Noise figure variation in a device with frequency!!

[abhijitrc]

hi,

we have seen in the data sheet of any device that noise figure of the device increases with the increase in the frequency, but the thermal remains constant over a wide frequency range! .Then how the noise figure of the device increases ???Plz clear the doubt.

 

[vfone]

The noise figure of a device increase with frequency (or decrease sometimes) because its noise figure doesn't depends only by the thermal noise.

 

[BigBoss]

How the Thermal Noise remains constant over a temperature range ?? It's impossible.It should be increased by temperature.
NF is generally increased by frequency,it's normal and expected.( Except some GaAs devices at lower frequencies because of flicker noise)

 

[mtwieg]

I doubt the datasheet actually specifies the thermal noise... it's hardly even worth documenting since it just depends on input impedance and temperature.

 

[jiripolivka]

I doubt the datasheet actually specifies the thermal noise... it's hardly even worth documenting since it just depends on input impedance and temperature.

Most active devices exhibit frequency variation of gain as well as noise figure.Many active devices like LNAs are optimized for a chosen band, so their NF can be 2-3 dB or less e.g. at 800 MHz while t grows below and above that optimal band like its gain does.
Only pure resistors and waveguide terminations have a flat thermal noise.

 

[abhijitrc]

How the Thermal Noise remains constant over a temperature range ?? It's impossible.It should be increased by temperature.
NF is generally increased by frequency,it's normal and expected.( Except some GaAs devices at lower frequencies because of flicker noise)

Since thermal noise remains constant over a wide frequency range, i mean to say that device noise figure of a device should not vary with frequency! Is it the coupled noise due to which the noise figure changes!1

- - - Updated - - -

Most active devices exhibit frequency variation of gain as well as noise figure.Many active devices like LNAs are optimized for a chosen band, so their NF can be 2-3 dB or less e.g. at 800 MHz while t grows below and above that optimal band like its gain does.
Only pure resistors and waveguide terminations have a flat thermal noise.

Sir ur statement reflects that since the input impedance of the device changes with frequency, hence there will be a variation in the thermal noise accordingly---- so the noise figure. There is an effect of coupled noise also!!! Plz do suggest me!!

 

[jiripolivka]

I do not know what you name the "coupled noise". Device NF depends on its input impedance but also on its gain, output impedance, etc.

 

[abhijitrc]

I do not know what you name the "coupled noise". Device NF depends on its input impedance but also on its gain, output impedance, etc.

I mean to say from the coupled noise is that noise is coupled to the input by the device capacitive effect, sir i want to know -- is it the device impedance changes with frequency that reflects the variation of the thermal noise !!, if the noise will increase gain will decrease automatically.

 

[abhijitrc]

i will let u know!

 

[godfreyl]

Thermal noise is not the only kind of noise in electronic circuits. There are other kinds too e.g. flicker noise, shot noise, popcorn noise etc.

 

[jiripolivka]

i will let u know!

The noise behavior of microwave active elements has been analyzed by many researchers.
In general, cascaded elements like in a receiver are described by FRiis' formula.
Individual components like low-noise amplifiers have also been analyzed in many papers but still there are people who fail to understand the basics.

Thermal noise, P=kTB, only depends on ambient temperature and bandwidth. For frequencies up to almost one terahertz, the power spectrum is considered flat.

Passive elements contribute by their loss (loss equals NF) if they are purely resistive and matched to line impedance. Reactive components do not generate noise but affect the frequency transmission.

Active elements generate inherent noise which usually exceeds thermal noise. Such noise contributions are described by their noise figure. NF indicates as a ratio the noise contribution of a device ABOVE the thermal noise.
Technology has advanced during last 50 years. Microwave masers achieved low noise figures by cooling while new MESFET devices achieve the same low NF now at ambient temperature.

I recommend the Noise Figure Primer by Agilent for the basics. Experimental observations sometimes differ from predictions. Do make experiments and find your solutions. Modeling and simulations often fail the reality.

 

[abhijitrc]

The noise behavior of microwave active elements has been analyzed by many researchers.
In general, cascaded elements like in a receiver are described by FRiis' formula.
Individual components like low-noise amplifiers have also been analyzed in many papers but still there are people who fail to understand the basics.

Thermal noise, P=kTB, only depends on ambient temperature and bandwidth. For frequencies up to almost one terahertz, the power spectrum is considered flat.

Passive elements contribute by their loss (loss equals NF) if they are purely resistive and matched to line impedance. Reactive components do not generate noise but affect the frequency transmission.

Active elements generate inherent noise which usually exceeds thermal noise. Such noise contributions are described by their noise figure. NF indicates as a ratio the noise contribution of a device ABOVE the thermal noise.
Technology has advanced during last 50 years. Microwave masers achieved low noise figures by cooling while new MESFET devices achieve the same low NF now at ambient temperature.

I recommend the Noise Figure Primer by Agilent for the basics. Experimental observations sometimes differ from predictions. Do make experiments and find your solutions. Modeling and simulations often fail the reality.

Sir i do agree with the statement that----Active elements generate inherent noise which usually exceeds thermal noise. Such noise contributions are described by their noise figure. NF indicates as a ratio the noise contribution of a device ABOVE the thermal noise. but i want to know why the NF of the devices increases with frequency , currently we are using infenion make HEMT device CFY67 whose typical NF value at 6 GHz ~ 0.7dB where as at 17 GHz it is near about 115 GHz, !!! What i guess it is due to input impedance of the same device varies with frequecy, is it or something else !!

 

[jiripolivka]

The noise generation in active devices is a complex problem. As you compare a FET over more than one octave, consider only channel resistance and stray capacitance. These alone increase with frequency as you observed.

 

[pstuckey]

Sir i do agree with the statement that----Active elements generate inherent noise which usually exceeds thermal noise. Such noise contributions are described by their noise figure. NF indicates as a ratio the noise contribution of a device ABOVE the thermal noise. but i want to know why the NF of the devices increases with frequency , currently we are using infenion make HEMT device CFY67 whose typical NF value at 6 GHz ~ 0.7dB where as at 17 GHz it is near about 115 GHz, !!! What i guess it is due to input impedance of the same device varies with frequecy, is it or something else !!

In general, Noise Figure degrades the signal to noise ratio. So a component could have an increasing noise figure by either increasing the noise level or by decreasing the signal (if noise level remains constant).

For instance a passive 10 dB pad has a 10 dB noise figure, but it does not create noise that is 10 dB above the thermal noise floor. It attenuates a signal by 10 dB, while the thermal noise remains constant (assuming room temperature). This degrades the SNR by 10 dB and therefore the 10 dB Noise Figure. It's likely that your device gain is rolling off considerably at higher frequencies, and you're seeing much less signal, but still the same thermal noise level at the output.

- - - Updated - - -

Sir i do agree with the statement that----Active elements generate inherent noise which usually exceeds thermal noise. Such noise contributions are described by their noise figure. NF indicates as a ratio the noise contribution of a device ABOVE the thermal noise. but i want to know why the NF of the devices increases with frequency , currently we are using infenion make HEMT device CFY67 whose typical NF value at 6 GHz ~ 0.7dB where as at 17 GHz it is near about 115 GHz, !!! What i guess it is due to input impedance of the same device varies with frequecy, is it or something else !!

In general, Noise Figure degrades the signal to noise ratio. So a component could have an increasing noise figure by either increasing the noise level or by decreasing the signal (if noise level remains constant).

For instance a passive 10 dB pad has a 10 dB noise figure, but it does not create noise that is 10 dB above the thermal noise floor. It attenuates a signal by 10 dB, while the thermal noise remains constant (assuming room temperature). This degrades the SNR by 10 dB and therefore the 10 dB Noise Figure. It's likely that your device gain is rolling off considerably at higher frequencies, and you're seeing much less signal, but still the same thermal noise level at the output.

 

[jiripolivka]

In general, Noise Figure degrades the signal to noise ratio. So a component could have an increasing noise figure by either increasing the noise level or by decreasing the signal (if noise level remains constant).

For instance a passive 10 dB pad has a 10 dB noise figure, but it does not create noise that is 10 dB above the thermal noise floor. It attenuates a signal by 10 dB, while the thermal noise remains constant (assuming room temperature). This degrades the SNR by 10 dB and therefore the 10 dB Noise Figure. It's likely that your device gain is rolling off considerably at higher frequencies, and you're seeing much less signal, but still the same thermal noise level at the output.

- - - Updated - - -



In general, Noise Figure degrades the signal to noise ratio. So a component could have an increasing noise figure by either increasing the noise level or by decreasing the signal (if noise level remains constant).

For instance a passive 10 dB pad has a 10 dB noise figure, but it does not create noise that is 10 dB above the thermal noise floor. It attenuates a signal by 10 dB, while the thermal noise remains constant (assuming room temperature). This degrades the SNR by 10 dB and therefore the 10 dB Noise Figure. It's likely that your device gain is rolling off considerably at higher frequencies, and you're seeing much less signal, but still the same thermal noise level at the output.

I do not know for what frequency band the CFY67 was designed. If for 6 GHz, it is optimized for this frequency and some bandwidth around it. It is then a mistake to assume its NF will stay low at 17 GHz. Find the specifications and follow them.
For 17 GHz there are other low-noise devices, try Hittite, Triquint, etc. All are specified for certain frequency band, and their NF is adjusted low just there. Below and above the NF grows.