17
\$\begingroup\$

When selecting a multimeter, what should one look for in terms of safety features and measurement capabilities?

What can you look for in the specifications to tell a good meter from a crappy one (besides price)?

I'm looking at cheaper multimeters for hobby use - so I don't need super high accuracy, and am not planning to measure any higher voltages than normal household power. But I want something that's a little better than the $10 one I have from Canadian Tire.

I'm not looking for specific product recommendations, just what to look for in choosing a multimeter.

\$\endgroup\$
7
  • \$\begingroup\$ Question is too vague. \$\endgroup\$ Aug 24, 2012 at 13:02
  • \$\begingroup\$ Tried to make it a little less vague. \$\endgroup\$
    – Grant
    Aug 24, 2012 at 13:12
  • 2
    \$\begingroup\$ If you don't what you need from a multimeter probably any one would do. \$\endgroup\$ Aug 24, 2012 at 13:40
  • 6
    \$\begingroup\$ @Bruno - Me no agree. It's not because you don't know about all the features that they can't be relevant. When you tell that repair engineer with his manual-ranging meter that they now also have autoranging meters, and explain what they do, he'll buy one on the spot. \$\endgroup\$
    – stevenvh
    Aug 24, 2012 at 14:49
  • 1
    \$\begingroup\$ When I got my first digital multimeter (they were pretty new then) I made sure to get one that measured to 0.01 ohm, since I was working with motors at the time. If you have specific areas of interest, read up on what types of measurements you're liable to want to do, and make sure to get a meter that will handle that. \$\endgroup\$ Aug 25, 2012 at 3:16

4 Answers 4

15
\$\begingroup\$

Accuracy. Which is something else completely than resolution. Your meter may have 4 digits, that's a 0.1 % resolution, but if its accuracy is only 1 % that last digit is useless.

Accuracy is given by two numbers, an absolute error and a relative error. The relative error is the one expressed in %, like 0.5 %. The absolute error is expressed in digits, like 2 digits. If you have a 0.5 % meter, +/- 2 digits, that means that a reading of "100.0" may as well be (100.0 + 0.2) * 1.005 = 100.7. Engineers fresh from uni often neglect or underestimate measurement error due to the number of digits the meter gives them.

The absolute error becomes less important when the reading gets larger, like for a 900.0 reading 2 digits are relatively less (0.022 %) than for a 100.0 reading (0.2 %).

RMS. If you need to measure non-sinusoidal waveforms you'll need that. Non-RMS meters assume your waveform is a sine, and will only produce correct results if it actually is.

Autoranging. You don't want to put your probe aside all the time to turn the knob.

USB interface. May sound as luxurious, but can be handy to log a whole series of measurements in the computer.

\$\endgroup\$
5
  • 3
    \$\begingroup\$ "that last digit is useless"...Not strictly true. If you're comparing voltage at two nodes, and you want to know which one's higher (but don't care so much the exact values) then the last digit can be useful. \$\endgroup\$
    – The Photon
    Aug 24, 2012 at 15:16
  • 2
    \$\begingroup\$ @ThePhoton - I guess that depends on what causes the error. If it's component tolerance, then I agree; then the error won't vary between measurements. If it would be caused by, let's say, noise, I'm not so sure. \$\endgroup\$
    – stevenvh
    Aug 24, 2012 at 15:20
  • 1
    \$\begingroup\$ Exactly. If the accuracy limitation is because of noise, you're right...the last digit is basically useless (but you should also see the display jumping around). If it's a systematic error, there's still some value to having the extra precision. \$\endgroup\$
    – The Photon
    Aug 24, 2012 at 15:32
  • \$\begingroup\$ In addition to stehenvh's answer you might want to check EEVblog #75 – Digital Multimeter Buying Guide (52min. running time). \$\endgroup\$
    – suha
    Aug 25, 2012 at 20:11
  • \$\begingroup\$ How to test accuracy of a multimeter once you get it (this can help you determine if the battery needs to be replaced; the accuracy drops when the battery level does): electronics.stackexchange.com/questions/78768/… \$\endgroup\$ Aug 12, 2013 at 22:53
6
\$\begingroup\$

Good things to look for if you may be tempted to poke at light switches, wall outlets power supplies etc or anything over 50V or that might be near something > 50V once in a blue moon.

  1. Safety - HRC fuses, MOVs, creepage/clearance distances, overlapping case halves.
  2. Separate sockets for A (uA/mA, A) and V - Safety.
  3. Jack alert - don't want to measure 240V using the A socket. Safety.
  4. Flexible test leads/probes with marked credible safety rating.

Good things to look for in general

  1. Clear display with good contrast - $1000 meter no good if can't read it.
  2. Stable stand - see above.
  3. Autoranging
  4. Accuracy - not as important as you might think but 0.5% better than 1%
  5. Microamps range - One day you might be curious about base current on a BJT.
  6. Touch-Hold (not Data Hold) - press button, look at DUT, connect probes, hear beep.
  7. Fast, latched continuity buzzer.
  8. True RMS - if you need accurate measurements of non-sinusoidal AC.
  9. Bar graph.

Things that may be unimportant

  1. Capacitance range - few measure down to 1 pF caps. Buy a $10 cap meter kit.
  2. Diode test - don't think you can test LEDs with it.

Bad things to look for and avoid

  1. Transistor test - invariably a sign of a cheap unsafe meter.
  2. Glass fuses (or no fuses)
\$\endgroup\$
3
  • \$\begingroup\$ Why transistor test is a sign of cheap DMM? I think it should be quite useful \$\endgroup\$ Nov 20, 2016 at 8:43
  • \$\begingroup\$ @Zeta.Investigator. 1) You only find transistor testers on the very cheapest of mutltimeters, more capable meters just never have them. 2) unlike a decent component analyser, you have to know in advance if what you are testing is a FET or a BJP or something else, you have to know if it is PNP or NPN, you have to know the exact pinout. 3) the only thing it tells you is hFE, which isn't always enough. 4) You can use the resistance range to check a BJT anyway. \$\endgroup\$ Nov 20, 2016 at 16:38
  • \$\begingroup\$ Thanks a lot. Can you check the chat. I've also asked a few questions there \$\endgroup\$ Nov 20, 2016 at 16:41
2
\$\begingroup\$

Autoranging, as stevenvh said, is very useful.

I always miss the ability to measure unusual things like capacitance when I don't have it. Some meters also have temperature probes, or ability to measure inductance.

\$\endgroup\$
1
  • \$\begingroup\$ Auto-ranging is often helpful, but being able to set a range manually can also be helpful. Most auto-ranging meters take a moment to change ranges; if a signal is usually zero volts but periodically jumps up to ten-ish volts for a half second, an auto-ranging-only meter may be unable to get a find the proper range and get a good reading before the voltage changes. Being able to lock the meter on the 0-19.99 scale may be very helpful in such cases. \$\endgroup\$
    – supercat
    Dec 31, 2014 at 5:35
2
\$\begingroup\$

Frequency & duty-cycle. Useful for sanity-checks on signals at the interface of a u-controller, timing software loops (complement an output bit on every loop execution), timing a function (raise & lower an output on entry & exit). I probably use this one function as much as all of the rest combined.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.