IMD Measurements Indoors vs Outdoors --- Results Compared

[Joseph Crowe]

I would like to share my latest YouTube video comparing IMD both indoors and outdoors. There are some cautionary conclusions for indoor measurement. It does raise some further questions, such a why the room reflection artifacts disappear after raising the test SPL to 100dB versus 95dB. It begs the question why the room reflection level doesn't rise in a linear fashion with the test signal SPL.

 

[pozz]

@Joseph Crowe have you tried measuring speakers with different sensitivities and comparing?

 

[restorer-john]

The attached ARTA images tell it all.

The noise floor inside is way lower than outside, not the other way around. You're also making declarations based on one Arta FFT with 9 averages (indoor) vs 21 averages (outdoor). The "noise" is what the averaging pushes down in an FFT- the longer the averaging, the more it pushes it down. It isn't the actual acoustic level of noise.

Not only that, as you can see the outdoor noise is much greater in the low frequencies as you would expect. And nobody measures an overal acoustic noise level at 1kHz only.

And you say "the sampling rate loses resolution" at low frequencies. It does not. It is the exact opposite. The FFT bin size and log scaling on the plot mean the low frequencies appear to jump around, that's all.

All the "differences" in the "intermodulation" tests can be explained again by the averaging. 12 vs 24. 12 vs 18.

 

[Joseph Crowe]

The attached ARTA images tell it all.

The noise floor inside is way lower than outside, not the other way around. You're also making declarations based on one Arta FFT with 9 averages (indoor) vs 21 averages (outdoor). The "noise" is what the averaging pushes down in an FFT- the longer the averaging, the more it pushes it down. It isn't the actual acoustic level of noise.

Not only that, as you can see the outdoor noise is much greater in the low frequencies as you would expect. And nobody measures an overal acoustic noise level at 1kHz only.

And you say "the sampling rate loses resolution" at low frequencies. It does not. It is the exact opposite. The FFT bin size and log scaling on the plot mean the low frequencies appear to jump around, that's all.

All the "differences" in the "intermodulation" tests can be explained again by the averaging. 12 vs 24. 12 vs 18.

The attached ARTA images tell it all.

The noise floor inside is way lower than outside, not the other way around. You're also making declarations based on one Arta FFT with 9 averages (indoor) vs 21 averages (outdoor). The "noise" is what the averaging pushes down in an FFT- the longer the averaging, the more it pushes it down. It isn't the actual acoustic level of noise.

Not only that, as you can see the outdoor noise is much greater in the low frequencies as you would expect. And nobody measures an overal acoustic noise level at 1kHz only.

And you say "the sampling rate loses resolution" at low frequencies. It does not. It is the exact opposite. The FFT bin size and log scaling on the plot mean the low frequencies appear to jump around, that's all.

All the "differences" in the "intermodulation" tests can be explained again by the averaging. 12 vs 24. 12 vs 18.

Are you suggesting that I didn't allow sufficient time for the averaging to take place? It's quite obvious by observing the averaging in real time to know when it has sufficiently lowered enough to not be a factor. The side band noise products are very obviously consistent enough to not be averaged away so this is an area that needs to be trusted that the test procedure was carried out correctly.

 

[Joseph Crowe]

@Joseph Crowe have you tried measuring speakers with different sensitivities and comparing?

I've tested quite a range of speakers so far. The test results don't suggest that sensitivity is a factor in IMD performance.

 

[audio2design]

Am I missing something? These two images are identical. Perhaps there are very slight differences, hard to tell at the resolution, but otherwise absolutely identical. That means a measurement error. Can't tell you the source since I didn't do it, but I am confident there is a problem in the measurement.

 

[Joseph Crowe]

Am I missing something? These two images are identical. Perhaps there are very slight differences, hard to tell at the resolution, but otherwise absolutely identical. That means a measurement error. Can't tell you the source since I didn't do it, but I am confident there is a problem in the measurement.

View attachment 134528

Yes, you are right! I've changed the graph to the correct image.

 

[audio2design]

It does raise some further questions, such a why the room reflection artifacts disappear after raising the test SPL to 100dB versus 95dB.

They don't disappear, they get buried under the distortion which is incorrectly labelled noise at higher volume levels, then again you could have some input/output crosstalk, etc. Either way, I don't think your conclusion is valid.

There is also a major shift in the outdoor noise floor at 20Hz on your last post though the shape of the noise floor seems the same. Looks like about an 8db change. Either way, the similarity of the distortion (noise) at 105db between indoor and outdoor tells me that a new source is dominating and that is likely distortion.