Measurement with a Multimeter

• Voltage across, current throughA voltmeter goes in parallel with the thing you are measuring; an ammeter goes in series, which means breaking the circuit. Putting an ammeter in parallel with a supply is how most beginners blow a meter fuse. The distinction follows directly from the definitions of voltage and current.Voltmeter: high impedance, in parallel. Ammeter: low impedance, in series.
• The meter changes the circuit (burden)A voltmeter's 10 M input resistance is in parallel with your circuit — fine on a 10 k node, disastrous on a 10 M one. An ammeter has a shunt resistor, so it drops voltage: its 'burden voltage' can be hundreds of millivolts, which is enough to stop a low-voltage circuit working while you measure it.Measurement error from voltmeter loading ~ R_source / R_meter. Burden voltage = I x R_shunt.
• Reading the accuracy specA meter's accuracy is written as a percentage of reading plus a number of counts (digits). Both terms matter, and the counts term dominates at the low end of a range. A '0.5% + 3 counts' meter reading 1.000 V on a 6.000 V range is accurate to about +/- 8 mV, not +/- 5 mV.Uncertainty = (% of reading x reading) + (counts x resolution of the range).
• True RMS vs average respondingA cheap meter measures AC by rectifying and scaling, which is only correct for a clean sine wave. On a PWM or switching waveform the reading is simply wrong. A true-RMS meter integrates properly, but still has a specified crest factor and bandwidth limit.
• Continuity, diode test and resistance in circuitContinuity beeps below some threshold, typically 10-50 ohm — it is not a short test. Diode mode drives a known current and reads the forward drop, which is how you identify unknown semiconductors and check for blown junctions. Resistance readings taken in-circuit are almost always meaningless because parallel paths exist.
• Safety and CAT ratingsMeter category ratings (CAT I to CAT IV) describe the transient energy the meter can survive at that point in a mains installation. For low-voltage bench work this rarely binds, but you should still know your leads' rating and never probe with two hands on a live circuit.

ProblemYour meter is specified 0.5% of reading + 2 counts, and has 4 digits (a 6000-count meter). On the 6.000 V range it reads 3.281 V on a rail specified 3.3 V +/- 2%. Is the rail in spec?

1. 1
   Range resolution: a 6000-count meter on the 6.000 V range has 1 mV resolution.
2. 2
   Percentage term: 0.5% x 3.281 = 16.4 mV.
3. 3
   Counts term: 2 counts x 1 mV = 2 mV.
4. 4
   Total uncertainty = 16.4 + 2 = 18.4 mV, so the true value lies between 3.263 V and 3.299 V.
5. 5
   The rail's spec window is 3.3 V +/- 2% = 3.234 V to 3.366 V.
6. 6
   The entire uncertainty interval falls inside the spec window, so the rail passes regardless of where in the interval the true value sits.
7. 7
   Now suppose the spec were +/- 0.5% (3.284 to 3.317 V). Your measurement interval (3.263 to 3.299 V) straddles the lower limit, so the meter cannot answer the question. You would need a more accurate instrument or a comparison against a known reference.

AnswerIn spec, with confidence. But the same meter could not verify a +/- 0.5% rail — the measurement uncertainty is larger than the acceptance window. Always compare your instrument's uncertainty to the tolerance you are testing.

Tool: sigrok-cli / PulseView with a supported USB multimeter (or any bench DMM)

• Check the sigrok supported-hardware list for your meter. Many cheap USB/serial DMMs (UNI-T, Brymen, Fluke 187/189, Metex-protocol meters) are supported.
• Install sigrok-cli and PulseView (macOS: brew install sigrok-cli pulseview; Debian: sudo apt install sigrok-cli pulseview).
• List detected devices: sigrok-cli --scan
• Log 60 samples: sigrok-cli --driver <yourdriver>:conn=<port> --samples 60 --output-format csv > rail.csv
• Plot rail.csv in any tool you like. Look at the spread: that is your meter's noise plus the rail's real variation.
• If your meter is not supported, do the same thing manually: take 20 readings over 10 minutes and record them. Compute mean, min and max.
• Now load the rail (switch on the thing it powers) and repeat. The difference between the two datasets is load regulation, measured rather than assumed.

• EEVblog multimeter tutorialsfreevideo~90 minSearch the channel for 'multimeter tutorial' and 'burden voltage'. The burden voltage episode in particular changes how you measure current forever.
• Keysight / Fluke free application notes on DMM accuracyfreereference~45 minLook for the material on reading accuracy specifications and on making low-current measurements. Vendor-neutral principles, free to read.
• sigrok supported hardware and getting startedopen licencetoolGPL toolchain for logic analysers, oscilloscopes and multimeters. Your open-source instrumentation backbone for the rest of the track.
• Your own meter's manualfreedoc~30 minGenuinely required reading: find your meter's accuracy table, its input impedance, its current shunt values and its fuse ratings. You cannot do the project without them.

Characterise your own multimeter, then use it to measure a circuit whose true values you know from calculation. The point is to quantify the gap between your instrument and reality.

Deliverables

• An instrument datasheet summary: your meter's model, its accuracy spec for DC volts, DC amps and ohms on each range, its input impedance, and its current-range shunt resistances (measure them if not published).
• A burden voltage measurement: build a circuit drawing a known current, measure the voltage across the meter's own leads while it is in current mode, and report the burden voltage on at least two current ranges.
• A loading experiment: build two dividers producing the same nominal voltage, one from 1 k resistors and one from 10 M resistors. Measure both. Explain the discrepancy quantitatively using your meter's input impedance.
• A measurement report on a real circuit (the LED panel from the electricity skill is ideal): for every node, the calculated value, the measured value, the meter's uncertainty at that reading, and a pass/fail verdict.
• One case where you can honestly say 'my meter cannot resolve this' and a statement of what instrument you would need.

Acceptance criteria — how you know it is good enough

• Burden voltage is measured, not quoted from a datasheet, on at least two ranges.
• The 10 M divider discrepancy is explained with a calculation using the meter's stated input impedance, and the predicted reading matches the observed one within 10%.
• Every reported measurement has an uncertainty derived from the meter's own accuracy spec, using both the percentage and the counts terms.
• At least one measurement is correctly identified as beyond the meter's resolving power.
• No measurement was taken with the meter in the wrong mode; if you blew a fuse, that is written up as a finding, not hidden.

If you want to push further

• Measure the same 1.5 V battery with your meter and with a second meter. Where they disagree, determine which is closer to correct without a calibrated reference (hint: measure a stack of known references and look for a consistent slope or offset).
• Measure a PWM waveform's 'voltage' with an average-responding meter and compute what the true RMS and true average should be. Explain the error.

• I connect voltmeters in parallel and ammeters in series, every time, without thinking about it.
• I can predict how much a voltmeter will load a high-impedance node.
• I can measure and explain burden voltage and its effect on a current measurement.
• I can convert a meter's accuracy spec into a plus-or-minus figure for a specific reading.
• I know when my instrument is not good enough to answer the question being asked.
• I use diode mode to identify and test semiconductors.