INA126UA2K5
Product Specifications
INA126UA2K5
Texas Instruments — Power Management & Signal Chain
Product Overview
The INA126UA2K5 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
INA126UA2K5: the A grade SOIC body on a large reel
The INA126UA2K5 is the A grade SOIC body on a large reel, which is the carrier a production build uses when the channel needs the tighter offset bin and the volume justifies a reel. The die, the body and the electrical specification match the tube code exactly; the differences are the carrier and the moisture level that comes with it, so a line that has qualified the tube part does not need to requalify the reel part.
The electrical summary is the same as the rest of the family: gain from 5 to 10,000 with one external resistor, 175 µA per channel, 25 nA of bias current, 3 µV per degree of drift, and a supply window from plus or minus 1.35 V to plus or minus 18 V. On a production board the two rows that decide the build are the bandwidth at the working gain and the offset bin, in that order, because bandwidth cannot be trimmed and offset can.
The INA126UA2K5 rows as the addendum publishes them
| Part class | Micropower instrumentation amplifier, two op amp, single channel |
| Vendor status | Active, production |
| Data sheet | SBOS062D, revised December 2025 |
| Grade | A suffix, tighter offset and gain error window |
| Gain equation | G = 5 + 80 kΩ divided by RG |
| Input offset voltage | Family row 250 µV maximum, tighter on the A bin |
| Offset drift | 3 µV per degree maximum |
| Input bias current | 25 nA maximum |
| Voltage noise | 35 nV per root hertz at 1 kHz, 24 on the alternate chip origin |
| Bandwidth | 200 kHz at a gain of 5, 9 kHz at a gain of 100, 1.8 kHz at 500 |
| Supply range | Plus or minus 1.35 V to plus or minus 18 V |
| Quiescent current | 175 µA typical, 200 µA maximum per channel |
| Body | SOIC, D outline, 8 pins |
| Carrier | Large reel, quantity per the addendum row |
| Moisture level | Level 3, 260 degrees, 168 hour floor life |
| Temperature range | Minus 40 to plus 85 degrees |
| Tube sibling | INA126UA, 75 pieces per tube, level 2 |
175 µA per channel and the supply window that comes with it
At 175 µA per channel typical and 200 µA maximum, INA126UA2K5 is built for a product that runs from a battery or from a loop supply, and the supply window of plus or minus 1.35 V to plus or minus 18 V is what makes that possible. The bottom of the window is low enough that a pair of fresh alkaline cells in series will run the part directly, and the top is high enough for a plus or minus 15 V industrial rack. Nothing in between needs a regulator, which is the point: a channel that does not need a regulator does not need the quiescent current of one either.
The current is quoted per channel and with no load, so the total for a dual INA2126 is twice the figure. What the number does not include is the reference divider. A single supply channel needs a mid rail reference, and a 10 kΩ divider across 5 V draws 250 µA, which is more than the amplifier itself. A 100 kΩ divider draws 25 µA and needs a capacitor to hold it stiff; that trade is made once per board rather than once per channel, and it decides whether the microamp budget survives.
250 µV of offset and 3 µV per degree of drift on a bridge channel
The offset row for this family is 250 µV maximum on the plain parts and tighter on the A grade, and the drift row is 3 µV per degree maximum. Against a 10 mV full scale bridge those numbers are 2.5 per cent of range and 0.03 per cent per degree, so the offset is trimmed at the bench and the drift is lived with. Against a 100 mV full scale the offset is a quarter of a per cent and the drift is negligible, which is the region where the part is comfortable and where most pressure and flow transmitters actually sit.
The comparison that matters is against the amplifier the part replaces. A general purpose op amp pair built into an instrumentation front end can easily carry 1 mV of offset and 10 µV per degree, and at that level the drift, not the bridge, sets the annual recalibration interval. Moving to a trimmed instrumentation amplifier buys the interval back. The other half of the budget is the bridge excitation: a 5 V excitation on a 350 Ω bridge draws 14 mA, which is two orders of magnitude more than INA126UA2K5 draws, so the current budget is set by the sensor rather than by the amplifier.
The 80 kΩ term, 0.1 per cent resistors and the resistance you forgot
Gain accuracy on INA126UA2K5 is set by the external resistor and by the 80 kΩ term inside the equation, and the internal term is trimmed. What is not trimmed is the resistance from the RG pins to the resistor itself: the socket, the solder joint, the trace and the resistor lead all sit in series with RG, and at a gain of 1000 where RG is 80 Ω a single ohm of that path is 1.25 per cent of gain error. That is why the vendor note says sockets and wiring resistance matter at high gain, and why the resistor should be placed within a few millimetres of the pins.
The resistor itself should be a 0.1 per cent metal film part with a 25 ppm per degree coefficient, and it should be the same type on every channel of a multi channel board so that the channels track rather than each drifting its own way. A 1 per cent resistor at a gain of 100 contributes 1 per cent of gain error before temperature, which is larger than the amplifier gain error row. The cost difference between a 1 per cent and a 0.1 per cent resistor is a few cents, and it is the cheapest accuracy in the whole channel.
Bandwidth that falls with gain: 200 kHz at 5, 9 kHz at 100
The bandwidth row is quoted at three gains on this family: 200 kHz at a gain of 5, 9 kHz at a gain of 100 and 1.8 kHz at a gain of 500. That is a steep fall, and it is the price of running the input stage at 175 µA. For a pressure or temperature channel sampled a few times a second it is irrelevant. For a vibration channel or a multiplexed data acquisition front end it is the number that decides whether the part can be used at all, and it has to be read at the gain the channel actually runs rather than at the gain in the headline.
Settling follows the same curve: 30 µs to 0.01 per cent at a gain of 5, 160 µs at a gain of 100 and 1.5 ms at a gain of 500, with a 4 µs overload recovery from a 50 per cent overdrive. On a multiplexed front end the settling time, not the bandwidth, sets the channel scan rate, because the amplifier has to recover from the previous channel before the converter samples. A 16 channel scan at a gain of 100 therefore runs at a few kilohertz at best, and a board that assumes otherwise will show crosstalk that looks like a converter fault.
Single supply operation and the mid rail that is not optional
INA126UA2K5 runs from a single rail as long as the input signal sits inside the common mode window and the output has somewhere to swing. On a 5 V single supply the input window stops roughly 1 V below the positive rail and the output stops about 0.75 V short of it, so the usable span is smaller than the supply suggests. A bridge with a 2.5 V common mode and a plus or minus 20 mV signal fits easily; a sensor that sits at 0.2 V does not, and a channel that has to measure near ground needs either a negative rail or an input level shifted before the amplifier.
The mid rail reference is the second half of the problem. The REF pin has to be driven from a low impedance, because any resistance in series with it turns into a common mode error and a gain error at the same time. The vendor quotes 8 Ω as the point at which a typical part starts to lose common mode rejection, which is low enough that a divider alone is not a reference: the divider has to be buffered, or the reference has to come from a shunt regulator with a capacitor across it.
Plain grade against A grade on the same die
The A suffix on the PA, UA and EA rows is a tighter offset bin, not a different silicon. Both grades share the 3 µV per degree drift, the 25 nA bias current and the 175 µA supply current, and they differ in the maximum offset and in the gain error window. On a channel with a 10 mV full scale the plain grade consumes 2.5 per cent of the range before trimming and the A grade consumes half of that, which is the difference between a trim circuit that is a convenience and one that is mandatory.
Because the two grades are pin compatible, the choice is made on the specification rather than on the board, and that makes substitution easy in both directions and dangerous in one. Shipping the A grade where the plain grade was ordered costs money and raises no objection. Shipping the plain grade where the A grade was ordered passes incoming inspection only if nobody measures the offset, and fails at the first temperature cycle if somebody does. The grade should therefore sit on the purchase order line rather than in a note.
Buying INA126UA2K5
The reel is justified by volume and by the feeder, not by the price alone. A 2500 piece reel on a product that builds a few hundred boards a year is years of stock under a level 3 floor life, and the cost of that stock is usually larger than the price difference. On a build that does consume the reel, the A grade removes a trim operation, and a removed trim is worth more than the grade premium on any volume.
The other half of the buying decision is the second source. This is a micropower part with a specific gain floor and a specific bandwidth curve, and the parts that look similar rarely match both. Our amplifier cross reference page lists what we compare when a second source is needed, and the drop in replacement verification page sets out the bench checks before a substitution is approved.
Questions that arrive with this part number
Whether the reel code and the tube code are the same silicon is the first question and they are, with the carrier and the moisture level as the only difference. The second is whether the A grade can be replaced by the plain grade in a shortage, and it can only if the specification has margin or a trim, because the offset window widens and nothing else changes.
The third question is the gain resistor again. At a gain of 100 the resistor is 842 Ω and a 1 per cent part contributes 1 per cent of gain error, which is larger than the amplifier contribution. A 0.1 per cent part with a 25 ppm coefficient is the correct choice on a calibrated channel, and it costs less than the grade premium.