INA133U2K5
Product Specifications
INA133U2K5
Texas Instruments — Power Management & Signal Chain
Product Overview
The INA133U2K5 from Texas Instruments is a power management / signal chain IC delivering high efficiency, precision regulation, and robust protection. Texas Instruments (TI) is a global leader in analog ICs, power management solutions, and embedded processors.
Superb Automation partners with leading brands including Qorvo, TI, ADI, XILINX, NVIDIA, Intel, ST, MACOM, Renesas, and Coilcraft to provide complete BOM and supply-chain solutions.
Key Features
- High-efficiency power conversion with low quiescent current
- Wide input voltage range with precision output regulation
- Integrated over-current, over-temperature, and short-circuit protection
- Small footprint packaging for space-constrained designs
- AEC-Q100 qualified options available for automotive applications
- Flexible sequencing and soft-start control
Key Specifications
| Input Voltage | V |
| Output Voltage | V |
| Output Current | A |
| Switching Frequency | kHz |
| Efficiency | % |
| Quiescent Current | µA |
| Topology | — |
| Package | — |
| Operating Temp | °C |
| Protection | — |
Applications
Request Quote — In Stock, 100% New & Original, Full Warranty
INA133U2K5: the plain grade on a 2500 piece reel, where the chip origin decides the slew rate
The INA133U2K5 is the plain grade of the fast difference amplifier from a 2500 piece large reel, the catalogue spelling without the slash. It is the production carrier for a part whose electrical rows are fixed but whose speed is not: the same order code arrives with a 5 V per microsecond slew rate from one fabrication flow and 20 from the other, and the reel label carries the flow rather than the part number.
The fixed rows are 450 µV of offset referred to the output, 0.05 per cent of gain error, 80 dB of minimum rejection, 1.5 MHz of small signal bandwidth, 5.5 microseconds of settling and 950 µA of supply current on a plus or minus 2.25 V to plus or minus 18 V window. The network is 25 kΩ per input, so a low impedance source is required for the gain error to mean anything.
The INA133U2K5 rows as the addendum publishes them
| Part class | High speed unity gain difference amplifier, trimmed 25 kΩ network |
| Vendor status | Active, production |
| Data sheet | SBOS115B, revised January 2026 |
| Addendum row | INA133U/2K5, 2500 pieces on a large reel |
| Grade | U bin |
| Gain | 1 V/V fixed |
| Input offset voltage | 450 µV maximum at plus or minus 15 V |
| Offset drift | 5 µV per degree maximum |
| Gain error | 0.05 per cent maximum |
| Common mode rejection | 80 dB minimum |
| Input impedance | 50 kΩ differential, 25 kΩ common mode |
| Bandwidth | 1.5 MHz small signal |
| Slew rate | 5 V per microsecond on SHE, 20 on RFB |
| Voltage noise | 57 nV per root hertz on SHE, 20 on RFB |
| Settling | 5.5 µs to 0.01 per cent |
| Supply range | Plus or minus 2.25 V to plus or minus 18 V |
| Quiescent current | 950 µA typical |
| Body | SOIC, D outline, 8 pins |
| Carrier | 2500 pieces, 330 mm reel |
| Moisture level | Level 3, 168 hour floor life |
| Dual sibling | INA2133UA, 14 pin SOIC, 50 pieces per tube |
1.5 MHz at unity gain from a difference amplifier
INA133U2K5 is the fast member of the unity gain difference amplifier group: 1.5 MHz of small signal bandwidth against the 300 kHz of the INA132, with the same on-chip resistor network and the same eight pin body. The bandwidth is what a servo feedback path or a string inverter current loop needs, because a 20 kHz loop with 60 degrees of phase margin wants an amplifier an order of magnitude faster than the loop, and a 300 kHz part leaves no margin.
The network is four 25 kohm resistors trimmed as a set, giving a differential input impedance of 50 kohm and a common mode impedance of 25 kohm. Gain is fixed at 1 V/V, gain error is 0.05 per cent maximum on the U bin and 0.1 per cent on the UA bin, and nonlinearity is 0.001 and 0.002 per cent of full scale. At unity gain the offset is quoted referred to the output, so the input and output numbers are the same.
Two chip origins, and 5 against 20 volts per microsecond
The same part number can arrive from two fabrication flows, and the vendor labels the difference by chip site origin. On INA133U2K5 the slew rate is 5 volts per microsecond on the SHE flow and 20 volts per microsecond on the RFB flow, a factor of four on a row that no distributor listing shows. Voltage noise moves the other way: 57 nanovolts per root hertz at 1 kHz on SHE and 20 on RFB, and the common mode rejection minimum moves from 80 to 74 dB between the U and UA bins on top of that.
For a servo loop running at a few kilohertz either flow is fine. For a channel that has to settle in a few microseconds after a multiplexer step, the 5 volt per microsecond flow is a 20 microsecond edge on a 10 V step and the RFB flow is 5 microseconds, which is the difference between a working scan rate and a specification that is met only on paper. The flow is printed on the reel label and in the lot record, not on the part.
A common mode range of twice the supply, minus three volts
The input common mode range of INA133U2K5 extends beyond both rails: from 2 times the negative rail plus 3 V to 2 times the positive rail minus 3 V. On a plus or minus 15 V supply that is minus 27 V to plus 27 V, and on a plus or minus 5 V supply it is minus 7 V to plus 7 V. The absolute maximum input is 80 V, so the part survives a common mode it cannot measure, which is the useful half of the specification on an industrial rack where a transducer cable can pick up a fault voltage.
That range is what allows the part to sit directly across a shunt in a motor leg or a battery string, where the common mode is neither near ground nor inside the supply. It also means the inputs are not protected by the supply rails: a 40 V common mode on a plus or minus 5 V supply is inside the input window and outside the supply, and the output will do something the loop may not expect. The window has to be checked against the supply, not assumed from it.
450 microvolts on the U and 900 on the UA
The offset window on INA133U2K5 is 450 microvolts maximum on the U bin and 900 microvolts on the UA bin at a plus or minus 15 V supply, doubling to 750 and 1500 microvolts at plus or minus 5 V, with a drift of 5 microvolts per degree on both. The supply dependence is unusual and worth reading twice: the same part is twice as loose on a 10 V total supply as on 30 V, so a 5 V single rail application starts with a 1.5 mV window on the UA bin.
Because the offset is referred to the output and the gain is one, a 900 microvolt offset is 900 microvolts at the output, which against a 2 V full scale is 0.045 per cent and against a 100 mV shunt signal is 0.9 per cent. On a shunt monitor that is a real error and it is why the U bin exists; on a servo feedback path it is trimmed by the loop. Long term stability is 0.3 microvolts per month on both bins.
950 microamps, 5.5 microseconds and the dual INA2133
Quiescent current on INA133U2K5 is 950 microamps typical, which is five times the INA132 and the price of the 1.5 MHz bandwidth. Settling to 0.01 per cent takes 5.5 microseconds, short circuit current is 32 mA source and 25 mA sink on the SHE flow and 70 mA both ways on RFB, and the output is stable into 1000 pF. Output swing stops 1.5 V short of each rail under a gain error condition, which on plus or minus 15 V leaves plus or minus 13.5 V of usable range.
The single channel is the eight pin SOIC. The dual INA2133 is the fourteen pin SOIC with two independent channels at 8.65 by 6 mm, sold in 50 piece tubes, and it is specified over minus 55 to plus 125 degrees rather than the minus 40 to plus 85 of the single. On a multi channel board the dual saves one package and one reel but costs board area, and the wider temperature range is worth having when the board sits near a heatsink.
The 25 kohm network loads the source and unbalances it
Each input of INA133U2K5 sees 25 kohm to the amplifier, and the differential pair presents 50 kohm. Across a 100 ohm shunt that is a 0.2 per cent gain loss, which is larger than the 0.05 per cent gain error row and has to be calibrated. Across a 10 kohm source the loss is 17 per cent and the whole measurement is a divider, which is the point at which a buffer belongs in front of the part rather than after it.
The network is ratio matched, so the loading is common to both inputs and cancels as long as the source impedances match. A protection resistor in one leg, a connector with different contact resistance, or a filter capacitor on one side breaks that match and converts common mode into a differential error. On a board the two input traces should be the same length, the same width and the same distance from any plane, which is ordinary practice for a differential pair and easy to forget on a DC channel.
Buying INA133U2K5
The reel is the right carrier for a line that builds in thousands, and the moisture level is level 3, so the floor life clock is 168 hours once the bag is open. On a part with no active shelf life limit that is the only schedule constraint, and it is the one that gets missed when a reel is opened for a sample and put back on the shelf.
The grade and the flow are the two specification decisions. The U bin buys half the offset window, and the flow buys a factor of four on slew rate and a factor of three on noise. Neither is on a distributor page, so both have to be written on the requirement rather than assumed from the code. Our traceability report page covers how the lot record carries the flow back to the label.
Questions that arrive with this part number
Whether the reel spelling without a slash is a different part is the first question, and it is not: it resolves to INA133U/2K5. Whether the slew rate on the data sheet is the slew rate in the box is the second, and it is not always: 5 and 20 V per microsecond are the same part number from two flows, and the difference is a factor of four on a 10 V edge.
The third is the settling time, 5.5 microseconds to 0.01 per cent, which is what makes the part usable in a multiplexed scan and is quoted independently of the flow. The fourth is the load: 25 kΩ per input is a real load on a 10 kΩ source, and the divider loss is larger than every amplifier error row on the sheet combined.