AMC1303M0520DWVR
Product Specifications
AMC1303M0520DWVR
Texas Instruments — Power Management & Signal Chain
Product Overview
The AMC1303M0520DWVR 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
AMC1303M0520DWVR: 50 mV of shunt, an uncoded bitstream and 20 MHz on the clock
AMC1303M0520DWVR is the member of the AMC1303 family with a plus or minus 50 mV input range, an uncoded CMOS bitstream output and an internal clock of 20 MHz, in the eight pin DWV body. It measures the voltage across a low value shunt on the high voltage side of a drive and puts a one bit stream on the low voltage side across a capacitive barrier certified for reinforced isolation.
The three letters and digits that follow the family name are the whole specification. The M means the output is uncoded, so the receiver takes both DOUT and CLKOUT and does its own sampling. The 05 means the linear range is 50 mV, which pairs with a 4.9 k differential input resistance. The 20 means the internal clock is 20 MHz, which doubles the data rate against the 10 MHz members and halves the time between a current change and the sample that carries it.
The figures below come from SBAS771D, revised October 2024, and from the orderable addendum in the same document. They are the ones we check against a customer current sense requirement before quoting, because the difference between this part and the 250 mV member is not accuracy in the abstract, it is which shunt the customer can buy and which filter the receiving device can run.
What the M and the 05 change about the board
The M removes the Manchester encoder from the output path. DOUT becomes a plain bitstream at the modulator rate and CLKOUT carries the clock that the modulator used, so the receiver needs two traces instead of one and a sampling clock that is synchronous with the data. On the low voltage side of the barrier that is usually free, because the digital isolator that carries the signals across has spare channels and the microcontroller or FPGA has a serial port that can be clocked externally.
The 05 changes the shunt. A 50 mV full scale range means a 1 milliohm shunt at 50 A, or a 10 milliohm shunt at 5 A, and it means the shunt dissipates 2.5 W at 50 A on the 1 milliohm part. The 250 mV member would allow a 5 milliohm shunt at the same current and a quarter of the dissipation, so the narrow range part is chosen when the current is small and the shunt can be large, and the wide range part is chosen when the current is large and the shunt has to be small.
The 20 changes the rate. At 20 MHz with a sinc3 filter at an oversampling ratio of 256 the output data rate is 78 kSPS and the filter settles in three output periods, which is 38 microseconds. That is fast enough for a current loop closed at 20 kHz with margin for the filter delay, and it is the reason this member exists alongside the 10 MHz version.
The numbers that decide a current sense front end
| Function | Isolated delta sigma modulator for shunt based current measurement |
| Linear input range | plus or minus 50 mV for this order code |
| Differential input resistance | 4.9 k for the 50 mV members |
| Input current at full scale | 10.2 microamps |
| Output format | Uncoded CMOS bitstream with a separate clock output |
| Internal clock | 20 MHz nominal for this order code |
| Output data rate | 78 kSPS with a sinc3 filter at oversampling ratio 256 |
| Dynamic range | 85 dB at the 78 kSPS output rate |
| Offset error | plus or minus 50 microvolts maximum on the 50 mV members |
| Offset drift | plus or minus 1 microvolt per degree Celsius maximum |
| Gain error | plus or minus 0.2 percent maximum |
| Gain drift | plus or minus 40 parts per million per degree Celsius maximum |
| Common mode transient immunity | 100 kilovolts per microsecond typical |
| Reinforced isolation | 7000 V peak per DIN EN IEC 60747-17, VDE 0884-17 |
| Basic isolation | 5000 V RMS for one minute per UL1577 |
| Working voltage | 1500 V RMS |
| Analog supply | 3.0 V to 5.5 V on AVDD |
| Digital supply | 2.7 V to 5.5 V on DVDD |
| Ambient range | minus 40 to plus 125 degrees Celsius |
| Package | SOIC, DWV, 8 pins, 5.85 mm by 11.5 mm |
| Carrier | Large tape and reel, 1000 pieces, part marking 1303M052 |
| Moisture level | Level 3, 260 degrees, 168 hours floor life |
| Lead finish | Nickel palladium gold |
50 mV full scale and the 4.9 k input resistance
A 50 mV range with 4.9 k of differential input resistance draws 10.2 microamps at full scale, which through a 10 milliohm shunt develops 102 nanovolts and is invisible. The burden is not the issue at this range; the issue is the offset, because 50 microvolts of offset on a 50 mV range is 0.1 percent of full scale and that is the entire error budget of a channel that has to hold one percent.
The offset drift is the number that gets forgotten. One microvolt per degree Celsius over a 100 degree swing is 100 microvolts of apparent offset at the end of the range, which is 0.2 percent of full scale, twice the initial offset. A channel that is trimmed at 25 degrees and then run at 85 degrees therefore has to be trimmed again or corrected in firmware, and the correction is a straight line only if the drift is linear, which the specification guarantees as a maximum rather than as a shape.
The gain drift of 40 parts per million per degree Celsius adds 0.4 percent over the same swing, and it applies to the reading rather than to the offset. A 50 A channel at 0.4 percent is 0.2 A, and a protection threshold set at 55 A moves to 55.2 A at temperature. That is usually acceptable and it is worth writing down, because the alternative is a customer who measures the threshold at room temperature, measures it again after a soak, and reports a failing board.
20 MHz of clock, 78 kSPS and the delay that follows
The 20 MHz clock sets everything downstream. The modulator produces one bit per clock, the sinc3 filter consumes 256 of those bits per output sample, and the result is 78 kSPS with a filter that settles in three samples, so 38 microseconds from a step at the input to a settled output. A control loop running at 20 kHz has a 50 microsecond period, so the measurement arrives with most of a period of delay, which is the real bandwidth limit of the channel.
Doubling the clock is not free. The bitstream rate doubles, the receiver has to sample at 20 MHz, and the digital isolator that carries DOUT and CLKOUT has to pass a 20 MHz square wave without excessive jitter. A board that runs the isolator at its limit and then asks why the current reading has noise usually finds the answer at the clock edge rather than in the modulator.
A higher oversampling ratio trades rate for noise. At an oversampling ratio of 512 the data rate halves to 39 kSPS and the noise falls by about 9 dB for a third order filter, which is the trade to make when the loop is slow and the resolution matters. The modulator does not care; the filter on the other side is a firmware setting.
Uncoded output means the filter takes the clock
With the uncoded member the receiver has to sample DOUT on CLKOUT, and the two traces have to arrive with matched delay. On a board where the isolator has a channel to channel skew of a few nanoseconds that is easy; on a board where DOUT goes through a different isolator than CLKOUT it is not, and the resulting errors look like intermittent bit slips at temperature. The rule is to keep both signals on the same isolator package and the same side of the barrier.
The Manchester coded member of the family removes that requirement at the cost of a decoder in the receiver. Neither is better in the abstract: a microcontroller with a flexible serial port takes the uncoded part, an FPGA with spare logic takes the coded part, and a product that changes processor late in the program often changes the modulator order code with it.
One practical note on the clock output: it is a CMOS output driving a trace across a barrier, so it is the loudest signal on the high voltage side of the board. Routing it beside the shunt traces couples the clock into the measurement, and the resulting error appears at the clock frequency rather than at the switching frequency, which makes it easy to identify and easy to fix by moving the trace.
Isolation and the safety numbers that go with it
The barrier is rated for 7000 V peak of reinforced isolation under DIN EN IEC 60747-17, 5000 V RMS for one minute under UL1577, and 1500 V RMS of working voltage. The first two are type test numbers that appear on a certificate; the third is the continuous rating that decides whether the part is suitable for a 400 V or a 690 V drive, and it is the one to quote in a safety file.
Common mode transient immunity of 100 kilovolts per microsecond is what keeps the bitstream clean while an inverter leg switches. It is a typical figure rather than a guaranteed one, so a project that sits close to the limit should be tested rather than calculated, and the test is a switching bridge with the modulator powered from the same floating supply as the high side gate driver.
The DWV body gives 5.85 mm of width and the creepage that goes with it. A project that needs more for a higher pollution degree changes package rather than adding a slot, and a project that needs less can use the smaller body if the family offers one, which it does not: all eight members are in the same DWV package.
Ordering: 1000 pieces, level 3, marking 1303M052
The reel carries 1000 pieces with nickel palladium gold lead finish, part marking 1303M052, and a moisture sensitivity level of 3 with a 168 hour floor life. Level 3 is the same rating as the rest of the family, so a line that already handles the 250 mV member needs no new bake rule, and a line that has only handled level 1 parts needs one before the first build.
Sourcing for this family runs through our ADC and DAC cross reference, where the isolated modulator alternatives are compared on input range, output format and body width, and through our component traceability report, which records the date code and the reel for every isolated part we ship. Lead time for the DWV body is the item that moves most often on this family, and our component lead time solutions page lists the alternatives we hold stock on when the franchised channel is quoting a long date.
On inspection, the channel is verified with a known current and a known shunt at two temperatures, and the bitstream is checked for a stable output rate rather than for a particular code. That is the same procedure we use on the 250 mV member and the same one the drive builders use at end of line, and it is the reason we keep a reference shunt and a current source in the lab rather than relying on a socket test.