AMC1035DR
Product Specifications
AMC1035DR
Texas Instruments — Power Management & Signal Chain
Product Overview
The AMC1035DR 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
AMC1035DR: a delta sigma modulator for shunt and voltage sensing
AMC1035DR is a precision delta sigma modulator in an 8 pin SOIC body. It takes a bipolar plus or minus 1 V differential input, runs from a single 3.0 V to 5.5 V supply, and puts out a bitstream that a microcontroller's digital filter turns into a number. It also carries an internal 2.5 V reference with 5 mA of sink and source capability, which is what makes a ratiometric measurement possible without a second reference device on the board.
The part is not an ADC in the usual sense. It has no serial register map, no conversion result register and no configuration bus. The output is a stream of ones and zeros at the modulator clock rate, and the resolution comes from the filter that follows it. With a sinc3 filter at an oversampling ratio of 256 the document quotes 16 bits and a dynamic range of 87 dB at 82 kSPS, which is the pair of numbers a motor control engineer actually needs.
The figures below come from SBAS837B, revised April 2020, and from the orderable addendum at the back. The document is unusually explicit about the limits of the input structure, and those limits are where most integration problems on this part begin.
The input structure sets the rules
The specified linear full scale is plus or minus 1 V differential, with clipping starting at 1.25 V. The common mode range is not a single number: it depends on the supply and on the differential signal. At 3.0 V to 4 V supply with a full scale differential input the common mode has to sit between roughly 0.85 V and 2.15 V. At 4.5 V to 5.5 V supply the window widens. The document prints a second window for the case where the differential input is zero, and it is wider still.
What that means in practice is that the input is not a general purpose differential pair. It is a buffered sampling stage with a charge pump that lets a single supply handle a bipolar signal, and the buffer's headroom is the constraint. A shunt at the top of a 48 V rail cannot connect directly; the common mode has to be brought into the window first, either by a divider or by an isolated stage.
The input resistance is high, 1.6 Gohm typical differential, which is why a divider can be used without loading it. The input bias current is in the low nanoamp range and drifts by about 5 pA per degree, so a 10 kohm divider adds a fraction of a millivolt of offset, which is inside the 0.5 mV maximum offset error of the part.
| Function | delta sigma modulator with differential input |
| Input range | plus or minus 1 V linear full scale, clipping at 1.25 V |
| Input resistance | 1.6 Gohm typical differential |
| Supply range | 3.0 V to 5.5 V single supply |
| Reference output | 2.5 V nominal, 5 mA sink and source |
| Clock input | 9 MHz to 21 MHz external, continuous |
| Manchester mode | 9 MHz to 11 MHz clock window |
| Resolution | 16 bits with a sinc3 filter at ratio 256 |
| Dynamic range | 87 dB at 82 kSPS data rate |
| Offset error | 0.5 mV maximum |
| Offset drift | 6 microvolts per degree maximum |
| Gain error | 0.25 percent maximum, 45 ppm per degree drift |
| Ratiometric gain drift | 15 ppm per degree |
| Integral nonlinearity | plus or minus 12 LSB |
| Common mode rejection | 104 dB at DC, 88 dB at 10 kHz |
| Package | SOIC, D body, 8 pins, 4.9 mm by 3.9 mm |
| Operating range | minus 40 to plus 125 degrees Celsius |
| Thermal resistance | 120 degrees per watt junction to ambient |
The clock is a system decision, not a device setting
There is no internal oscillator. The modulator needs a continuous external clock between 9 and 21 MHz, and the data rate scales with it. At 20 MHz with a sinc3 filter at ratio 256, the output rate is about 78 kSPS, and the document's 82 kSPS figure corresponds to a slightly different combination. Halve the clock and the data rate halves with it, and the noise floor improves by roughly 3 dB for every doubling of the oversampling ratio.
A clock that stops is the failure mode to plan for. The document warns that the clock must be applied continuously, and if it stops the output stops with it. A watchdog on the receiving side that counts bitstream edges is cheaper than a diagnostic pin, and on a motor drive it is the difference between a controlled stop and a runaway.
Manchester coding is the other option. With the coding pin pulled high the bitstream carries its own clock, which removes the setup and hold requirement at the receiving device and makes the interface tolerant of a slower isolation path. The trade is a narrower clock window, 9 to 11 MHz, and half the effective data rate. On a board where the modulator and the controller sit on opposite sides of an isolator, Manchester is usually the right choice; on a board where they share a ground, it is not needed.
Ratiometric measurement and why the reference matters
The internal reference is 2.5 V and the input full scale is 1 V, so the reference is not the full scale. That is deliberate: the reference is there to supply the sensor bridge or the divider, not to set the conversion range. When the same 2.5 V feeds the divider and the reference input, supply drift cancels in the ratio, and the document quotes a ratiometric gain drift of 15 ppm per degree against 45 ppm per degree in absolute mode.
That is a factor of three in thermal drift for the cost of one wire. On a current sensor that has to hold accuracy over a cabinet temperature swing, it is the difference between a 0.5 percent and a 0.2 percent error budget.
The reference also has a current limit worth respecting. 5 mA is enough for a divider and a bridge, not enough for a chain of sensors. Past that, the reference sags and the gain error moves with the load, which is the sort of fault that looks like a mysterious nonlinearity until somebody measures the reference pin under load.
Where the part sits in a drive or an inverter
The document lists motor drives, photovoltaic inverters, uninterruptible power supplies, industrial transport systems, and AC voltage and temperature sensing. The two jobs are current and voltage, and the input structure handles both, with different external parts.
For current, a shunt of a few milliohms produces tens of millivolts at rated current, which is well inside the 1 V range and leaves plenty of headroom for a fault transient. The limiting factor is the common mode, not the range: a shunt in a half bridge sits on a node that swings with the switching, so the modulator needs either an isolated supply or a careful divider with a differential filter.
For voltage, the input resistance allows a divider of a few hundred kilohms, and the 1 V range maps a 400 V rail through a 400 to 1 divider with margin. The same ratiometric trick applies if the divider is fed from the reference, though at 400 V the divider current dominates and the reference is not the natural source.
Our ADC DAC cross reference page covers the parts we hold against this function, and the incoming inspection plan page describes the sample we pull for offset and gain on each lot.
Three questions buyers ask about this string
Can it replace an AMC1305 or an AMC1306? Same family concept, different pinout and different clock arrangement. The replacement has to be checked pin by pin, not by family name.
Does it need an isolated supply? No, but it needs a common mode inside the window. If the shunt sits on a floating node, the supply has to float with it, which is the job of an isolated DC to DC converter, not of this part.
Is the 16 bit figure a guarantee? It is the resolution with a specified filter. The accuracy is set by the offset, gain and linearity rows, which are 0.5 mV, 0.25 percent and 12 LSB respectively. Resolution and accuracy are different promises and the page keeps them apart.