Simplified REW Setup and Use (USB Mic & HDMI Connection) Including Measurement Techniques and How... - Page 37

 

Hi Sanjay, I just wanted to get back to you on this.  A few posts above you can see the uncorrected results of relocating the sub to the middle of the front wall vs. my previous location shown both with and without Audyssey correction.  A noticeable improvement in the 40-90Hz region!

 

As for the L and R speakers, I measured them at 40.5" from left and right walls.  Unfortunately, 28" puts me right in front of the aluminum frame of my AT screen so not doable with my current setup.  I think going outside the frame might introduce some boundary effects with the side walls as I would have to be only a few inches from the wall on both sides (this is doable though).  Also, I could move them closer together if you feel any significant benefit could result but then I would be reducing the sound stage as well?

 

Anyway, thanks again for your feedback!  It had some meaningful results in my current setup and I'll get some post Audyssey measurements with the sub in the center of the front wall as soon as possible.

 

Hi Feri, the L+R+Sub was just to show AJ what my measurements looked like with my new UMM-6 mic and SPL calibration in REW.  I followed his recommendations from below and posted my full mdat file for anyone to review:

 

Probably best would be to start and show measurement #1 out of AJ's table, eh?

The REW Guide version 2.1 has been released

 

Changes in this version:

 

- Windows Audio Playback Device Configuration not required if using HDMI

- Replaced sample measurement graphs with examples using full 15-20,000Hz range, and specified 1/6 smoothing as default

- Added section “Measurement Guidelines”

- Minor cleanup for consistency

 

See the link in my signature to upen the Guide.

  

Edited by AustinJerry - 2/9/13 at 5:49pm

Didn't get the mic today, guess its due to the east coast storm .

Glad it helped. Understand that my suggestion wasn't based on some special knowledge but merely pointing out something already in your chart:



As an exercise, notice that IF you could move your sub to the midpoint of room length, you'd be able to address that peak you measured at 31Hz:



I know that's probably not feasable, but I wanted to show you that the information about the peaks and dips (in your recent measurements) was already in your chart.
The reason I mentioned 28" from each wall is so that your speakers would be at the 1/6th, 3/6th and 5/6th divisions of room width to deal with the dip you measured at 121Hz:



You have to accept the fact that your room set-up won't allow you to solve all your modal problems by placement alone (like arranging your L/C/R speakers along the room length to knock out the peak at 93Hz). At least you nailed a couple of major dips with a single subwoofer. Maybe Audyssey can do something with the dip at 121Hz.
Was almost tempted to tell you to move your L/R speakers to the quarter points of room width, 42" from the side walls, to address what the chart said would be a modal problem at 81Hz:



But your recent measurements don't show a peak or dip at 81Hz. Well, there is kind of a ripple close by. So you might try moving your L/R speakers 1.5" inward (centering them at 42" away from the side walls rather than your current 40.5") and see if that helps in the next round of measurements. Considering how small a move it is, can't hurt to try.

 

Impressive difference. When you say 'middle of the room' I am assuming you mean 'middle of the wall'?

 

EDIT: I see from subsequent posts that you did mean middle of the wall.

Edited by kbarnes701 - 2/10/13 at 3:05am

 

It's a good point. I don't think there is much value in trying to compare the REW Audyssey-off graphs directly with the Audyssey-on graphs except for a way to show what Audyssey is doing when it applies the secret sauce. My aim will be to get my Audyssey-off measurements as good as I can and then to run Audyssey and then to measure again to see the final result. As I will never listen without Audyssey on, the final graph will be my 'real in-room' result, and it will be as good as it can be because I optimised before running Audyssey, using the measurements REW provides to make whatever adjustments/tweaks I can. Mostly any adjustments I can now make will be via room treatments - I have gone as far as I can meaningfully go with hardware, speakers and subs - so all I am left with is the room. I have already applied treatments, albeit perhaps not in the most scientific way, and achieved big improvements. With REW and the help of the guys here, I am hoping that I can rearrange treatments as necessary or add/remove treatments to improve the final sound. I think I will be done then and will be able to concentrate on the benefits of all this: watching movies at home. I will only be able to make further improvements if I move house and get a better/bigger/more suitable room.

 

I would love it if someone could explain to me how Sanjay looks at that chart and then can determine precisely where to put the speakers for best results. He is clearly right, as your measurements show - but what is the analysis process of the chart?

 

EDIT: Just saw Sanjay's post above. I still don't really understand the interpretation of the charts - Sanjay, if you have time a short tutorial would put me out of my misery ;)

I am obviously not Sanjay, but have benefitted from his explanation of the Standing Waves model on several occasions.  Allow the student to demonstrate whether he has understood the lessons!

 

After entering the room's dimensions into the input boxes, the model calculates and displays the room's standing waves, with each colored line representing a standing wave.  Where the line touches the bottom of the graph, there is a null at that frequency, and where the line is at the top of the arc, there is a peak at that frequency.  Looking at jkasanik's room, for example, there are standing waves at 40Hz, 81Hz, 121Hz, and 161 Hz (each standing wave is a multiple of the fundamental standing wave, e.g. 40x1, 40x2, 40x3, etc.).

 

If we were to look at jkasanic's frequency response curve, we would look for response peaks and dips that happen to coincide with the room's standing waves.  So we would look for peaks and dips associated with the room's width at 40Hz, 81Hz, etc., and for peaks and dips associated with the length at 31Hz, 62Hz, etc., and the same for the room's height.

 

Once we have identified these response peaks and dips, we then look for an opportunity to place a speaker into the null associated with that standing wave.  The theory is that by exciting the null, the peak or dip can be eased.  So, for example, jkasanic's room had a problem at 40Hz, so moving the sub into the null associated with the width's standing wave at 40 Hz (the mid-point of the front wall), that issue was made better.  jkasanic's subsequent measurements clearly demonstrated that the placement did indeed make things better.

 

The standing waves model also predicts recommended speaker placements.  For two subs along the front wall, the model suggests placing them at the 1/4 and 3/4 points.  For three speakers across the front wall, placing them at 1/6, 3/6, and 5/6 is often a good choice.  For the depth of the room, try and place the front speakers 1/6 of the room depth away from the front wall.  All of these placements are at nulls associated with standing waves.

 

In my listening room, my two subs on the front wall follow the 1/4, 3/4 placement.  The other two subs are co-located at the room's midpoint.  So all four subs are at standing wave nulls.

 

Phew. Thanks Jerry. I am in a hurry right now, and the above needs careful digesting. I will study it later and if I am still in the dark, will come back to you. I have avoided using these charts in the past because of my lack of understanding and have relied on trial and error measurements to fine tune things. It would be a lot easier to be able to 'get in the ballpark' first and then use REW for the final say.

 

PS. Thanks for the updated Guide which continues to go from strength to strength. Still no sing of my mic - 2 weeks now since despatch. I expect it will arrive this week.

How does one use the axial standing wave calculator when they have a vaulted ceiling? My room is 25.5 L x 19.4 W x 8 to 14 H. The vault is along the width wall.

 

Great explanation!  Makes perfect sense and I understand now why Sanjay was implying the suggested locations were practically staring me in the face!   This thread is really starting to get interesting!  I can't wait for Mr. Barnes' mic to arrive so he can play too!

 

That is actually a good question, and the unfortunate answer is that the model probably doesn't apply, at least in the room height direction.  Every discussion I have read pertaining to room acoustics mentions that most theories go out the window when it comes to odd-shaped rooms.  So many modern floor plans are very open, creating some challenging issues with respect to acoustics.

 

To state the question a little differently, is there a way, using our measurement tools, to determine the standing waves in an irregularly-shaped listening area?  I don't know the answer, but I hope we discover it along the way.

 

I am re-posting the link to the Harmon white paper that describes how to place speakers to cancel room modes.  This should be required reading, at least the pages recommended.

I have messed with REW in the past using the legacy connections, but it was difficult to see if i was getting good readings. I did buy an SMS-1 for my sub and it helped a lot. It certainly helped with getting the crossover right. I think one of the reasons i go the crossover right was the phase adjustment it provides in 15 degree steps.

I may have to buy a UMM-6 to join the fun. I have a laptop with a HDMI output also. Sounds like this would be easy to get going with REW again.

I am enjoying this thread!

That was my thinking. It is now so much easier to get started, and there is willing help (and Jerry's ever-improving Guide) to guide us along the way. I am still waiting for my mic and can't wait to get my hands on it.

 

You and me both!  2 weeks and counting. I wish now that I had asked Herb to send it Fedex for an additional charge - it is only a few bucks more than USPS and Fedex gets thing to me from the States in 5 working days or sometimes even less. Herb doesn't offer a choice of carriage methods on his site, so I just clicked and bought, but with hindsight, I would ask if he would use Fedex. He doesn't even need to do anything as I can get Fedex to pick up from him without any intervention on his part other than handing the package to the driver. 

There is no way to do that using simple spreadsheets and such. As I explained before, there is a method called computational fluid dynamics where the room is fully modeled including the properties, and true shape of all the objects/walls in the room. Using that, we can then excite the room using arbitrary number and location of subs and see how it impacts the room at every location. Here are some pictures from my article that covers low frequency optimization using various techniques including this one:



On the left we have a single subwoofer in the left corner. This is by the way what the simple mode calculators assume. That the sub is in the extreme one corner and hence "excites" all the modes equally. If you have your sub currently in a different location the predictions are not accurate and won't relate to your measurements. On the right we see what happens when three subwoofers are used, two on one wall and one in the ceiling. It is a little hard to interpret there so let's slice that into the one plane we are interested in: the ear height (this is for a different frequency):



Ideal response would have all the ears with the same color (which represents the sound SPL level). We see that the single sub on the left flunks that test completely. The three sub configuration does far better with all the seats having similar (yellow orange) color.

Note how the room has a soffit and hence is not rectangular.

As good as this technique is, it is expensive to run. The software costs is tens of thousands of dollars and it takes an experienced operator to get the model to converge. The one I am showing is from Keith Yates and he provides that service as a stand alone project but it still costs a few thousand dollars. If you are building a high-end theater this is the only way to go as it provides high accuracy results.

Another alternative which unfortunately is not available outside of Harman products is to use dynamic analysis and optimization using multiple subwoofers. This process starts with putting the subs somewhat optimally if you can (OK if you cannot) and let a computer analyze all the combinations of sub level, delay and filtering to find the optimal solution for multiple seats. There is a bit of that in the above article but much deeper dive with many actual measurements/examples is in this article: computer optimization of room acoustics: http://www.madronadigital.com/Library/Computer%20Optimization%20of%20Acoustics.html

Back to mode calculator, be mindful of another important factor. The modes do not have zero bandwidth as the tools show. In reality, the modes have about 5 Hz of width (for most of their energy). This means that they can easily overlap and when they do, they can change the combined response that the microphone picks up and REW shows. In English , this means that often you will have peaks and valleys that cannot be translated into actual modes easily. So even in perfectly rectangular room, don't be surprised if you can't match the measurements to the mode calculator.

Also, note that as frequencies go up, we care less about these modes as they start to pack closer and closer to each other. I think you had posted a graph asking about modes at a few hundred hertz. While I like the Harman spreadsheet you have been using, for this part of the analysis Ethan's mode calculator visualizes the situation better. Here is a snapshot for a random room:



Note that this shows the same thing as the Harman spreadsheet. The difference is the graphical representation. A few observations:

1. As I noted, when the frequencies get higher, we wind up with many modes. When that happens, we no longer see the massive peaks and valley contributions from each. They sort of cancel out each other's effects. The point where this happens depends on dimensions of the room. If you have a huge space, the point where that occurs is way down in lower frequencies. If I for example multiply the above dimensions by 10, the point where that occurs is less than 20 Hz! This is why the bass anomalies we have is due to our small rooms. Large rooms such as concert halls and such have far better bass response. The smaller the room, the worse the problem.

2. Note that in my sample room above there is a single mode at around 28 Hz and that is it. What this means is that we don't care about room modes below that frequency. There just isn't more of them to worry about! So if you see variations there, it is a function of your sub and such.

3. Look at the two modes right next to each other right past 50 Hz. Recall that I said modes actually have bandwidth of few hertz and are not those clean spikes. When you add that assumption you can see that the two modes will combine there and will result in a different waveform than what you expect there as one mode can be a peak and the other a null. Or two peaks.

4. Now look at the modes at 300 Hz. You see them packed tightly showing the effect I talked about earlier in that there is enough of them to mix as to create an average that is more or less independent of all the individual modes. A simple EQ to adjust their levels is probably all that you would want to do with them.

OK, this got long . Hopefully it is useful addition to your excellent basic explanation of what is going on.

To add to Jerry's terrific explanation: If you've ever watched the game show Jeopardy, then you'll know that host Alex Trebek doesn't really know all the answers, he's just reading off cue cards. I did the same thing: analyzing that chart just required me reading the numbers printed there.



In the above graph of the width modes in Joe's room, notice that the middle of the graph has the number 7 printed near the bottom. His room is 14 feet wide, so the middle of the room is 7ft. Right below the number 7 it says 40Hz. So if he has a single sub, placing it in the middle of the front and/or back wall will address the problem (in this case a dip) around 40Hz.

You can try the same thing with the room's length modes:



At the middle of the above graph, it shows that there will be a modal problem at 31Hz. Above that number it says 9¼ feet, which is the middle of his room length. So placing a sub at the midpoint of one or both side walls will address the problem (in this case a peak) around 31Hz.

So the chart gives you two sets of numbers at the bottom of each (length, width, height) graph: one for number of feet in from the left wall, and the second tells you the modal frequency you're addressing at those locations.

For 2 subwoofers, I told Joe that they should be at 42 inches from the side walls, because the chart showed his 2nd width mode at 3½ feet:



For 3 speakers, I told Joe that the L/R speakers should be around 28 inches from the side walls, because the chart showed his 3rd width mode around 2¼ feet:



At this point, it looks like Jerry and Joe both understand not only how to read the mode calculator but also how to match it up to peaks and dips in their measurements. Once you get used to this Keith, it will become intuitive. Next time someone tells you their room is 16 feet wide, you'll be asking them if they have any peaks or dips at 35Hz or 70Hz. (Fantastic trick at cocktail parties, makes you an instant chick magnet.)

Finally, notice that the calculator assumes a perfectly rectangular room with perfect reflectivity. But in real life, rooms aren't exactly rectangular and reflections lose energy because things like wood and drywall absorb some. Which is why in Joe's graph the big dip around 40Hz isn't followed by one centered precisely at 80Hz (close enough though). The precise numbers on the charts aren't as important as recognizing the patterns in the measurements (but the chart remains a useful predictive tool).

Thanks to both for Sunday school lesson, awesome resource and very inspiring for me . Now where is my mic!!!

No update from USPS Track & Confirm website, still show Status: Depart USPS Sort Facility, Location: Hartford CT 06101, Date & Time: February 7, 2013 (with no time), and Expected Delivery By: February 9, 2013.

When the mic arrives, do you have to sign for the package?

I didn't have to sign for mine but after my local post office "misplaced" some certified mail and then managed to deliver it to me 2 weeks later without a signature, it's anybody's guess!!

I was looking at Joe's measurements and indeed there are peaks around 31Hz, 62Hz and 93Hz (just like the mode claculator predicted).

For those not using the calculator, the easy way to figure out modes is by dividing the speed of sound (in feet) by a room dimension (in feet). Using Joe's room as an example: 1130 ÷ 14 = 80.7 (round up to 81).

But it turns out his first room mode is not at 81Hz, because a room dimension doesn't need to match an entire wavelength in order to resonate. Half the wave bouncing from one wall to another and back completes a full wavelength and is enough to excite a resonance. So first mode is at 40Hz, then multiples thereof.

Same with length: 1130 ÷ 18.3 = 61.7 (round up to 62). So first room mode is at 31, and then multiples thereof.
For residential sized rooms, it is neat how this works out, with the first and second order modes typically in the subwoofer range and third order mode and above in the speaker/satellite range. As the Church Lady used to say: how convenient.

I remember her saying "isn't that special"!  In this case, not so much!

 

Thanks again Max...I picked up a DR Pro 259 low profile boom mic today at my local GC!  I missed the $10 off sale but after completely disassembling the one in the store and then not being able to get it back in the box (still can't find that extra 2"...I know, that's what she said!), I felt obligated to purchase.  Anyway, highly recommend this low profile mic stand based on it's design.  The issue I was having with mine is the boom pivot wouldn't stay tight.  Crappy design compared to the DR Pro model.  Now I don't have to worry that the mic is moving during measurements and no additional counterweight required when the boom is fully extended!

 

 

 

Brilliant - thanks Sanjay.  (I can hardly wait to be in invited to my next cocktail party ;)).

Jerry, FWIW I've had some trouble with the guide in setting the UMM-6 mic level:

 

I've played around with this quite a bit to see what effect different values have on measurements and the only issue I can find is that if this value is set too low, then you will get this warning:

 

 

I know you mention in the guide this might need to be changed later once we get to REW but all I could find is this reference which assumes a good headroom value is obtained (apologies if I've missed it!):

 

 

 

If changing the level of the USB mic is an acceptable means to avoid the headroom error (which I think is based on my limited testing of changing this value where higher levels result in the same FR curve while lower levels below the headroom threshold result in unpredictable curves).  Also, I've found that 75 is the minimum level I can use to get a good measurement (vs. the 60 you recommend...and I think 75 was at least a previous suggestion in an older rev of the guide?).  Anyway, just wanted to let you know of my findings.  Back to measuring...

^ The point is to adjust the Windows Audio level until you get an acceptable level in REW for the measurements.  I am actually using 100% on this level adjustment, so I recommend that you try the same.  I'll edit the online guide so it shows a recommendation of 100% as well.

 

Please let me know if this fixes the problem.

The REW Guide version 2.2 has been released

 

Changes in this version:

 

- Changed Windows Audio Recording Device level to reflect value of 100% (page 23).

 

See link in my signature to download a copy of the Guide.

Quote:

 

Ok, I took some more measurements today of the L and R at 42" from either side wall.  I also moved them as far forward as possible but perhaps this muddied the water since I added another variables (I noticed that 2 1/4' was a null and I was only about 20" so moved both L and R forward about 4"):

 

Left 1/6th smoothing:

 

 

 

Right 1/6th smoothing (noticed a nasty dip at around 111 Hz but not as evident with 1/6th smoothing applied):

 

 

Center just as a comparison between yesterday's measurments and today's measurements:

 

 

So my conclusion is no big improvements with the slight change to the L and R speakers moving in 1.5".  I guess I need to get my sub relocated to the mid front wall and rerun Audyssey to see where I stand!  Looking forward to understanding how room treatments can improve my results!

 

EDIT:  Graphs updated to 1/6th smoothing

Edited by jkasanic - 2/10/13 at 4:00pm