Temperature Control In Speaker Measurements

[Dave Zan]

I believe it has to do with collision rate between water molecules (remember, they are bent,...

I have a recollection of the way the specific heat of polyatomic gases will increase as temperature increases because new modes become accessible as the quantum threshold is passed.
So I understand (more or less) the way translation, rotation and finally vibration modes become activated.
The fact that water molecule is bent will make a difference to this I expect.
I still have no idea how this plays out as the humidity is varied.

Best wishes
David

 

[j_j]

What I remember is that at a certain point, the energy put into the rotation around the smallest moment stops being as lossy because it transfers the rotation to other molecules as they become more common in the air, rather than going into heat, so first loss, then less loss. (but never back to "dry air".

 

[Dave Zan]

What I remember is that at a certain point...rotation around the smallest moment...(but never back to "dry air".

Your memory may not be quite correct- seems to be dominated by vibrational mode relaxation, in the frequencies of interest to us it's O2, N2 at lower frequencies, "catalysed" by H2O.
And at lower frequencies the "dry air" actually has more loss than wet - another behaviour I had no idea about.
Probably because losses at the lower frequencies are too low to matter in "normal" practical audio.
So it's even less intuitive than I realised, not very useful but fun to have learned.

Best wishes
David

 

[j_j]

Your memory may not be quite correct- seems to be dominated by vibrational mode relaxation, in the frequencies of interest to us it's O2, N2 at lower frequencies, "catalysed" by H2O.
And at lower frequencies the "dry air" actually has more loss than wet - another behaviour I had no idea about.
Probably because losses at the lower frequencies are too low to matter in "normal" practical audio.
So it's even less intuitive than I realised, not very useful but fun to have learned.

Best wishes
David

Oh, intuition is "mostly useless" for sure. And not sure what you mean by "low" frequencies, but really, really dry air tends to go back to better transmission at most frequencies, but only dry-as-a-bone.

But yes, intuition is "mostly harmless". And my memory is from about 1972/3 so it is entirely likely to have faded a bit, since that's the last time I used that particular data for a test answer.

 

[Dave Zan]

....not sure what you mean by "low" frequencies, but really, really dry air tends to go back to better transmission at most frequencies, but only dry-as-a-bone.

By "Low" frequency I meant the usual audio use.
By coincidence the scale of frequencies over which the complexity occurs is just about the same as human perception.
Red line is N2 contribution, clearly dominant at "low" frequencies below, say 200 Hz.
Too small to matter to a normal "sound reproduction" system but fun to have learned about, thanks for the push.
In the second picture there's a narrow band just above 200 Hz where the absorption....
Well, you can see for yourself, kind of hard to summarise the effect of humidity in that plot!

Best wishes
David

 

[hardisj]

I did some experimenting yesterday in my car (yes, car audio does have more than just boom boom). I set up using REW and a loopback with a Behringer ECM8000. I measured in both 50*F and then 80*F. The mic was set up at my seated position, where the left side speakers are roughly 3 feet away and the right side speakers are roughly 4-5 feet away. What I expected was:
a) The speed of sound would increase with the higher temp but when I took the time to make individual speaker adjustments based on the temperature change that the overall delta between left and right sides would not matter.
b) The short distance of microphone to speaker would mean no real change in frequency response because there wasn't enough distance for the sound waves to be impeded by the "medium" (air temp/humidity).

Results:
a) I was wrong by about 0.03ms. So, I made a little spreadsheet and saw why. Ok. No big deal. Time Delay settings varied due to temperature/humidity adjustments can easily be remedied with math. Next.
b) The FR changed quite drastically. The higher temp measurement showed a drop of about 2-3dB over the entire passband. This had me thinking surely something I did was wrong. So I re-tested today. And that's when I realized that maybe it was the mic itself. So, I tested this theory by heating the mic itself up but keeping the car temp inside at about 50F. Sure enough, the "mic warmed up" measurement resulted in a drop in amplitude over the entire frequency response range. I then changed mics to the UMIK1 and tested the theory again. This time heating the mic up didn't change the response. So I proceeded with the testing again and in this case I had as much as a 2dB drop in the "warm car" portion of the experiment.

What I learned from that is:
1) My ECM8000 is very sensitive to temperature changes. I don't know if that's due to age (I've had it for about 10 years) or if it's a design issue. I'd like to test this further but I just ran out of time this weekend and I'd prefer to use a different XLR microphone so I can properly do a loopback measurement for other reasons.
2) Aside from the ECM8000 tests, based on the UMIK1 results the temperature does indeed effect the response but in my case it was only below 40hz and above 1khz.

These results are absolutely not universal. But I think there's enough information here to at least make people caution how and why they're taking and using their measurements. Even though the results weren't what I expected I did see it as a learning experience.

I'm putting together a video showing the results differences and walking through all the steps I went through to get the measurements. If you guys are interested I'll post it here once I finish it.

- Erin

 

[hardisj]

And to be clear, even though there are only a few measurements in the screenshot above, I have 5 different files of measurement sets (REW caps you at about 40 or so, apparently). I kept trying different ideas to see if I could make sense of why I was getting such drastic changes relative to what I expected and have probably 90-something individual speaker-drivers and "side" (left/right) responses. It's a LOT of data to go through and make sense of. But I summed it up pretty well above.

 

[j_j]

Complicated is the only way to put it.