Noise Figure and Noise Floor

Hi all,

I want to compute the noise figure and the noise floor for my reciever.

for the NF i use the Friss equation.
But for the noise floor i d'ont found any formula.

With an amplifier the No floor is : Noise output = Noise input + Gain + NF
But with an rf attenuators the NF is equal to the attenuation value. So 
Noise output = Noise input + Gain + NF don't work when Noise input 
 > -174dBm/Hz

Someone can help me ?

For example
input signal : power -50 dBm / noise floor -174 dbm/Hz
first  amplifier Gain 30dB NF 1.25dB
second attenuators Gain -3dB NF 3dB
third amplifier Gain 10dB NF 0.8dB

what is the final noise floor ?

>So 
>Noise output = Noise input + Gain + NF don't work when Noise input 
> > -174dBm/Hz

Actually it does. Personally I believe that everybody who does receiver
system design has to fall into that pit once.

There is one equation I have to remember, I can derive everything else
incl. Friis' formula from it:

------------------------------------------------------------------------
Input-referred noise contribution by the device is (F-1) kBT 
referred to the input
------------------------------------------------------------------------
F: noise figure (linear scale)
k: 1.3806e-23 J/K
T: Temperature, and assume for simplicity's sake that the source
temperature is the same as the attenuator.

The INPUT-REFERRED noise contribution of the attenuator is (F-1) kBT. The
source noise is kBT. So the total input-referred noise is F kBT.

Example: Let's consider a 20 dB attenuator: F is 100. The INPUT-REFERRED
noise contribution is 99 kBT, the source contributes 1 kBT => total 100.
The attenuator scales it down by a factor of 100. The total output noise
is kBT, as thermodynamics demand for matter at temperature T.

To answer your example question, calculate for example as follows:
- Determine the equivalent noise figure using Friis' equation
- Determine the INPUT-REFERRED noise contribution and add 1 (kBT)
- Scale with the gain and you've got the noise floor at any point in the
chain.

Hope that clarifies.

Cheers

Markus

PS: The reason why I have capitalized "input-referred" three times: it is
REFERRED to the input, although this number is a power / power density, it
is NOT a physical quantity I could measure with a power meter.

thanks for your answer,
i've understood now

"mnentwig" <mnentwig@elisanet.fi> a &#4294967295;crit dans le message de news: 
6u-dnZ5ETJpNsK_anZ2dnUVZ_gudnZ2d@giganews.com...
> >So
>>Noise output = Noise input + Gain + NF don't work when Noise input
>> > -174dBm/Hz
>
> Actually it does. Personally I believe that everybody who does receiver
> system design has to fall into that pit once.
>
> There is one equation I have to remember, I can derive everything else
> incl. Friis' formula from it:
>
> ------------------------------------------------------------------------
> Input-referred noise contribution by the device is (F-1) kBT
> referred to the input
> ------------------------------------------------------------------------
> F: noise figure (linear scale)
> k: 1.3806e-23 J/K
> T: Temperature, and assume for simplicity's sake that the source
> temperature is the same as the attenuator.
>
> The INPUT-REFERRED noise contribution of the attenuator is (F-1) kBT. The
> source noise is kBT. So the total input-referred noise is F kBT.
>
> Example: Let's consider a 20 dB attenuator: F is 100. The INPUT-REFERRED
> noise contribution is 99 kBT, the source contributes 1 kBT => total 100.
> The attenuator scales it down by a factor of 100. The total output noise
> is kBT, as thermodynamics demand for matter at temperature T.
>
> To answer your example question, calculate for example as follows:
> - Determine the equivalent noise figure using Friis' equation
> - Determine the INPUT-REFERRED noise contribution and add 1 (kBT)
> - Scale with the gain and you've got the noise floor at any point in the
> chain.
>
> Hope that clarifies.
>
> Cheers
>
> Markus
>
> PS: The reason why I have capitalized "input-referred" three times: it is
> REFERRED to the input, although this number is a power / power density, it
> is NOT a physical quantity I could measure with a power meter.
>
>