Atmospheric Noise Levels

N7AIG

I'm doing some noise studies in preparation for some upcoming FMT runs (frequency measuring tests -- complete with Rubidium clocks, GPS disciplined oscillators, and all other variety of craziness). The panadapter displays are derived from FFT's. Hence, signal levels represent average power over the duration of some number of samples. There is inherent tension between the desire for accuracy of narrowband signals (e.g., CW), and estimates of the noise floor. An FFT is not simultaneously ideal for both of these at the same time. So the compromise is usually made in favor of narrowband signal reporting.

I can make the noise floor corrections on my own as long as I know the duration of the FFT's being used. Can I assume that this is indicated by the "Buffer Size" on the DSP/Options panel of PowerSDR Setup? I see it defaulting right now to 2048 samples at 48 kHz sample rate. And hence the FFT bin width is 48 kHz divided by 2048 = 23.4 Hz?

Eric-KE5DTO

I believe we always use 4096 sample FFTs in PowerSDR.  If you choose a buffer size that is less than this then they are collected until we have enough.

N7AIG

I should apologize for even asking. Looking at the waterfall display shows a band noise level of roughly -130 dBm while a 1 kHz CWU bandpass shows an S-meter reading of -114 dBm. Hence, your FFT analysis bins have an effective bandwidth of somewhere around 25 Hz. That being the case, with a sample rate of 48 kHz for the Flex-1500, you must be using FFT's with 2048 samples, assuming you are using Cooley-Tukey with 2^N samples.

Tim - W4TME

The S-meter displays the integrated power within the receive filter.  It will be in just about all cases be higher than the display which is displaying power in the individual FFT bins.

N7AIG

Yes, of course, I understand that. What I'm pointing out to you is that, unless you are doing something very strange, your numbers work out to show that the Flex-1500 PSDR really is using a 2048 point FFT for its pan adapter and waterfall display.

Tim - W4TME

RRR.  Got it.