Measuring Noise Figure using a Spectrum Analyzer - The Gain Method

The gain method of noise figure measurement is explained and demonstrated. The only assumption made here is that you have a signal generator and a spectrum analyzer. No noise figure meter or calibrated Y source is involved.

I discuss a little bit about the Y-factor method and why it is better.

For all I know, everyone who watches this video hates it. Clicking "LIKE" is how you set the record straight on that. If you found this to be useful, please click on "LIKE" so that I know that this content is valued.

Please consider clicking on "Like" so that I know what sort of content is being appreciated. I don't ask for support for developing this content, just the satisfaction of knowing that it is appreciated.

#NoiseFigure
#MicrowaveTest
#TestAndMeasurement

[stephenremillard]

As soon as I post any video I always see something that I wish I had said. I would like to add the little fact that -174 dBm/Hz is the ambient noise only at 290 Kelvin. In the vicinity of room temperature it goes up (becomes less negative) by about 0.07 dB for every 5 degree Celsius increase in temperature.
Also, at 10:40 I mention too briefly the need to account for cable loss in the gain measurement. So Let me re-emphasize it here. Careful that your gain does not include cable loss.

[Berk-lfge]

Thank you so much. You are a fantastic teacher

[springer]

I have a variable gain setup with an attenuator followed by an amp. For our "low gain" (0dB) setting, there is a -22 attenuator feeding the amp. In that case, I cannot simply terminate with a 50 ohm terminator, because then the amplifier wouldn't "see" the -22 attenuator. Therefore is it a fact that I can't use the "gain method" for this setup?

[kapilgavali]

The center frequency is 20MHz while measuring the noise power of around - 137 dBm. It should have been 40 MHz right?

[jasonconerty]

Use spec an often at work to verify Noise Figure or estimate noise power at DUT output. Calibration is critical. Always expect some error when testing via this method

[j.w.]

I will be using pyramidal horn antennas with an LNA and a spectrum analyzer to scan in a 360 degree azimuth to look for possible interference in the X and S bands (2GHz and 8GHz). One of the requirements is to "Calculate cold sky noise temperature at 10 degree elevation." Would you (or anyone) perhaps know how one does this? Is this simply measurable somehow? Google doesn't get me too far...😕