In a discrete measurement system, the Nyquist criteria requires you to sample your signal at a rate that is at least twice the bandwidth (or span) you want to measure. In praxis values like 2.50, 2.56 (or higher if oversampled) are used.
So when you expect your span to be 20 kHz you should sample at a rate of at least 2.5*20 = 50 kHz. The same story applies to your "number of lines", so if you request your measurement span to be 800 lines wide, you should acquire data blocks of 800 * 2.5 = 2000 samples.
I have modified your VI to reflect that. If needed you can change the constant value "2.5" to a higher value (like 5.00 or 10.0 depending on the board you are using) to reduce effects of aliasing on your measurement results.