Johnson noise is parasitic for electrical engineers in signal analysis,but useful for thermal engineers in measuring temperature.Researches measure absolute temperature by Johnson noise thermometer.because it does not require pre-calibration and has no constraint on the physical properties of the temperature sensor.These merits make Johnson-noise thermometry a strong candidate in various applications.However,it is very challenging in accurately measuring the tiny Johnson noise in the exist of other noises.Based on this analysis,passive and active Johnson-noise thermometry are theoretically and experimentally analyzed.To address the shortcomings above,we develop two circuits to measure the Johnson-noise precisely.The first circuit employs a precise preamplifier and second one is based on a homemade analog amplifier circuit.We verify the first circuit by varying the sensor resistance,the bandwidth,and the temperature.The former agrees with the theory to within 0.7%,while the latter is better than0.3%,as compared to separate measurements using thermocouples(K Type),and the measurement time are both 4 seconds.The homemade analog amplifier circuit realizes the replacement of precise amplifier.In addition,an experiment is proposed to extract the Johnson-noise and thus the temperature while the sensor is driven with electrical current.Both time domain and frequency domain analyses are employed to estimate the systematic uncertainties.Using the Fourier transformation of time-domain signals and observed spectral density,it is found that the low frequency noise of the amplifier,such as the flicker noise,is responsible for the uncertainty of the Johnson-noise thermometry at low frequencies.Based on this analysis,the uncertainty could be reduced to 0.1% if replacing the spectral density of the low frequency regime of the precise preamplifier with the spectral density of the high frequency regime.Likewise,the uncertainty of a single output analog circuit could be potentially reduced to 3% and a dual-output analog circuit could be potentially reduced to 0.2%. |