Re: SDR operated with a noise source


I don't see here that you understand noise at all well, especially the characterization and measurement of noise. First off, S9+30 dB is in practical terms no more descriptive than, "You have a bodacious signal, OM." My experience with giving calibrated signal level reports and how poorly they were received suggests a properly calibrated S9 point is about as likely as a purple unicorn poaching your breakfast cereal.

However, -43 dBm is a fairly large number. If it really is 10 GHz wide then it's level in a 1 Hz bandwidth is about -143 dB. Yeah, that's a fair number to survive insertion of some attenuators for an accurate measurement process. (Boltzmann noise, which you cannot go below, is about -174 dBm, -204 dBm. Note: That's dBm not S "-7" or something. S-Units are "good buddy" measurements for somebody interested in any degree of precision.) Including or not including the bottom 100 MHz is not easily measured compared to leaving out the bottom 100kHz or bottom 1 Hz. The formula for Boltzmann noise is kTB, k = Boltzmann's constant, T is temperature degrees Kelvin, and B is the bandwidth in question.

Hopefully you see the "oops" above. I left out your S9+30 measurement. I don't know the bandwidth you used for measurement. Let's figure you used 10k kHz for AM measurements. Move that up to a 10 GHz bandwidth and you are talking about 60 dB more noise power over that whole 10 GHz bandwidth - it becomes a bit under 1 watt, actually +23 dBm. That is a ridiculously high power level for a noise generator. To measure a noise figure you want a noise source roughly equal in power to the noise figure you wish to measure. Since it would be generating noise at -83 dBm/Hz it's not good for noise figure measurement.

That thumbnail analysis is why I figure you are groping in the dark. There are a lot of good articles to read on the web. Some will probably be at a good starting level for gaining a quick study level of knowledge. And even the highly technical ones are not all that hard to understand when you extract the fundamental numbers and formulas.


On 20210115 03:53:01, Allan Isaacs wrote:

That’s more or less what I expected. The SDR meter provides a jolly useful measure of change and not anything absolute.

I began to get interested in this topic when discussing the Tiny 2.8 Spectrum Analyser that has no tracking generator unlike my Rigol.


I built a noise source that seems to work OK and provides, after some iterations, a reliable output from around 40KHz to a couple of GHz so should enable the Tiny SA (or other things, even SDRs) to check things like filters.

I’ve been using my Rigol as well as my SDR and both give the same subjective results (in other words both provide equal filter test characteristics) but with the Rigol giving accurate power readings or attenuation figures.

One point of interest is the total power from the noise source. Although the power level is not especially high at any given frequency, for example it shows as S9 +30dB on a decent communications receiver from 100KHz to 30MHz, it develops 1mW (0dBm) as measured by my HP power meter (10MHz to 10GHz). If the power meter spec covered 0-10GHz that 0dBm figure would register a higher output, perhaps as high as +10dBm? That means that it would be a bad idea to further amplify the noise output for fear of accidentally damaging a spectrum analyser which typically has a max input rating of 20dBm.

Allan G3PIY


From: [] On Behalf Of jdow
Sent: 15 January 2021 10:25
Subject: Re: [SDR-Radio] SDR operated with a noise source


Until you calibrate it there is no way SDRC can be thought of as being calibrated.

Without a thorough of what Simon is displaying you cannot take the spectrum levels as meaningful.

Without an even more detailed knowledge of the S-Meter function you cannot track it back to anything meaningful, pretty much the same as your average rice box. (I have managed to completely fool Simon's S-Meter tool by 20 dB or more. Unless he changed it the tool is merely a "keep the bloody ignorant hams whining about wanting an S-Meter shut up.)

SDRC is very pointedly not a measurement system giving precise readings of power. And it probably cannot be converted into such a thing without a gawdawful amount of work.

But, you are welcome to try.


Join to automatically receive all group messages.