What is the BEST way to measure your room responce? (1 Viewer)

[Anthony_Gomez]

I just picked up TrueRTA level 4 in my quest to properly EQ my room. The software is rather nice btw!

I already have a pair of 10" ID helmholts resonators to "help" tame the 63hz room resonance that I get, and will be using the BFD to take care of the rest from 20-100hz.

Here are the "tools" I have at my disposal:
*1/3 up to 1/24th octave measurements
*pink noise
*sine sweeps
*sine tones at any frequeny
*averaging over X # of measurements
*peak hold
*quick sweep

I will be using the RS spl meter until I get my panasonic mic all done.

any advice?

 

[Chris Tsutsui]

Dips in the response should not be filled. It is laughable how a 10db increase needs more than 10 times the acoustic power and then you'd have tons of bass in other parts of the room. Do not increase volume at dips with equalization!

What people use is parametric equalization (filters) with the proper measuring equiptment. (High resolution equiptment of 1/10 octave or better). A very vague 1/3 octave graph can easily miss the details that will still cause bass irregularities. So my advice to you is measure as precisely as possible given you are willing to spend the time.

In case you didn't know, pink noise is a variant of white noise. Pink noise is white noise that has been filtered to reduce the volume at each octave. This is done to compensate for the increase in the number of frequencies per octave. Each octave is reduced by 6 decibels, resulting in a noise sound wave that has equal energy at every octave.

1. Do you have 1 sub you are tuning?
2. How many listening locations do you want to find the best response?
3. Are you willing to move speakers/listening seat around?

Just some food for thought, good luck

 

[Stephen Dodds]

I use TrueRTA level 4 with the Behringer ECM8000 mike, and I also have an external 1/3 RTA to check with.

The quickest and easiest method would seem to be using the Quick Sweep that comes with TrueRTA once you have done the PC Sound system calibration. Alternatively I have used a 20-20 sine sweep (log) that I have burned to a CD so as to include the whole system.

I tend to use the 1/6 octave resolution. The 1/24 is just too despressing sometimes.

I do the measuring with the mic at my usual seating position.


Steve

 

[VinhT]

I use the quick sweep function of TrueRTA level 4 with the analog RS SPL meter. Works great.

I put the meter on a tripod at the listening position, with the mic pointed at the TV.

 

[Anthony_Gomez]

Chris:
1)I don't plan on filling 10db dips maybe about 5-6db max. I have plenty of power to afford that.
2)I can measure up to 1/24th, so that will not be an issue..
3)I am currently running dual mono subs. I am not sure if I want to go stereo (will image better with a 100hz XO), but mono will work better for room accoustics.
4)I am just equalizing a love seat..so a small location
5) The subs have a little bit of moving (dual as speaker stands), and the seating can move forward or back about 6"
Stephen/ Vinh:
I am not looking to completely excite the room with pure long signs, and not looking at "under"exciting the room with pink noise. Does the quick sweep fit in between these two?
Thanks

 

[Stephen Dodds]

I believe so. Seigfried Linkwitz uses a similar burst tone and unless you have software that offers different windowing possibilities it would seem to be the best compromise. I use fast sweeps for the same reason (5 secs).

Steve

 

[Hank Frankenberg]

Has anyone seriously tried/compared ETF 5?
http://www.etfacoustic.com/
If you compared the two and bought TrueRTA instead of ETF 5 was your decision based on TrueRTA level 4 being $50 less than ETF 5?
Please respond only if you compared the two and bought one over the other. Thanks

 

[Brian Bunge]

Hank,
I haven't compared the two, but screw you I wanna reply anyway! I will tell you that Mike Dzurko of ACI personally told me that he compared the accuracy of ETF to MLSSA (which at the time he bought it was a multi-thousand dollar software package). He said that they were within a fraction of a dB of each other. He felt that for the price it was money well spent.

 

[Richard Greene]

A spectrum analyzer, whether 1/3 octave or even one-octave bands, used with pink noise will help you understand how the general shape of a frequency response curve correlates with what you hear.
While most people will enjoy a flat mid-range frequency response, what sounds good below 100Hz. and above 2kHz. varies a lot among people.
For bass below 100Hz.
- SPL greatly affects hearing ability -- at lower volumes it becomes increasingly difficult to hear deep bass ... so the SPL of test tones should match the SPL you prefer for music.
- A slow frequency sweep gives you the best picture of steady-state bass frequency response including fully excited room modes (expect +/- 10dB or worse = that's reality ... and be thankful our ears do some 1/3 octave smoothing if the bass peaks are not too far apart)
Place the microphone exactly where your ears would be
(I measure for the right ear and move the mic to measure for the left ear and then average the dB's -- for some frequencies moving the microphone by 6" can change the measurement by up to 6dB based on my own experiences over 20 years).
* Test CD with very slow frequency sweep:
1Hz. to 100Hz. sweep over five minutes = great for locating "room noises" and CD also includes 1/6 octave steady state sine waves for setting parametric EQ ...
while the portion of the sweep from 1 to 20 Hz. is great for checking for speaker enclosure leaks, please leave the volume low for the first 30 seconds of this five minute long sweep as loud tones below 20Hz. may be life-ending for some weenie drivers
***** click on "Test CD's" at:
http://www.stryke.com/
- Steady state sine wave tones will be needed to set up a parametric EQ -- I also use the Behringer 1124 -- set with 1/6 octave tones (fast) or 1/12 octave tones (may be more accurate)
Please be aware you can not equalize bass for an entire room -- you can only reduce bass peaks at one seating position (sometimes for two side-by side seats) for wall-to-wall room modes ... and at a variety of seating positions for floor-to-ceiling room modes. You can not lift output of standing wave-related bass troughs, so don't even try = a complete waste of amplifier power.
- If you use 1/6 octave tones to set your EQ, doublecheck the results with a more accurate slow sweep because 1/6 octave tones will not quite fully excite most room modes ... and also use the slow sweep to locate rattling walls, pictures, knick-knacks, etc. (every object has a resonant frequency and a slow sweep will rattle everything in your house, if loud enough, even your dog) because "room noises" are usually far more audible than harmonic distortion.
- Use tone bursts to check bass transient response. Unless you have a room full of bass traps or a dipole (directional) subwoofer, the quality of bass transient response in a typical home listening room is likely to be disappointing. Fortunately compressed dynamic range music tends to mask this problem ... and rock music is often quite compressed.
***** Tone burst CD:
http://www.linkwitzlab.com/burst-cd.htm
Above 100Hz.
I prefer using 1/3 octave pink noise and a sound meter. 1/3 octave spectrum analyzers are expensive and seem to have somewhat different results than 1/3 octave pink noise + sound meter ... but I can't recall the reason for
this.
Above 2kHz.
High frequency hearing damage due to loud noises plus the natural aging process (especially for men) affects the preferred high frequency response curve ... along with how far your ears are located from the tweeters (air is a great high frequency absorber) ... so it's barely worth measuring treble unless you have enough experience to know exactly what shape frequency response curve is pleasing to your ears.
It's easier to set treble by ear (it seems difficult to correlate what you hear and prefer with microphone-based measurements above roughly 2kHz.).

 

[Shawn Solar]

So Richard, if the ear smooths bass frequencies approx 1/3 octave do you think a 7db drop in a 35-42hz trough will be audible? likewise how about 15db drop between 66-80hz? I am almost debating moving the room back around the other wall just to smooth the bass.

 

[Wendell R. Breland]

Anthony Gomez - If you have access to SMPTE standards, then look at SMPTE 202M and SMPTE 222M. Make sure to use the the modified X curves. The X and N curves are intended for rooms of 5300^3 or more.
Chris Tsutsui, Pink noise is white noise filtered at a slope of 3 dB per octave, not 6 dB.

 

[Mark Seaton]

Hi guys,

While we do somewhat integrate peaks through a filter, and for upper frequencies a rough estimate for our hearing is 1/6th octave resolution, you do want to take high resolution measurements at low frequencies if you plan to EQ with a parametric. The issue is really with the nature of the peaks in the response. With less than 1/12th octave resolution, you could easily mistake a broad peak for a single mode as opposed to two distinct resonances and do as much harm to the response as you do good.

Also realize that you first want to get the flattest response possible BEFORE EQ. Modes are time related problems, so realize what we are really doing with an EQ. Once you have done all you can to get rid of the peaks, if there are any left standing, you want to simply deprive them of energy. This does most certainly improve the transient response, but not to the same extent as if it was not there in the first place. I would suggest you check out Harman's main website and read through some of their white papers, definitely an interesting read and good for more ideas and tools in the pocket.

Regards,

 

[Chris Tsutsui]

Chris Tsutsui, Pink noise is white noise filtered at a slope of 3 dB per octave, not 6 dB.

It will be noted in my log.
I tried ETF 5 the other day but it was using the Mic input because I didn't have an adapter. (And the mic used is very cheap). I sort of waisted my 2 trials just messing around.
Using an SPL meter and test tones suddenly seemed a whole lot easier.
I was wondering if anyone took an RTA measurment and compared it to a Radio Shack SPL meter that had compensated numbers? I've heard from some respectable names that the Radio Shack meter is far from accurate and am worried I may have to keep playing with a program like ETF5 until I can figure things out.
Of course the Bass sounds superb as it is now.. I just think it's fun getting technical and perhaps add a Para EQ to my xmas list.

 

[BruceD]

Couple things;

I found the ETF website tutorial and program feature set (including frequency waterfall charts, adjustable windowing of the FFT analysis, phase charts, etc.) to be invaluable in helping me understand how to "tune my room".

If you are going to use ETF, the Radio Shack SPL meter's RCA output connector (or a MIC+preamp output) needs to be plugged into a "LINE" input on your PC, NOT the "MIC" input.

 

[David.N]

Just my two cents..... I like to spend time playing with speaker placement and adjusting the EQ, ALMOST as much as other folks on this site. Here is my question for you. After you spend all the time and money on SPL meters, room acoustics, EQ's, calibration CD's etc... Do you take into account real-ear acoustics. I bring this to your attention because I am a licensed audiologist and spend a good portion of my day measuring SPL at the ear drum. 99.8% of people with non-surgical ears have a resonant peak at 3000Hz. It ranges, of course, due to ear canal size and shape. It begins increasing SPL by about 10dB @ 2000Hz, 16dB at 3000Hz and starts to fall to 13dB @ 4000Hz. So for a "true flat response" you may have to start filter notching around 3 kHz... hope this doen't keep you awake

 

[Anthony_Gomez]

David, Most people include a BCC at around 3k in the speaker XO.

The corrections I am looking at are in the subwoofer area.

Thanks

 

[Mark Seaton]

David,

This is a common confusion of our hearing sensitivity vs. reproduction of an acoustic event. For reproduction, we need to re-create the exact sound pressure level to the ear, not match the sensitivity of our hearing. There is no filter to sounds in real life. If we record those accurately(big if...), then we want and exact reproduction. Where hearing sensitivity curves come into play is if we want to play back sounds at different levels than they would naturally occur at. Now there's a loaded statement... Even then, we are not really concerned with the shape of the curve, but rather how the curve CHANGES with level.

Regards,

 

[Manuel Delaflor]

Richard Greene,
Excellent post. I will keep it for reference, if you don't mind.
Chris Tsutsui,

I was wondering if anyone took an RTA measurment and compared it to a Radio Shack SPL meter that had compensated numbers?

I just measured against a Bruel&Kjaer and my initial guessing (about my doubts regarding the "compensation" numbers over the Internet) is confirmed. The "calibration" files are wrong at least regarding my SPL Meter. Unfortunately I dont have the numbers here (I will get them next week) but from memory I can say that the response was not that bad. From about 200Hz to 2Khz it is pretty flat (about +/-1dB compared to the reference. Above 2Khz the FR was higher than reference for about 3dB, about 5khz at around 10Khz ant then if fall sharply than reference to 20Khz. On the lower FR it was about 3dB down from reference to around 70Hz and about 5dB to 30Hz... please excuse me if my numbers are not correct at this time, but my point is only to show a light about the so called "calibration" files.

 

[BruceD]

Manuel,

I believe what you measured is exactly what a "C" weighting curve is supposed to represent.

Remember, the RS SPL meter's preamp modifies the mic's input to fit either an "A" or "C" weighted curve.

Neither of these curves is flat from 20Hz-20kHz.

You can modify the preamp to produce an unweighted response, the circuit mods are available on the web.

 

[Manuel Delaflor]

BruceD
That's interesting. Im not sure in wanting to modify my SPL Meter. Im considering a calibrated mic. But of course, in the meantime, I will use my calibrated SPL meter!
I used "C" weighting, but the results from about 2Khz to 10Khz are higher than the reference. On the lower FR it was just as expected.