David I. Bergman

Dynamic error correction of a digitizer for time domain metrology

A method for numerical correction of distortion in a digitizer used for metrology applications is described. Investigation of the digitizers error behavior in the phase plane leads to the development of an analytic error model that describes the digitizers distortion behavior. Of particular significance is the models ability to describe nonlinear error in the fundamental spectral component manifested as amplitude and frequency-dependent gain and phase error. When fitted only to the harmonic distortion content of the digitizers output data, the model generates an amount of fundamental that correctly accounts for the error in the digitizers gain that is not due to linear system response. The model is therefore able to improve not just the total harmonic distortion (THD) performance of the digitizer but its ac root mean square measurement accuracy as well. At 1 MHz, the model linearizes the digitizer to 70 /spl mu/V/V over a range of 1 to 8 V and reduces harmonic distortion by >20 dB. It is believed that this is the first time that results of this nature have been reported in the literature.

A low noise latching comparator probe for waveform sampling applications

A new latching comparator probe is described The probe is being developed as part of an effort to augment voltage measurement capability in the 10 Hz to 1 MHz frequency range. The probe offers an input voltage range of 10 V, input impedance of 1 M/spl Omega/ and rms noise referred to the input as low as 55 /spl mu/V. The probes 3 dB bandwidth is approximately 20 MHz. Total harmonic distortion is as low as -93 dB at 50 kHz. Gain flatness is within /spl plusmn/10 /spl mu/V/V from 100 Hz to 100 kHz. Improved step settling performance is achieved using a technique that minimizes circuit thermal errors. The probe Is input range can be extended with a frequency-compensated, 1 M/spl Omega/ input impedance attenuator allowing measurement of pulses in the microsecond regime up to 100 V. The attenuator can be compensated further with a digital filtering algorithm to achieve gain accuracy better than 100 /spl mu/V/V.

Phase-Plane Derived Distortion Modeling of a Fast and Accurate Digitizing Sampler

Continued efforts to model the distortion behavior of custom-designed digitizing samplers for accurate measurement of dynamic signals are reported. This work is part of ongoing efforts at the National Institute of Standards and Technology (NIST) to advance the state of the art in waveform sampling metrology. In this paper, an analytic error model for a sampler having a -3-dB 6-GHz bandwidth is described. The model is derived from examination of the samplers error behavior in the phase plane. The model takes as inputs the per-sample estimates of signal amplitude, first derivative, and second derivative, where the derivatives are with respect to time. The models analytic form consists of polynomials in these terms, which are chosen from consideration of the voltage dependence of the digitizer input capacitance and the previously studied error behavior in a predecessor digitizer. At 1 GHz, an improvement in total harmonic distortion from -32 to -46 dB is obtained when model-generated sample corrections are applied to the waveform. The effect of timebase distortion in the sampling system is also accounted for and corrected. The inclusion of second-derivative dependence in the model is shown to improve the models fit to the measured data by providing fine temporal adjustment of the fitted waveform

Pulse parameter dependence on transition occurrence instant and waveform epoch

The pulse parameters of amplitude and transition duration are dependent on the waveform epoch and transition occurrence instant of the pulse transition in the waveform epoch. The primary explanations for the observed variations are the pulse aberrations and settling behavior of both the pulse generator and the measurement instrument (sampling oscilloscope). Measurement results are included for three pulse generators and two sampling oscilloscopes.

NIST program for traceable power and energy measurements under non-sinusoidal waveform conditions

In response to industry requests to have calibration services for distorted power instrumentation, NIST is developing a new sampling system to provide this service. This development effort is aimed at providing calibration of instruments that will make measurements in accordance with the IEEE trial use standard 1459-2000 on power measurements in distorted and unbalanced conditions. The system will make use of several NIST developed instruments and sensors.

Low thermal error sampling comparator for accurate settling measurements

A new sampling comparator design employing a signal-dependent biasing scheme is described. The dynamic bias significantly reduces signal-induced thermal error in the comparator. The circuit design approach is applicable to comparators intended for use in equivalent-time, successive approximation analog-to-digital conversion where required bandwidths may exceed 1 GHz and digitizing resolution may be as high as 16 bits. The technique is well suited for high accuracy settling measurements where thermal tail error can undermine the achievable settling response of an otherwise high bandwidth sampler. The new comparator design is a logical follow-up to previous work in which front-end bias on/off switching was employed. A prototype circuit has been fabricated in a 1.5 /spl mu/m BiCMOS process. In the prototype device, the technique reduces settling error of 300 ns from 800 /spl mu/V/V from dc to 1 MHz.

NIST Sampling System for Accurate AC Waveform Parameter Measurements

This paper summarizes efforts at the National Institute of Standards and Technology (NIST) to develop a waveform sampling and digitizing system with accuracy comparable to that of an ac-dc thermal transfer standard for ac voltage measurements over the frequency range of 10 Hz to 1 MHz. In the frequency range from 1 kHz to 1 MHz, the samplers gain flatness is better than that available from the best commercial digital multimeter. In ac-ac comparisons referenced to 1 kHz, the system agrees with a NIST-calibrated thermal transfer standard to within 17 muV/V from 20 Hz to 100 kHz for measurements made at both 1 and 0.1 V. The samplers excellent dynamic linearity and flat input impedance are also discussed