Testing multimeters and measurement errors

A study 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, all of them are connected in parallel: The


Fluke 87-V is a high-quality automatic multimeter capable of calculating the effective (rms) value of “true rms” 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 under-reads a few percent, but this will be repaired over time. In tests, it is used at the “3v” measurement limit.

A visual monitor of the signals fed to the devices will be using a Lecroy 9354TM digital oscilloscope. It is also shaggy's, but it 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:



pkpk - full amplitude swing of the
RMS signal - root mean square
freq - 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 the 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 the readings of the 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 various frequencies We

supply 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 starting from 3 kHz. At the same time, inexpensive multimeters slipped this barrier - unless of course they did not assume that from the very beginning their error was as much as 16% downward. At 20 kHz, their readings cannot even be called evaluative, so only Fluke and the switch are adequate. That passed 50 kHz is still about the case, but it is already meaningless to measure higher frequencies.

Rectangular test current

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 of rectangular 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 at Hertz was a bit overstated, but leveled closer to the point at 3 kHz. Higher frequencies pulled only Fluk and arrow.

A rectangular signal with a constant component

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

Inexpensive multimeters have shown themselves very effectively, for them the frequency barrier seems to have lost relevance. While normal devices to the last try to work the brain with a processor to squeeze something adequate - simple 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

send 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 for 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 a frequency of 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,