TPS53647RTAR
Product Specifications
TPS53647RTAR
Texas Instruments — Power Management & Signal Chain
Product Overview
The TPS53647RTAR 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
TPS53647 four phase D-CAP+ controller with PMBus for ASIC and point of load rails
The TPS53647RTAR is a driverless multiphase step down controller in the 40 pin 6 mm by 6 mm QFN body with the thermal pad on the underside. It carries no power switch of its own. Each phase drives a TI NexFET smart power stage, the CSD95372BQ5MC in the reference circuit, over a single PWM line, and reads that stage back over a current sense line, so the controller never sees the inductor current directly. That split is the whole point of the part: the loop stays in a small signal package while the heat sits in the stages, and the number of stages can be changed without changing the controller.
The family is built around a 4.5 V to 17 V conversion rail, 1 to 4 phases, and an 8 bit boot voltage from 0.5 V to 2.5 V in 5 mV steps. The boot level is latched at start up either from a resistor pinstrap or from the non volatile memory inside the part, which means a board can come up at the right voltage before any host processor is awake to talk to it. A 1.8 V or 3.3 V PMBus bias is accepted on the same pins, so the management bus does not need its own translator.
Key parameters taken from the SLUSC39B document
| Conversion input | 4.5 V to 17 V recommended, 12 V nominal, 19 V absolute maximum, so a 12 V bus with a 10 per cent transient stays inside the window |
| Bias rails | V5 at 4.5 V to 5.5 V and an internal 3.3 V LDO on V3R3, both to be bypassed with at least 1 microfarad of ceramic |
| Phases | 1, 2, 3 or 4, with CSP2 to CSP4 tied to V3R3 to disable the phase that is not fitted |
| Boot voltage | 0.5 V to 2.5 V in 5 mV steps, 8 bit, from pinstrap resistors or from NVM |
| Switching frequency | 300 kHz to 1 MHz closed loop, with the default row at 315 kHz minimum, 350 kHz typical and 385 kHz maximum |
| Reference output | 1.70 V typical, 1.685 V to 1.717 V, brought out on VREF to feed the REFIN pin of the power stages |
| Output tolerance | 6 mV typical and 6.5 mV maximum in the 0.61 V to 0.995 V band at 0 to 85 degrees, 0.6 per cent from 1 V to 1.52 V, 1 per cent in the VR12.5 band up to 2.5 V, and 0.8 per cent over minus 40 to plus 125 degrees |
| Current limit | 16 levels per phase, chosen by one resistor from OCL-R to VREF or by NVM, with the fault trigger at 125 per cent of IMAX by default |
| Standby current | 6.1 mA typical and 7 mA maximum on V5 with ENABLE low |
| Temperature | Minus 40 to plus 125 degrees ambient, 150 degrees junction maximum |
Setting the load line with one resistor
Current is summed in the analogue domain. Each stage reports 5 mV per amp, the controller adds the phases and produces a voltage on ISUM equal to the output current times 5 milliohm times the internal transconductance times the resistor from ISUM to VREF. A zero ohm resistor gives a flat load line, which is what a rail feeding a digital core usually wants. A finite resistor gives a droop that is proportional to current, which is what a rail feeding a processor with a strict voltage positioning requirement wants. Because the resistor sits at the summing node and not in the power path, the load line can be trimmed on the bench without touching the copper.
The maximum current is a separate setting. IMAX is an 8 bit value derived from the voltage on the F-IMAX pin, and the transfer is IMAX equal to the pin voltage divided by VREF times 256, minus 0.5 A. That half amp offset at the bottom is easy to miss when a divider is copied from a spreadsheet: at low IMAX settings the offset is a larger share of the total than at high ones, so the divider should be solved from the equation rather than scaled from a similar board.
PMBus telemetry and fault reporting
The part reports input voltage, output current, output power, temperature and fault status over PMBus, and the same bus configures the loop compensation, the slew rate and the boot level when NVM is used. Phase current imbalance is detected and reported, which is the practical way to find a stage that has stopped conducting before the rail collapses under load. VR_RDY, VR_FAULT and VR_HOT are separate pins, so a system that cannot read the bus still has three hardware lines to work with. Prebiased start up is supported, so a rail that is already charged through a body diode does not get pulled down when the controller enables.
Where the four phase version fits
A four phase controller at 350 kHz with 5 mV per amp sensing suits an ASIC or FPGA core rail in the 60 A to 200 A range, and also suits a networking line card where a 12 V bus has to become 0.9 V at high current with a fast transient response. The D-CAP+ loop with undershoot and overshoot reduction exists for exactly that load step: when a core wakes from a low power state, the current step arrives in a few hundred nanoseconds and the output capacitors alone cannot hold the rail. Phase shedding drops phases at light load, which matters on a line card that idles at a few amps for most of its life, and adds them back when the load returns.
For a procurement team, the useful point is that this is a controller only. The BOM carries the controller, the power stages, the inductors and the output capacitors as separate lines, so a shortage on one power stage can be worked around with a pin compatible stage from the same family while the controller stays put. Read more on power management cross reference and on pin to pin cross reference.
What to check before a purchase order
Three fields decide whether a quoted reel is the right one. The first is the speed and reel suffix, since the same body ships on a large reel and a small reel and the two are not interchangeable on a high volume line. The second is the NVM image, because a part configured for a different boot voltage will start a board at the wrong level even though the marking looks correct. The third is the PMBus address, which is set by a resistor to VREF on the ADDR-TRISE pin and must match the host map. Our sourcing desk checks these three against the customer BOM before quoting, and when a lead time moves we show the alternative stage and the alternative controller side by side rather than a single exchange part. See component lead time solutions for how that comparison is presented.