How does a digital multimeter work? Experienced electricians use it but may not know it!

In the more than ten years of working in the marketing of basic measuring instruments, I have met many engineers and discussed test-related technical issues with them countless times. But what is interesting is that engineers are most concerned about some basic issues. After all, their main business is definitely not test and measurement technology. The most common questions engineers ask are about accurate DC and AC measurements. They often experience some confusion, such as what is the measurement error, why the digital meter measurement display is unstable, why the measurement results of different digital meters are very different, and the AC effective value measurement results are unreliable. I will write a series of articles to discuss these issues with you.

In this article, I will take Agilent's 34401A and 34410A, two high-performance digital multimeters, as examples. 34401A was a product of HP in 1993 and is still the world's largest-selling 6.5-digit digital multimeter, with nearly 100,000 units in China. 34410A is the first LXI-compliant digital multimeter.

First, let me introduce the working principle of a high-precision digital multimeter. A 6.5-digit digital multimeter has very high accuracy and resolution. For example, if you measure 5VDC, its resolution can reach 1UV. When reading, we hope that only the last digit will jump. If the second-to-last digit or even the third-to-last digit jumps, that is, only 3 or 4 digits of the 6-digit display are stable, then the 6.5-digit meter will become 5.5 or even 4.5 digits. So what causes the unstable measurement results?

If the input 5VDC bias is stable, the first reason for the large measurement uncertainty is noise. Generally, there are two types of noise, namely series mode noise and common mode noise.

Cross-mode noise is the noise existing in the circuit of the device under test, as shown in the following figure:

The measured signal is input from the front end. In fact, for most instruments, the front end is the most valuable part and the most important indicator to measure the level of instruments from different manufacturers. The signal is conditioned by the front end and converted into a signal amplitude suitable for ADC. The AC RMS in the figure is a dedicated circuit, which is used to calculate the effective value of the input AC signal. In the latest digital meters, such as 34410A, this circuit no longer exists.

Unlike oscilloscopes, high-precision digital meters use dual-integration ADCs. This ADC is characterized by extremely high resolution and superb noise suppression capabilities. It is suitable for high-resolution and high-precision measurements, but the speed is relatively low. For example, a 6-and-a-half-bit digital meter uses a 22-bit ADC, and an 8-and-a-half-bit digital meter uses a 28-bit ADC. The internal working principle of the digital meter is shown in the figure below.

Vi is the measured voltage after front-end conditioning, and Vref is the internal reference power supply. First, the switch (indicated by red) is switched to the Vi terminal, and Vi charges the capacitor in the integrator. The charging time is an integer multiple of the public frequency period, that is, 20ms and its integer multiples, in order to suppress the power frequency noise (as shown in the figure below).
After charging, the voltage on the capacitor is equal to the average value of Vi. At this time, the switch is switched to Vref, and under the control of Vref, the capacitor discharges at a fixed slope. At the same time, the discharge time is recorded by the internal counter. Vi can be calculated using the discharge time and slope. In this process, the process of charging and discharging the capacitor can eliminate high-frequency noise. The suppression method for 50Hz power frequency noise is shown in the figure below:

If the charging time is 20ms (one power frequency cycle, i.e. 1PLC) or its integer multiple, the common frequency noise can be suppressed. Therefore, for high-precision measurement, 20ms is necessary. Of course, if the measurement time is longer, such as 10PLC, a higher noise suppression ratio will be obtained. But this will affect the measurement speed, especially in high-precision data acquisition or automated test systems. Therefore, test speed and accuracy are actually a contradiction. In actual use, a compromise should be considered.

Different digital multimeters have different suppression ratios for common frequency noise in the same measurement time. For example, the traditional 34401A, if a 200ms measurement time is selected, the power frequency suppression ratio is 60dB. For the new 34410A product, the power frequency suppression ratio can reach 120dB in a 40ms measurement time. Some engineers may have problems if they buy some old goods imported from the United States on the second-hand market, because the United States has a 60Hz power frequency cycle. If the

power supply power frequency cycle is unstable, the common frequency noise suppression ratio will also be reduced. The following figure shows the relationship between the power frequency noise suppression ratio of 34410A and the power grid frequency. It can be seen from the figure that if the power frequency period deviates by 1Hz, the power frequency noise suppression ratio will drop by 60dB.

The above discussion focuses on the structure of digital multimeters and the suppression of series noise. It can be seen that in order to ensure the stability and repeatability of the readings, we must consider reducing and suppressing the input noise, reasonably set the measurement time according to the requirements of measurement speed and accuracy, and choose a suitable digital multimeter.