Abstract
Lock-in amplifiers (LIA) are used to measure signals from noisy sources or detectors and for measuring AC-mode devices such as pyroelectric detectors. In infrared radiometry, LIAs are often used in both detector and radiometer calibrations where gain linearity directly impacts the measurement uncertainties. In applications such as BRDF measurements where large signal ratios are used to determine reflectance factors, knowledge of LIA linearities is critical since large signal ratios much be known with very low uncertainties.
At present, most commercial LIAs are limited by their 16-bit or 18-bit amplitude resolution, maximum input voltage of 1 V and nearly 1 % gain nonlinearities. To overcome these limitations, we have built a software-based LIA using a commercial 24-bit, FPGA-based digitizer with maximum input voltages of ±18 V.
The 24-bit digitizer was calibrated for DC voltage to 0.002 % using a voltage source whose outputs are measured using a DCV-calibrated DMM. Using a sine-wave, function generator at 5 Hz and 10.5 Hz with this 24-bit software-based LIA, we calibrated a commercial LIA over 9 gain settings from 1 mV to 1 V. We find that the commercial LIA is nonlinear to about 0.5 % over this range with nonlinearities in gain primarily coming from within-gain nonlinearities. The commercial LIA can be corrected using these measurements such that uncertainties in gain linearities can be reduced by a factor of 50 to 0.03 % (k=2). Even lower uncertainties in gain linearity of 0.004 % (k=2) are possible when the 24-bit digitizer is used directly as a LIA in place of a commercial LIA. We also show how the higher signal resolution using the 24-bit digitizerbased LIA results in resolving blackbody fluctuations which could not be measured using the 16-bit commercial LIA.
Development of a software-based lock-in amplifier with linearity of 0.004 % (k=2) using a 24-bit, FPGA-based digitizer
Lock-in amplifiers (LIA) are used to measure signals from noisy sources or detectors and for measuring AC-mode devices such as pyroelectric detectors. In infrared radiometry, LIAs are often used in both detector and radiometer calibrations where gain linearity directly impacts the measurement uncertainties. In applications such as BRDF measurements where large signal ratios are used to determine reflectance factors, knowledge of LIA linearities is critical since large signal ratios much be known with very low uncertainties.
At present, most commercial LIAs are limited by their 16-bit or 18-bit amplitude resolution, maximum input voltage of 1 V and nearly 1 % gain nonlinearities. To overcome these limitations, we have built a software-based LIA using a commercial 24-bit, FPGA-based digitizer with maximum input voltages of ±18 V.
The 24-bit digitizer was calibrated for DC voltage to 0.002 % using a voltage source whose outputs are measured using a DCV-calibrated DMM. Using a sine-wave, function generator at 5 Hz and 10.5 Hz with this 24-bit software-based LIA, we calibrated a commercial LIA over 9 gain settings from 1 mV to 1 V. We find that the commercial LIA is nonlinear to about 0.5 % over this range with nonlinearities in gain primarily coming from within-gain nonlinearities. The commercial LIA can be corrected using these measurements such that uncertainties in gain linearities can be reduced by a factor of 50 to 0.03 % (k=2). Even lower uncertainties in gain linearity of 0.004 % (k=2) are possible when the 24-bit digitizer is used directly as a LIA in place of a commercial LIA. We also show how the higher signal resolution using the 24-bit digitizerbased LIA results in resolving blackbody fluctuations which could not be measured using the 16-bit commercial LIA.