The LM7805 is a three-terminal linear positive voltage regulator producing a fixed 5 V output at up to 1.0 A, with a guaranteed output band of 4.75–5.25 V over an input range of 7–20 V and load currents from 5 mA to 1 A. It comes in a TO-220 package with internal current limiting, thermal shutdown, and safe-operating-area protection, so it survives output shorts and overloads without any external protection circuitry. Its native habitat is a rectified-mains supply: ripple rejection is specified at 120 Hz — the exact frequency a full-bridge rectifier puts on 60 Hz mains — with a 62 dB minimum guaranteed for inputs of 8–18 V. It is not a low-power part: quiescent current runs up to 8 mA, and being a linear regulator, it burns the difference between input and output voltage as heat. If your design needs a clean, protected, fixed 5 V rail from a roughly 8–20 V source and can afford the thermal budget, this part is about as simple as power design gets.
The numbers below come straight from the part's datasheet; where the datasheet is silent — and it is silent on a few important things — that gap is flagged explicitly rather than papered over.
The Five LM7805 Numbers Engineers Misread
Before wiring anything, internalize these five specs. Each one constrains a design in a way that isn't obvious from "it makes 5 volts."
1. The output is 5 V ±5%, not 5.000 V. The guaranteed output is 4.75–5.25 V across VI = 7–20 V and IO = 5 mA–1 A, tightening to 4.8–5.2 V at a 25 °C junction. If your load needs tighter than ±5%, the LM7805 is the wrong regulator — and even within spec, you must design downstream parts against the 4.75 V worst case, not 5.0 V.
2. The regulation numbers don't include self-heating. The datasheet states load and line regulation are "specified at constant junction temperature" and measured with pulse testing at low duty cycle — heating effects must be accounted for separately. So the headline ≤100 mV load regulation (5 mA–1 A) and ≤100 mV line regulation (7–25 V) figures are isothermal values. Under continuous real load, the output also drifts −0.8 mV/°C; over the full 0 to +125 °C operating range, that's roughly 80 mV of sag from cold to hot junction. Stack that on the −250 mV low-end tolerance and a hot, fully loaded regulator can sit well below 5 V while still being "in spec."
3. The input window is narrower than the absolute maximum suggests. Absolute maximum input is 35 V — never exceed it, even transiently. But the output is only guaranteed for VI = 7–20 V. Between 20 V and 35 V the part won't immediately die, but you're outside the rated operating envelope. The lower bound is set by dropout, typically 2 V: below roughly 7 V in, the output follows the input down and regulation is gone. Practical operating window: 7–20 V, with 35 V as a hard ceiling.
4. There are two line-regulation specs, and which one applies depends on where your input sits. Line regulation is ≤50 mV over VI = 8–12 V but only ≤100 mV over the wider 7–25 V span. The datasheet's own test conditions and ripple-rejection spec (62 dB min at 120 Hz, VI = 8–18 V) reveal the intended source: a rectified transformer supply landing in the 8–12 V range. Keeping your input in that band buys you the tighter spec; a wide-wandering input gets the looser one.
5. Peak current is not a rating. Continuous output is 1.0 A; peak current is 1.6 A typ. Treat 1.6 A as a transient/overload ceiling, never a design point — and note the datasheet conditions even the 1 A: "If adequate heat sinking is provided, they can deliver over 1.0 A output current." The full rated current is contingent on thermal management, not free.
A few supporting numbers worth keeping in mind: output noise is ~42 µV typ over 10 Hz–100 kHz, quiescent current is ≤8 mA (changing ≤0.5 mA with load, ≤1.3 mA with input), and short-circuit current folds back to ~200 mA typ — though that figure is specified only at VI = 35 V, so at your actual input voltage the foldback level will differ.
Wiring Up the LM7805
The part has exactly three external connections — INPUT (pin 1), GND, and OUTPUT (pin 3) — with no enable, standby, or logic pins, so there are no pull-ups or configuration straps to worry about. The basic circuit is nearly trivial:
- Input capacitor Ci = 0.33 µF and output capacitor Co = 0.1 µF. These aren't suggestions — the datasheet's entire Electrical Characteristics table is referenced to a test circuit with C1 = 0.33 µF and C0 = 0.1 µF at IO = 500 mA, VI = 10 V. These are the values the part was characterized with, making them the safest starting point. Place them close to the pins with short ground returns — that placement rule is engineering practice, not stated in this datasheet.
- Bulk capacitance for ripple-heavy inputs. The datasheet's ripple-rejection test circuit uses a 470 µF capacitor with a 120 Hz source — evidence that large bulk storage on the input rail is expected when the input is rectified mains rather than a clean supply.
- Input-side clamping if the source can spike. The 35 V absolute maximum is the one limit with no internal protection behind it. If your input comes from unregulated rectified mains or can see hot-plug transients, add an input clamp — a practice recommendation, since this datasheet specifies no such component.
- No protection circuitry needed externally. Thermal overload, short-circuit, and SOA protection are all internal. A brief output short will not destroy the part.
The part is also usable with external components for adjustable outputs — the description explicitly allows it, and the internal block diagram shows a feedback divider (R19/R20) that external resistors can interact with. However, this datasheet provides no adjustable-output circuit or resistor values, so that path is mentioned but not specified here.
Known limitation: This datasheet publishes no thermal-resistance (θJA/θJC) values, no ESR or capacitor-type requirements, no maximum output capacitance, and no grounding/layout guidance. Heatsink sizing and capacitor-type decisions must be made from general practice and validated on the bench — they cannot be derived from this document.
The Thermal Bill (How much heat will the LM7805 generate?)
This is the part of the design that actually takes work. Being a linear regulator, every volt of headroom times the load current becomes heat:
P ≈ (VIN − 5 V) × ILOAD
At 12 V in and 1 A out, that's about 7 W — real power that the TO-220 tab and whatever it's bolted to must carry. The junction must stay within 0 to +125 °C, and the datasheet's own wording makes heatsinking a precondition, not an option, for full output current. Because this datasheet publishes no thermal-resistance number, you cannot formally compute a heatsink size from it — budget conservatively (a substantial heatsink for anything near 1 A with more than a few volts of headroom) and verify empirically.
There's one counterintuitive flip side: keeping the junction cooler isn't just about shutdown margin, it also keeps the output voltage higher, because of the −0.8 mV/°C drift. Thermal design and output accuracy are the same problem here.
Testing and Verifying the LM7805
Start by replicating the datasheet's own reference point — IO = 500 mA, VI = 10 V, Ci = 0.33 µF, Co = 0.1 µF — so your readings are comparable to the characterization conditions rather than an arbitrary supply voltage.
- Static output sweep. Sweep the load from 5 mA to 1.0 A and confirm the output never falls below 4.75 V (the guaranteed band over VI = 7–20 V). At 25 °C junction and nominal conditions, expect 4.8–5.2 V.
- Line regulation. Sweep the input and confirm ≤100 mV of output shift over 7–25 V, and ≤50 mV if you hold the input to 8–12 V.
- Load regulation. At constant junction temperature, confirm ≤100 mV over 5 mA–1 A, and ≤50 mV over the tighter 250–750 mA band.
- Quiescent current. Measure ground current at 25 °C; it should be ≤8 mA, varying ≤0.5 mA with load and ≤1.3 mA with input. A reading well above 8 mA usually indicates a wiring or layout fault, not a bad part.
- Dropout check. Under full load, lower the input until the output falls out of regulation and confirm your minimum valid input sits around 7 V (5 V out plus the 2 V typ dropout). This validates brownout margin and is not captured by the constant-temperature regulation figures.
- Ripple rejection. Only meaningful at its test condition: inject 120 Hz ripple with the input in 8–18 V and confirm ≥62 dB. At other frequencies or input voltages, the 62 dB figure does not apply — the datasheet's own test uses a 120 Hz source with a 470 µF cap.
- Thermal soak. Run continuous worst-case load long enough to reach steady state, then measure the hot output — remember the regulation specs were pulse-tested at constant junction temperature. Add the −0.8 mV/°C drift for the measured temperature rise and confirm the part stays below thermal shutdown.
- Protection demos. Short the output and confirm it folds back (the level will differ from the 200 mA typ spec, which is valid only at VI = 35 V), survives, and recovers. Also push a transient past 1 A to confirm the 1.6 A typ peak-current capability, and drive it toward thermal shutdown to confirm the protector trips and recovers — proving your heatsink keeps normal operation below the trip point.
Troubleshooting the LM7805
Output is low (below 4.75 V). First check the input: below ~7 V you've left the guaranteed window and the 2 V typ dropout means the output tracks the input down. Then check the load: beyond 1.0 A continuous the output can sag (1.6 A is only a transient peak). If both are in range and the output is still low, check that the part isn't hot — see below.
It runs hot or shuts down intermittently. This is the classic 7805 field failure, and it's the thermal protector doing its job. Compute (VIN − 5 V) × ILOAD honestly — with worst-case input ripple peaks, not the nominal wall voltage — and compare the resulting junction temperature against the 125 °C limit. Cycling shutdown that correlates with warm-up is the classic signature of an undersized heatsink. Since this datasheet gives no θJA, the fix is empirical: bigger heatsink, lower input voltage, or less load current.
Output reads low and sags further as it warms. This isn't a fault — it's the −0.8 mV/°C drift on top of a part whose regulation specs were pulse-tested at constant junction temperature. Measure the output hot under real continuous load; that's the number your load actually sees.
Output shifts as the input wanders. Line regulation is ≤100 mV over 7–25 V but ≤50 mV only over 8–12 V. If you see more than 100 mV of shift, the input has likely left the 7–25 V window or the part is hot.
Output collapses under load and stays collapsed until power cycle. Suspect a real short or overload: the short-circuit protection folds current back (200 mA typ, but that value is specified only at VI = 35 V, so expect a different level at your input). An in-circuit short can hold the rail at the folded-back level rather than 0 V. Unload the output — if the rail recovers, the regulator is healthy and your load is the problem.
Unexpected battery drain while "off." Quiescent current is ≤8 mA regardless of load — a standing ~40 mW at 5 V. That's fine for an always-on mains-powered rail and disqualifying for a battery design; no wiring fix will remove it.
Ripple getting through that you didn't expect. The 62 dB rejection only holds at 120 Hz with the input in 8–18 V. Out-of-band ripple (switching noise, high-frequency garbage) gets much less attenuation. Characterize the ripple at the input pin first before blaming the regulator.
The fastest triage step: reproduce the datasheet's reference point — VI = 10 V, IO = 500 mA, Ci = 0.33 µF, Co = 0.1 µF, 25 °C. If the output measures within 4.8–5.2 V there, the IC is almost certainly fine and your problem lives in the surrounding conditions: input voltage, load, temperature, or layout.
Part page: LM7805.