reflection arrives exactly 180deg out of phase and of exactly the same amplitude as the direct signal, you will get total phase cancellation, boosting the EQ at that freq, even by a million dB will still result in total phase cancellation at that listening location. In practice, we rarely see a reflection exactly 180deg out of phase AND of exactly the same amplitude, as there is usually some loss of energy in the reflection and therefore we typically get only partial cancellation, rarely more than about 60dB. Therefore EQ can have an effect, however I would expect a -30dB dip to usually require considerably more than 30dB of boost to level out because even with significant loss of amplitude in the reflection, boosting the direct sound by 30dB will almost always still increase the amplitude of the reflection by some amount, thereby increasing the amount of phase cancellation by that same amount. In my experience, adding say 30dB of boost to a -30dB dip would likely result in the dip being reduced to around -20dB or -10dB but still a significant dip. I've only played around with boosts of that size though, in practice it wouldn't be worth in effect loosing 30dB of system gain, even if 30dB of EQ would solve the problem and of course assuming you had a PEQ capable of 30dB of band gain. Additionally, as you mentioned, even if you did have a suitable EQ, it's use did completely solve the problem and you could afford to loose 30dB of system gain, it's still impractical above about 600Hz-800Hz anyway!

G

Put another way, phase cancellation's a proportional phenomenon, but so are volume relationships. You get a frequency that is cancelled out 90%, now double the power. The new perceived power is 10% of double, which is still double that of the original.

I challenge you to the following experiment:
1. Find a low frequency in your speaker system with the worst null
2. Play a sine tone at the frequency and put a dB meter at your listening position
3. Now crank up the sine tone 6dB in volume (via adjustment of the signal generator itself) and note the change in the dB meter's reading.

I'm ready to eat a few unpleasant large objects if the dB meter doesn't register almost exactly 6dB change in volume.

 

[gregorio]

I challenge you to the following experiment:
1. Find a low frequency in your speaker system with the worst null
2. Play a sine tone at the frequency and put a dB meter at your listening position
3. Now crank up the sine tone 6dB in volume (via adjustment of the signal generator itself) and note the change in the dB meter's reading.

I'm ready to eat a few unpleasant large objects if the dB meter doesn't register almost exactly 6dB change in volume.

 
I've done an essentially identical experiment: Pink noise, take an FR plot (using REW and a measurement mic), identify a freq with a 12dB dip, then add 12dB with a PEQ at that freq and remeasure. Result: About an 8dB dip instead of a 12dB one. To get the dip to zero couldn't be achieved with the PEQ I had as it maxed out at +18dB and even double the 12dB gain was not enough to flatten it! This is just one of easy more than 100 similar examples.
 
G