Milliamp (mA) and amp (A) are two units that describe the same quantity—electric current—yet they sit at opposite ends of the practical scale. Confusing them can quietly wreck a battery, overheat a PCB trace, or cause a sensor to read 1 000× too high.
Choosing the correct unit is not academic pedantry; it is the fastest way to cut power drain by 90 % or to stop a copper trace from fusing like a filament. The following sections move from atomic-level definitions to real-world schematics, giving you a checklist you can apply before the next soldering iron ever reaches temperature.
What Milliamp and Amp Actually Measure
An ampere is 6.24 × 10¹⁸ electrons rushing past a point every second. A milliamp is simply one thousandth of that flow, yet at 1 mA you still have 6.24 × 10¹⁵ electrons per second—more than enough to kill many modern chips if they flow the wrong direction.
Think of current as water through a pipe: an amp is a fire-hose blast, while a milliamp is an eyedropper. Both can fill a bucket, but the time scale and the pipe diameter must match the intended task.
Engineers use milliamps when the number stays below 1 A because it keeps datasheets readable; 0.035 A becomes the cleaner 35 mA, eliminating leading zeros and reducing transcription errors.
Unit Symbol Conventions That Prevent Costly Typos
Always leave a space between the number and the unit: “350 mA” not “350mA.” A missing space once caused a satellite contractor to misread 350mA as 350 mega-amps during a copy-paste, vaporizing a $200 000 test harness.
Capitalization matters: “M” means mega, “m” means milli. A single keystroke shift can turn a 5 mA LED into a 5 MA lightning bolt on paper, and purchasing will happily order a 1 000 000× overrated power supply if you let them.
On schematics, add the unit in the net label itself—“LED_20mA” instead of “LED.” The annotation travels with the trace, so layout engineers never confuse a 20 mA backlight with a 2 A motor line.
Converting Between mA and A Without a Calculator
Move the decimal three places: 2 300 mA becomes 2.3 A left, 0.075 A becomes 75 mA right. Do it aloud once and you will never need a browser tab again.
For mental estimates, remember that 500 mA is half an amp, so a 5 V USB port rated at 2.5 A can supply five 500 mA loads in parallel before sagging.
When you need to add mixed units, convert everything to milliamps first; summing 300 mA, 0.8 A, and 1 200 mA becomes 300 + 800 + 1 200 = 2 300 mA, which is easier to scan for errors.
Spreadsheet Formulas That Catch Unit Drift
In Excel, use =IF(RIGHT(A1,2)=”mA”,VALUE(LEFT(A1,LEN(A1)-2))/1000,A1) to normalize a column that mixes “A” and “mA” text. Conditional formatting turns any cell red if the result exceeds your regulator’s 3 A ceiling.
Google Sheets supports =REGEXEXTRACT() to strip the unit, letting you build a BOM that auto-flags when the total load crosses 80 % of your supply rating. The same sheet can plot heat rise versus current density so you see the trace width bottleneck before copper turns into a fuse.
Why Datasheets Prefer Milliamps for Low-Power Devices
A BLE radio that draws 3 mA in TX looks innocent in milliamps, yet written as 0.003 A the leading zeros invite someone to misread it as 0.03 A. Keeping the value in milliamps aligns with the 1×–999× range human brains parse fastest.
Microcontrollers list deep-sleep current as 0.9 µA, but the next line bumps to 2.1 mA during RAM retention. The three-order-of-magnitude jump is instantly obvious when both lines use milli- or micro-units, preventing the “missing m” mistake that would inflate battery life by 1 000×.
When you compare vendors, normalize everything to coulombs per operation. A 2.1 mA burst for 200 µs costs 0.42 µC, letting you rank processors on an apples-to-apples scale regardless of how each marketing team labels the current.
Color-Coding Current Ranges on Schematics
Establish a palette: < 1 mA = green, 1–50 mA = yellow, 50 mA–1 A = orange, > 1 A = red. Apply the color to the line itself in Altium or KiCad, and the eye catches overloaded nets before the first prototype spin.
Pair the color with a rule that forces trace width calculators to run automatically. An orange 800 mA net on an inner layer must be 0.3 mm minimum; if the autorouter drops to 0.2 mm, a DRC violation halts the release, saving a $500 respin.
Practical Examples: From Coin Cell to Motor Drive
A CR2032 cell rated 220 mAh can source 0.5 mA for 440 h, but at 20 mA its internal resistance tanks the voltage below 2 V in 30 min. Knowing the crossover curve lets you decide when to duty-cycle a sensor rather than oversize the battery.
WS2812B LEDs specify 60 mA at white full, so a strip of 60 pixels wants 3.6 A—already above USB-C 3 A. Designers who forget to convert milliamps to amps often wonder why their laptop port browns out.
A 24 V stepper with 1.2 Ω winding resistance draws 20 A if driven straight from the supply, but a chopper driver limits current to 1 A peak. The difference between seeing “1 000 mA” on the datasheet and “1 A” on the power budget determines whether you spec a 25 W or 250 W supply.
USB-C Power Negotiation in mA Steps
USB-C PD advertises current in 10 mA granularities up to 5 A. A tablet that requests 1.43 A actually asks for 143 units of 10 mA, so firmware engineers must send 0x8F in the PDO packet. Misreading the spec as 1.43 mA yields 0x00, and the charger stays at 5 V 0.5 A, doubling charge time.
Test this with a PD analyzer: set the sink to 1 500 mA, then drop to 1 490 mA and watch the bus voltage rise 40 mV as the cable IR drop shrinks. Ten milliamps matter when the battery charger’s dropout difference is only 150 mV.
PCB Trace Width and Via Current Density
1 oz copper handles 35 mA per mil of trace width at 10 °C rise; therefore a 100 mil outer trace carries 3.5 A safely. Compress that into a 4 mil inner layer and you hit the same temperature at 140 mA—an 8× penalty for going internal.
Via farms are worse: a single 0.3 mm via plates to 1 mΩ, so 1 A dissipates 1 mW, but 10 A burns 100 mW and can reach 100 °C. Use the mA-to-via count chart in IPC-2152 to decide when to add parallel vias instead of widening the trace.
Run a thermal camera on your first article: if a 500 mA LED line glows 20 °C above neighbor traces, you misjudged copper weight. Bump the outer layer to 2 oz or parallel a 0.2 mm trace to drop 10 °C without rerouting the entire board.
Saturation Current of Ferrite Beads
Ferrite beads derate above their DC current rating; a 120 Ω @ 100 MHz bead rated 200 mA may present only 20 Ω at 400 mA. Convert the load profile to milliamps peak, then overlay the bead’s impedance curve to verify EMI suppression still works at worst-case burst.
If your MCU dips 50 mV during a 350 mA LTE transmission, the bead is saturating. Swap to a 1 A part and the ripple collapses to 15 mV, passing radiated emissions without adding capacitors that would consume board area.
Battery Life Math: Coulombs, not Amp-hours
Capacity in mAh is a marketing wrapper; the physics unit is coulombs. A 1 000 mAh pack holds 3 600 C, enough to run 1 mA for 3.6 million seconds—precisely 1 000 h. Converting to coulombs lets you mix sleep, active, and transmission chunks without unit mismatch.
Build a state machine in firmware that budgets micro-coulombs per sensor reading. If a BLE advertisement costs 120 µC and you have 3 600 C left, you can transmit 30 million times—far more than the 10 000 times a simple mAh division would suggest, because it ignores the duty-cycle savings.
Log actual coulombs with a coulomb counter such as the LTC2941; it integrates current in µC directly, removing the mA-to-A conversion that a resistor-plus-ADC approach would require. The result is battery-life prediction within 2 % instead of the typical 20 % guess.
Self-Discharge Normalized to Nanoamps
A Li-ion cell loses 2 % per month at 25 °C; for a 1 Ah cell that is 20 mAh ÷ 30 ÷ 24 ≈ 28 µA self-discharge. Compare that to a 15 µA RTC block and you realize the battery leaks almost twice the current your circuit uses—pivoting the architecture toward a super-capacitor backup.
Current Sensing Resistors: mA Resolution on an A-scale Load
A 10 mΩ sense resistor drops 10 mV per amp, giving 100 mV at 10 A—easy for a 12-bit ADC with a 1.2 V reference. The same resistor yields only 10 µV at 1 mA, buried in op-amp offset. Choose the resistor value last, after you know both extremes.
Place a switchable gain stage: 60 dB for the sub-mA region, 0 dB above 100 mA. A GPIO-controlled analog switch flips the op-amp feedback network, so one resistor covers six decades without sacrificing resolution or burning watts.
Calibrate with a precision current source, not a multimeter. A 0.1 % accurate source set to 1.000 mA lets you quantify the 5 µV offset that would otherwise masquerade as 0.5 mA error—half your sleep budget.
Kelvin Footprints That Eliminate Solder Resistance
Force-sense Kelvin contacts remove the 0.5 mΩ variability of a sloppy solder joint. At 10 A that uncertainty is 5 mV, equal to 500 mA measurement error. Use 0.3 mm sense traces that branch directly from the pad center, keeping them away from the high-current copper pour.
Protection Circuits: Fusing at mA Precision
Traditional 1 A hold fuses blow at 2 A, too coarse for a 200 mA sensor rail. Use a 250 mA PTC fuse with 50 mA trip tolerance, then place a 200 mA current limiter in series; the dual stage prevents nuisance trips during 150 mA inrush yet stops a dead short in 20 ms.
eFuse ICs such as the TPS25942 allow 1 % current limit programming. Set the 20 mΩ sense version to 350 mA for a 5 V rail, and the part shuts down at 351 mA ±2 %. Replace the sense resistor with a 2 mΩ part and you can rescale to 3.5 A for the motor rail using the same footprint, saving BOM variants.
Remember foldback: once an eFuse trips, it clamps current to 30 % of the limit, so a 300 mA rail drops to 90 mA. Verify your LDO can start under that reduced current, or it will latch forever in a brownout condition.
Crowbar vs. Clamp on Overcurrent
A crowbar shorting the supply trips a upstream fuse, dumping zero current into the load but requiring replacement. A clamp limits voltage and lets 50–100 mA leak, enough to keep a MCU alive for diagnostics. Decide based on whether a field trip is acceptable or if telemetry must report the fault before shutdown.
Thermal Runaway: When mA Becomes A in Seconds
A MOSFET with 10 mΩ RDS(on) passing 1 A dissipates 10 mW, trivial at 25 °C. Double the current to 2 A and power quadruples to 40 mW; if thermal resistance is 50 °C/W the junction rises 2 °C, barely measurable.
Now push 5 A—125 mW—and the 6 °C rise increases RDS(on) by 30 % for a typical silicon part. The feedback loop drives current higher until the device latches off or melts, all within milliseconds. Simulate this with a SPICE temperature sweep; the knee occurs around 4 A, warning you to parallel two FETs long before the lab fills with magic smoke.
Apply the same thinking to LEDs: a 20 mA indicator allowed to reach 40 mA doubles junction temperature in 30 s, dropping luminous flux 15 % permanently. A simple thermistor on the same PCB can PWM the LED once die temperature crosses 80 °C, preserving both lifetime and battery.
Trace Fusing Current in Milliamps
A 0.1 mm outer trace in 1 oz copper fuses at ~3 A, but at 300 mA it survives indefinitely. The difference is only one order of magnitude, so never trust “it looks thin”—run the IPC calculator for every power net, then save the result as a fabrication note.
Field Debugging: Measuring Microamps to Amps Without Range Switching
Clamp meters resolve 10 mA on a 40 A scale—useless for a 200 µA sleep current. Instead, use a 1 Ω precision resistor in series with the battery and measure the millivolt drop with a scope: 200 µV equals 200 µA, while 200 mV equals 200 mA, giving you six decades on one 1 mV/div screen.
Zero-drift op-amps such as the OPA388 amplify 100 µV to 10 mV, shifting the signal into a handheld DMM’s sweet spot. Power the op-amp from a coin cell so its 1.9 µA quiescent current does not corrupt the reading, and place the entire fixture inside a shielded Pomona box to kill thermocouple noise.
When you need isolation, use a Bluetooth ADC like the Mooshimeter. It floats, samples at 8 kHz, and logs directly to your phone, letting you watch a 3 A motor spike and the subsequent 50 µA sleep tail on the same graph without touching the range knob.
Current Probes That Auto-Scale
Active current probes with 1 mA to 30 A dynamic range cost less than a single respin. Calibrate with a 1 kHz square wave and a 10 Ω resistor; the flat-top current should read exactly V/R. If the scope shows 1.2 % error, store the offset in probe memory so later field captures remain traceable to NIST standards.