2nd year laboratory script E17 Johnson Noise Department of Physics and Astronomy 2nd Year Laboratory E17 Johnson Noise Scientific aims and objectives To determine an accurate value for the Boltzmann constant, k, by measuring the electrical power generated from thermal fluctuations. • To compare the accuracy of k measured by assessing the power generated at different temperatures and with that measured using different resistances • Learning Outcomes • To appreciate differences in language and methods of original scientific investigations • To be able to use Excel to numerically evaluate an integral from raw data. • Apparatus • • • • • • • • • Safety instructions Unit comprising of amplifier, attenuator, resistor holder, and R.M.S voltmeter Signal generator Oscilloscope Frequency meter Metal-film resistors Liquid nitrogen Dry ice Ice Hotplate September 2010 • • V1.0 Gloves and goggles must be worn when handling liquid nitrogen. Care must also be taken when handling dry ice and hot water Page 1 2nd year laboratory script E17 Johnson Noise Task 1 - Pre-session questions The original scientific paper by Johnson (Phys. Rev. 32 (1928) 97-109) is provided. This paper is the information source with which to answer the pre-lab questions. The questions will help you pick out the key information from the paper and interpret it into a method that you will apply using the modern equipment available in the 2nd year lab. Task 2 – The experiment Boltzmann’s constant can be determined by measuring the voltage produced across a conductor with resistance R at temperature T. However because these voltages are very small they must be amplified and it is the amplification process that adds complications to the measurements. The thermal fluctuations result in voltages with many different frequencies. The average of these different frequencies can be measured using an RMS meter, but the integrated gain of the amplifier must be calibrated by measuring the gain of known frequencies. A 100:1 attenuator allows a 10 mV sine wave (4 - 11 kHz) that is amplified by 105 to be measured on an oscilloscope with 10 V peak-to-peak range. Once the amplifier has been calibrated and its integral gain determined, the effect of temperature and resistance on voltage can be measured. The measurements need to be corrected for the internal noise of the instrument, V02 , by short-circuiting with a copper wire of essentially zero resistance. A variety of metal-film resistors provide a method of changing R, and ice, dry ice, liquid nitrogen and a hot-plate for use with a beaker of water provide a method for changing T. When investigating the effect of T, the temperature stable resistance rods, rather than the metal foil resistors should be used. Two different resistance rods are available. You must also become familiar with the modifier M that is used to correct the impedance of the conductor where the conductor’s resistance is similar to the input resistance of the RMS meter. In general when M < 1.001 a value for k can be determined. When M > 1.001 a value for k will be much less precise, but the capacitance of the meter can be estimated. Task 3 – Reporting You should report your experimental method in your own words showing how you determine the integral gain of the amplifier and your method for determining k. You should report your experimental precision of any value of k you determine. You should also be able to provide values of k, and a comparison and explanation of the differences in precision for k determined using both resistance and temperature methods. You should compare your experimental value with both the accepted value for k and Johnson’s original paper. September 2010 V1.0 Page 2 2nd year laboratory script E17 Johnson Noise Appendix of supporting information 1. 2. 3. 4. 1. Electrical noise due to thermal fluctuations Determination of integral gain Compound errors table Pre-lab questions Electrical noise due to thermal fluctuations The average noise power from a conductor of resistance R is given by Pf = V f2 4R = kT∆f (A1) Where Pf is the fraction of the total power having component frequencies between f and f + ∆f , V f2 is the mean square of the fluctuating potential, k is Boltzmann’s constant, and T is the absolute temperature of the conductor. Thus: V f2 = 4kTR∆f (A2) When a resistor is connected across the input terminals of an amplifier and the amplifier has with a frequency dependant voltage gain A(f), the noise voltage at the amplifier output is given by V 2 = V02 + 4kTR ∫ [A( f )] df ∞ 2 (A3) 0 Additionally, one must also consider the noise contribution from the amplifier itself. Consider the following diagram that illustrates the addition components within the amplifier that contribute to the noise signal. Clearly, Vi = IZ i and V f = I ( R + Z i ) , and thus the input and output voltages (Vi and Vf respectively) are related by September 2010 V1.0 Page 3 2nd year laboratory script Vi = Vf (1 + R E17 Johnson Noise (A4) Zi ) Where the impedance Zi comprises of the parallel combination of the input capacity Ci and the input resistance Ri. Briefly, the voltage across the amplifier Vi is equal to the output voltage Vf modified inversely by a term that depends on the resistance outside the amplifier and the impedance (based upon the internal resistor and capacitor) of the amplifier. Suitable manipulation of this complex modifier leads to the modified version of A3, namely 4kTR ∫ [ A( f )] df ∞ V =V + 2 2 0 2 0 (A5) M Where the modifier M is given by 2 R 2 M = 1 + + (2πfCi R ) Ri (A6) At low resistor values, R (typically below 33 kΩ), the modifier, M, tends to unity and therefore the modified equation A5 simplifies to that given by A3. Note that in the current experimental setup, Ri = 10 MΩ. It is in this low resistance region that the linear relation between voltage and the variable (resistance or temperature) allows a determination of k. The higher resistance regime however allows one to estimate the input capacity Ci. The frequency used to make an estimate of Ci should be that at the peak gain of the amplifier. 5. Integral Gain The integral gain, defined by I = ∫ [A( f )] df ∞ 2 (A7) 0 and can be determined by evaluating the area under an amplitude-frequency curve. Various methods exist for doing this, but the preferred technique available to you is the trapezium method. The curve is approximated by a series of trapeziums of width ∆t. The total area is thus the sum of each trapezium over the whole frequency range. You should use your experimental data points to determine the trapezium widths and then use excel to perform the calculation of area for each trapezium. September 2010 V1.0 Page 4 2nd year laboratory script 3. E17 Johnson Noise Compound errors f x Proportional error ∆f 2∆x = f x ∆f n∆x = f x xn ∆f = nx n −1∆x x+ y (∆f ) 2 = (∆x) 2 + (∆y ) 2 (∆f ) 2 ( ∆x) 2 + (∆y ) 2 = f2 ( x + y) 2 xy (∆f ) 2 = y 2 (∆x) 2 + x 2 (∆y ) 2 (∆f ) 2 ∆x ∆y = + f2 x y x/ y 4. Absolute error ∆f = 2 x∆x 2 (∆f ) 2 = (∆x) 2 x 2 + 4 (∆y ) 2 2 y y 2 2 2 2 (∆f ) 2 ∆x ∆y = + f2 x y Pre Lab questions 1. In the abstract of Johnson’s 1927 Phys Rev. paper he defines an equation relating the noise current in a conductor to temperature and resistance. This is his equation: is it: a. Equivalent to V 2 = V02 + 4kTR ∫ [A( f )] df ∞ 2 0 b. Approximately equivalent to V 2 = V02 + 4kTR ∫ [A( f )] df ∞ 2 0 4kTR ∫ [ A( f )] df ∞ c. Equivalent to V = V + 2 September 2010 2 0 2 0 M V1.0 Page 5 2nd year laboratory script E17 Johnson Noise 4kTR ∫ [ A( f )] df ∞ d. Approximately equivalent to V = V + 2 2 0 2 0 M e. Not equivalent to any of these equations. 2. Can you write down a more precisely equivalent equation to the ones identified above? (If you can identify the approximation that has been made, can you discuss why and how appropriate it might be – you might want the return to this point following the experiment) (page 99 in the paper might help) 3. Figure 3 shows Johnson’s calibration plot for his amplifier – how the gain varies with input frequency. Did Johnson make an appropriate number of measurements to determine the integral gain of the amplifier? How much more precise would the determination have been if Johnson doubled the number of data points? (a) Approximately 10 times (b) Approximately 5 times (c) Approximately the same On page 101 Johnson describes the resistances that he used to determine k. 4. Why is he using these funny materials – why not just use metal film resistors? (a) They were not available in 1927 (b) He wanted to see if the resistor material affected the thermal fluctuations (c) He was just a curious kind of guy. September 2010 V1.0 Page 6