how to measure input and output referred noise of amplifier in practice

[gouzou]

As said, I want to measure the input and output referred noise of an amplifier in practice.

I have some thoughts and want to share them to see if they hold true.

Assume a single ended input, single ended output amplifier.

Should i do it like,
1) short the input
2) measure the power spectral density (which is actually noise) vs freq at output with a signal analyzer (units dbm/hz), which gives the output referred noise
3) translate (2) into nV/rootHz, divide by gain and take the input referred noise?

Thanks

 

[johnjoe]

Which measurement devices you use for noise measurements? Noise Figure meter or just SA?

 

[FvM]

The input would be terminated rather than shorted.

 

[gouzou]

Which measurement devices you use for noise measurements? Noise Figure meter or just SA?

I could use both. What are the options for each?

- - - Updated - - -

The input would be terminated rather than shorted.

If I use a 50 ohm termination, shouldn't I take its johnson noise into account as input to my amplifier?
Then, the output would be the resistor's johnson noise amplified plus the input referred noise.
That's why I thought about shorting the input.

 

[johnjoe]

The easiest way would be the noise figure meter. Ah ok, maybe not so easy because you have to setup your measurement setup carefully. But it is faster.

If you want to use only an SA maybe read some good Agilent app notes about noise measurement. Theory and practice is well explained.

- - - Updated - - -

And shorten the input is not a good idea. I think you want to measure some rf lna. They are designed for a minimum noise figure with a defined impedance, normally 50 ohm. If you short your amplifier at the input you will get something but for sure not the right noise figure

 

[gouzou]

Ok, so I have two options (please confirm/add/correct):
1) Using a spectrum analyzer, I measure a 50-Ohm terminated amplifer, get noise power in dbm, apply correction factors for detector and equivalent resolution bandwidth, compute noise power density in dbm/hz. This is the output referred noise, right? How i get the input referred noise?
2) Using a noise figure meter, I measure the NF(db) of the amplifier. Translating to noise factor F, F=1+Na/Ni, Ni is kT, I get Na which is input referred noise voltage/power. Am I right? How i get the output referred noise?

 

[johnjoe]

1) Using a spectrum analyzer, I measure a 50-Ohm terminated amplifer, get noise power in dbm, apply correction factors for detector and equivalent resolution bandwidth, compute noise power density in dbm/hz. This is the output referred noise, right? How i get the input referred noise?

You have to know the gain of the amplifier. Then you can refer your output noise to the input.

2) Using a noise figure meter, I measure the NF(db) of the amplifier. Translating to noise factor F, F=1+Na/Ni, Ni is kT, I get Na which is input referred noise voltage/power. Am I right? How i get the output referred noise?[/QUOTE]

I added you two app notes, where everything is described in detail.

 

[gouzou]

Ok, thanks.