Friday, November 11, 2011

Comprehensive Look at Sources of Error in DC Voltage Measurements Part 2

This is part 2 of a 2 part post that takes a comprehensive look at all of the factors that can lead to errors in a DC voltage measurement with a DMM and how to eliminate them so you can achieve the highest accuracy possible in your measurement. In part 2 we will cover the following topics: loading errors,  power-line noise, injected current noise, and ground loop errors. If you are a seasoned DMM measurement veteran and you feel I missed something in the following sections please add it as a comment.

Loading Errors Due to Input Resistance — Measurement loading errors occur when the resistance of the DUT is an appreciable percentage of the DMM’s own input resistance. The figure below shows this error source. To reduce the effects of loading errors, and to minimize noise pickup, see if your DMM allows you to set its input resistance to a higher value. For instance, Agilent 34401A’s input resistance can be set from 10 M to > 10 G for the 100 mVdc, 1 Vdc, and 10 Vdc ranges.

Ri should be much larger than Rs or loading error will be a factor in the measurement
Power-Line Noise — This type of noise is caused by the powerline voltage signal (50 Hz or 60 Hz) being coupled onto the measurement setup either from the DUT, the DMM, or both. This noise appears as an AC ripple summed on top of the DC level you are measuring. To eliminate this common noise source DMM designers use integrating or averaging measurement time settings that are integer multiples of the powerline noise's period. Remember if you integrate over a sine wave you get zero. This is typically called normal mode rejection or NMR. If you set the integration time to an integer value of the powerline cycles (PLCs) of the spurious input, these errors (and their harmonics) will  average out to approximately zero. For instance, the Agilent 34401A provides three integration times to reject power-line frequency noise (and power-line frequency harmonics). When you apply power to the DMM, it measures the power-line frequency (50 Hz or 60 Hz), and then determines the proper integration time. The table below shows the noise rejection achieved with various configurations. For better resolution and increased noise rejection, select a longer integration time.

Noise Caused by Injected Current — Residual capacitances in the DMM’s power transformer cause small currents to flow from the LO terminal to earth ground. The frequency of the injected current is the power line frequency or possibly harmonics of the power line frequency. The injected current is dependent upon the power line configuration and frequency. With Connection A (see figure below), the injected current flows from the earth connection provided by the circuit to the LO terminal of the DMM, adding no noise to the measurement. However, with Connection B, the injected current flows through the resistor R, thereby adding noise to the measurement. With Connection B, larger values of R will worsen the problem.

The measurement noise caused by injected current can be significantly reduced by setting the integration time of the DMM to 1 power line cycle (PLC) or greater.

Ground Loop Error — I have done two posts on this topic and the links for them are below. The second being the more thorough one.

Understanding Ground Loop Error in Voltage Measurements

If you think I missed anything or if you have a question please leave it in a comment