SNR function returns low values

Question

db 2022-5-14

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1719345-snr-function-returns-low-values

回答： Jerome Blair 2022-7-7

Hi,

I'm trying to use the snr function with data acquired from a spectrum analyzer but the value returned by the function is way too low. The spectrum analyzer is connected to a DAC that is generating a 100MHz sine wave and the goal is to characterise the DAC using the data acquired from the spectrum analyzer. The spectrum analyzer does measurements with an RBW of 10KHz and the data is saved as CSV file. This is what the data looks like, the line horizontal line should be the noise floor(average of the dbm values excluding the 100MHz signal), I don't know if I'm calculating the noise floor correctly but graphically it does make some sense. The biggest peak is at 100MHz and the remaining peaks are harmonics as expected. Subtracting the manually computed noise power from the signal power yields a value of approximately 87 dB so I was expecting a similar value from the snr function.

The problem starts when I use the snr function on the same data using this command

snr(db2pow(dbm),freqs, 10000,'power')

dbm is the array that stores the dBm values and freqs is the array that stores the correcponding frequencies. I set the rbw value to 10000 as the spectrum analyzer did the measurements with an rbw of 100KHz but I don't know if it's supposed to be that way. Anyway the function returns 41.76 dB which seems too low as the noise power seems to be at around 85dB and increasing the RBW value in the function somehow increases the SNR whereas increacing the RBW reduces the noise floor when using a spectrum analyzer. Sorry for the long and verbose text but can anyone please point out what I'm doing wrong?

Thanks.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Jerome Blair 2022-7-7

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1719345-snr-function-returns-low-values#answer_1001950

The noise power is not the level of the level of theh noise on the spectrum analyzer--it is the sum of the energies at all excluded (signal and harmonics) frequencies. There are about 25000 of these frequencies (Max_frequency/RBW), so the snr is about 44 dB lower (10*log10(25000)) than teh observed value.