Post Processing Filters

I need some guidance. I want to post process a data set to put an audio A-weighting on it. I have 200 data points with the level and frequency for each; both the DUT as well as an A-weighted filter. I am getting a value close to what I get when I run an A-weighted measurement by its own but not quite there. Is there a major flaw to my math here, is there a better way of doing this?

Normally you add the curve to your results (or subtract). So for 1kHz freq you let it as it is since it's reference (0dB in the table) and then, for 2kHz you add 1.2dB to your results, for 40Hz you subtract 34.6dB, etc. Following a table you can find anywhere, usually in 1/3 oct or 1 oct bands. Easy to build on exel or similar. Not much there. Once you have all your values you add as usual with dB. Take level from dB, sum=10^(20Hz.lvl[dB]/10)+10^(22Hz.lvl[dB]/10)+...+10^(20kHz.lvl[dB]/10) and then 10log(sum)=A-w.lvl[dB]
Notice it's always with 10 and not with 20 or squared since you want to add already squared, as power, not voltage or energy, not pressure. No need to use the reference, since all will keep the same, divide and then multiply once summed, no reason to do that unless you want the energy or power involved but doesn't make sense since it doesn't has a physical meaning, once A-wighted.

JS

What sort of signal are you measuring (noise?), and how were the narrow-band measurements done?

Samuel

Thank you for the reply,  I thought that my Audio Precision did not have an A weighted card.  I opened it up and found out that I was wrong so I won't need to do this post processing.  But thank you so much for the information!  I was going to see if a swept band pass or a swept frequency response of the noise.  If that did not work I was going to do some investigation into using a FFT.