TPS546D24ARVFR
Product Specifications
TPS546D24ARVFR
Texas Instruments — Power Management & Signal Chain
Product Overview
The TPS546D24ARVFR 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
TPS546D24ARVFR: 40 A PMBus buck converter in a 7 by 5 mm QFN that stacks to 160 A
TPS546D24ARVFR is a 40 A synchronous step-down converter in the 40 pin LQFN-CLIP body that runs from a 2.95 V to 16 V power input, carries 4.5 milliohm and 0.9 milliohm integrated switches, and reports output current and die temperature over PMBus in a LQFN-CLIP body of 40 pins. The TPS546D24A is the 40 A member of the TPS546x24A family, in a 7 mm by 5 mm by 1.5 mm QFN with a clip lead frame. Input is 2.95 V to 16 V on the power stage and 2.95 V to 18 V on the analogue rail, output is 0.5 V to 5.5 V from a pin strap or 0.25 V to 5.5 V from a PMBus command. Up to four devices stack with current sharing for a single 160 A output, and twelve switching frequencies from 225 kHz to 1.5 MHz are available, eight of them from pin straps.
Order code and package
| Order code | TPS546D24ARVFR |
| Power stage input voltage | 2.95 V to 16 V |
| Analogue input voltage | 2.95 V to 18 V |
| Output voltage from pin strap | 0.5 V to 5.5 V |
| Output voltage from PMBus | 0.25 V to 5.5 V |
| Rated output current | 40 A |
| Stacking | two, three or four devices, up to 160 A shared |
| Switching frequency | 225 kHz to 1.5 MHz, twelve settings |
| Package | 40 pin LQFN-CLIP, 7 mm by 5 mm by 1.5 mm |
| Document | TPS546D24A data sheet, SLUSC21 class |
Electrical parameters that decide the application
| High side switch resistance | 4.5 milliohm |
| Low side switch resistance | 0.9 milliohm |
| Analogue supply current, not switching | 12.5 mA typical, 17 mA maximum |
| Analogue UVLO, enable | 2.5 V to 2.7 V |
| Analogue UVLO, disable | 2.09 V to 2.3 V |
| Analogue UVLO hysteresis | 250 mV |
| Delay from analogue UVLO to PMBus ready | 8 ms at a 3 V analogue rail |
| Power input turn on, factory default | 2.75 V |
| Power input turn on, programmable range | 2.75 V to 15.75 V in 0.25 V steps |
| Power input turn on accuracy | plus or minus 5 percent |
| Output accuracy at 1 V, 0.5 V and 1.5 V settings | better than 1 percent from minus 40 C to 150 C junction |
| Thermal resistance, junction to case bottom | 1.0 C per W |
| Thermal reference board | eight layer, 2 oz copper per layer |
| Junction temperature range | minus 40 C to 150 C |
Points to check before this part goes on a BOM
Forty amps through a 0.9 milliohm low side switch is 1.44 W of conduction loss before the duty cycle is applied, and the junction to case bottom resistance of 1.0 C per W makes the package look harmless until the board is considered. The case is the board: the eight layer reference board with 2 oz copper is what turns 1.44 W into a workable rise, and a two layer board with 1 oz copper puts the same heat into a much smaller volume. The 40 A figure is a package and copper statement as much as a silicon one.
Stacking changes the thermal problem without changing the conduction total. Four devices at 40 A each still dissipate 1.44 W each, so the loss is unchanged, but it is spread over four packages and four copper areas, and the phases interleave so the input ripple cancels at the switching frequency and appears at four times it. The DC current sharing error between phases is what decides whether one device runs hotter than the rest, and that is set by the internal current sense matching rather than by the external components.
The split rail is the reason this part can run from 12 V at high current. The power stage sees the 12 V bus while the analogue rail can be fed from 5 V, which keeps the internal linear regulator from dropping the difference and burning it inside the package. A single 12 V feed still works, and the analogue current of 12.5 mA at 12 V is only 150 mW, but the gate drive and the internal regulator are happier with a 5 V analogue rail, and the 4 V minimum for switching is the constraint that decides whether the split is worth it.
The pin strap sets the output in 0.5 V to 5.5 V steps and the PMBus command sets it from 0.25 V to 5.5 V, so the strap is a coarse hardware default and the bus is the fine control. The MSEL pins pick which PMBus defaults are loaded, which means the strap resistor network encodes a whole configuration, not just a voltage. A wrong resistor therefore does not simply move the rail, it selects a different set of defaults for UVLO, frequency and fault response, and the resulting behaviour can look like a fault in the converter rather than a mistake in the bill of materials.
Telemetry without an external sense resistor is the feature that sells this part, and the accuracy of the internal current sense is the term to check. The datasheet quotes output accuracy of better than 1 percent over the whole junction range for the voltage loop; the current reading is a separate specification with its own offset and gain terms. A board that uses the PMBus current for a power capping loop, rather than for display, needs that current specification read at the operating point rather than assumed equal to the voltage accuracy.
The analogue rail has its own UVLO at 2.5 V to 2.7 V with 250 mV of hysteresis, and PMBus communication is only ready 8 ms after that threshold. A hot plug controller that asserts an enable as soon as the bus is present will be talking to a device that has not finished booting, so the sequencer has to wait for the ready flag rather than for the rail. The factory default power input turn on of 2.75 V is low for a 12 V system and is normally reprogrammed, which is another reason the strap configuration has to be treated as a specification rather than a default.
Where this order code fits
This order code belongs on core and memory rails in servers, switches and storage, on 12 V to 1 V point of load stages in telecom line cards, and on any rail that needs current and temperature telemetry without an external sense resistor. It is chosen when a single stage has to deliver tens of amps into a load that also needs margining and voltage adjustment at run time, because the PMBus interface replaces both the trim network and the monitoring circuit. Teams comparing it with a neighbouring part start from our Power Management Cross Reference page, and teams replacing a part that has gone end of life start from the OEM Procurement page.
Questions buyers ask
Does the 40 A rating require the split rail? No. The power stage accepts 2.95 V to 16 V with the analogue rail tied to it, and the analogue current of 12.5 mA at 12 V is 150 mW, which the package can carry. The split rail exists to keep the internal regulator cool and to allow the analogue rail to sit at 5 V while the power stage runs from a bus that may be anywhere from 3 V to 16 V. On a 5 V only system the question does not arise.
How is current shared between stacked devices? Each device senses its own output current and the devices communicate over a shared bus so that the loop reference is common, which is what makes a single address cover the whole output. The sharing accuracy is set by the internal sense matching, and the phase interleaving is what cancels input ripple. Four phases at 225 kHz each give an effective input ripple at 900 kHz, which is why a stacked rail needs less input capacitance than four independent converters.
Can the output be changed while the rail is live? Yes, within the PMBus command range of 0.25 V to 5.5 V, and the family supports margining and adaptive voltage scaling for exactly that purpose. The constraint is the load, not the converter: a load that cannot tolerate a step of tens of millivolts during a command change needs the change made with the rail disabled, and the telemetry is what confirms the new level actually arrived.
Is a 0.5 V output reachable from a 12 V input? The duty cycle is 4.2 percent at 12 V to 0.5 V, and at 1.5 MHz the period is 667 ns, so the on time is 28 ns. Whether that is inside the minimum on time is a datasheet term to check at the chosen frequency. Falling back to 550 kHz doubles the on time to 76 ns, which is the usual route for a low output from a high input on this family.
SUPERB supplies this converter against the exact order code, with the revision, date code and packing stated on the offer, and can hold a schedule against a production forecast. When a rail is moving to a second source, the Component Lead Time Solutions page sets out the checks that keep the substitute from failing at temperature.