The noise figure of an amplifier tells us the input referred noise in a 1 Hz
noise bandwidth. A 0 dB Noise Figure (NF) tells us the amp input noise is
Convert the noise power of the amp output and the input noise of the RX into
linear units (watts), add the powers, and convert back to decibels. Now
refer it to the amp input.
To start, you need to specify the noise bandwidth. Given your example, and
neglecting the 30% modulation as we don't need this, you say the RX S/N
ratio is 10 dB given a signal of -101 dBm. The RX input noise is -111 dBm.
For now let's say the RX noise bandwidth is 6 kHz.
The amp input referred noise, given the 3 dB NF, has a noise power of -133
dB in 6 kHz. Here is how this is figured:
-174 dBm + 3 dB + 10LOG(6000) = -133 dBm. Remember, this is input referred.
The noise at the 20 dB gain amp output is -133 dBm + 20 dB = -113 dBm = 5.0
The RX input noise is -111 dB = 7.94 fW.
5.0 fW + 7.94 fW = 12.94 fW = -109 dBm. This is the new noise referred to
the RX input. But we want to know the noise referred to the amp input. It is
-109 dBm - 20 dB = -129 dBm.
So, the RX noise of -111 dBm is now -129 dBm with the amp. This is a
decrease in noise of 18 dB. What is the new NF?
-129 dBm - 10LOG(6000) - (-174 dBm) = 7 dB.
What is the NF of the receiver alone? -111 dBm - 10LOG(6000) = -149 dBm
gives us the input referred noise. NF = -149 dBm - (-174 dBm) = 25 dB.
On 4/25/09, Andy Ikin <firstname.lastname@example.org> wrote:
> Hello Folks,
> Is the following, a valid method to either measure amplifier excess noise
> or make a meaningful comparison?
> Assume one measures the S/N of a AM Rx to be say 10dB S/N at 30 percent
> modulation for an input signal -101dBm.
> Then one re-measures the Rx S/N with a 20dB noiseless pre-amp. Would the Rx
> sensitivity, 10dB S/N now be a theoretical -121dBm????
> If so, then, if the 20dB amplifier had a NF of 3dB, the measured Rx
> sensitivity for 10dB S/N would be 3dB less i.e. -118dBm!!!!!
> For example, the well respected DXeng RPA-1 HF Pre-amp. has gain of 15.8dB
> at 500kHz. This amplifier increases the RX sensitivity 10dB S/N to -112dBm.
> If the amplifier was noiseless then the RX sensitivity for 10dB S/N would be
> -116.8dBm! So the difference between the noiseless amplifier and the
> measured sensitivity is 4.8dB. Can I assume that 4.8dB is the NF??? I would
> have expect a NF of over 4dB as this amplifier uses resistor feedback.
> I also have a transformer feedback amplifier, gain 14.6dB at 500kHz. This
> amplifier increases the RX sensitivity 10dB S/N to -113dBm. If the amplifier
> was noiseless then the RX sensitivity for 10dB S/N would -115.6dBm. So the
> difference between the noiseless amplifier and the measured sensitivity is
> 2.6dB. Can I assume that 2.6dB is the NF???
> The Rx is an NRD525. A HP 8568B Spectrum Analyser and Marconi 2019A signal
> generator is used for amplifier gain measurement and Rx sensitivity.
> I assume that there is probably a flaw in the above rational, but the
> figures don't seem too "way out"!!!
> Antennaware mailing list
Antennaware mailing list