It's all about the Software

Dear Simon,
First of all, thanks for witing this superb software.

Over these last few days, I finally succeded measuring the noise figure of some VHF preamps I have here. The results seem quite accurate and maybe some users of SDR Console could also be interested in measuring their own preamplifiers.

I used the so-called "gain method" as explained in a document I got online from the ITU.
This method requires to measure the gain of the preamp followed by the RTL SDR. This is quite easy using a calibrated signal generator and SDR Console as it is now.

The "gain method" also requires to calculate/measure the noise density (in dBm/Hz) of the noise floor, having the preamp input terminated with a 50 ohms load.

To obtain the noise densiity, the level of the noise floor, measured in mW must be divided by the width of the passband.
What I found with my copy of SDR Console is that the level of the noise floor as displayed by the s-meter, remains the same if I try changing the filter width. I have no way to know which filter width the software uses to display the noise floor level.
That would be useful to know.

Also, the s-meter display fluctuates quite a bit (2-3 dB, sometimes a bit more) when looking at the noise floor, which is of course only normal.
Accuracy when measuring the noise floor is crucial when you are trying to determine a noise figure. I think it would be very useful to have the option to integrate the level of the noise floor over several seconds, maybe 10 seconds or even more. This would probably allow an accuracy of a tenth of a dB?

Another suggestion for an improvememt of SDR Console: Many spectrum analyzers offer the possibility to display directly the noise density. Maybe this could be offered as an option in the s-meter settings of SDR Console?

Regards,

Yvan

Are you sure the S-Meter is calibrated? As mentioned elsewhere I have spoofed it into giving a reading VERY much at variance from what is present.

If you are using the spectrum display you are not measuring in any bandwidth you can quickly calibrate. It varies with the width of the spectrum. How many FFT bins live within each pixel? And "is Simon summing the squares if the I and Q values for each spectrum entry for each pixel?" (no) (I keep twisting his arm to do this. Yes, it is not perfect; but, it is arguably better than what is being done. It is also calibratable for a given FFT weighting function.) As it is, you cannot do reliable noise figure work on SDRC. You can tell if A is better than B, however. But I get the drift that this is not what you want.

{^_^} (Been working with noise figure professionally since about 1960 or so.)

Yvan,

What is Noise Density - how do I calculate that?

In a given bandwidth with no signal present, as with a dummy load or cooled dummy load, measure the noise power within a given bandwidth. The noise density is the measured power in dBW / Hz. This can be fumbled around to a noise figure by comparing with the same filter with an added signal or an added noise source at a calibrated level.

{^_^}