5. The Digital Multimeter // Electronic Measurements and Instrumentation // bhaswanth

The DMM is the universal tool. Even in side-channel and fault-injection work, you reach for one constantly: voltage rails, power consumption, continuity, fuse checks, sense-resistor verification.

5.1 Resolution: digits and counts

DMM resolution is quoted in two slightly inconsistent ways. Counts is the maximum number on the display: a 6000-count DMM can display up to 5999 (the leading digit is half-digit because it can only show 0 or 1, not 0-9). Digits is a related convention: $3\frac{1}{2}$ -digit means 3 full digits plus a half digit (0 or 1), giving 1999 counts (a typical $19.99$ V or $1.999$ V reading). $4\frac{1}{2}$ digits gives 19999 counts; $5\frac{1}{2}$ , 199999; $6\frac{1}{2}$ , 1999999.

The Fluke 87V is a $4\frac{1}{2}$ -digit / 20000-count handheld. The Keithley DMM7510 is a $7\frac{1}{2}$ -digit / 9999999-count benchtop. Resolution is not the same as accuracy: a high-resolution DMM with a poor reference will show many digits, but only the first few are trustworthy. Always pair a resolution claim with the matching accuracy spec.

5.2 Architecture

plaintext

   Inputs    Front-end                Conversion           Display
                                                                   
   V─o   ┌──[Input attenuator]──┐                                  
   I─o──►│                       ├──►[ADC]──►[μC]──►[LCD]          
   Ω─o   └──[Range relays]───────┘    │                            
                                  (dual-slope                      
                                   integrating                     
                                   or sigma-delta)

The ADC is the heart. Two architectures dominate.

Dual-slope integrating ADC. The input charges a capacitor for a fixed time $T_1$ ; then a known reference of opposite polarity discharges it; a counter measures the discharge time $T_2$ . The unknown voltage is $V_x = V_{ref}(T_2/T_1)$ . Integration over $T_1$ rejects noise: pick $T_1$ to be a multiple of the line period (16.67 ms for 60 Hz, 20 ms for 50 Hz) and you get strong rejection of mains hum (50 dB or more). This is why bench DMMs offer "NPLC" (number of power-line cycles) settings: more NPLCs equal more averaging and more rejection.

Sigma-delta ADC. Massively oversamples a 1-bit comparator, then digitally filters. Used in newer instruments and audio gear because it gives 24+ bits of resolution at moderate sample rates with simple analog hardware.

5.3 Input impedance

Bench DMMs typically have 10 M $\Omega$ on most ranges. A truly high-impedance scope front end ("auto" or ">10 G $\Omega$ " mode) is offered on some bench units; this matters when measuring high-impedance nodes like reference voltages on bias networks. Be aware: low DC ranges of cheap meters can have varying input impedance. Always check the spec.

5.4 Specific products

Fluke 87V. The reference handheld. $4\frac{1}{2}$ digits, true-RMS, 20 kHz AC bandwidth, CAT IV 600 V safety. On every electrician's belt and in every serious lab. About 0.05% basic DC accuracy. Survives drops, dust, water, and questionable line voltage. The "if you only own one DMM" answer.

Fluke 289. Logging DMM with graphical display. Can record values over time, useful for catching intermittents. Same accuracy ballpark as 87V, with extra software.

Keysight 34461A / 34465A / Keithley DMM6500. Benchtop $6\frac{1}{2}$ to $7\frac{1}{2}$ digit DMMs. 0.0035% basic DC accuracy on the 34465A. USB and LAN for automation. Used for production test, calibration labs, characterization work.

Keithley DMM7510. $7\frac{1}{2}$ -digit, 18-bit digitizer mode (1 MHz), graphical touchscreen. The premium hybrid: it's a DMM and a slow scope. Used in solar cell testing, battery characterization, low-current chip work.

For a hardware-security lab, a Fluke 87V plus a 34465A or DMM6500 covers 99% of measurements. The handheld for portable poking, the bench for slow-and-precise.