Originally Posted by jevansoh - this is the (slightly) better formatted version of Post 2. Jason - if you go into the editor and copy this, you should be able to edit it back into your own post.
Thanks to AustinJerry and all the rest of you for asking a lot of great questions, which I've answered and added to below.
To get us started, below are some FAQ’s, do’s, do not’s, how’s, and why’s on a variety of subjects that both come up often and unfortunately, don’t come up often enough.
Q: Did I configure my graphs properly? Are the horizontal and vertical scales set properly? What are the various settings in the "Controls" panel for each type of graph?
A: You will use the “Controls” and “Limits” function located at the top right of your screen in REW most often. There are different settings for different graphs and different reasons for each one which is explained below.
- Frequency Response – You need to set the top and bottom limits to fit the “unsmoothed” frequency response graph on your screen, but with as much resolution as possible. It is easy to make any graph look “good” and by not showing much resolution almost anything can be made to look flat. Pretty graphs are not what we’re after though. A good way to start is to look at the highest peak on the frequency response.
Let’s say it’s 85db for example. Click “Limits” and set the “Top” limit to no more than 90db so the highest point in the graph will be close to the top of your screen but you have a bit of room to spare with nothing spilling over the top. Then for the bottom, set it to where the bottom of the longest spike, which will always be in the upper frequencies past about 2khz or so is shown and you’ve maxed out your resolution. Then verify that on the left hand side of your screen that the graph lines are in increments of 5db. Less than this is too much resolution and more isn’t enough.
The standard is 5db and is what most people are used to looking at. If you see a lot of comb filtering and are measuring an untreated room and without EQ you may find it impossible to get within these parameters on the upper end of the frequency spectrum. As long as you show the highest peak and have about 60db showing below this, it’s okay if you can’t quite fit all the vertical lines/comb filtering on your screen, but make certain you don’t clip the top/peaks off. As for the left and right limits, these will differ depending on what you’re working on or sharing at the time. You can set the “Left” limit to 20hz for instance, and the “Right” limit to 20000hz for instance, to show the entire audible range of the frequency response, but more commonly you’ll either be measuring the effects of bass trapping and want to focus more on the lower frequencies or measuring the effects of different placement of acoustical panels, speaker placement, listener placement, etc, and be focusing more on mid and high frequencies. To show more resolution you can set the left and right limits to just show 10hz to 300hz for example to get the maximum resolution in your graphs for modal frequencies.
TIP: ALWAYS MEASURE FULL RANGE, IE: 10HZ – 20000HZ AS YOU CAN MANIPULATE THE GRAPHS LATER TO ONLY SHOW THE PORTION YOU WANT. NEVER LIMIT MEASUREMENTS TO ONLY 10HZ – 200HZ FOR EXAMPLE, AS IT ONLY TAKES A FEW MORE SECONDS TO DO A FULL MEASUREMENT SWEEP AND THAT DATA CAN BECOME INVALUABLE LATER.
- Waterfall aka Cumulative Spectral Decay – This is a much more important graph than a simple frequency response and one we’ll surely be talking about it more detail. It’s also much more important we get this one right when sharing it with others as just like the frequency response, almost any waterfall graph can be made to look good, but we want the real scoop and not pretty graphs to sell acoustical treatments.
This graph shows us frequency against time (there is a lot more to the explanation than this, but we’ll start here) and is used mainly to view lower frequencies, below 300hz or so. Recommended graph settings for left/right limits are 20hz – 300hz. If you have a subwoofer that goes lower than 20hz with substantial output, it is still not necessary to show frequencies below 20hz unless you have a very large room (longer than 30ft or more) that can modally support these lower frequencies. This graph is used to look at modal decay and can show ringing and why you may have one-note bass. It can also be used to show what your Room EQ software has done and if it’s made any of the ringing “worse” as sometimes happens. As for the “Top” and “Bottom” limits, you want to follow the same advice as above in the frequency response, showing 60db. You need to take all of your measurements with as little noise in the room as possible, but this measurement especially needs to show 60db above the noise floor, or as close to it as possible. For instance, the average American living room has an approximately 40db noise floor. That means your measurements need to be taken at about 100db which is very loud.
TIP: WEAR EAR PLUGS WHEN TAKING MEASUREMENTS. Taking measurements at this level will do a few things. First, and most importantly, it will get you above the noise floor so you can get accurate decay times and secondly it will show you how flat your subwoofer truly is down at 15-20hz when played at this volume. Lot’s of people brag they have “Flat” sub response down to 10hz, but at what level? 60db? 70db? You cannot hear or feel 10hz at 70db but at 100db if you have a sub-woofer(s) that can perform with reasonable distortion at this level, you have something to be quite proud of, along with a literally shaking foundation. If you cannot run a measurement at 100db and you cannot get your noise floor below 40db your measurements aren’t totally invalid, but you may not be able to get the full picture or capture your true in room decay times to 60db.
TIP: THE RT60 TAB/MEASUREMENT IS USELESS FOR SMALL ACOUSTICAL SPACES, WHICH IS EVERY “HOME” IN THE WORLD. THERE IS NO RT60 IN SMALL ROOMS, WHICH IS WHAT WE’RE DISCUSSING ON THIS THREAD. WE WILL NOT DISCUSS RT60 FURTHER AS IT IS MEANINGLESS, HOWEVER, DECAY TIMES IN MODAL FREQUENCIES ARE VERY IMPORTANT AND IS A BIG TOPIC ALONG WITH A HARD PROBLEM TO SOLVE.
- In short, the goal for decay times for an average sized living room or dedicated home theater in the USA is somewhere in the range of 300ms to 500ms for frequencies between 500hz and 20khz and a good rule of thumb is not more than 450hz for modal frequencies. The ultimate goal is to make the entire spectrum not differ more than 10% full range, but that’s hard to do because unless you know the tricks of the trade (which we’ll talk about on this thread) you will find that as you add more bass trapping to get the decay times down for the lower frequencies, you overdeaden/overdampen the upper frequencies. There are several workarounds for this and we’ll go into much more detail in the thread on how to compensate for this.
- ETC – Envelope/Energy Time Curve – The settings for this graph aren’t as precise, but for what we’ll mostly be looking at, I recommend showing from 0 – 40ms unless your room is longer than 30’ or so and again, show at least 60db on the left side. What we want to see here depends on the acoustical model you’ve chosen. However, a good rule of thumb, to begin with at least, is to have no spikes above -20db between 0ms and 20ms. Anything under 20db is fine. There are several nuances to this though. All the spikes should decay as evenly as possible and be spaced as evenly as possible. Having no spikes above -20db (from reference set as 0db) is effectively anechoic, or “without echo” and this is a good thing, to a point. However, this initial time delay (sometimes referred to as ITG or initial time gap) needs to be terminated at some point or you’ll have a rather dull, dead, and lifeless sounding room. A good termination point is somewhere past 15ms and no more than about 25ms. This should be as close to or above -12db as possible to terminate this initial time delay gap. This is critical in a LEDE/RFZ room and we’ll go into much more detail in the thread on the how-to. The ETC is often very misunderstood and overlooked, and this is very unfortunate as it is the most critical of all tools and by it being time based you can easily use it to calculate and solve for most other problems, so this is my personal pick for a desert island graph where if I could only pick one…
Q: How should the speakers be configured before taking these measurements? Do I measure one speaker at a time for ETC? Does the waterfall need all subs as well as the main speakers in the measurement? Is Audyssey on or off? Do I take measurements only at the MLP, or at several spots in the room?
A: There are different configurations for different types of measurements. You should always measure the full spectrum starting with the lowest frequency your subwoofer(s) will play to and going up to 20000hz no matter what you will be analyzing. Using graph limiting, explained above, you can look at whatever range you wish, but it’s good to have all the data available.
- Frequency Response – Measure L+Sub then measure R+Sub and overlay the two. NEVER measure together for this. It will show a lot of comb filtering and not give you true results. TIP: Comb Filtering doesn’t actually exist!! It’s simply the name we give to the visual effect we see on graphs! When you overlay the L+Sub and R+Sub graphs they should not deviate by more than +/- 2db from each other. If they do, you need to alter the speaker placement and/or listening position until there is as little deviation between the L and R speakers as possible.
TIP: DO NOT RELY ON ROOM CORRECTION SOFTWARE FOR THIS AND ALWAYS MEASURE WITHOUT ROOM CORRECTION ENGAGED!
- Waterfall – This is the only measurement where you do actually want to measure both the Left and Right speakers along with all subwoofers, together. You’re looking at the combined response in the modal region on this graph, so you need every speaker playing that will be playing frequencies below 300hz at the same time, which is the Left, Right, and Subs.
- ETC – See Frequency Response – Same thing applies – Overlay Left and Right and if spikes (equal to reflections) exist in one speaker but not the other, at the same point in time before 20ms or so (early reflections) then you don’t have symmetry and at the very least need to reflect, diffuse, or absorb the “different” spikes/reflections.
- You will want to take two sets of measurements each time you make a change. The great thing about REW is you can look at many different graphs, set them up to view different parameters at a time, IE: Full range or limited range (as long as you always ‘measure’ full range to begin with) and manipulate the graphs to suit your needs any time in the future. Each time you measure, you should ALWAYS first measure with NO EQ then, without moving the mic, measure WITH EQ.
Then you should move the mic a few inches to the left, and repeat, then to the right, repeat, up, down, etc. You can then look in the bottom left hand corner of your screen and click the “Average” button which will spatially average all of your measurements together. So you always want to take several measurements around your MLP and both with and without EQ for comparison and to see exactly what that EQ is really up to.
TIP: ALWAYS NOTATE YOUR MEASUREMENT BEFORE PROCEEDING TO THE NEXT MEASUREMENT. TRUST ME, YOU WILL “NOT” REMEMBER THE DETAILS LATER AND WITHOUT PROPER NOTES ON EACH MEASUREMENT, THE IMPORTANT MEASUREMENT CAN BECOME MEANINGLESS QUICKLY.
TIP: IF YOU WANT MORE RESOLUTION/DATA ON YOUR SCREEN, AT THE TOP OF THE MEASUREMENT NOTES ON THE LEFT HAND SIDE OF YOUR SCREEN, CLICK “COLLAPSE” AND THE AREA DESCRIBING EACH MEASUREMENT WILL SHRINK WHILE THE ACTUAL GRAPH WILL EXPAND. IF YOU WANT TO READ YOUR NOTES MORE CLEARLY LATER, SIMPLY CLICK THE SAME BUTTON AGAIN, WHICH WILL NOW READ, “EXPAND.”
Q: What do the waterfall graphs show? Does the Waterfall show any resonances that need to be looked at, or does the graph show something that is "good enough"?
A: Teaching you to understand how to interpret all the graphs in REW is one of the main goals of this thread.
In essence, the waterfall graph, if configured properly as detailed above, will show you certain frequencies that take longer to decay than others. If, for instance, 40hz “rings” out to 700ms but most of the rest of the frequencies stop ringing (this looks like mountains coming forward/towards you on your monitor) at 300ms, then you have a modal problem at 40hz and need to treat it with passive EQ, IE: Bass traps.
TIP: IF YOU CANNOT SEE YOUR MOUNTAINS FULLY DECAY ON YOUR MONITOR AND INSTEAD, YOU ONLY SEE “FLAT” PEAKS THEN YOU HAVEN’T SET THE “CONTROLS” TAB PROPERLY, LOCATED NEXT TO LIMITS. YOU WANT TO SET YOUR CONTROL TABS “TIME RANGE (MS) TO WHATEVER LEVEL IS NECESSARY TO SEE THE LONGEST MOUNTAIN PEAK FULLY DECAY INTO THE FLOOR. DEPENDING ON HOW STRONG THE PEAK IS AND WHETHER OR NOT YOUR ROOM IS TREATED, THIS COULD EASILY EXCEED 700-800MS. AS YOU ADD MORE TRAPS OR MOVE SPEAKERS/SUBS AROUND, THIS WILL GET SMALLER, AND THE GOAL IS TO GET ALL THE FREQUENCIES WITHIN 10% OR AS CLOSE AS POSSIBLE TO EACH OTHER AND BE ABLE TO EVENTUALLY VIEW EVERYTHING ON THE GRAPH AT NO MORE THAN 600MS MAXIMUM, PREFERABLY CLOSER TO 400MS.
Q: Once I know what I am looking at, what are the various methods to address the problems? Should I go out and buy a roll of pink fluffy?
A: Once you understand that there will certainly be peaks in the ETC, ringing in the waterfall, and uneven frequency response both before and after Room EQ software is engaged, you have to determine where to begin.
The best place to begin is with moving your sub(s) and listening position, even just a few inches at a time, being sure to measure and notate the exact locations of each speaker and your listening position on each measurement.
TIP: BE SURE TO “SAVE ALL MEASUREMENTS” IN REW – THIS WAY THE .MDAT FILES CAN BE SHARED WITH OTHER USERS AND YOU CAN RECALL THEM LATER.
Then, once you’ve found the best speaker and listening position, the best place to begin as far as treatment is concerned is with bass trapping. Although there are several different types of traps, IE: Helmholtz, Membrane, Slotted, etc, the most common (and affordable) type you will see for sale is of the fuzzy breed also referred to as “Pink Fluffy.” This is basically household insulation like R30 or specialty insulation most often used by HVAC contractors for air conditioning duct work, commonly referred to as “OC 703” or “OC 705.” These types of insulation/bass traps are resistive and worked based on the relationship between the Gas (air) Flow Resistance and the thickness.
TIP: 4” “BASS TRAPS” COMMONLY SOLD FOR $75 - $250EA ARE “NOT” BASS TRAPS!!! AT BEST THEY ARE BROADBAND ACOUSTICAL PANELS.
A common question is, “Is thicker better?” NOT ALWAYS. For instance, a common design touted amongst the forums is the “SuperChunk.” This is commonly prescribed as a minimum of 17” x 17” x 24” triangles stacked up in the corners of your room. Unfortunately, while this does work, it would work much better and be much cheaper if a much lower GFR material, for instance R-30 which is approximately 5000 Rayl/s were used instead of the commonly used OC 703 (or variant) which is somewhere above 30000 Rayls.
We’ll talk much more about this on the thread, but a good rule of thumb is, if you have the space, thicker traps (from about 8” and thicker) work better with lighter, fluffier, common insulation like R-30 attic insulation but if you have less than 8” to work with, OC 703 or Roxul Safe N Sound (20,000 Rayl/s or so and now available at most Lowe’s stores) with a higher GFR work better in the lower frequencies.
It’s all relative and there is no one size fits all answer, but I assure you that thicker/bigger is not always better. For the folks that spent a lot of money on OC 703 to make triangles/SuperChunk style, pull it out…Use it for broadband panels, and replace with “Pink Fluffy” as 17” thick traps made from OC 703 (36,000 Rayl/s or so) or heaven forbid someone that went all out and used OC 705 (60,000 Rayl/s or so) and made the 24” thick (34” wide) larger variant have simply wasted a lot of money. While these “traps” will perform, they won’t perform nearly as well as a properly sized trap vs its GFR will.
A great tool which is free, easy to use, and requires no download/installation to see exactly how your traps perform (you can model 4 different sizes/types of trap at the same time to see what will work best with the space and budget you have) is at - This is a very powerful Porous Absorber Calculator
, is very accurate, and is what I use in designing/building Porous traps (traps made from insulation).
TIP: ANY “BASS TRAP” YOU WILL SEE FOR SALE BETWEEN $75 AND $200 IS NOTHING MORE THAN INSULATION WRAPPED IN CLOTH. IF YOU USE THIS FORUM, USE THE CALCULATER MENTIONED ABOVE, AND HAVE EXTREMELY BASIC DIY SKILLS, YOU CAN VERY EASILY BUILD ALL OF YOUR OWN ACOUSTIC PANELS AND BASS TRAPS FOR ABOUT 5% TO 10% OF THE COST OF PURCHASING, PLUS THEY WILL BE CUSTOM FITTED AND YOU’LL HAVE LEARNED A LOT. IF, HOWEVER, YOU HAVE MORE MONEY THAN TIME, WELL… YOU PROBABLY AREN’T READING THIS TO BEGIN WITH.
I plan to add to this post as other topics, FAQ’s, etc come up, so please check back to the first page often as there will be updated information appearing here as time permits and if requested, or as long as the thread stays active.
Thanks for reading and I hope this helps…