Loading…
TO-92 · LM35
ActiveAnalog Temperature Sensor (Centigrade)
LM35
Texas Instruments · TO-92 · Analog Temperature Sensor (Centigrade)
Price on Request
Get a competitive quote within 24 hours. Volume discounts available. Shenzhen stock ships in 1-2 business days.
Overview
The LM35 is Texas Instruments' precision centigrade temperature sensor and one of the longest-lived analogue temperature parts still in production. Its defining characteristic is the output: a voltage linearly proportional to Celsius temperature at a wafer-trimmed 10 mV/°C scale factor, referenced to ground. Because the device is calibrated directly in Celsius, the designer never subtracts the large constant offset that Kelvin-calibrated sensors impose — at 25 °C the output is simply 250 mV, and at 0 °C it is 0 mV. Trimming happens at wafer level, which is why the part costs so little while still holding ±0.5 °C accuracy at 25 °C on the LM35A grade.
The datasheet splits the family by temperature range and grade. The LM35 and LM35A are rated for the full -55 °C to +150 °C range, the LM35C and LM35CA for -40 °C to +110 °C, and the LM35D for 0 °C to +100 °C. Accuracy at 25 °C is ±0.5 °C maximum on the A and CA grades (±0.2 °C typical) and ±1 °C maximum on the standard LM35, LM35C and LM35D parts (±0.4 °C to ±0.6 °C typical), with the tested limit over the full range reaching ±1.5 °C to ±2 °C depending on grade. Sensor gain is held to 9.9 mV/°C to 10.1 mV/°C on the A grade and 9.8 mV/°C to 10.2 mV/°C on the standard grade, nonlinearity is ±0.18 °C to ±0.3 °C typical, and load regulation is ±0.4 mV/mA typical. Supply range is 4 V to 30 V, quiescent current is 56 µA typical with 67 µA to 80 µA maximum at 25 °C and about 105 µA to 158 µA at temperature extremes, self-heating is 0.08 °C in still air, and the output impedance is 0.1 Ω for a 1 mA load — low enough to drive a cable or an ADC sample-and-hold directly.
Two circuit details trip up most users. First, the basic configuration only measures positive temperatures: the output cannot go below ground, so measuring -55 °C requires either a negative supply or the datasheet's full-range circuit, which connects a resistor R1 from the output to -VS with R1 = -VS / 50 µA so the device can sink current and swing negative. Second, the output is absolute rather than ratiometric — accuracy comes from the wafer-level trim, while your ADC reference sets only the resolution: at 10 mV/°C a 12-bit converter with a 3.3 V reference yields 0.8 mV per LSB, or about 0.08 °C per code, which comfortably oversamples the ±1 °C device accuracy. Packages include hermetic TO-CAN (TO-46), plastic TO-92, 8-lead SOIC and TO-220, with the TO-92 being the volume part. Alternatives are worth naming on the BOM. The TMP36 is pin-compatible in TO-92 and adds a 500 mV offset so it reads negative temperatures from a single supply, but runs only to 5.5 V. The LM34 is the Fahrenheit equivalent at 10 mV/°F. For a digital output, the DS18B20 gives 1-Wire multidrop and ±0.5 °C accuracy, and I2C parts such as the TMP102 or SHT31 remove analogue layout concerns entirely. The LM35 remains the right choice where you already have an ADC, need a wide supply range up to 30 V, or must drive a long cable with a low-impedance analogue output.
Sourcing note: TO-92 temperature sensors are a favoured target for remarking, and the LM35 is no exception — counterfeit parts typically show the wrong slope, excessive quiescent current or nonlinearity well outside ±0.5 °C. Verify the 10 mV/°C slope at two known points (ice bath at 0 °C and a calibrated block), measure supply current against the 56 µA typical figure, and check laser marking and lead finish. We stock TI-channel LM35DT (TO-92) and LM35DR (SOIC-8); MOQ is 2,000 pcs on tape-and-reel for the TO-92 or 100 pcs loose, with 1-3 day dispatch.
Key Features
- 10 mV/°C output calibrated directly in Celsius — no Kelvin offset subtraction
- ±0.5 °C accuracy at 25 °C on LM35A/LM35CA grades, ±1 °C on standard parts
- Wide 4 V to 30 V single supply with under 60 µA current drain
- 0.1 Ω output impedance drives cables and ADC sample-and-hold inputs directly
- TO-92, TO-CAN, SOIC-8 and TO-220 package options across the family
Typical Applications
- Power supply, battery pack and motor thermal monitoring
- HVAC controllers, thermostats and appliance temperature feedback
- Cold-chain, incubator and environmental datalogger front ends
- Analogue temperature setpoints and thermal protection loops
- Educational and prototyping boards with an MCU ADC input
Specifications
- Output Scale Factor
- 10 mV/°C
- Accuracy at 25 °C
- ±1 (LM35A ±0.5) °C
- Temperature Range (LM35 / LM35C / LM35D)
- -55 to 150 / -40 to 110 / 0 to 100 °C
- Supply Voltage
- 4 to 30 V
- Quiescent Current (Typ, 5 V)
- 56 µA
- Nonlinearity (Typ)
- ±0.3 °C
- Self-Heating (Still Air)
- 0.08 °C
- Output Impedance (1 mA Load)
- 0.1 Ω
- Pins
- 3
- RoHS
- Compliant
Procurement Tips
Put the exact grade and package suffix (for example LM35DT or LM35CZ/NOPB) on the BOM, and require a two-point slope test report on incoming lots — it is the cheapest way to keep remarked TO-92 temperature sensors off your production line.
Frequently Asked Questions
What is the difference between the LM35 and the TMP36?
Both are TO-92 analogue sensors with a 10 mV/°C slope, but the TMP36 adds a 500 mV offset so that 0 °C reads 500 mV — which lets it measure sub-zero temperatures from a single supply. The TMP36 runs from 2.7 V to 5.5 V and covers -40 °C to +125 °C, while the LM35 needs 4 V to 30 V and, in its basic circuit, only reads positive temperatures unless you add a negative rail or the R1-to-minus-VS arrangement. Choose the TMP36 for low-voltage single-supply designs that go below freezing; choose the LM35 for wide supply range, low output impedance and better absolute accuracy.
How do the LM35, LM35A, LM35C, LM35CA and LM35D grades differ?
By temperature range and accuracy. LM35 and LM35A are rated -55 °C to +150 °C; LM35C and LM35CA are rated -40 °C to +110 °C; LM35D is rated 0 °C to +100 °C. The A grades guarantee ±0.5 °C at 25 °C with tighter gain (9.9 to 10.1 mV/°C) and lower nonlinearity, while the standard grades allow ±1 °C at 25 °C and 9.8 to 10.2 mV/°C. The LM34 is the Fahrenheit sibling at 10 mV/°F. Specify the exact suffix on the BOM, because the TO-92 markings are easy to misread.
How do I measure temperatures below 0 °C with an LM35?
The basic circuit cannot, because the output is referenced to ground and cannot swing negative. Two options: use the datasheet full-range circuit, which connects a resistor R1 from VOUT to a negative supply with R1 = -VS / 50 µA (for -VS = -5 V that is 100 kΩ), giving an output of -550 mV at -55 °C and +1500 mV at +150 °C; or run the device from split supplies. If a negative rail is unacceptable, switch to the TMP36, which handles -40 °C from a single supply thanks to its 500 mV offset, or use a digital sensor such as the DS18B20.
My LM35 reads 1 °C to 3 °C high. What is wrong?
Check four things. Self-heating: the part dissipates supply current through itself, so at 30 V the die warms measurably — 0.08 °C in still air is specified at low supply, and a hot PCB or an enclosed TO-92 with no airflow makes it worse. Loading: keep load current under 1 mA or use a buffer, since load regulation is ±0.4 mV/mA (about 0.04 °C per mA). ADC scaling: with a 3.3 V reference and 12-bit converter each LSB is 0.8 mV or 0.08 °C, so a mis-set VREF or a shared multiplexer with a high-impedance source can add real error. Finally, sensor placement: a TO-92 measures its own die temperature, so mount it away from heat-generating components and thermally couple it to the medium you care about.
Should I use the LM35 or a digital sensor such as the DS18B20 or SHT31?
Use the LM35 when you already have a spare ADC channel, need a wide supply range up to 30 V, must drive a long cable with a low-impedance analogue output, or want the simplest possible BOM with no bus protocol. Use a digital sensor when you need multidrop addressing (DS18B20 on 1-Wire), an I2C bus shared with other devices (SHT31, TMP102), immunity to cable-induced noise, or combined humidity and temperature. Analogue also means your accuracy depends on ADC reference quality; a digital sensor carries its own calibration.
How do I detect counterfeit or out-of-spec LM35 parts?
Test the slope, not the marking. Measure output at two calibrated points — an ice bath at 0 °C should give 0 mV and a 25 °C block should give 250 mV within the grade tolerance — and confirm the difference is 10 mV/°C. Measure supply current: genuine parts sit near 56 µA at 5 V and 25 °C, while clones often draw several hundred microamps. Check nonlinearity across the range and inspect laser marking, TO-92 lead finish and date-code traceability. We screen every lot with a two-point slope test and can supply the records.
What is the MOQ and lead time?
MOQ is 2,000 pcs on tape-and-reel for LM35DT (TO-92) or LM35DR (SOIC-8), or 100 pcs loose for prototyping and small-batch builds. TI authorized channel lead time is typically 6-12 weeks; our Shenzhen stock ships in 1-3 business days. Specify grade and suffix on the order — LM35DT, LM35CZ/NOPB and LM35AH are not interchangeable in a design that depends on the accuracy or temperature range.
Alternatives & Equivalents
View full cross-reference table →Need LM35 pricing and availability?
Submit an RFQ — we reply within 24 hours with stock, pricing, lead time and packaging options.
Request a Quote →