Site Links
Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Categories
Playing with REW's RTA averaging function
This will be a thread to discuss and post graphs using REW's RTA to develop a better understanding of how the room impacts the apparent frequency response balance at the listening position (or listening window), especially below 1kHz. I have the current version of REW loaded on my laptop with the loopback and calibration settings working properly. I have also followed @dcibel s instructions for making RTA averages and I seem to have it all working properly. My first average, using my Zonker speakers as test subject, seemed to work very well and is repeatable. I'll post graphs later today. I will also post a copy of dcibels instructions from the other thread and a reference to the youtube video linked by @tktran
Comments
Here is a link to the youtube video:
I tried to cut and paste dcibels instructions, but the graphic images would not transfer. Here is a link to the other thread with this important setup information:
https://diy.midwestaudio.club/discussion/2182/mac-crossover-review-thread#latest
Posted the instruction as an article on HTGuide for easy future reference:
https://www.htguide.com/forum/articles/do-it-yourself-diy/933194-how-to-measure-with-rew-for-room-eq
As background, the speaker under test is a small bookshelf with 8" woofer and 1" soft dome tweeter called the Zonkers. Full contruction thread is located here:
https://diy.midwestaudio.club/discussion/1513/the-zonkers-an-isobarik-8-two-way/p1
Room background: Testing will be done in my great room, which measures approximately 16 feet wide, 29 feet deep, with an average ceiling height of 12 feet above the speaker. The ceiling is sloped, 9 feet high at the speaker end and 16 feet high at the upper end, which leads to an upper loft area. The speaker under test is located approximately 9 feet from the left wall, 7 feet from the right wall, 6 feet 10 inches from the rear wall and 22 feet 2 inches from the far wall. The tweeter is 5 feet above the floor and the microphone is being rotated in an approximate 18" circle, 45" above the floor and 8 feet from the speaker.
Here is the average RTA for this position. I performed the test 3 times, with approximately 40 averages per test:
For reference, here is my VituixCAD spin measurement for the same speaker, taken in the same position and rotated 180 degrees. Note that this was taken at 1 meter, so the high frequencies are probably somewhat puffed up at this distance.
For my next test, I am going to lower the speaker to a tweeter height of 41.5" (instead of 60"). I want to see how changing the speaker's distance to the floor boundary alters the average RTA measurement.
With the response as-is, 2 bands of PEQ to knock down the peak at 750Hz and bring up the level from 100-250Hz and give it a listen. If you wanted to take a step further you could look at dropping the level a bit around 2-3kHz (low Q) to provide more of a constant slope of response through mid-high end, experiment that way and see if you prefer it or not.
Next test. Everything the same (volume settings, distances, mic rotation, etc), except I lowered the speaker, placing it on my 27" high stands. The tweeter is now 41" above the floor (the woofer is 7.125" below the tweeter, so the woofer is now 33.875" above the floor).
Next test. Everything the same, except I lowered the speaker once again, this time placing it on my 13" high stands. The tweeter is now 26.75" above the floor and the woofer is 19.625" up.
Next test: Everything the same, except i lowered the speaker again, this time placing it directly on the floor. Tweeter is now 12.75" above the floor and woofer is 5.625" high.
The point of all this testing is to note the changes as I lower the speaker. The worst case is the 27" high stand, which places the woofer 33.875" above the floor. This creates a huge suck out at 250Hz. But notice how this suck out fills in when I switch to the 13" high stand. Interesting.
Tomorrow, I am going to move all my measurement gear (amp, cables, laptop, UR22, mic, etc) into another room and repeat this same set of measurements, using the same speaker and settings, to see what happens. This room is 14 feet wide by 20 feet long with a flat 7.5 foot ceiling.
Something I want to mention with mic orientation (mostly because the above linked video holds the mic 90 degrees to the speaker), is that even though measurement microphones are omnidirectional, they're not perfectly omnidirectional, so you will very likely notice a difference in response slope >5kHz if you complete this process with the mic facing forward vs the mic facing sideways as shown in the video. Issues in the top end should really be addressed by the crossover anyway, with room EQ applied <1kHz, usually <600Hz. I'll add this detail to the instruction at HTGuide.
I have always wondered about this, as I have seen several build threads where the author made all the measurements with the mic pointing straight up, 90 degrees to the speaker. All of my measurements, however, are made with the mic pointing straight at the speaker. I have my mic mounted in the end of a 26.5" long x 1" diameter PVC pole. For these tests, I held the pole at arms length as I rotated it in an 18" circle, pointing it straight at the speaker. This kept my body from influencing the measurement.
@4thtry are you listening to the speaker with music in each position to see what your brain prefers or just to try and detect any obvious difference in tonality?
Yes. As I lower the speaker to the floor, the sound becomes "warmer." I can clearly hear the lower mid bass frequencies filling in.
Found this vid by Erin (Erin's Audio Corner). He talks about the moving mic method vs the single mic method to generate RTA averages.
https://www.youtube.com/watch?app=desktop&v=fO9QSfyBkWY
Pointing your mic AT the speaker is suitable if you are using only one or 2 speakers at the listening position. When you start to have a multichannel setup like eg. Cinema, automobile, home theatre to Bette capture what is happening at the listening position(s), it’s preferable to point the mic equally to all speakers; so a good way is directly up (or down). Or course you should load the appropriate mic calibration file for 90 degrees use.
Now you wave your mic around the LP like a magician and calibrate each and every speaker.
Here’s Charlie doing it again for some commercial studios.
From-
https://www.kaliaudio.com/immersive
Good on ya Chuck for getting rid of those nasty NS10’s that Bob made viral and fixing em’ studios up with some neutral monitors for less than the price of an outboard…
It's really just to use best judgement based on speaker location relative to mic. Like I mentioned, change in response from mic orientation will occur 5kHz+, not going to influence result from <1kHz much. Not everyone has 90 deg calibration.
Position of your body relative to the mic is more important IMO. Best to stand to the side with arm extended, to keep furthest away from mic. Alternatively if you have a boom stick, use that to get even further away for least influence from body reflections. It's also important to keep steady hand, avoid adjusting your grip, jostling the cable, etc. which can influence measurements in the bass range significantly. Many measurement mics are quite sensitive to being touched, just listen to audio recording while poking at your mic for your own verification, or at least run RTA with exponential 0.5 averaging, short FFT, and no signal gen and just watch the SPL jump as you adjust your mic. Keep decent SPL for measurement to reduce influence of background noise and other external influences. I aim for ballpark 80-85dB, loud enough that you have to raise your voice to talk over the measurement noise.
Some quick comparison of body proximity to my own mic. Used a small bookshelf speaker with 4" woofer for this, and low SPL so respones <35Hz is just noise floor really. Meaurement is single point without averaging just to compare the effects of someone standing near the mic.
Red trace is with me 6ft away from the mic standing against the back wall. I then took measurements at arms length as if I would be holding the mic, but without holding the mic, just extending my arm as a ruler stick. Easy to replicate at home for your own comparison and peace of mind.
Division to show only the difference. About +/- 1.5dB swing in this basic comparison.
From this I would suggest, that standing to the side provides less influence that directly behind the mic, which I assume is what most people will do at home. I would also tend to think that taking an average measurement moving the mic in a circle would reduce the effects further. Perhaps later on I will complete this comparison.
My Sonarworks SoundID mic only has calibration txt files for 0 degrees and 30 degrees. I'm using the 0 degree one. I have my mic mounted on the end of a 1" diameter PVC pole. I point it straight at the speaker, like a ruler stick, and then rotate in an 18 inch circle. One thing that I have found that is very helpful is to use one hand to rotate the mic pole and then use the other hand to hold and left click my laptop mouse. This way, the mic is always rotating in the proper 18" circle, at the correct distance, when I left click the mouse to start a new RTA measurement. When I see the number of averages hit 40, I click the left mouse button again to stop.
Next test: So, I moved my mic, amp, laptop, UR22, Zonker speaker, etc., to a smaller room to perform the same set of 4 measurements at different speaker to floor heights.
Background info: Mic distance is the same: 8 feet from the speaker, rotating it in an 18 inch circle at an average height of about 40 inches above the floor. This room is approx 14 feet wide by 20 feet deep. The ceiling is flat and measures 7.5 feet. Some furniture and boxes along the walls, but the center part of the room is fairly open. Walls and ceiling are covered in drywall; the floor is carpeted. I located the speaker 72 inches from the rear wall, 115 inches from the left wall, and 45 inches from the right wall. I toed the speaker in slightly, pointing to the listening position, similar to where the speaker would be placed as the right channel of a stereo speaker setup.
Here is the REW RTA average with the tweeter 60" above the floor (woofer 52.875" high):
Here is the REW RTA average with the tweeter 41" above the floor (woofer 33.874" high):
Here is the REW RTA average with the tweeter 26.75" above the floor (woofer 19.625" high):
And, finally, here is the REW RTA average with the tweeter 12.75" above the floor (woofer 5.625" high) (speaker sitting directly on the floor):
In looking at these 4 different speaker heights, the averages above 1kHz or so are pretty much the same. But below 1kHz, I can clearly see a frequency response depression moving throught the graphs as I re-position the speaker higher and higher. At a woofer height of 52.8" the dip is centered at 150Hz. At a woofer height of 33.8" the dip starts pushing up into the 250Hz region. At a height of 19.6" the dip moves up to about 300Hz. And with the speaker directly on the floor (woofer height of 5.6") the depression moves up into the 600-800Hz area. I wonder what is causing this change. It appears to be very similar to what happened in the larger room. Tomorrow, I'll load these four graphs onto a single overlay and post it. This will make the differences easier to see.
To make it easier to compare, I summarized the averages, shown above, for the 4 speaker heights (woofers 5.6, 19.6, 33.8, and 52.8 inches above the floor) and overlayed the curves on a single graph. Note that all measurement settings are the same for each graph and averages (amp vol, UR22 settings, mic distance & rotational height above floor, etc.). Only the speaker distance to the floor changed for each set of 4 measurements in a particular room.
Below is the small room comparison (20x14x7.5 feet) with the speaker positioned in the room 72" from the rear wall, 115 inches from the left wall, and 45" from the right wall. The speaker (Zonker) was toed in and pointing to the mic, the tip of which was being rotated in an 18" circle at a height of about 40 inches. The mic tip was 8 feet from the speaker and 7 feet from the right wall.
Below is a summary of the "Great room" comparison, which measures approximately 16 feet wide, 29 feet deep, with an average ceiling height of 12 feet above the speaker. The ceiling is sloped, 9 feet high at the speaker end and 16 feet high at the upper end, which leads to an upper loft area. The speaker under test was located approximately 9 feet from the left wall, 7 feet from the right wall, 6 feet 10 inches from the rear wall and 22 feet 2 inches from the far wall. For each measurement, the microphone is being rotated in an approximate 18" circle, 45" above the floor and 8 feet from the speaker.
When I look at the above two comparison overlay graphs, what strikes me is how similar the four curves are from about 2kHz to 20kHz. They are also very similar from 20 to 100Hz, but only when I look at the 4 curves, one room at a time. There is a huge difference, however, from 20-100Hz, when I compare the large room to the small room. Also of note is that the relative SPL level (not the absolute level) changes significantly for the two rooms. The overall SPL increases by about 5dB in the smaller room, at all frequencies.
Between 100Hz and 1kHz, the height of the woofer with respect to the floor signifiantly impacts the averages. The curves look somewhat jumbled, but there seems to be an overall trend within the jumble. Closer to the floor seems to result in more SPL in the 100-300Hz area and less in the 400-800Hz area. Further away from the floor seems to result in less SPL in the 100-300Hz area and more in the 400-800Hz area.