A limited comparison of five Digital MultiMeters - technical measurements

Discussion in 'Portable and Other Gear Measurements' started by atomicbob, Jun 8, 2020.

  1. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Sep 27, 2015
    Likes Received:
    Dislikes Received:
    Trophy Points:
    On planet
    The closer one looks, the less certain the answer.

    A question arose concerning specifications on a particular DMM, Keysight u1231a, to accurately adjust Eddie Current amplifiers with 50 KHz AC heater voltage. I acquired a Keysight u1231a for evaluation and decided to also include a number of DMMs in my lab in this limited comparison.

    Each of the DMMs evaluated were tested for the following three attributes
    1. True RMS AC Frequency Response
    2. 1000.000 ohm accuracy
    3. 10.000 Vdc accuracy

    True RMS AC Frequency Response
    01 HF_FR_setup_small.jpg
    A Siglent SDG2042X signal generator is adjusted to produce as close to 2.000 Vrms sine output as possible. Then a frequency sweep using 1-2-5 sequence from 0.1 KHz to 200 KHz and a few select higher frequencies to 400 KHz are measured and observed DMM readings recorded. From the readings frequency response plots were generated.

    1000.000 ohm accuracy
    02 34465A_small.jpg
    Keysight 34461a and 34465a have 4-wire (Kelvin) measurement capability. Using the special Pomona 6303 4-wire probes an ESI SR1 1000.000 ohm laboratory reference is measured and DMM observation recorded. Handheld DMMs evaluated do not have 4-wire capability and observations will include a slight error due to measurement probe resistance.
    03 SR1_small.jpg

    Keysight 34461A and 34465A have data logging features allowing the collection of multiple observations over a period of time. Using BenchVue the following 5 minute recording of 34465A observations for the SR1 1K ref resistor demonstrate how device 1/F noise impacts readings when attempting higher measurement precision.
    07 20200608-3 SR1-1K 34465A 5min 100uOhm Div.png
    Averaging 300 readings over 5 minutes results in an estimation of 1000.0013 ohms with the following environmental conditions:
    Temperature: 20.0 °C ambient, 21.4 °C SR1 top surface.
    Relative Humidity: 46%
    Barometer: 1014 hPa

    10.000 Vdc accuracy
    04 SR1_Vref10_small.jpg
    10V Precision DC Voltage Reference, Vref10-r4 shown above

    02 20200601 Vref10-r4-SJH1 10V ref 34465A 30min 2uV Div.png
    Shown above is a 30 minute data log of 34465A observations from the Vref10 10.000Vdc standard. The Vref10 has an accuracy of 0.0025% and temperature coefficient of ±5µV/°F (±0.9ppm/C) typical, ±7µV/°F (±1.3ppm/C) maximum. 34465A 1/F noise will dominate and limit the precision of observations.

    Three handheld DMMs included in this limited evalution:
    05 Handheld_DMMs_small.jpg


    03 20200606 DMM True RMS HF performance + 10V + 1K ref.png
    Visual readings from each DMM display were recorded in the spread sheet shown above. Number of digits past decimal point reflect respective DMM precision capabilities for the 1K and 10Vdc measurements. During the 10V std ref test two observations were recorded on four of the DMMs as their respective displays alternated between the two values.

    04 20200604 DMM true RMS HF perf.png
    1.9 Vrms is approximately -0.5 dB relative to the 2.0 Vrms stimulus, the cutoff I would use, though purely arbitrary. Keysight uses -3 dB for the bench meter specifications and achieves 350 MHz, meeting the specification. Using my arbitrary criteria I would not venture above 300 KHz. Fluke 189 makes good on the 100 KHz spec claimed. Keysight u1231a was not clear about Vrms frequency response specifications. It would appear to be trustworthy to 200 KHz using my criteria, which again is arbitrary.

    The Neotek NT8233D does not claim True RMS response and does not specify AC Vrms frequency. It is included as an example of inexpensive DMM limitations. It appears to be trustworthy to a little over 2 KHz.

    Keysight 34465A and 34461A results were very consistent, as expected for the caliber of instrument they represent. If the Vrms frequency response were plotted for both the data points overlay so tightly as to be indistinguishable on the graph above. Therefore only one was plotted.

    Links to more comprehensive comparisons with additional links to useful information, much from Fluke:
    Last edited: Jun 8, 2020
    Joshvar, Gaspasser, famish99 and 14 others like this.

Share This Page