I have taken noise data with a spectrum analyzer. Unfortunately, it doesn't seem able to plot the noise in the units of V/Sqrt(Hz) over frequency (Hz).
What I have acquired instead is dB vs frequency. I know the measurement bandwidth (100Hz), the resolution (# of data points) and reference power (10 dBm).
Is it possible to convert my data into the units of (V/Sqrt(Hz))?
Default for this instrument is measuring in \$dBm\$ (dB referenced to 1mW at 50 ohms). Page 87 in the manual shows how to set the Y-axis to various units. However, it looks like this analyzer doesn't support spectral density.
To go from \$dBm\$ to \$dBV\$ (dB relative to 1 volt) in a 50 ohm system: $$ dBV = dBm - 13 $$ To go from \$dBV\$ to \$dBV/Hz\$: $$ dBV/Hz = dBV - 10 \; log(noise\_bandwidth) $$
To go from \$dBV/Hz\$ to \$ V/\sqrt{Hz}\$: $$V/\sqrt{Hz} = 10^{dBV/Hz \over {20}}$$
The problem is figuring out your noise bandwidth, which is not the same as resolution bandwidth unless the filter is a brick wall. Analyzers that give a spectral density readout know the relationship between resolution BW and noise BW. As a first guess, you can use the resolution bandwidth as your noise bandwidth if the bin filter is steep walled. You should actually calibrate your analyzer against a known noise source if you desire accuracy to figure out the ratio between noise_bandwidth and resolution_bandwidth.
Another thing is the Video filter. The Video filter alters noise measurements. Again, you need to use a calibrated noise source to determine the effects of video bandwidth. As a guess, I would make the video filter bandwidth as wide as possible (you need to experiment to understand this instrument). The analyzers I use, I disable video filtering for accurate noise measurements.
