Measurement inconsistency

[HiImJeff]

Hi all

Not sure what is not enough info, and what is too much, but here goes.

My home studio was designed (not from the ground up) by a professional acoustician. Lots of panels and diffusors.

Even after he took my measurements (not rew) and gave me the all clear, I am, being a perfectionist, looking to see if I can get even better. I have a DBX rta-m routed to a Focusrite Scarlett (figured this would be better than my BAE 1073), it is pointed upward, it is equidistant to both tweeters. I have a pair of Dutch and Dutch 8Cs, and a done the rew calibration process as per the instructions.

Even if everything remains the same (same mic position, same levels, etc), I am not expecting the measurements to be 100% identical every time, but many times the measurements are noticeably different. Can someone explain to me why this is, and if there is anything I can do to alleviate this?

Thanks in advance.

 

[Keith_W]

In the example you posted, it is what I would expect from a minor change in microphone position. I don't see anything unusual here.

Also, please do not clip out the vertical scale when posting measurements.

 

[HiImJeff]

Hi, and thank you.

The mic has not moved.

 

[Keith_W]

Please post the .MDAT.

 

[HiImJeff]

Att. B/c I didn't save the last one, I had to take another one. This one is closer, but still not the same.

Thanks for the help.

 

[Keith_W]

Are you SURE the microphone was not moved?

You can see that the impulse responses are aligned. Let us examine this reflection that I circled. The time-lag between the two reflections is 0.0159ms. Using the speed of sound, d = c * t (d = distance in m, c = speed of sound 343m/s, t = time in seconds) the distance is 5.45mm. This means the mic was moved 5.45mm between measurements.

Tiny changes in microphone position can result in a different measurement, and this is most evident in high frequencies were wavelengths are very short (20kHz = 17.2mm), so the small movements are relatively "large" relative to the wavelength.

The other possible explanation is poor signal-to-noise ratio. I have already looked, and you're not suffering that problem.

Are you using a proper microphone tripod?

 

[ddude003]

Is there possibility human(s) or animal(s) in the room that are in different places from measurement to measurement?

 

[HiImJeff]

Yes, I am positive the mic has not moved. No pets, I live alone.

Is it possible that my chair - which is in front of the mic - not being in exactly the same position could have affected this?

And me - I was in the same general area, but in exactly the same spot, maybe not. But there, I am off to the side and back.

 

[John Mulcahy]

Anything that has moved in the room will change the result.

 

[HiImJeff]

Got it.

I took two more measurements. One last night, and the other this morning. Added a 7 sec delay to the start, to give myself time to exit the room and close the door, so my positioning doesnt affect anything. Didn't touch the mic, didnt touch the chair. Ddn't turn the computer nor the monitors off. I even made sure that the mouse was in exactly the same spot and position. The only thing that moved is the door. Which was completely closed before 7 sec.

 

[Keith_W]

Very good!

 

[HiImJeff]

Very good... meaning?

 

[Keith_W]

Meaning that your measurements are virtually identical. Those tiny little perturbations are normal. It's only like 0.1-0.2dB here and there.

Don't forget that loudspeakers and rooms are imperfect LTI (linear time invariant) devices. Voice coils can heat up, the speed of sound changes depending on temperature, humidity, and air pressure, you may get a different frequency response if you measure at a different volume, etc. I wouldn't worry about those little variations that you see.

 

[HiImJeff]

K, thanks.

It's only like 0.1-0.2dB here and there.

K, but I am seeing more in certain spots. Below is All SPL zoomed in, and I am seeing approx 13 db @54. What am I missing?

 

[ddude003]

I think we are down to splitting hairs... As Keith implies there are some things beyond our control to make a perfect match between measurements taken days apart... If you really want to spit hairs then take several measurements and create an average of those several measurements... BTY it looks like you have a null at 54.5hz ish... Do not try and boost your way out of a null... Better to squash everything else down to a room curve... If you don't add some physical room correction... Ah, DSP... FIR filters and Convolution... Or buy some more subwoofers... Are three to many, four not enough...

BTY I make measurements at 75dB... Do you ever listen above 75dB???

 

[John Mulcahy]

K, but I am seeing more in certain spots. Below is All SPL zoomed in, and I am seeing approx 13 db @54. What am I missing?

Two things. Low levels in the measurement will be more affected by background noise, and deep notches in the response are due to cancellations and are very sensitive to the smallest change in the sound along any of the paths that are combining to generate them. Two exactly equal magnitude but opposite phase signals combining would create an infinitely deep notch, only filled by the measurement's noise floor. Changing one magnitude by just 0.1 dB would change the notch from infinitely deep to about -39 dB.