The LM324 is the default choice when a design needs multiple general-purpose op amps on a single supply and cost matters more than precision. Four amplifiers, one 14-pin package, working from 3 V to 32 V without a negative rail, at one of the lowest prices per amplifier in the industry. That is the entire value proposition — and it comes with specific limits that the datasheet states but designers keep rediscovering the hard way, audio being the most common casualty.
This guide covers the pinout, the datasheet numbers that actually drive design decisions, the internal architecture that explains those numbers, the LM124/LM224/LM2902 family structure, and cross-manufacturer equivalents.
Table of Contents
What Is the LM324?
The LM324 is a quad operational amplifier: four independent, internally compensated op amps sharing one power supply pin pair. It is the four-channel sibling of the dual LM358 — same amplifier core, same electrical behavior, different channel count and package. Both descend from the LM124 that National Semiconductor introduced in the 1970s; Texas Instruments acquired National in 2011, which is why TI now hosts the original part, while onsemi, STMicroelectronics, and several Asian manufacturers produce compatible versions.
Two design decisions define the part. First, the input stage works down to the negative rail: with a single supply, it can amplify signals referenced directly to ground. Second, the whole device idles at almost no quiescent current — roughly 700 µA total for all four amplifiers, thanks to a minimally biased output stage — which is why the part survives in battery and always-on designs fifty years after introduction, and also why it distorts audio. Both decisions come straight out of the internal architecture, covered below.
LM324 Pinout
The four amplifiers are arranged symmetrically around the supply pins: V+ on pin 4, GND (or V−) on pin 11. The pinout is identical across DIP-14, SOIC-14, and TSSOP-14 packages.
| Pin | Function | Pin | Function |
|---|---|---|---|
| 1 | Output 1 | 8 | Output 3 |
| 2 | Inverting input 1 | 9 | Inverting input 3 |
| 3 | Non-inverting input 1 | 10 | Non-inverting input 3 |
| 4 | V+ | 11 | GND / V− |
| 5 | Non-inverting input 2 | 12 | Non-inverting input 4 |
| 6 | Inverting input 2 | 13 | Inverting input 4 |
| 7 | Output 2 | 14 | Output 4 |
Routing note: amplifiers 1–2 sit on the left half, 3–4 on the right, mirrored. Keeping high-impedance input traces away from adjacent outputs matters more here than on a dual package — crosstalk between channels is a layout problem, not a silicon problem, but the quad package puts more opportunities for it on the board.
Unused channels should not be left floating. Wire each spare amplifier as a follower with the non-inverting input tied to a mid-supply divider or ground (within the input range), otherwise the open inputs can drift and oscillate, wasting current and coupling noise into active channels.
LM324 Datasheet and Key Specifications
Numbers below are typical values from the TI datasheet for the standard LM324 at 25 °C; verify limits against the specific manufacturer and grade you are buying.
Supply range: 3 V to 32 V single supply (±1.5 V to ±16 V dual). The wide range is real, but the part is most at home between 5 V and 24 V industrial rails.
Input common-mode range includes ground. Inputs work from 0 V (in fact to about 0.3 V below ground) up to roughly 1.5–2 V below V+ — TI specifies V+ − 1.5 V, onsemi V+ − 1.7 V; design to the tighter figure when the source is unknown. Signals referenced to ground need no level shifting — the core reason to pick this part on a single supply.
Output swing does not reach the positive rail. The output drives to within about 1.5 V of V+ under load, and down to near ground when sinking is light. On a 5 V supply the usable output window is roughly 0 V to 3.5 V. This is the spec that quietly breaks designs migrated from 12 V to 5 V or 3.3 V: headroom that was irrelevant at 12 V becomes a hard ceiling at 3.3 V. For rail-to-rail output on low supplies, a CMOS replacement (see equivalents below) is the correct move, not a workaround.
Gain-bandwidth product: ~1 MHz (1.2 MHz on the newer B grade). Slew rate: ~0.5 V/µs. These are 1970s numbers. At a gain of 10 the usable bandwidth is on the order of 100 kHz; large-signal response is slew-limited well before that. Sensor conditioning, thermostats, current-sense amplification, and comparator duty are all comfortably inside the envelope; anything past low-frequency audio is not.
Input offset voltage: 2 mV typical, 7 mV maximum (standard grade). With a gain of 100, worst-case offset alone produces up to 0.7 V of output error. Precision work needs the A-grade, the modern LM324B, or a genuinely precise part — offset trimming around a standard LM324 costs more than buying a better amplifier.
For absolute maximum ratings, temperature curves, and package dimensions, the TI LM324 datasheet covers the full LMx24 family in one document.
LM324 Internal Architecture: How It Works
Every specification above traces back to two circuit decisions inside the die. Understanding them replaces a page of datasheet memorization.
The input stage is a PNP differential pair. Each amplifier’s inputs feed emitter-coupled PNP Darlington transistors. A PNP input conducts correctly when its base sits at or slightly below the emitter — which is what allows the common-mode range to include the negative rail and extend about 0.3 V below it. The price is paid at the other end: the PNP pair needs headroom below V+, which is where the input ceiling of roughly 1.5–2 V under V+ comes from. The same architecture appears in the LM358 and the LM2902; it is the family’s defining feature, not an LM324 quirk.
The output stage is class-B with a small current sink. The output is driven by an NPN Darlington that sources current toward V+ and a sink transistor that pulls toward ground, with only a ~50 µA internal current sink bridging them instead of a proper class-AB bias network. This is why quiescent current is measured in microamps per channel — and why there is a dead zone at the moment drive hands over from the sourcing transistor to the sinking one. Feedback corrects the dead zone at DC and low frequency; at audio frequency, the 1 MHz GBW is not enough loop gain to hide it.
Single supply capability, microamp idle current, ground-sensing inputs, no positive-rail swing, crossover distortion: one input-stage decision and one output-stage decision produce the entire profile. The LM324 is not an old part that happens to have limitations — it is a coherent set of trade-offs that traded speed and swing for supply flexibility and power, and it has held that niche for fifty years because the trade is still worth making in exactly those applications.
Is the LM324 Good for Audio?
No — not in the signal path, not without understanding exactly what you are trading.
The class-B output stage described above hesitates every time the signal crosses the source-to-sink handover point: crossover distortion. On a scope it appears as a step or kink at each crossing; in a speaker it is clearly audible grit on low-level signals, and negative feedback at audio frequencies is too slow (0.5 V/µs) to fully correct it.
The standard mitigation is to force the output stage into class-A operation: add a pull-down resistor from the output to ground (single supply) or to V−, sized so the output continuously sources a milliamp or two through it, keeping the sourcing transistor always active and moving the dead zone out of the signal path. It works for headphone-level and small-signal audio. It also burns the current the part was chosen to save — at which point a TDA2030 for power audio or an NE5532 for line-level audio is usually the better answer. The LM324’s legitimate audio roles are non-signal-path: microphone bias, VU-meter drivers, tone-detection comparators.
LM324 vs LM358
The LM358 is electrically the same amplifier — same PNP input stage, same class-B output, same specifications per channel. The differences are packaging and arithmetic:
| LM324 | LM358 | |
|---|---|---|
| Amplifiers | 4 | 2 |
| Package | 14-pin | 8-pin |
| Cost per amplifier | Lower | Higher |
| Power dissipation capacity | Higher (larger package) | Lower |
| Board area | Larger | Smaller |
Choosing between them is a channel-count and board-area decision, not a performance one. Three or four amplifiers in the design: LM324. One or two: LM358 saves board space. There is no configuration where one outperforms the other electrically — a point worth knowing because both parts attract “which is better” debate that the schematics do not support. Full treatment of the dual part in our LM358 design guide.
LM124 vs LM224 vs LM324 vs LM2902
The four part numbers are one die at four temperature grades — the original National Semiconductor grading system, still in effect:
| Part | Grade | Operating range |
|---|---|---|
| LM124 | Military | −55 °C to +125 °C |
| LM224 | Industrial | −25 °C to +85 °C |
| LM324 | Commercial | 0 °C to +70 °C |
| LM2902 | Automotive | −40 °C to +85 °C (LM2902V: −40 °C to +125 °C) |
Identical function, different guaranteed temperature window, different price. Specify the cheapest grade whose window covers the application with margin. Two sourcing implications. First, the LM2902 frequently costs about the same as the LM324 while carrying the wider range — for outdoor equipment, specify it instead. Second, TI’s newer LM324B / LM2902B generation offers tighter offset (3 mV max), improved EMI immunity, and higher output drive, at a compatible price. New designs that would default to the standard part should default to the B version; verify current availability per package before committing a BOM.
The dual-channel equivalents follow the same ladder: LM158 (military), LM258 (industrial), LM358 (commercial), LM2904 (automotive).
Package suffixes answer the other recurring comparison question — LM324N vs LM324D and similar pairs are the same silicon in different packages. On TI parts, N is the 14-pin plastic DIP, D/DR is SOIC-14, PW is TSSOP-14. Suffix conventions are per-manufacturer, not universal: the same letter can mean different packages at TI, onsemi, and ST, so resolve any suffix against that manufacturer’s own ordering guide rather than assuming. In procurement this matters twice — once for footprint, once at incoming inspection, because the four temperature grades are visually near-identical and remarked counterfeits exploit exactly this.
LM324 Example Circuits
Single-supply sensor front-ends. Thermistor bridges, photodiode transimpedance stages at modest bandwidth, load-cell pre-amplification before an ADC. Ground-sensing inputs mean the sensor and the amplifier share one rail with no bias network.
Four-channel comparator banks. Window comparators, battery-voltage monitoring across multiple thresholds, over-temperature flags — one package handles four thresholds. Two caveats: the LM324 has no hysteresis unless you add positive feedback, and its recovery from output saturation is slow. For hard comparator duty at speed, the LM339 — the quad comparator designed alongside this family, with open-collector outputs — is the correct part; the LM324 comparator is a convenience when the fourth amplifier is already free.
Active filters below ~50 kHz. Sallen-Key and multiple-feedback topologies for sensor anti-aliasing and mains-frequency noise rejection. Four amplifiers cover a fourth-order filter plus buffering in one package.
Current sensing on the low side. Amplifying the drop across a shunt referenced to ground is exactly the input range this part was built for.
Common Problems With the LM324
Nearly every field problem with this part maps to one of five causes:
Floating unused channels. The most common quad-specific mistake. Open inputs drift, the amplifier oscillates or rails, supply current rises, and noise couples into the working channels. Fix: wire every unused amplifier as a grounded-reference follower.
Output stuck below the positive rail. Not a defect — the V+ − 1.5 V ceiling from the architecture section. Shows up as “the op amp won’t output 5 V from a 5 V supply.” It never will; redesign the signal range or move to a rail-to-rail part.
Crossover distortion on AC signals. Covered above; the pull-down resistor is the fix, a different amplifier is the better fix.
Oscillation with capacitive loads. Driving cable runs or MOSFET gates directly makes the output stage ring or oscillate. An isolation resistor (47–100 Ω) in series with the output solves most cases.
Offset-dominated error at high gain. 7 mV worst-case offset times the closed-loop gain appears at the output as a DC error that looks like a drifting sensor. At gains above ~50, budget the offset explicitly or specify the A/B grade.
LM324 Equivalents and Replacements
The LM324 is a multi-source commodity, which is precisely the information the original manufacturer’s page will not give you. The main production sources:
- Texas Instruments — LM324, LM324A, LM324B; the reference datasheet.
- onsemi — LM324, LM324A; long automotive supply history.
- STMicroelectronics — LM324 in industrial-focused packages.
- SG Micro — SGM324, a pin-compatible domestic Chinese version increasingly common in cost-driven consumer BOMs.
For drop-in replacement in an existing design, any of these cross directly at the same grade; datasheet limits differ at the margin (offset, output drive), so re-verify only the parameters your design actually stresses.
For new designs on 3.3 V or 5 V rails, the pin-compatible CMOS upgrades solve both classic limitations at once — rail-to-rail output and no crossover dead zone: TI’s TLV2374, Microchip’s MCP6004, and equivalents from multiple Asian manufacturers. They cost more per unit and give up the 32 V rating; below 6 V they are better parts in every respect that matters.
Availability across this family is broad and stable — it is one of the most stocked op amps in distribution. If you are sourcing LM324 variants, automotive LM2902 grades, or the CMOS replacements in production quantity, request a quote and we will confirm current stock, manufacturer options, and date codes against your BOM.