A question about RF receivers...

Sounds like something is wrong. Noise figure can't be negative.

Must be + then.. not my strong subject :(

If can be if there's some averaging involved somewhere, and since he's measuring at IF that's very much possible.

But I suspect that what's really going on here is that while his measurements are calibrated to some relative standard, they aren't calibrated to an absolute standard and hence he's really getting something propotional to noise factor rather than an absolute noise factor measurement. I.e., "engineering units" noise factors.

(Common example: Someone testing a narrowband device with a wider-band noise figure meter, which can lead to "negative" noise figures. Arguably this is an error, although for production testing I can see someone not caring and just adjusting their liimt lines accordingly...)

That makes more sense...

Noise figure effectively tells you how much noise the circuitry in your receiver is adding to whatever the incoming "intended" signal already has: If you had a perfectly noise-free receiver (0dB noise figure), you could exactly amplify and duplicate any input signal no matter how small it was. (...but note that "any input signal" will already have a certain amount of at least thermal noise on it, and over-the-air signals have plenty of additional means of picking up more noise...)

Under certain simplifying assumptions, what a receiver with, say, a 5dB noise figure does is that it takes an incoming signal that has, say, a 20dB SNR and reduces the SNR to 20-5=15dB. So clearly lower noisre figure receivers are desirable, and you're testing front ends to make sure they don't add an unacceptable amount of noise to the incoming signal, thereby degrading the incoming SNR too much.

---Joel

Maybe receiver technology has gotten better since you were in school?

Perhaps the final drive amps get cold the more output power, too.

sks

o the

se

Maybe you are reading the level of the noise rather than a noise figure. Then negative values are possible, i.e. the level is less than the reference 0dBm point.

Thanks guys, that's a good explanation.