The LM358 is the workhorse dual op-amp - it's been on every engineer's bench for 40+ years. But the original LM358 has real limitations: 0.3 V/µs slew rate, crossover distortion on the output, and it's not rail-to-rail on either input or output. If your circuit is hitting any of these walls, you have options.
The two most common upgrade paths are the LM358B (TI's direct silicon revision — same price, better specs, same footprint) and the TL072 (JFET input, 20× faster slew rate, lower noise). TI positions the LM358B as the drop-in replacement for existing LM358 designs. The TL072 is the right choice for new designs that need speed and don't require single-supply operation down to ground.
But here's the thing: there's no universal "best" replacement. The LM358's superpower is its input common-mode range extending to V− (ground in single-supply circuits). Many alternatives lose that.
Pick the wrong one and your DC-coupled single-supply amplifier stops working at low input voltages. This guide sorts out which one fits your circuit.
These three dual op-amps share the same SOIC-8 / DIP-8 pinout (industry-standard dual op-amp footprint: 1-OUT A, 2-IN A−, 3-IN A+, 4-V−, 5-IN B+, 6-IN B−, 7-OUT B, 8-V+). The LM358B is a silicon revision from TI — same design, better process. The TL072 is a fundamentally different part: JFET inputs, faster, lower noise, but its common-mode input range does NOT include V−.
| Parameter | LM358 | LM358B | TL072 |
|---|---|---|---|
| Manufacturer | TI / ST / onsemi / multi-source | Texas Instruments | TI / ST / multi-source |
| Input Stage | Bipolar PNP | Bipolar PNP | JFET |
| Supply Voltage Range | 3V–32V (single) / ±1.5–±16V | 3V–36V (single) / ±1.5–±18V | 7V–36V (single) / ±3.5–±18V |
| Gain-Bandwidth Product | 0.7 MHz | 1.2 MHz | 3 MHz |
| Slew Rate | 0.3 V/µs | 0.5 V/µs | 13 V/µs |
| Input Offset Voltage (max) | 7 mV (9 mV LM358A) | 3 mV | 5 mV (10 mV "H" grade) |
| Input Bias Current (typ) | 45 nA | 35 nA | 65 pA |
| Quiescent Current (per amp, typ) | 0.5 mA | 0.3 mA | 0.7 mA |
| CMRR (typ) | 85 dB | 95 dB | 100 dB |
| Input Common-Mode Range | V− to (V+)−1.5V | V− to (V+)−2V | (V−)+1.5V to (V+)−1.5V |
| Output Swing (to V−) | ~5 mV (with pull-down) | ~5 mV (with pull-down) | ~1.5V (JFET limit, not rail) |
| Output Short-Circuit Current | 40 mA (typ, sourcing) | 40 mA (typ, sourcing) | 26 mA (JFET limit, lower) |
| Input Voltage Noise (1 kHz) | 40 nV/√Hz | 40 nV/√Hz | 18 nV/√Hz |
| THD+N (1 kHz, G=1) | ~0.02% (crossover notch) | ~0.01% (reduced crossover) | ~0.003% (JFET, linear) |
| Package | SOIC-8, DIP-8, TSSOP-8 | SOIC-8, VSSOP-8, TSSOP-8 | SOIC-8, DIP-8, TSSOP-8 |
| Pinout | Standard dual op-amp (1OUT/1IN−/1IN+/V−/2IN+/2IN−/2OUT/V+) | Standard dual op-amp | Standard dual op-amp |
| Price Reference | $0.04–0.12/unit | $0.06–0.15/unit | $0.08–0.20/unit |
Per the TI LM358B datasheet (Rev D, SLOS068D), the LM358B is a direct silicon revision sharing the identical pinout, application circuits, and output stage topology. Per the TI TL07xH datasheet (Rev K, SLOS080K), the TL072 differs fundamentally: JFET input stage with picoamp bias currents but a common-mode range that sits ~1.5V above V− — incompatible with true ground-referenced single-supply circuits.
The LM358, LM358B, TL072, NE5532, RC4558, and most other dual op-amps share the identical pinout (industry standard dual op-amp). This means physical replacement is a direct solder-in — no PCB change needed for any of the alternatives in the table above. The only layout consideration: the TL072 and other JFET op-amps can oscillate with capacitive loads above ~100 pF. If your PCB has long traces from the op-amp output, add a 100 Ω series resistor at the output pin for isolation. The LM358 is more forgiving here.
| Part Number | Type | Key Difference vs LM358 | When to Consider |
|---|---|---|---|
| LMV358 | Rail-to-rail output, bipolar | RRO: output swings to within 50 mV of both rails. 5.5V max supply. 0.21 mA/ch Iq. | Low-voltage single-supply (3.3V/5V), driving ADC inputs. Not a drop-in above 5.5V. |
| NE5532 | Low-noise bipolar | 5 nV/√Hz noise, 9 V/µs slew, ±3–±20V. Higher bias current (200 nA). Inputs clamped with diodes — differential input limited to ±0.7V. | Audio preamps, active filters. Caution: input clamping diodes break differential circuits where the two inputs see different voltages. |
| RC4558 | General-purpose bipolar | 3 MHz GBW, 1.7 V/µs SR. Essentially an LM358 with higher speed and split-supply bias. Common-mode to V−+2V (not ground). | Middle-ground between LM358 and TL072. Better speed, cheaper than TL072, but loses single-supply-to-ground. |
| TLV9302 | CMOS, rail-to-rail output | 1 MHz GBW, 3 V/µs SR, 150 µV offset (max!), 10 pA bias. 5.5V–40V supply range. Wide supply CMOS. | Precision on a budget. Handles 40V rails, picoamp bias, microvolt offset. Best modern CMOS alternative at the LM358 price point. |
| OPA2202 | Precision bipolar (TI) | 1 MHz GBW, 0.35 V/µs, 200 µV max offset, super-beta inputs. SOIC only, no DIP. | DC precision apps (current sensing, instrumentation). Not fast, but extremely accurate. |
| LT1013 | Precision bipolar | 1 MHz GBW, 0.4 V/µs, 150 µV offset. Different SOIC pinout! — die rotated 90°. | DC precision. Warning: SO-8 pinout is NOT standard dual op-amp. Only DIP-8 version is pin-compatible. |
| MCP6002 | CMOS, RRIO | 1 MHz GBW, 0.6 V/µs SR, 1 pA bias, RRIO. 1.8–6V supply. Low-voltage only. | Battery-powered, low-voltage (3.3V/5V). Pin-compatible. RRIO means full ADC range on 3.3V rails. Cannot handle 12V+ rails. |
| AD8656 | CMOS, low-noise RRIO | 28 MHz GBW, 11 V/µs, 2.7 nV/√Hz. 2.7–5.5V only. $0.80–1.50/unit. | Precision audio / measurement on low-voltage rails. High-speed CMOS. Not for ≥6V rails. |
This is the LM358's most notorious quirk — and the most common topic on StackExchange and EEVblog when engineers troubleshoot LM358 circuits.
The LM358 has a class-B push-pull output stage. When the output current crosses zero (i.e., transitions from sourcing current to sinking current or vice versa), both output transistors briefly turn off simultaneously. The op-amp's output goes high-impedance for roughly 2–3 µs while the loop fights to restore control. The result: a small but measurable distortion glitch at every zero-crossing. At low frequencies (DC, 60 Hz) this is invisible on a scope. At 10–20 kHz, it becomes roughly 1–10% distortion depending on the load — very audible and easily visible on an FFT.
The fix, discovered by Bob Pease decades ago and confirmed on StackExchange threads to this day: add a pull-down resistor from the output to V− (or GND in single-supply). 10 kΩ is the standard starting value.
This forces one output transistor to stay on all the time — the output stage never enters its "off" zone, and the crossover glitch disappears. The cost: an extra 330 µA of quiescent current (3.3V / 10 kΩ). In most circuits, that's a rounding error.
For AC-coupled circuits driving only positive loads (e.g., an ADC input buffer), a pull-down to ground works. For bipolar loads, connect the resistor to V− (the negative supply, or ground in single-supply) — so the output current always flows in one direction. Per the TI LM358 datasheet (Rev K, SLOS068K), output swing to V− improves to within 5 mV with a pull-down resistor sinking at least 50 µA.
The LM358B and modern CMOS alternatives (TLV9302, MCP6002) have improved output stages that reduce — but don't completely eliminate — this issue. Only JFET op-amps (TL072) and true rail-to-rail CMOS output stages are inherently free of crossover distortion under all load conditions.
| Parameter | Details |
|---|---|
| Primary Upgrade | LM358B (SOIC-8 / DIP-8, TI) — direct spec-for-spec upgrade |
| Best JFET Alternative | TL072 (SOIC-8 / DIP-8, TI / ST) — 20× faster, lower noise |
| Best Low-Voltage Alternative | LMV358 (SOIC-8, TI) — rail-to-rail output, 5.5V max |
| Condition | New, original manufacturer packaging, full traceability |
| Lead Time | In stock, ship from Shenzhen |
Contact ICMASS for current pricing on your specific quantity. We stock the full LM358 family (LM358, LM358B, LMV358) and the TL072 in SOIC-8 and DIP-8 packages. Volume pricing for the LM358B typically ranges from $0.06–$0.15/unit depending on quantity. If you need help validating a substitution before committing to production, we can provide sample quantities and engineering cross-reference support.
A: Yes, for the right applications. If your circuit is DC or low-frequency (under 1 kHz), single-supply (5–12V), and cost-sensitive, the LM358 remains a solid choice — it's one of the cheapest, most widely second-sourced op-amps on the planet. For new designs where you'd use an LM358 today, use the LM358B instead — same price, better specs, same footprint. For anything involving audio, high-speed sensors, or precision measurement, start with a TL072 or a modern CMOS op-amp.
A: In split-supply circuits (±5V or higher): almost always yes. The pinout is identical. The TL072 is faster, quieter, and has 700× lower bias current. In single-supply circuits where your input signal needs to reach ground (0V): no. The TL072's JFET inputs need at least 1.5V of headroom above V−. A signal at 0.1V DC will be clipped. For single-supply-to-ground, use an LM358B or a rail-to-rail op-amp instead.
A: Add a 10 kΩ pull-down resistor from the output to V− (or GND in single-supply circuits). This forces one output transistor to stay conducting, preventing the output stage from ever entering the high-impedance "dead zone" where the distortion glitch occurs. Cost: ~330 µA extra quiescent current. Confirmed fix across decades of application notes and StackExchange discussions. If you need inherently distortion-free output, use a TL072 (JFET, class-A output at low currents) or a modern CMOS RRO op-amp.
A: The LM358 can only drive about 50 pF of capacitive load at unity gain before its phase margin collapses. A large capacitor directly on the output (like a 100 nF bypass cap) looks like a short at high frequency and causes oscillation. Fix: add a 100–470 Ω series resistor between the op-amp output and the capacitor to isolate the load. This is standard practice for all op-amps, but the LM358 is particularly sensitive because of its slow, multi-stage internal compensation.
A: For 3.3V and 5V rails, the LMV358 (rail-to-rail output, 5.5V max) or MCP6002 (RRIO, 6V max) are the most popular drop-in alternatives. Both are pin-compatible with the LM358 in SOIC-8. The LMV358 is a direct TI part in the same product family. The MCP6002 adds rail-to-rail input capability and 1 pA bias current (CMOS). Both are cheaper than the LM358 at volume. Caution: neither handles voltages above 6V — if your rail is 12V, these are not options.
A: You can, but you shouldn't. The LM358's differential input voltage rating is ±32V (the full supply range), so it can handle large differential inputs without damage — unlike many precision op-amps with input clamping diodes. But the recovery time from saturation is slow (tens of microseconds) because the internal transistors saturate and need time to come out of it. A real comparator (LM393, LM339) has an open-collector output optimized for fast saturation and recovery. For low-speed threshold detection (power-good, battery-low), the LM358 works fine. For anything requiring speed or clean edges, use a comparator.
A: It can oscillate, inject noise into the active channel, and draw unpredictable supply current. The unused op-amp's inputs float to unknown voltages, and the open-loop gain (100 dB+) amplifies any stray coupling. Never leave an unused op-amp floating. Proper termination: connect the unused channel as a voltage follower — output tied to IN−, IN+ tied to a mid-supply reference (or ground in single-supply if common-mode range permits). This keeps both inputs at a defined voltage, the output stable, and current draw predictable. Per the TI LM358 datasheet (Rev K), this is the recommended termination.
A: The LM358 is one of the most counterfeited ICs in the world — it's cheap, high-volume, and easy to re-mark. Quick checks: (1) Measure supply current — a genuine LM358 draws 0.5 mA per amplifier (1 mA total for the dual) at 5V, no load. Counterfeits often draw more or less by a wide margin. (2) Check output swing — with a 5V supply and 10 kΩ load to ground, the output should swing from ~5 mV to ~3.5V minimum. Counterfeits often can't pull within 200 mV of ground. (3) Check slew rate — apply a 10 kHz square wave at unity gain. Rise/fall time should be in the 5–10 µs range. Counterfeits often slew 2–5× slower. (4) Marking quality — genuine TI/ST parts use laser marking with consistent font and alignment. Ink marking on a "new" part is a red flag.





