TPS546D24RVFR
Product Specifications
TPS546D24RVFR
Texas Instruments — Power Management & Signal Chain
Product Overview
The TPS546D24RVFR 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
TPS546D24RVFR 40 A single stage that reaches 160 A in a four device stack
The TPS546D24RVFR is the high current member of the PMBus buck family: 2.95 V to 16 V on PVIN, 2.95 V to 18 V on AVIN, 40 A continuous from a 7 mm by 5 mm package, and 160 A when four devices share one output. The power stage is 4.5 milliohm high side and 0.9 milliohm low side, which is what makes 40 A possible in a body this small without a parallel FET pair on the board.
Control is fixed frequency current mode with input feedforward and selectable internal compensation. The PMBus interface runs at 1 MHz with telemetry for output voltage, output current and die temperature, plus AVS and margining. Output is 0.6 V to 5.5 V by pin strap and 0.25 V to 5.5 V through VOUT_COMMAND, so the same part covers a 0.9 V core rail and a 3.3 V I/O rail with no board change beyond the strap resistors.
The internal 5 V LDO can be overdriven from an external 5 V rail on VDD5. On a 12 V input at 40 A, that is worth about 1.4 W of package dissipation moved off the die, and it is the difference between holding full load at 70 degrees ambient and folding back on thermal.
Key parameters from the SLUSDJ0A document
| Input range | 2.95 V to 16 V on PVIN, 2.95 V to 18 V on AVIN, 4 V minimum on VDD5 for switching |
| Output current | 40 A per device, 160 A with four devices stacked and sharing current on one address |
| Output voltage | 0.6 V to 5.5 V by pin strap, 0.25 V to 5.5 V through PMBus VOUT_COMMAND |
| Power FETs | 4.5 milliohm high side and 0.9 milliohm low side |
| Switching frequency | 12 settings from 225 kHz to 1.5 MHz, eight pin strapped, with frequency sync in and sync out |
| Sense accuracy | Differential remote sense with an internal divider for under 1 per cent output error over 40 to 150 degrees junction |
| Package | 40 pin LQFN, 7.00 mm by 5.00 mm by 1.5 mm, 0.5 mm pitch |
How it differs from the 10 A member of the family
The footprint, the pin strap scheme and the PMBus command set are shared with the 10 A device, so a board built for 10 A can be re populated for 40 A if the inductor, the input bank and the copper under the pad were sized for it. That is the reason the two are sold into the same socket: a rack server can start at 10 A per phase and move to 40 A per phase by changing one reel and the passive bill.
The difference that bites is the current limit setting and the inductor saturation rating. A 40 A part asked to run a 10 A inductor will hit valley current limit on every load step and then hiccup, and the fault looks like a control problem rather than a magnetics problem. When we quote a substitution between the two, we ask for the inductor part number before we quote.
What 160 A looks like on a board
Four devices at 40 A need a common output plane with a symmetric sense point, because the differential sense amplifier on each device reads the voltage at the point it is connected to, and 160 A across a few hundred microohm of plane is tens of millivolts of error. The document's stack arrangement puts the sense lines at the load, not at the converter, and routes the back channel on a short pair away from the switch nodes. A copy of the reference arrangement is faster than re inventing one.
Thermal planning follows the same logic. The four bodies should not sit in a row downwind of each other on a narrow board. Spreading them across the plane and letting each have its own copper island keeps the die temperatures within a few degrees, which is what the current sharing loop assumes.
Where this part fits
Rack servers, data centre switches, active antenna systems, baseband units and ASIC, SoC and FPGA core rails are the uses named in the document. The 40 A device is usually the last stage before the processor, so the questions that reach our desk are about second sources and about lead time rather than about the part itself. See power management cross reference and component lead time solutions.
What to check before a purchase order
First, the order code letter. TPS546D24, TPS546B24 and TPS546A24 sit in one family table, and the reel label is the only place the difference shows. Second, the RVF body is the 40 pin LQFN, and a quote for the same string in another body will not fit. Third, moisture level: these 40 pin bodies are typically level 3 with a 168 hour floor life once the seal is broken, so a line that opens a reel and then stops for a week needs a bake before the next build.
See date code management for how we handle floor life and bake records on the reels we ship, and one stop BOM sourcing for the full rail group.
Questions that come up on this part
Can a 40 A device replace a 10 A device without a board change, is the first question and the answer is yes for the body and no for the magnetics. The second is whether the PMBus address has to change when the stack size changes. It does not, because the back channel makes the stack answer as one address, but every device in the stack must be strapped for the same position table.