Testing multimeters as well as measurement errors

A study was made of the work of digital multimeters in the mode of an alternating current voltmeter and a switch instrument. In normal and abnormal modes, on currents of various shapes - both of symmetrical polarity, and in the presence of a constant component.

Content of publication:

• Description of the instruments used, and their initial calibration
• Test for sinusoidal current of various frequencies
• Rectangular current test
• DC Rect Current Test
• Test with arbitrary waveforms, incl. pulsed
• Significant conclusion
• Voting

The list of experimental devices, they are all connected in parallel:

')
The Fluke 87-V is a high-quality automatic multimeter capable of calculating the effective (rms) true rms value of measured currents and voltages.
UT-70C - workhorse, dragged everywhere and everywhere. Released by the popular firm Uni-T, also automatic, but not “true rms”.

And the main characters of the study are the low-cost MAS-830L device from Mastech, and the completely rootless DT-832, which are usually filled with buckets for delivery. I rented them from different places to avoid possible glitches of a specific single instance.

A switch-type voltmeter of alternating current B3-10A of Soviet production, released in 1969, also participates in the experiments. This is a good quality device. This copy slightly underestimates by several percent, but it will be repaired over time. In tests, it is used at the “3v” measurement limit.

A visual monitor of the signals supplied to the devices will be using a Lecroy 9354TM digital oscilloscope. He is also shaggy, but still works fine.

Under the waveform of the signal is the statistics of its parameters. The most interesting for this study are those that are highlighted in the photo brightness:



pkpk - full amplitude swing
RMS - RMS value
freq - the frequency of the signal under study, or its pulses

In the average column we observe the average value of the parameter, low and high - min. and max its values ​​within a sample, sigma is the standard deviation. We will use only the data from the column average .

Calibration

We supply on digital multimeters 220 v from the socket. The switch voltmeter is disabled for now. he has not yet received post-purchase prophylaxis.



Also let's calibrate on the constant, including let's see what the gauge shows. Serving 2.5 v from the power supply. The oscilloscope is a bit overpriced - as it turned out compared to fluke.



According to this template, all photos are organized in the future: first, the oscillogram, under it readings of instruments.

Now making sure that the devices are working, we start the tests. Signals are supplied from low-voltage GSS type G3-36A. Of course, he is not a digital camera, but even better is closer to real conditions.

Sinusoidal alternating current of different frequency

We supply a voltage of 2.5 v at frequencies of 30 Hz, 300 Hz, 3 kHz, 20 kHz, 50 kHz, and 150 kHz.

The first, strangely enough, began to merge UT70C from 3 kHz. At the same time, inexpensive multimeters slipped this barrier - unless, of course, not to assume that from the very beginning their error was as much as 16% downward. At 20 kHz, their testimony cannot even be called estimated, so only Fluke and the pointer were adequate. Which passed 50 kHz is still around, but it is already meaningless to measure higher frequencies.

Rectangular current test

This mode, like all further ones, is non-standard for non-“true rms” devices, but we will still conduct research. We supply approximately 2.5 v square voltage at 30 Hz, 3 kHz, 30 kHz, and 100 kHz.

The readings of cheap multimeters have become more adequate at frequencies up to 3 kHz. But the UT70C on Hertz was a bit overstated, but leveled closer to the point at 3 kHz. Higher frequencies pulled only Fluk and arrow.

DC signal with a constant component

Let's see how the instruments behave at them at frequencies of 300 Hz, 3 kHz, 50 kHz, and 200 kHz.

Inexpensive multimeters proved to be very effective, for them the frequency barrier seems to have lost relevance. While normal devices to the last try to work the brain with a processor in order to squeeze something adequate - simple ones up to 200 kHz tritely show the amplitude value of the signal. Now it is clear what the seekers of superunit technologies admire, and why they prefer cheap devices. After all, it is easiest to get a party through them ...

We give signals of complex shape

Which are obtained by distorting the rectangular voltage by coils and capacitors.

On the first signal with the main frequency of 5 kHz - adequate readings only from Fluk and the needle instrument.
Short bipolar pulses normally digest Flück (and of course the oscilloscope too). But cheap devices practically do not see them. UT-70C gives an error of more than half the current value, and the arrow is also considerable.

The third experiment at 30 kHz is a better result than the previous one, but the error is nonetheless noticeable.
In the fourth experiment, a constant current was applied again. Cheap multimeters and this time gave an amplitude value, so even with some excess.

Upon completion of any research, it is necessary to draw a conclusion.

Updated: I will add two comments from readers,