UCD90120ARGCR
Product Specifications
UCD90120ARGCR
Texas Instruments — Power Management & Signal Chain
Product Overview
The UCD90120ARGCR 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
UCD90120ARGCR: 12-Rail Power Supply Sequencer and Monitor With ACPI Support
UCD90120ARGCR is a twelve rail sequencer and voltage monitor with a 12-bit converter on chip in a 64-pin VQFN body of 64 pins. It reads every rail on a 400 us cadence, keeps four thresholds for each, and holds up to twelve fault entries in flash, so the reason a board reset survives the power cycle that caused it.
Order code and package
| Ordering string | UCD90120ARGCR, VQFN-64 (RGC), 9.0 mm by 9.0 mm, tape and reel |
| Rails monitored | twelve, MON1 to MON13 with gaps, four thresholds each |
| Digital I/O | 26 GPIO, twelve of them with PWM capability |
| Interfaces | I2C, SMBus and PMBus, plus JTAG for configuration |
| Logging | nonvolatile error and peak value logging, twelve fault detail entries |
| Margining | closed loop on ten rails through the margin output |
Electrical parameters that decide the application
| Converter | 12 bit, 2.5 V internal reference at 0.5 percent |
| Sample cadence | all rails every 400 us, fourteen conversions at 3.89 us |
| MON1 to MON9 range | 0 V to 2.5 V |
| MON10 to MON13 range | 0.2 V to 2.5 V |
| Input leakage | 0.5 nA typ, 1.5 nA max |
| Input impedance | 0.5 percent to 1 percent of the 3 Mohm divider |
| Supply current | 40 mA typ on V33D, 50 mA while writing flash |
| IO supply | 3.3 V on V33DIO, 2 mA typ |
| Clock outputs | 15.259 kHz to 125 MHz, independent per output |
| Watchdog | programmable timer with a system reset output |
Points to check before this part goes on a BOM
One step of the converter is 610 uV on the 2.5 V range, so a 0.9 V core with a plus or minus 3 percent window has 44 steps of headroom, while the 0.5 percent reference tolerance is 12.5 mV, twenty steps, close to half the window. A monitor with these numbers can flag a rail that is still inside its own specification, so the thresholds belong against the measured reference and not against the nominal divider.
Fourteen conversions at 3.89 us take 54.5 us of the 400 us period, 13.6 percent of the time, and the rest is spent waiting. A rail that sags for 100 us between two samples is invisible: the log shows a clean sequence, the processor still resets, and the fault is attributed to firmware.
The 40 mA at 3.3 V is 132 mW in a 9 mm body, present whenever the board is plugged in, and it rises to 50 mA while the part writes a fault entry to flash. That write is the only moment the current changes, which makes it a useful marker when a supply rail shows a notch in a long capture.
MON10 to MON13 start at 0.2 V rather than 0 V, so a negative rail or a shunt of a few tens of millivolts cannot be read directly; it needs a level shift into the window. Four channels have that floor and nine do not, which decides which rails can carry a current sense resistor.
Rail states are selected by up to three GPIs, so eight combinations, and the sequencer runs on time, on rail dependency or on those pin states. A board that changes its power order at run time is therefore possible without a PMBus write, which is what makes the ACPI use case work.
Address selection uses two pins strapped to ground, left open or biased, so four of these devices can share one bus without a hub. The 26 GPIO pins are the resource that runs out first: twelve rails need twelve enables, which leaves fourteen for resets, interrupts, cascades and the clock outputs, and a board with its own sixteen rails plus a mezzanine reaches the limit before it reaches the rail count.
The peak value log records the highest reading each monitor has seen, and the fault detail log records twelve entries in the order they occurred. A rail that glitches repeatedly therefore fills the log with its own entries and pushes out the earlier fault from a different rail, so the log is read oldest first and cleared before a test, not after a failure.
Where this order code fits
Telecom line cards, storage shelves, ATE racks and servers use this part where a dozen rails have to come up in a fixed order and a fault has to be readable after the fact. A shelf controller that changes its power order between operating modes uses the pin selected states for that, and a rack with twelve rails of its own plus a mezzanine reaches the GPIO limit before the rail limit. Boards that read the fault log after a field return are the reason the twelve entries live in flash rather than in RAM. 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 One Stop BOM Sourcing page.
Questions buyers ask
Does it sequence rails or only monitor them? Both. Twelve rails are monitored and the 26 GPIO pins drive enables in a sequence set by time, by rail dependency or by pin state, with the same device holding the watchdog and reset.
Can a 0.6 V rail be measured directly? Yes on MON1 to MON9, which span 0 V to 2.5 V. A 0.6 V rail uses about a quarter of the range and loses resolution, so a divider that maps it near 2.0 V uses the converter better.
What happens to the log on power loss? Up to twelve fault entries and the peak value per monitor are held in flash, so they survive the power cycle and can be read over PMBus on the next start.
Is a separate 3.3 V rail needed for the I/O? V33DIO is a separate pin drawing 2 mA, so the I/O can run at 3.3 V while the core of the device is supplied from the same rail; it is one rail with two pins, not two rails.
SUPERB holds and sources this line for production and repair demand, quoted against the exact ordering string with date codes, packing and country of origin stated on the offer. When a part is moving, the Component Obsolescence Management page sets out the checks that keep a substitute from failing on the bench.