It was a steep learning curve for me to understand noise! And I think for most people.
I saw my predecessor within a year or so after I took his job and he asked me, "Are you able to think in the frequency domain yet?"
He hit the nail on the head, as most people are used to thinking in the time domain and not frequencies. All of my previous test experience was in the time domain - force over time, displacement over time, etc.
That is probably where untrained people start making mistakes, as understanding frequencies helps us understand how humans actually perceive sound. And it gets even more complicated from there if you start taking measurements.
A-weighting filters what the microphone records in the real world to represent what humans perceive.
So if Formi or anyone else posts A-weighting, then you need to realize that the data was manipulated in a controlled fashion based on individual frequencies to match human hearing.
A-weighting attenuates what the microphone detected for frequencies that we are not good at hearing and slightly increases it where we hear well. Our hearing is great around the frequencies for human speech, for example. Makes sense, right?
The Z-weighting is unmolested. It's the sound pressure in the real world with no attenuation or manipulation of the data.
Humans don't hear well at lower frequency. See how the blue line for A drops off drastically from 1000 back towards zero? Since we don't hear well there, any sound is attenuated/reduced to represent our difficulty hearing those frequencies.
So a hypothetical can with most of its sound energy from 1 to 1000 Hz would have a lower reported noise than one with most of its energy from say 1000 to 2000, using the A filter. That could be misleading though, right?
With Z, you cut the bullshit and show what happened in the real world. No filtering, no attenuation.
So, A-weighting can be misleading if there's a lot of energy being attenuated as damage can still be done to systems even if we don't hear it so good!
Put another way, if Can1 had lower reported values than Can2 with A-weighting, the astute person would say, "What about Z-weighted values, as I don't care about the filtered values?"
View attachment 934834