Firstly a big thank you to
for such a comprehensive guide. It’s commendable how much effort he has put in this guide and the other ones- not only authoring but also painstakingly updating them.
My adventure with REW started many years ago but it’s always been limited to the SPL response and using them to tweak crossover settings. Now, with hdmi and this excellent guide, I am learning to make sense of the waterfall, ETC, impulse response and how to tweak the room not just the electronics. Since the frequency curves always get more coverage, I thought I would post some of my findings with the decay in my room.
Below is the waterfall for left+right+sub with no EQ in place. My noise floor is between 45-50 db, so I ran the measurements loud as u can see. My speakers are 104db efficient so going loud wasn’t a problem- especially since I was outside!
First the spectrogram- right of MLP
As you can see there is a problem frequency at 55 hz and then at a smaller level at 100hz. Also if you examine for horizontal uniformity, there is a problem at about 68hz- seems to be a null of some sorts. I am not worrying too much about the problem region of 15-25 hz.
From the guide- “
(… avoid frequencies below 20Hz, which always exhibit excessive ringing). If the line is at or below 450ms, bass ringing is generally “under control”. Also look at the horizontal uniformity of the Spectrogram colors. If the colors are general smooth horizontally, the decay is balanced across the frequency spectrum. A rough, jagged line would represent an opportunity for better control of bass ringing at certain frequencies. “
I took 3 measurements at the MLP for this and though they do not look identical you can see that the problem is apparent in all of them. The next 2 are for mid and left of MLP.
For comparison here is the SPL for the three positions. Black is the average of the three. It’s interesting how the graph is consistent till about 80 and then fluctuates so much!
Next we look at the waterfall. The highest level was about 120, so I took the lower limit about 40db down to 80. Choosing the lower limit is kind of tricky since it totally changes the graph. Here are two extreme examples where lower limit is 80db and 40db. Same measurements just different lower limits.
Obviously it takes much, much longer for the level to fall to 40db. As I see it, the noise floor of my room is around 40-50 db, at max I would listen at 75db reference, so for my usual listening the range would be about 80-40 db (average reading- disregarding transient peaks).
However, because the noise would skew the lower part of the waterfall, I shift everything up by 30db, 120-80db. The guide also has details on this and recommends at least 90db. It doesnt really matter though to nitpick to get the exact values- the problems will be apparent enough.
From the guide- “…For a Waterfall graph to show meaningful information, the original measurement level needs to be high enough. A general rule is that there should be at least 40dB between the measurement level and the noise floor. If you are using 45dB as the noise floor, then your mic should be calibrated to at least 85dB, and 90dB or 95dB provides even more headroom (although much louder).”
The average for the freq response posted above is about 103db from the SPL.
So back to the waterfall and you can see the same problems as in the spectrogram.
Here is the same curve for 300ms.
Reasoning being that the lower frequencies take more time to decay compared to the mids. I don’t remember where I read this- perhaps the guide again! It said that the standard for lower freq was 450 and mids was 300ms.
Ok, so this means that I need to take care of the ringing at 55hz and 100 hz, the null at around 65hz. This region is around the crossover region- so playing with the crossover point, slopes and phase could take care of the problem. Also, more than the room, this is a known problem with my subwoofer itself- it has ringing at 54hz and at 100. My crossover setting was 55hz for these readings.
Now we look at decay. From the guide- “
Again drawing from the Paul Spencer articles, a good objective for the Decay graph is to have all bass resonances -20dB within the first 160ms. “
Nothing new to see here. Same results consistent with the waterfall and spectrogram.
Subwoofer is a DTS-10, mains is JBL 3772, sides are the jbl 8330. It’s a dedicated media room, with bass traps on front corners, top side walls, and some on rear walls.
My mains are 104db and currently I am getting a hiss from the fronts. Its apparent in the lower passages- this is what it looks like with amps switched on and no inputs to them (I forget if the projector was on during this time).
The red line is the threshold of hearing (from wiki- search for loudness curve), so the left hand side doesn’t matter.