Abstract

As we know some of the components' performance will change with temperature: e.g. amplifier. Also from the datasheet, IF switch and digital attenuator we used also has this temperature dependency. That could be another source of 1/f noise.

This table just summarize effect of the thermal noise to the IF system and the requirement for temperature variance of each components.

Why

See how the temperature change can results in 1/f noise.

How

Step1: For the amplifier, dGain/dTemp is calculated from the plot on datasheet. and assume linear performance across the whole temperature range. Here is detailed data reference: MKID_IF_thermal_analysis_datasheet.pdf

Step2: Once we know dGain/dTemp, we can calculated the resulted SNR for that component by this equation: calculate_SNR_from_dG_dT.pdf

Step3: here is summaried results for temperature variance requirement for amplifiers used in the IF system: MKID_Receiver_Chain_Temp_Fluctuation_Requirement.pdf

•    (reference for this calculation: MKID_Receiver_Chain_Suggested_Design_3rd_Aug_Add_Temp_Fluctuation.xls)

Other things need to note about this calculation:

* One thing to note is this gain fluctuation from temperature change apply to both signal and noise. It does not actually reduce the SNR, but it can introduce slow 1/f noise with temperature chagne.

* If we know d_Temp / d_time , that can help figure out where 1/f frequency going to be: e.g. If we can control the temperature within 0.1 C, and if dTemp/dt = 0.0017 C/second (0.1 degree C per min), Then we would see a 1/f at 0.0042Hz with around 12 dB higher than the white noise floor

* Unlike the VCO noise or amplifier used for mixer LO(those noise apply to both transmitting and receiving side, it is possible they cancel each other), some IF amplifier only used in the receiver chain. Its 1/f noise might not be canceled.