Anyone here familiar with REW who could look at my graphs for any insight?

Hi all, I was toying with the idea of implementing Dirac Live in my listening room. While I did eventually get it measured and implemented on a "secondary" PC in my room, I wasn't able to get it functional on my dedicated Audio PC. But, I had all my stuff handy - mic stand, mic and preamp, and some time and figured why not - so I fired up REW (Room EQ Wizard for the uninitiated) and got some results. These graphs are from the entire system, mic'd where my ears would be in my listening chair. I don't know if they're good, bad, or otherwise based on experience, but from a little Google-fu they seem to be ok, specifically the Waterfall plot.

No Smoothing


Var Smoothing:


RT60


Waterfall (this is probably the best graph):


I will eventually get each speaker and sub mic'd individually and will use this as a much more precise way to level match my left/right channels. Basic setup is Audio PC > USB out > Aqua La Voce DAC > SE RCA to Bel Canto Pre1 > XLR out to amp. Speakers are 3-way front ported, sitting atop ported 12" subs crossed over at 45 Hz. Mains run full range, usable output down to 35 Hz. Room is 12.5 x 9.5, 9' ceiling unfinished with pink stuff and Rockwool insulation between the joists in the ceiling. Room treatments are (2) 2' x 4' x 4" thick OC703 panels in the front corners, angled to the wall. Rear wall is treated with one 2' x 4' x 6" thick OC703 panel.

To me, it looks ok. The dip at 50 Hz is perplexing, as is the spike at 150 Hz. I can tame the spike at 150 with parametric EQ, but not the issue at 50 Hz so I'm guessing that's a room mode issue.

The funny thing was how sensitive the mic placement was. Thankfully (luckily?) at my seated position, it's by far the flattest response I was able to get.

Thanks in advance for any thoughts, criticisms, or questions!
Todd

Edit - I used this link for basic tips on what, and how to measure: http://realtraps.com/art_measuring.htm

 

Reducing the Y-axis on the FR graphs would be helpful to make the peaks and dips in the response more obvious easier to see (e.g. make the lower cut-off point 40dB and the upper cut-off point 110dB, and make the intervals 5dB instead of 20dB). Your reverberation times are VERY good though (do you already have room treatments?).

 

I was playing around with the axis and scaling options after I posted, but I didn't make the interval 5 dB instead of 20, so I'll try that tonight.

I do have Room treatments, yep:

Thanks for the tips!

 

Another interesting graph is the Spectogram. This can be found on the right next to waterfall, if you hover your mouse in that area.

As well as having the range of 40-110dB as suggested, you could do only 20-500Hz. If the dip at 50Hz is very narrow I wouldn't worry about it too much. The peaks at about 160Hz, 290Hz and 450Hz are more of a concern. Moving speakers, altering room treatment or EQ could help.

 

I looked at the Spectogram but had no idea how to interpret it. I can read up some more on that. I'm going to mic the left and right separately in the next few days to see if those spikes are consistent, or just one particular channel is causing the issue.

 

You want to adjust the graphing to focus on the 300Hz and below range, and use the full resolution—no smoothing.

 

The REW signal starts roughly at 0ms and in my case mostly stops by 200ms, similar to the Waterfall plots. However you can see at 40Hz and 80Hz there was sound before the signal and it continues afterwards. I interpret that as traffic noise (I live on quite a busy road). When I had a quite loud amplifier hum at 150Hz that showed up in the same way.

On the other hand, some of the 40Hz sound seems to be generated in the room - the red bit. In fact I did have a boomy bass at 42/43Hz, so that is what it is.

http://www.roomeqwizard.com/help/help_en-GB/html/graph_spectrogram.html

 

So on the advice of home theater shack moderators, I ended up measuring left and right independently, including sub/main independently. What an eye opener. I was able to EQ the right channel to a much smoother line, at least from 20-200 Hz. It gets pretty spikey after that but only for very short intervals and honestly, those seemed to change from measurement to measurement. I drove my wife crazy tonight running constant 20-1000 sweeps Anyway, I figured ok, right channel is done - let's see if the EQ settings I applied for both channels help the left out just as much.

NOPE.

So that was another hour, doing that side completely separately. It dropped off a cliff at 45 Hz. Oddly enough, gaining down at about 42 and 52 Hz by 3,4 dB actually solved that issue. Weird, but it was repeatable. Same thing at 90 Hz on the left side - a big dip from about 70-110, I was able to smooth that out quite a bit, then another one at about 160. Definitely seems like I have a mode issue around the 40 Hz area but I tried changing speaker and treatment positions with no real effect. Changing the mic location forward or back by 18" made a huge difference in how smooth the response was, so ultimately I am sitting forward a few inches from where I was.

I didn't grab most of the data tonight, I had to run all this from the Audio PC which is enough of a hassle and it's getting late. Needless to say, "trusting thy ears" is great and I'd never discourage anyone from doing so - ignorance is bliss when it comes to this stuff. But, put yourself out there and measure your room - you may just be surprised and learn a thing or two in the process!

Todd

 

Well, ****. Here's the thing - I'm a big believer in "trust your ears" but apparently, I should ask for advice more often. I posted up my findings and measurements on HomeTheaterShack and a kind gentleman informed me earlier today - you're doing it wrong. Turns out I "calibrated" my sound card using my microphone, not using a loopback which is honestly unnecessary in most cases. So, I was coloring the measurements drastically. The advice was to remove that specific calibration (but keeping my Mic calibration file in place) and yea, it wasn't exactly "easy" to tune but in a lot of respects it was.

In the end, I was able to get a much better picture of what was going on in the room. I EQ'd every speaker individually and got the following curve:



I'm not sure if it's *supposed* to look that jagged, but it's better than what it started, at least much more linear from 20-200.

Surprisingly, once I mic'd everything together, I had a really big suckout at 50Hz. I was able to correct that but didn't grab an image. Spectrograph and Waterfall all looked good but I didn't take screenshots of those, I can if anyone is interested. Decay looks great as well, lively but not too much so.

All in all, I'm glad I went through this, it's been great fun and a nice learning experience. I'm usually not so patient as to measure everything 50 times each. The absolute best end result was playing around with the listening position and I can confirm, once that was nailed down to the best response possible, my imaging seems more lifelike and the little details that seemed to have been clouded by uneven bass response are now clear and present. Dynamics are great, even though I have all my treatments back in the room (I had thought previously that having all the treatments in the room was over-damped, turns out I needed better placement, which REW showed me).

 

Very interesting stuff. Couple questions. What mic did you use? Did you use electronic eq to effect changes?
I'm going to look in to this.

 

Hi timind, I used a Dayton Audio EMM-6 mic and a MXL MicMate USB preamp, set to medium. I'm using the "JRiver WDM audio driver" in Windows so that any audio playback routes through JRiver's audio engine. In turn, I'm using Room Correction and Parametric EQ within JRiver to manage my EQ curve. The great thing about using the WDM driver is that you can observe all of your changes basically in real time, especially if you set up a 20-200 (or 20-whatever) sweep set on repeat in JRiver, and use the "RTA" function within REW.

 

It would pay to look at what your early reflections are doing. This is more important than freq response. Post your ETC plot and well have a look. You only want to got out as far as 20ms or so.

 

A moderator on the REW forums mentioned this (I was able to upload my measurement file there):

In your .mdat file, the second-to-last plot shows very high distortion, > 10%. It might have been something rattling when the measurement was made. It appeared at a resonant frequency during that measurement. Something to check on.

You have a couple of big reflections at 3.7 mS and 5.2. mS. That is a bit on the early side, but they match extremely closely between L and R sides. Depending on arrival path, they could be hurting soundstage and imaging or they might be just fine.

Is the plot you're referring to called "ETC" in REW? The very little googling I've done so far suggests I would use the Impulse response graph to find reflection info, but that info could be outdated by now.

Thanks for the help and I would agree 100% with you - FR is a good first start but personally I'm more concerned with reflection and reverb being that I'm in such a small space.

 

Thanks Todd. I'm going to look into Jriver. Although I have to say this sounds more computer complex than I can handle. I downloaded one of the audio software packages a few years ago. Can't remember which one but it wasn't worth the effort to my ears.

 

Which part sounds too complex, JRiver, REW, or the combination of the two? I'm a tech nerd by trade so I sort of gravitate to these sorts of solutions anyway, but if I'm pulling back from that a little - JRiver CAN be complex but doesn't have to be. REW seems overwhelming (I was very hesitant to restart this process, last time I used it was to EQ a very weird upstairs room and it was a royal pain to do) but by using the EMM-6 and Mic Mate, plugged directly into a garden variety USB port, the setup was actually quite painless. Interpreting the graphs and resolving issues? That was the fun part, but it was time consuming and gets back to the idea that JRiver CAN be complex - it will be in some ways if you go this route.

 

Xposted this in another thread but not looking to derail that into a separate conversation. Here's my latest measurement, ETC curve with all treatments in place:

 

It has been a while since I've used REW but I'll take a shot and ask, before taking a measurement, did you calibrate the sound level reading of the software?
Do you have an spl meter to validate the sound level in the room?

 

I did calibrate the SPL level, yes. I've used two SPL meters in the room but one of them only goes as low as 40 dB and with my TV off, the door closed, and nothing playing, it's quieter than 40 dB in here. I've been trying to figure out a way to get the ambient noise level using my EMM-6 mic on the stand but haven't figured that out yet. If I use the freebie SPL mic on my iPhone, it registers a very quiet 16-17dB in the room in the above conditions. And it should be pretty quiet - there are no moving parts in the room at all. Both PCs are in a different room, no fan, no ventilation ducts, three concrete walls and a concrete floor with laminate flooring and an area rug - it's a really quiet space. Frankly though, I don't trust the free SPL meter and mic on my phone though it has been handy for showing changes in noise floor as I removed things and added treatments. It's not anechoic by any stretch (from your other thread) but it's a room where you can hear every part of your breath when nothing is running.

I think that sharp, quick reflection is coming off the front panel of my TV - it's a glass front - otherwise I really can't think of what it could be.

 

No problem. As long as we are all sharing/learning. Is this a sweep or a burst? Looks like you fall to the noise floor almost instantaneously and get a first reflection at 5ms? Hard to read the graph.

 

Maybe wall or floor. I have a phone app and a dedicated SPL meter and there is a 15 db difference. Disappointing.

 

Let's define calibrating. By calibrating, are you playing pink noise from REW and entering the value shown on the SPL meter at the listening position into the calibration window on REW?
http://www.roomeqwizard.com/help/help_en-GB/html/inputcal.html

 

Yes, but how accurate is your SPL meter?

 

I'm playing a 20-20khz sweep, the standard "256k length" sweep in REW.

Yes, that's what I'm doing. My mic has a cal file that I've applied. I have not done anything with sound card loopback which I understand to be important when it comes to measuring distances of reflections accurately but that's about all I know about that calibration piece.

In this case the "SPL meter" is an onscreen display that shows what REW thinks the SPL level should be based on microphone input. I raise the volume on my preamp to match what the display says. Alternatively you can enter the number that pops up in the display into the "Calibrate" field and that supposedly gives REW what it needs to operate. I'm certainly no expert at this but that's my understanding.

As for the "twin peaks" on the "both" graph - according to an Admin at HomeTheaterShack, that happens when the speakers are not completely equidistant from the mic. As they get further apart, that spike does so as well.

 

I'm at the wrong computer at the moment but I figured out part of the problem - I didn't think pointing the mic straight ahead vs pointed at the ceiling would make a huge difference but the latter is definitely the more realistic looking plot, the ones I posted earlier were with the mic facing forward. I'll post it up soon, and I also figured I'd try to match up the two spikes since it should be easy and it was, had to pull the right speaker in about an inch and up about half an inch and they're dead on now. So that's neat Now the spike I'm seeing, I'm almost positive it's the TV and honestly I'm fine with that, not much I can really do about that anyway (I generally browse here and other forums while listening at night and the TV is my monitor).